Biomechanic Ballet

BIOMECHANICAL PERSPECTIVES ON
CLASSICAL BALLET TECHNIQUE AND

IMPLICATIONS FOR TEACHING PRACTICE
Rachel Evelyn Ward
A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy
School of Risk and Safety Sciences

Faculty of Science
University of New South Wales
Sydney, Australia
September 2012
P)..EASE TYPE
THI; UNIVERSITY OF NEW SOUTH WALES
l',toslsiOI!.isurtatlun Sheet
SuuumtL! or Fanuly name Wa1d
Otht;H name/3: ElJolyn
Abbrevootion for de~ree a _given In the UniVersity calendar PhD
Sohoor. Stlloot ol Rls~ &nd Safety Sciences
Faculty. FaO<Jity of Science
TIUB! Biometllanfcal poll!pediVas On classical ballet techniqUe and

Implications for teac~tng practice
Abstract 350 words maximum: (PLEASE TYPE)

Classical ballet ls an 1111 foun wen known I(U 115: vury dllittnct ilnd precise movement styla. A review of retsvant. lltetature led (o lt'le tde'nlifitatlOn oJ
four fundamental blornechanlcal pn'nelple& of class.lcal ballet tedmlqll '"al!gnm:en,; platement", 1utnoul"'-, .ond "extellstorf .
The capacity to excuta technique cor(ectJy 45 a
rundame:nt~l
pretequ!"ite for a SiJCGe$$ful career-as a professional clnSsical ballal aa"~' Given
thiS. 111e avenlgC<J )Ctncmauc aata flom pfofc:ss1r.ma1 Oal!et danccw; pttrtiJttrlmg core ballet -slepa provide-& a practfcsl blo,.,echanlcal
ban~l't'lill'tl'
of
' correct' lochnlquo_ Ti'oroo-dlmoMional (30) IUII-I)ody motion analysis Of 14 b3llet steps ms osoo to eompme the J>CffOI1ll3hC<! ul p1uftr.~sionat
classiCal balll da"''""' (N12} wltll that ol non-profess>anaf baUel dance IS (Na lA}, snd 10 invesVgato the '"""' of agreement between practical
execut1on of the steps Willi ll'le lfleoretlcaf kfeals. Profess1onal danwrs did demonstrate ku12mat1c vau~b!es 1elal1ng to "turnou1 and extension in
lll)reomonl wltn
O> lhooroUcal pnnclpl"". howavor some delaUo~s bel\llee~ praclica and !Mol)' were o\lseM!d In the orea of "olign'mcnl" and
"pletceiJU.:Ut'" ih1s reiiult 11as nHphcellun~ lo1 thu teachmg mulhoOs c:une11Uy used m c:lsss1cal ballet ~nstructJon,
Having eslsblltihed lfte praf:1tcal kinematic pnnclplos of ' correl:1' l'ia$slcal l)allet tetllnlque, the aeeuraey of qualllatlve analysis
~lnwnect
ballet ttochnque
compared lo
qvahl~llvo
qu-an~tatlve
wa~ IUW!Illgah~d with
ot correcl' and
respt:ct to 'experience In classical ballet lnstrudJon_ Teachers qualitative re.spon.lut$ ware.
data oblaincd rrorr, 30 mc>l!on oua\ySts. A fJ0$1Uve
rt:le~Uonstup
bt:twettn yeuf3' ot teaching eJ(penence and aCcuracy of
aMiys.iS Wa$ damo!ll'l'llt~d. however eV_M the mosl Olq)erlenced leatllers only ttehlel/ed modero1e aCC<Jracy Improvement In acoul<lcy
witt\ tm:reasiny tlxpeuonce appear'cd to .be SlOVI, anel pl~1eauW aft~r l O years These obS~tvallons ar~ ithpOttanllf\ 1Vtms ol ttevetopfng ctmiculs
and
~eaching
programs- ror dance teachers.
Glven lha relatively low accuracy ol qualltalfve analy$!$ demonstroted by e~perle nced
!~tilers.
a quantllallve assessment 1001 was developed In
which data tram mdtVJdUal dancerS' were compared to the practical gold 'Standard ~ of techntque. as de.monsuated by protessronal ballet dancers
Anatyse were conducted to detetmlne the minimum number ana opptoprlate combination of ballet steps for mctusion ln tne assessment tool that
wovld mO$! afficl~ntly enabfo 111 evaluation ol dancers againsT Ihe four fUndamonta1 prfni:llll of classical ballellochnlqua
Oeclaratlon rotaUng to dlsposiHon of project lltasllldissertaUon
I heceby grant to the Umvers1ty ol New South Wsles or its agents the nght to arch1ve and to make avallable- my thesra or dtssertation ln Whole or in
pan In Ihe University llb13ries ln all ronn of media, naw or horo ahBI known, stiblecl to ti'os proVftions of !lie Copynght Act 1968, I reL11n 1111proporty
r1ghl$: such-as patent rtghls. 1niSO feHJjn lhe right to use In f'uture works (~ucfl as an}~es or books) :01! or part of this thesiS or dlssertnti(m
I also autnorlse University Mlcrofolms 10 trse ihe 3SO wort;! <Jbstr<~<;l of my lhe~lsln Olsserilion Ahstra~ International (I his ill applicable to docloral
the" only) .
............;trdv.:. C..../1/tk.f~. . .
Signature
~~.--.-.. .-Wllness
. I z./ ?~(z_".'.~
;t~
The University ~ecognlsas lhallhete may be ~ceptlonal clraJmstances requiring rostrlctloOll on COPYif19 or condttions on use Requests for
raslnctlon for a p'enod of up to 2 YftiU$ must be mad~ 111 wrilutg. Requt!sls fo1-a longer petiod crrcSttlction may b(! considered In oxccptwnal
cltC1imstanees- and ra urre the a oval ol Uus Oean of Graduate RaseJJrch.
FOR OF ACE USE ONLY
Date or completion of requirements for Award:
THIS S~EET IS iO 8E GLIJEO TO TfiE' INSIDE FRONT COVER OF THE THESIS
COPYRIGHT STAT EMENT
'I hereby grant the Universi\Y of New South Wales or its agents the right to archive
and to make available my thesis or dissertation in whole or part in the University
libraries in all forms of media, now or here after known. subject to the provisions of
the Copyright Act 1968, I retain all proprJetary ri9hts. such as patent rights I also
retain the right to use in ruture worlts (such as articles or books) all or part of this
thesl s or dissertatiOn.
I also authOrise University Microfilms to use the 350 Word abstract of my ltlesis ln
Dissertation Abstract l ntema~onal (this Is applicable lo doctoral theses only)
I hal/e either used no subst.anlial portions of copyri.gtit material In my thesis or I
have obtained permission to use copyr[ght material; where perrnis~iOri has ~ol
been granted I t)ave applied/will apply for a partial restriction of !he digital copy of
my thesls or dlssertatlon.'
Signed ...... ... ...
.r/..~.:R.( .... /1/.~. . . .. . . .. . . .

I z.j ?/ ~vl'l.-
Date ~ " .. ~ ...
:::.~ ~
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . .
,"~,,
,_.
AUTHENTICIT Y STATEMENT
'I certify thai the Library deposit digttal copy is a direct equivalent of the !mal
offlorally approved verslon of my thesis. No emendation of content has occurred
and 11 there are any mmor variations In formatting, they are the result of the
conversion to digital format.'
Signed ... .. ...
&.~4./. . . .o/..C?.e:::.. ..................... . .

Cl./.~1 .1..~~!. "!-:-. .. ,....................................
Pate ......... ......
ORIGINALITY STATEMENT
'I hereby declare that this submission is my own work and lO the best of my knowledge II contains
no materials previously published or written by another persun. or substantial proportions of
matenal which have been accepted for the award of any other degree or diploma at I.JI\ISW or any
other educational institution, eJ<cepl where due acknowledgement Is rnaoe hi the thesis Any
contribution rnade to the fesearch by others, with whom I have worked at UNSW or elsewl1ere, Is
explicitly acknowledged in tlie Ulesis I alsc.> declare !hat the Intellectual content of this thesis Is the
product of rny own woril. e.xcept to the exHmt !hat assistance from others In the projecfs design
and conception or 1o styl'e, presentation anl'.t llngulsUc el(presston Is acknowledged:
Signed
./?r!vt.K......iz~~. . . . . .
oate .........
J.?:-./.'1./.?:~t.?-:: ............................
tl
Acknowledgements
ACKNOWLEDGEMENTS
There are many people who contributed to the completion of this project. To each of them
I am extremely grateful.
Thank you to my supervisors Associate Professor Andrew McIntosh and Dr John

Mullins. Andrew, I sincerely thank you for guiding and assisting me to completion of this
project. You have been a constant source of support and encouragement, often amidst
times of uncertainty about the future of the equipment, the laboratory, and the school.
Thank you for helping me stay focused on the task at hand during these difficult and
distracting times. My gratitude to you also extends beyond completion of this project.
Your mentorship during the years prior to commencement of his project was invaluable.
You laid the foundations for my academic career, instilling me with confidence,
motivation, enjoyment and appreciation of biomechanics, and the necessary skills for a
rewarding future. I am grateful for the interest you have always shown towards my
passion for dance, and for your constant belief in my abilities.
John, your input to this project has been truly invaluable and is deeply appreciated. Thank
you for bringing your expert perspectives on dance education to the project, and for
ensuring that the projects relevance and practical application remained a priority. Thank
you for many interesting and thought provoking conversations, and for your constant
availability and accessibility. I have enjoyed working with you, and I sincerely respect
and value the supervision you have provided.
To my esteemed friend and colleague, Ms Kim Traynor, thank you for sharing your
immense understanding of classical ballet with me and with this project. I am grateful for
your unceasing interest, encouragement, and enthusiasm. You are a source of great
inspiration for me and for the hundreds of ballet students who have the privilege of your
tutelage. The knowledge and skill you bring to the dance teaching profession are
unsurpassed. I hope many students will continue to experience your brilliance.
iii
Acknowledgements
I also extend heartfelt thanks to all the dancers and dance teachers who volunteered to
participate in this project. I appreciate your time and interest, and share in your
commitment to and passion for dance.
To Patrick, thank you for your unwavering love and support during these past few years. I
am truly grateful for the steadfast encouragement, the delicious meals, and the comic
relief, all provided at exactly the right moment. Thank you for riding out the difficult
times with me, and for your incredible loyalty and patience.
Finally, to my darling mum Lynne, to whom I owe everything. You have provided me
with so many opportunities, and for this I am eternally grateful. Thank you for making it
possible for me to pursue and fulfil my dance ambitions, and for also inspiring me
towards academic achievement. I appreciate your love and support more than I am able to
articulate. My love for you is infinite. Thank you.
iv
Table of Contents
TABLE OF CONTENTS
Copyright Statement........................................................................................................
Authenticity Statement....................................................................................................
Certificate of Originality..................................................................................................
ii
Acknowledgements.........................................................................................................
iii
Table of Contents............................................................................................................
List of Figures..................................................................................................................
List of Tables...................................................................................................................
xvi
CHAPTER 1 INTRODUCTION..............................................................................
1.1 Introduction..............................................................................................................
1.2 Aims of the thesis.....................................................................................................
1.3 Structure of the thesis..............................................................................................
CHAPTER 2 - LITERATURE REVIEW..................................................................
2.1 Foundations of Classical Ballet Technique...........................................................
10
2.1.1 Introduction..................................................................................................
10
2.1.2 The History and Development of Classical Ballet Technique......................
12
2.1.3 The Fundamental Principles of Classical Ballet Technique.........................
33
2.2 Quantitative Biomechanical Analysis in Dance....................................................

2.2.1 Introduction..................................................................................................
53
53
2.2.2 Instrumentation Used in Quantitative Dance Biomechanics Research.........
54
2.2.3 Purpose and Aims of Quantitative Dance Biomechanics Research..............
56
2.2.4 Dance Genres and Quantitative Dance Biomechanics Research..................
61
2.2.5 Quantitative Biomechanics Research in Classical Ballet Technique............
62
2.3 Qualitative Analysis of Human Movement.............................................................
79
2.3.1 Introduction....................................................................................................
79
2.3.2 Application of Qualitative Analysis of Human Movement............................
80
2.3.3 Reliability and Accuracy of Qualitative Analysis of Human Movement.......
92
2.3.4 Investigation of the Accuracy of Qualitative Analysis of Human Movement 96
Table of Contents
2.4 Summary of Literature Review and Project Rationale.......................................... 115

2.4.1 Summary of Literature Review....................................................................... 115
2.4.2 Project Rationale............................................................................................. 116
CHAPTER 3 - COMPARISON OF THEORETICAL AND PRACTICAL

EXECUTION OF CLASSICAL BALLET TECHNIQUE............
118
3.1 Introduction.............................................................................................................
119
3.2 Methods....................................................................................................................
121
3.2.1 Participant Recruitment................................................................................
121
3.2.2 Instrumentation.............................................................................................
122
3.2.3 Kinematic Measurement Variables...............................................................
138
3.2.4 Experimental Protocol..................................................................................
141
3.2.5 Data Extraction.............................................................................................
154
3.2.6 Statistical Analysis........................................................................................
157
3.3 Results.......................................................................................................................
159
3.3.1 Participant Demographics.............................................................................
159
3.3.2 Comparison of Variables Used to Define Start and Peak Time Points for
Battements...................................................................................................
160
3.3.3 Between Group Analysis of Fundamental Principles with Respect to Joint

and Segment Kinematics..............................................................................
161
3.3.4 Analysis of Fundamental Principles with Respect to Movement Categories

and Steps.......................................................................................................
184
3.4 Discussion.................................................................................................................
192
3.4.1 Alignment......................................................................................................
192
3.4.2 Placement......................................................................................................
195
3.4.3 Turnout..........................................................................................................
198
3.4.4 Extension.......................................................................................................
202
3.4.5 Analysis of Fundamental Principles with Respect to Movement Categories

and Steps.......................................................................................................
203
3.4.6 Limitations.....................................................................................................
204
3.4.7 Conclusions....................................................................................................
207
vi
Table of Contents
CHAPTER 4 ACCURACY OF QUALITATIVE ANALYSIS OF CLASSICAL
BALLET TECHNIQUE.....................................................................
209
4.1 Introduction..............................................................................................................
210
4.2 Methods.....................................................................................................................
213
4.2.1 Participant Recruitment.................................................................................
213
4.2.2 Development of the Qualitative Assessment Questionnaire..........................
214
4.2.3 Video Data Selection and Preparation...........................................................
216
4.2.4 Experimental Protocol...................................................................................
223
4.2.5 Data Analysis.................................................................................................
225
4.2.6 Statistical Analysis.........................................................................................
226
4.3 Results........................................................................................................................
229
4.3.1 Participant Demographics............................................................................... 229

4.3.2 Accuracy and Reliability of Identification of the Body Region Most in
Need of Correction.......................................................................................
231
4.3.3 Accuracy of Diagnosis of Correct or Incorrect Technique.....................
235
4.3.4. Accuracy of Identification of Type of Correction Required.....................
244
4.4 Discussion...................................................................................................................
248
4.4.1 Accuracy and Reliability of Identification of the Body Region Most in

Need of Correction.......................................................................................
248
4.4.2 Accuracy of Diagnosis of Correct or Incorrect Technique...................... 249

4.4.3. Accuracy of Identification of Type of Correction Required......................
254
4.4.4 Limitations......................................................................................................
255
4.4.5 Conclusions..................................................................................................... 256
CHAPTER 5 DEVELOPMENT OF A QUANTITATIVE ASSESSMENT

TOOL FOR CLASSICAL BALLET TECHNIQUE.......................
258
5.1 Introduction..............................................................................................................
259
5.2 Methods.....................................................................................................................
261
5.2.1 Data Extraction..............................................................................................
261
5.2.2 Consideration of Steps for Inclusion in the Quantitative Assessment Tool...
262
5.2.3 Analysis of the Effect of the Number and Combination of Steps Included...
265
5.2.4 Development of the Assessment Tool Report................................................
267
vii
Table of Contents
5.3 Results......................................................................................................................... 268
5.3.1 Consideration of Steps for Inclusion in the Quantitative Assessment Tool...
268
5.3.2 Comparisons of Dancer Rankings when Including Different Numbers of

Steps in the Assessment Tool.........................................................................
270
5.3.3 Analysis of the Effect of the Number and Combination of Steps on ZScores for Individual Kinematic Variables....................................................
272
5.3.4 Sample Assessment Tool Report....................................................................
278
5.4 Discussion...................................................................................................................
285
5.4.1 Identification of Steps for Inclusion in the Assessment Tool........................
285
5.4.2 Affect of the Number and Type of Steps Included in the Assessment Tool..
286
5.4.3 Comparison With Other Tools Used For Assessment of Classical Ballet
Technique......................................................................................................
287
5.4.4 Potential Use of the Quantitative Assessment Tool.......................................
288
5.4.5 Limitations.....................................................................................................
291
5.4.6 Conclusions....................................................................................................
291
CHAPTER 6 CONCLUSIONS AND IMPLICATIONS FOR TEACHING

PRACTICE..........................................................................................
293
6.1 Specific Conclusions and Implications...................................................................
294
6.1.1 Comparison of Theoretical and Practical Execution of Classical Ballet

Technique......................................................................................................
294
6.1.2 Qualitative Analysis of Classical Ballet Technique.......................................
301
6.1.3 Development of a Quantitative Assessment Tool for Classical Ballet

Technique......................................................................................................
304
6.2 General Conclusions.................................................................................................
307
6.3 Future Directions......................................................................................................
309
REFERENCES..............................................................................................................
311
APPENDICES................................................................................................................
335
Appendix A: Ethics Approval for Chapter 3 Comparison of Theoretical and

Practical Execution of Classical Ballet Technique............................. 336
Appendix B: Information and Consent Form for Non-professional Ballet
Dancers..............................................................................................
viii
337
Table of Contents
Appendix C: Information and Consent Form for Professional Ballet Dancers......
340
Appendix D: Dancer Questionnaire........................................................................
343
Appendix E: Chapter 3 - Additional Information on Cubic Spline Curve

Interpolation.......................................................................................
348
Chapter 3 - Additional Information on Woltring General CrossValidatory (GCV) Quintic Spline Routine................................................
349
Appendix F:
Appendix G: Chapter 3 - Graphical Outputs for Gait Data Displayed in Vicon

Polygon Authoring Tool................................................................... 350
Appendix H: Chapter 3 - Summary of Descriptive Statistics for Kinematic
Variables........................................................................................... 351
Appendix I: Chapter 3 - Summary of Mean Difference in Joint Angle for Steps
and Kinematic Variables for which a Statistical Significance was
360
Achieved...........................................................................................
Appendix J: Ethics Approval for Chapter 4 Accuracy of Qualitative Analysis of
Classical Ballet Technique......................................................................... 362
Appendix K: Information and Consent Form for Professional Ballet Teachers.... 363
Appendix L: Information and Consent Form for Student Ballet Teachers............ 366
Appendix M: Ballet Teacher Questionnaire...........................................................
369
Appendix N: Qualitative Analysis Questionnaire................................................... 373

Appendix O: Chapter 4 - Percentages of Accurate Responses, Type I and Type
II Errors............................................................................................
379
Appendix P: Chapter 5 - Correlation Matrix for Different Versions of

Quantitative Assessment Tool............................................................
380
ix
List of Figures
LIST OF FIGURES
CHAPTER 2
Figure 2.1
Influences of selected changes in ballet (from Hammond & Hammond,

1989, p21).................................................................................................
13
Figure 2.2
From Practica seu arte tripudii (1463) (from Bland, 1976, p33).............
16
Figure 2.3
Faune from Lullys Le Triomphe de lAmour (from Nordera, 2007,

p26)..........................................................................................................
18
Figure 2.4
The five positions of the feet (from Homans, 2010, p136)......................
19
Figure 2.5
The five positions of the feet as depicted by Feuillet in his

Chorgraphie in 1701 (from Hammond, 2007, p6, 7)..............................
20
Correct perpendicular alignment of the body (from Blasis, 1820, p18

& p26).......................................................................................................
34
Correct and incorrect alignment of the body for classical ballet

technique (from Vaganova, 1953, p25)....................................................
35
Figure 2.8
Correct and incorrect body alignment (from Lee, 1983, p207).............
36
Figure 2.9
The correct lengthened position of the spine (from The Royal Academy
of Dance 1997, p8)....................................................................................
37
Figure 2.6
Figure 2.7
Figure 2.10
In leg extension to the front or side, the body is expected to remain

close to the vertical (from Paskeva, 2005, p43)....................................... 38
Figure 2.11
Correct (left image) and incorrect (right image) placement of the

pelvis (from Lee, 1983, p209)................................................................... 40
Figure 2.12
Centred placement of the pelvis (from Grieg, 1994, p40)......................
40
Figure 2.13
Turnout (from Blasis, 1828, p12)..............................................................
43
Figure 2.14
Five positions of the feet (from Vaganova, 1953, p17)............................
44
Figure 2.15
Turnout Maximum outward rotation of the legs in the hip sockets

(from Lee, 1983, p211).............................................................................
45
Figure 2.16
Turnout turnout is a movement (from Grieg, 1994, p53)..................
46
Figure 2.17
Leg extension (from Blasis, 1830, p18).................................................
48
List of Figures
Figure 2.18
Leg extension (from Vaganova, 1953, p29-30).....................................
49
Figure 2.19
Leg extension (from Grieg, 1994, p88)..................................................
50
Figure 2.20
Leg extension (from Grieg, 1994, p105)................................................
51
Figure 2.21
Summary of publication years of quantitative dance biomechanics

research...................................................................................................... 54
Figure 2.22
Summary of study purposes of quantitative dance biomechanics

research...................................................................................................... 60
Figure 2.23
Summary of dance genres analysed in quantitative dance biomechanics

research...................................................................................................... 62
Figure 2.24
Summary of dance movement categories analysed in quantitative ballet

biomechanics research............................................................................... 70
Figure 2.25
Sample size for single group studies involving quantitative

instrumented biomechanical analysis of classical ballet technique..........
73
Sample size for multiple group studies involving quantitative

instrumented biomechanical analysis of classical ballet technique..........
75
A strategy for the qualitative analysis of human movement (from Arend

& Higgins 1976, p38)................................................................................
86
Model of qualitative analysis of sports movement (from McPherson

1996, p s86)...............................................................................................
88
Summary of percentages of studies conducted on the accuracy of

qualitative analysis in clinical practice, occupational health and safety
(OHS), sport, and dance............................................................................
99
Figure 2.26
Figure 2.27
Figure 2.28
Figure 2.29
Figure 2.30
Summary of percentages on studies that have used categorical scales,

interval scales, visual analogue scales (VAS), or numerical estimates for
qualitative analysis of human movement................................................... 104
CHAPTER 3
Figure 3.1
L-Frame device for static calibration......................................................... 124
Figure 3.2
240mm Space Bar Wand device for dynamic calibration.........................
Figure 3.3
Vicon Plug-in-Gait reflective marker placement for full body analysis

(from Vicon Plug-in-Gait Manual)............................................................ 127
xi
125
List of Figures
Figure 3.4
Reflective marker placement on participant.............................................. 128
Figure 3.5
Knee Alignment Device (KAD)................................................................
Figure 3.6
Marker placement and three-dimensional coordinate axis system for the

thorax segment........................................................................................... 132
Figure 3.7

pelvis segment...........................................................................................
129
132
Figure 3.8

thigh segment............................................................................................. 133
Figure 3.9

shank segment............................................................................................ 134
Figure 3.10

foot segment..............................................................................................
134
Example of starting position and end position for a series of 3 Cardan

angle rotations...........................................................................................
136
Directions and signs for kinematic measurement variables for thorax,

pelvis, and lower limb (adapted from Vicon Plug-in-gait Manual)..........
137
Figure 3.13
Grand pli in first position........................................................................
145
Figure 3.14
Grand pli in second position...................................................................
145
Figure 3.15
Battement devant (front)............................................................................ 146
Figure 3.16
Battement derriere (back).........................................................................
147
Figure 3.17
Rise in first position..................................................................................
147
Figure 3.18
Rise in second position..............................................................................
148
Figure 3.19
Saut in first position................................................................................. 149
Figure 3.20
Temps lev.................................................................................................
149
Figure 3.21
Glissade dessous (under)...........................................................................
150
Figure 3.22
Glissade dessus (over)...............................................................................
151
Figure 3.23
Jet ordinaire............................................................................................. 152
Figure 3.24
Grand jet elanc en avants......................................................................
Figure 3.11
Figure 3.12
xii
152
List of Figures
Figure 3.25
Pirouette en dehors...................................................................................
153
Figure 3.26
Pirouette en dedans...................................................................................
154
Figure 3.27
Thorax anterior/posterior tilt at start of step.............................................. 163
Figure 3.28
Thorax anterior/posterior tilt at peak of step.............................................
163
Figure 3.29
Thorax side tilt at start of step...................................................................
165
Figure 3.30
Thorax side tilt at peak of step..................................................................
165
Figure 2.31
Pelvic anterior/posterior tilt at start of step...............................................
167
Figure 3.32
Pelvic anterior/posterior tilt at peak of step..............................................
168
Figure 3.33
Pelvic obliquity at start of step..................................................................
169
Figure 3.34
Pelvic obliquity at peak of step.................................................................
170
Figure 3.35
Pelvic - thoracic transverse rotation at start of step..................................
171
Figure 3.36
Pelvic - thoracic transverse rotation at peak of step..................................
172
Figure 3.37
Hip transverse rotation at start of step.......................................................
174
Figure 3.38
Hip transverse rotation at peak of step......................................................
174
Figure 3.39
Ankle transverse rotation at start of step...................................................
176
Figure 3.40
Ankle transverse rotation at peak of step..................................................
176
Figure 3.41
Knee flexion/extension at start of step......................................................
178
Figure 3.42
Knee flexion/extension at peak of step.....................................................
179
Figure 3.43
Ankle dorsi/plantarflexion at start of step.................................................
180
Figure 3.44
Ankle dorsi/plantarflexion at peak of step................................................
181
Figure 3.45
Thorax anterior/posterior tilt for grand plie in first position..................... 187
Figure 3.46
Hip transverse rotation for grand plie in first position.............................. 187
xiii
List of Figures
Figure 3.47
Knee flexion/extension for grand plie in first position.............................
187
Figure 3.48
Pelvic-thoracic transverse rotation for battement jete front......................
188
Figure 3.49
Hip transverse rotation for battement jete front........................................
189
Figure 3.50
Knee flexion/extension for battement jete front........................................
189
Figure 3.51
Thorax side tilt for grand jete elance en avants........................................
190
Figure 3.52
Pelvic-thoracic transverse rotation for grand jete elance en avants.......... 190
Figure 3.53
Hip transverse rotation for grand jete elance en avants............................ 191
Figure 3.54
Knee flexion/extension for grand jete elance en avants...........................
191
Figure 3.55
Ankle dorsi/plantarflexion for grand jete elance en avants......................
191
Example of sagittal and frontal views played simultaneously for all

videos.......................................................................................................
217
Percentage accurate and inaccurate responses by group for

identification of which body region requires the most correction.........
234
Percentage accurate and inaccurate responses by step observed for

identification of which body region requires the most correction.........
234

identification of correct and incorrect technique...................................
236
CHAPTER 4
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Linear regression of years teaching experience against percentage

accuracy for identification of correct or incorrect ballet technique... 237
Figure 4.6

accuracy for identification of correct or incorrect ballet technique
for all teachers with 1-9 years experience.................................................
238

accuracy for identification of correct or incorrect ballet technique
for all teachers with 10 or more years experience.....................................
238
Figure 4.7
xiv
List of Figures
Figure 4.8

accuracy for identification of incorrect ballet technique for all
teachers with 1-9 years experience............................................................ 239
Figure 4.9

accuracy for identification of correct ballet technique for all teachers
with 1-9 years experience.......................................................................... 239
Figure 4.10

accuracy for identification of incorrect ballet technique for all
teachers with \HDUVH[SHULHQFH..........................................................
240

accuracy for identification of correct ballet technique for all teachers
with \HDUVH[SHULHQFH.........................................................................
240
Figure 4.11
Figure 4.12
Percentages of Accurate responses, and Type I and Type II Errors by

group.......................................................................................................... 241
Figure 4.13

identification of type of correction required..........................................
245
Percentage accurate and inaccurate responses by step observed for

identification of type of correction required..........................................
246
Figure 4.14
Figure 4.15
Percentage accurate and inaccurate responses by body region observed

for identification of type of correction required..................................... 246
CHAPTER 5
Figure 5.1
Flow chart depicting process for calculation of overall mean z-score per
step............................................................................................................
263
Figure 5.2
Flow chart depicting process for calculation of overall mean z-score per
dancer......................................................................................................... 266
Figure 5.3
Example of assessment results for a cohort of dancers for a specific

time point................................................................................................... 289
Figure 5.4
Example of hypothetical assessment results by time point for individual

dancer NP015............................................................................................
xv
290
List of Tables
LIST OF TABLES
CHAPTER 2
Table 2.1
Summary of quantitative biomechanics research studies that used (3D)

three-dimensional motion analysis and reported kinematic data.............
76
Table 2.2
Interpretation of measures of reliability or validity.................................. 94
Table 2.3
Summary of studies that have used categorical scales, interval scales,

visual analogue scales or numerical estimates for qualitative analysis
of human movement in various fields......................................................
104
Summary of conclusions from literature review of accuracy of

qualitative analysis of human movement.................................................
114
Table 2.4
CHAPTER 3
Table 3.1
Fundamental principles of classical ballet technique and associated

kinematic variables................................................................................... 139
Table 3.2
Participant demographics.........................................................................
159
Table 3.3
Results of the independent samples t-tests for global thigh sagittal

angle (thigh elevation) at the start and peak of the battement movement
160
Pairwise comparison of thorax anterior/posterior tilt for all steps and

start, peak, and start and peak combined.................................................
164
Pairwise comparison of thorax side tilt for all steps and start, peak,
and start and peak combined....................................................................
166
Pairwise comparison of pelvic anterior/posterior tilt for all steps and

start, peak, and start and peak combined.................................................
168
Pairwise comparison of pelvic obliquity for all steps and start, peak,
and start and peak combined....................................................................
170
Pairwise comparison of pelvic - thoracic transverse rotation obliquity

for all steps and start, peak, and start and peak combined.......................
172
Pairwise comparison of hip transverse rotation for all steps and start,
peak, and start and peak combined..........................................................
175
Table 3.4
Table 3.5
Table 3.6
Table 3.7
Table 3.8
Table 3.9
xvi
List of Tables
Table 3.10
Pairwise comparison of ankle transverse rotation for all steps and start,
peak, and start and peak combined........................................................... 177
Table 3.11
Pairwise comparison of knee flexion/extension for all steps and start,

peak, and start and peak combined........................................................... 179
Table 3.12
Pairwise comparison of ankle dorsi/plantarflexion for all steps and

start, peak, and start and peak combined.................................................. 181
Table 3.13
Summary of pairwise comparisons between groups at start of step,

using repeated measures general linear model.........................................
182
Summary of pairwise comparisons between groups at peak of step,

using repeated measures general linear model.........................................
183
Summary of pairwise comparisons between groups for start and peak

time points combined, using repeated measures general linear model....
183
Kinematic variable test matrix.................................................................
185
Mean z-score for each step for selected kinematic variables observed
during qualitative analysis data collection sessions.................................
221
Table 3.14
Table 3.15
Table 3.16
CHAPTER 4
Table 4.1
Table 4.2
Mean z-scores for all kinematic variables for all steps observed during
qualitative data collection sessions........................................................... 223
Table 4.3
Kappa value ranges and interpretation (from Landis and Koch 1977, p
165)..........................................................................................................
227
Table 4.4
Participant demographics.........................................................................
230
Table 4.5
Response counts for Group 1 (Teaching Experience = 0 years)..............
231
Table 4.6
Response counts for Group 2 (Teaching Experience = 1-9 years)........... 232
Table 4.7
Response counts for Group 3 (Teaching Experience \HDUV
Table 4.8
Response counts for all participants......................................................... 233
Table 4.9
Inter-rater reliability data for each group of teachers............................... 233
Table 4.10
Output for binary logistic regression of dependent variable accuracy

of identification of body region most in need of correction...................
xvii
232
235
List of Tables
Table 4.11

of diagnosis of correct or incorrect technique........................................
236
Table 4.12
Output for binary logistic regression of dependent variable occurrence

of Type I Error........................................................................................ 242
Table 4.13
Output for binary logistic regression of dependent variable occurrence

of Type II Error....................................................................................... 242
Table 4.14
Sensitivity and specificity by teaching experience................................
Table 4.15
Average difference between individual dancer joint angle data and

professional dancer group mean, and percentage of teachers that
243
detected error............................................................................................
Table 4.16
Results of independent samples t-test for effects of professional

performing experience on percentage accuracy.................................... 244
Table 4.17
Results of independent samples t-test for effects of participation in

teacher training program on percentage accuracy................................
244

of type of correction...............................................................................
247
Table 4.18
242
CHAPTER 5
Table 5.1
Overall z-scores and deviant kinematic variables for each step and
associated movement category................................................................. 268
Table 5.2
Dancer rank order by overall z-score across different numbers and

combinations of steps...............................................................................
271
Spearmans correlation coefficients (r) and p-values for correlation of

Assessment tool A with Assessment tool B.............................................
271
Table 5.3
Table 5.4
Dancer overall z-scores per assessment tool and kinematic variables

with z-scores 273
Table 5.5
Classical ballet steps included qualitative screening tools.......................
xviii
287
Chapter 1 Introduction
CHAPTER 1
INTRODUCTION
-1-
CHAPTER 1
INTRODUCTION
1.1 Introduction
The art form of classical ballet is well known for its very distinct and precise movement
style. The professional classical ballet dancer spends many years diligently learning,
practising and perfecting their skills. From a young age dancers undergo regular training
sessions, more commonly referred to as classes, in an effort to acquire the very
particular movement patterns and aesthetics that characterise the dance style. Typically, a
young dancer will spend a minimum of 8 years in a demanding training program if they
are to gain the skills necessary to advance to a professional level. From the age of about
10 years, young dancers aspiring to a professional career in classical ballet will undertake
at least 3 ninety minute classes per week. If, when they reach 14 or 15 years of age, they
are still serious about pursuing this passion, then it is not uncommon to enter into a fulltime training program for the final 3 or 4 years of training. This level of training usually
involves multiple classes per day.
Throughout this training process, it is the role and responsibility of the ballet instructor to
help the aspiring young dancer learn what constitutes correct classical ballet technique,
and how to execute it. Unlike many other physical activities, most sporting pursuits for
example, dance does not have an obvious measurable external parameter to assess a
dancers level of competency. In the sporting arena, success can be measured objectively
using variables such as running speed, jump height, or the number of baskets, goals or
tries scored. The execution of correct classical ballet technique however is determined
purely by the movement patterns and shapes the dancer makes with his or her body. This
may be the movement of the upper and lower limbs in relation to the torso, or the
movement trajectories of the whole body in the classroom or on the stage.
Although it is widely understood that other characteristics also influence the perceived
merits of a dancer, without correct and competent technique a classical ballet dancer will
-2-
not gain employment in a professional capacity. Aspects such as: musicality; projection
of personality and charisma; body shape and proportions; and the interpretive and artistic
abilities; all influence an audience members opinion of the appeal of a classical ballet
dancer. However, it remains the execution of correct technique that is the fundamental
prerequisite for a successful professional classical ballet career, and so it is essential that
these fundamental technical principles be clearly understood and meticulously mastered.
Without clear objective measures of skill competency, the young dancer in training must
rely on the movement analysis skills of their instructor in order to attain correct
technique. That is, they must rely on their teacher to have a thorough understanding of
what constitutes correct classical ballet technique, trust that their teacher is able to clearly
see how they are moving, and then be able to communicate effectively the details of any
corrections or alterations that are required. Basic to making effective corrections is the
teachers ability to see what is wrong and know what will help (Schlaich and DuPont,
1993, p53). (Schlaich and DuPont, 1993)
Outstanding teachers of classical ballet are known to possess many different attributes,
which not only include mastery of the technical materials of the class, but also various
personal qualities and values. Schlaich and DuPont (1993, p9-10) identify some of the
necessary characteristics of the expert dance teacher as being: a love of movement and
teaching; a physical knowledge of movement; a knowledge of the mechanics of the body;
interest in people; an ability to create a positive atmosphere for learning; ability to plan a
purposeful class; musical knowledge; ability to teach dancing with artistry; a trained eye,
and an ability to communicate clearly. Although these many different qualities are
required for effective dance teaching practice, a specific aim of this research thesis is to
investigate the characteristics and accuracy levels of the trained eye of classical ballet
teachers. Schlaich and DuPont (1993, p2) describe someone with a trained eye as
knowing what to look for, having the ability to spot problems, and having an awareness
to detail. It is therefore essential that effective teachers of classical ballet technique have
the ability to perceive the difference between skilled and unskilled movement. A
considerable proportion of professional classical ballet teachers would have themselves
undergone many years of classical ballet training, some of them would have achieved a
competent and strong enough technique, together with the necessary artistic attributes, to
-3-
gain employment as a professional ballet dancer. Regardless of their eventual

employment status within the dance profession, most classical ballet teachers have spent
many years, usually more than a decade, carefully watching the execution of classical
ballet technique. As students they would have watched demonstrations given by their
teachers, and throughout their lifetime would have attended many live professional
classical ballet performances and watched numerous dance films and videos. One would
expect therefore that all professional teachers of classical ballet should easily be able to
differentiate between a skilled and unskilled ballet dancer. But what are these parameters
or characteristics of the classical ballet movement vocabulary that one looks for in order
to identify a technically skilled ballet dancer, and just how good are ballet teachers at
perceiving them? Identification and measurement of the fundamental principles that
define the classical ballet technique, and the characteristics of the trained eye will be
the focal topics of this research project.
1.2 Aims of the Thesis

This research project initially aims to identify the fundamental theories and basic
biomechanical principle of correct classical ballet technique. To this end, a number of
highly regarded classical ballet texts outlining the theories and principles of the technique
will be examined. From this review of the literature a succinct list of theoretical
principles will be compiled. These principles will be described in terms of concise
measurable biomechanical parameters. Having identified these theoretical principles, and
their corresponding biomechanical variables, a sample of professional and nonprofessional ballet dancers will then be recruited to the study. In order to assess the level
of agreement between what is considered to be theoretically correct execution of ballet
technique, as outlined in the dance literature, and the practical execution of classical
ballet technique, both groups will perform the same set of basic classical ballet
movements while undergoing three-dimensional (3D) motion analysis and video analysis.
The analysis of human movement is a vastly complex field of research, encompassing a
variety of important components. In addition to joint movement patterns, motor control
and coordination (i.e. movement cause-effect & inter-segmental relationships) are some
-4-
of the other interesting areas of human movement research that are also relevant to the
execution of classical ballet technique. However, it should be noted that the focus of the
current research project will be constrained to investigation of the actual movement
patterns (i.e. kinematic data) of professional and non-professional ballet dancers, without
consideration of the muscle activation patterns (i.e. electromyography data), or joint
forces and moments (i.e. kinetic data) that control the resultant joint movements. While it
is understood that multiple motor control strategies may lead to a similar movement
outcome, this study will not attempt to perform any analyses or reach any conclusions
related to the motor control strategies employed by the participating dancers.
Kinematic difference between the two groups will be compared to determine the degree
to which professional ballet dancers execute classical ballet technique relative to the
theoretically
correct
principles
compared
to
non-professional
dancers.
The
biomechanical characteristics of classical ballet demonstrated by the professional ballet

dancers will be examined to determine if they are in accordance with the theoretical
principles discussed in the review of the classical ballet literature.
Given that the capacity to execute ballet technique correctly is a fundamental prerequisite
for a successful career as a professional classical ballet dancer, the kinematic data
obtained from the professional dancers can therefore be considered to be a practical
biomechanical benchmark of correct technique. Thus, having established the practical
kinematic principles of correct classical ballet technique, the next section of the study
will investigate the accuracy with which ballet teachers can visually assess and identify
whether or not a ballet dancer is executing correct technique. A group of professional
classical ballet teachers and a group of student classical ballet teachers will be recruited to
the study. The participating teachers will watch a number of videos of dancers performing
standard classical ballet steps. These videos will correspond to 3D motion analysis data
obtained during the first part of the study. The teachers will be asked to respond to
specific questions about the dancers technical execution of the steps. These responses
(i.e. qualitative assessments) will be compared to the quantitative data obtained from the
3D motion analysis. From these results it will be possible to determine the accuracy with
which the professional and student teachers identify certain kinematic characteristics of
classical ballet technique.
-5-
Upon consideration of the data obtained from the first two analysis phases of the thesis, a
tool for quantitative assessment of classical ballet technique will be developed. Given the
high expense, and long data collection and processing times typically associated with 3D
quantitative analysis, an assessment tool that is designed for routine or regular use, would
be useful, however it would need to be designed with maximum efficiency as a key
consideration. To this end, the final aim of the project is to develop a quantitative
assessment tool consisting of the minimum number of kinematic variables and the
minimum number and appropriate combination of ballet steps that can effectively detect
deviations in the kinematic variables that relate to the fundamental principles of classical
ballet technique.
Although this project is focusing purely on the art of classical ballet, a highly specialised
movement technique, the questions addressed within the study, and the conclusions
reached, may be of interest to other dance styles and other physical activities. All
movement instructors or coaches need to have a basic knowledge of the fundamental
biomechanical principles on which their movement discipline is based, and must rely
heavily on their qualitative analysis skills through use of observation. Thus, relevant
content knowledge and a well trained eye are crucial factors that ideally should be
developed by all physical education teachers and coaches. The results of the current
study therefore have the potential to not only assist dance instructors within the classical
ballet community, but can also provide an example of how similar research can be
applied to other dance genres and other movement disciplines. Accurate observation of
dance and sporting movements may save students from the frustrations of endless trial
and error in their attempts to acquire new skills, and reduce the risk of fatigue and
physical injury. The question of how accurately physical education instructors observe
skilled movement is therefore of great importance to all pursuits involving skilled
physical activity. An instructors ability to see what is wrong is the first step towards
effectively assisting a student to improve their physical technique, and consequently their
chance of achieving excellence in their chosen discipline. In addition, the methods
applied in the development of the quantitative assessment tool for classical ballet could
potentially be used for design of a similar tool for other dance genres or movement
disciplines. Thus, the results from this thesis may be also be useful and applicable
beyond the field of classical ballet instruction
-6-
1.3 Structure of the Thesis

The following section provides an overview of the thesis, giving a brief outline of the
contents of each chapter:
Chapter 2 - Literature Review

This chapter starts by examining the historical context that shaped the development and
refinement of classical ballet technique. Review of dance literature dating from 1820
through to 2010 enables identification of a list of fundamental biomechanical principles
that succinctly characterise the technical elements of classical ballet. Review of the extent
to which instrumented quantitative biomechanical analysis has been conducted on
theatrical dance styles, and reported in academic literature, forms the basis for the next
section of the literature review. Aims, instrumentation techniques, and protocols for
scientific studies involving quantitative analysis of various dance genres are discussed.
Additional focus is given to quantitative analysis of classical ballet technique. The
chapter concludes with a discussion of qualitative analysis of human movement.
Qualitative analysis is defined, its application in various contexts is discussed, and a
review of scientific studies that have investigated accuracy of qualitative analysis is
presented. Factors that affect accuracy of qualitative analysis of human movement are
identified and summarised.
Chapter 3 - Comparison of Theoretical and Practical Execution of Classical Ballet

Technique
This chapter presents a biomechanical analysis of classical ballet technique. Having
identified the fundamental theoretical principles of classical ballet technique, 3D motion
analysis of professional and non-professional ballet dancers performing core classical
ballet steps is conducted. The aim is to assess the level of agreement between practical
execution of classical ballet technique with the theoretical ideals presented in the dance
literature. The methods used, results obtained, and conclusions reached are reported.
-7-
Chapter 4 - Accuracy of Qualitative Analysis of Classical Ballet Technique

This chapter investigates the qualitative analysis abilities of professional and student
ballet teachers. The accuracy of qualitative analysis of correct and incorrect ballet
technique is investigated with respect to experience in classical ballet instruction.
Teachers qualitative responses are compared to quantitative data obtained from 3D
motion analysis. The methods used, results obtained, and conclusions reached are
reported.
Chapter 5 - Development of a Quantitative Tool for Assessment of Classical Ballet

Technique
This chapter describes the development of a quantitative tool for assessment of classical
ballet technique. The aim is to determine the minimum number and appropriate
combination of ballet steps for inclusion in the assessment tool, so that redundancy of
steps is identified and eliminated. Through this process, the quantitative assessment tool
that most efficiently evaluates dancers against the four fundamental principles of classical
ballet technique is developed. Methods and results of the analyses, and the final version
of the assessment tool are presented.
Chapter 6 - General Conclusions and Implications

This chapter summarises and discusses the main findings from Chapters 2 to 5. General
and specific conclusions are discussed, with focus given to their implications on dance
teaching practice. Potential future directions for this line of research are also outlined.
-8-
CHAPTER 2
LITERATURE REVIEW
-9-
CHAPTER 2
LITERATURE REVIEW
2.1 Foundations of Classical Ballet Technique

2.1.1 Introduction
In order for movement instructors to be effective in teaching any movement discipline or
art form, it is imperative that they possess a thorough knowledge and understanding of
the principles on which the movement form is based, and thus an internal image of the
desired movement response. For ballet teachers, this knowledge is typically acquired
through their training as dance students, and is perhaps further developed throughout a
professional performance career. In some cases, additional knowledge of theoretical
aspects of the discipline is acquired through completion of teacher training programs.
Whether it is through practical experience of the movement form, or through teacher
training programs, it is expected that classical ballet teachers fully understand the
fundamental principles of the technique of the art form. The term technique is used
widely in relation to skilled movement activities, but it is rarely defined. It is implicit in
the general understanding of technique, that if a skill is performed with a good
technique, rather than a poor technique, performance will be better (Lee, 2002, p814).
Berardi (2005, p77) states that good technique, rather than luck or chance, allows a
dancer to balance at will, pirouette quickly, perform with precision, and appear graceful.
It appears that those who use the word regularly (e.g. teachers, coaches, researchers,
writers) presume that their audience has an understanding of the term, however this may
not necessarily be the case. The term technique will therefore be defined here.
(Kent, 1984; Oxford Online Dictionary, 2011)
A general definition of technique is a way of carrying out a particular task (Oxford
Online Dictionary, 2011). Sport-related definitions of technique include: a pattern of
movement which is technically appropriate for a particular skill, and which is integral to
that skill (Kent, 1994, p443); the pattern and sequence of movements that athletes use
to perform a skill (Carr, 1997, p5) and, more complexly, the motion activity specified
by biomechanical principles of human motion which utilise motor features of movement
- 10 -
and body structure to obtain the best sports result (Bober et al. 1981, p502). Some
authors (Cooper et al., 1976, p10) avoid the term technique altogether and prefer to use
the term movement pattern. With respect to classical ballet, technique has been
described as: that manifestation of art from which a large part of the human personality
has been excluded (Mumford, 1952, p21; as cited in Hammond & Hammond, 1989,
p16); what the basic elements, positions, and movements are, and how they are to be
performed (Royal Academy of Dance, 1997, p5); the correct execution of the steps and
exercises (Paskeva, 2002, p2); the way in which dancers use basic physical movements
in class or performance (Berardi, 2005, p77); and the foundational principles and steps
on which ballet steps are built (Wikipedia, 2011). The sport-related definitions listed
above provide a more specific and consistent meaning for technique than the balletrelated statements. Collectively, the sport-related definitions suggest that technique
describes kinematic characteristics relating to the relative positions of body segments
during the execution of a movement task, whereas the definitions of ballet technique are
more varied and perhaps more vague, thus preventing a succinct and clear interpretation
of their meaning. So, it seems that despite its wide use within the classical ballet
community, the word technique does not have a clearly established meaning.
Consequently, for the purposes of this project, the word will be defined and used
according to the sports-related descriptions listed above, namely the relative position and
orientation of body segments. This definition therefore implies that technique can be
characterised by kinematic variables that can be visually perceived. (Bober et al., 1981;
Carr, 1997; Cooper et al., 1976; Mumford, 1952; Wikipedia, 2011)
Although, classical ballet has a very distinct movement style, the actual movements
themselves are not necessarily unique to the art form. In 1760 Jean Georges Noverre
(1727 1810), one of the great ballet masters of the 18th century, described the Seven
Movements of Dance, in terms of the quality or type of step (Lee, 2002). These
movements are defined as:
1. Bend (Pli)
2. Stretch (Battement)
3. Rise (Relev)
4. Jump (Saut)
5. Glide (Gliss)
- 11 -
6. Dart (Elanc)
7. Turn (Tour)
Noverres classification system is still referred to by many dance practitioners today
However, on consideration of each term it is clear that these types of movement are not at
all unique to classical ballet. There are many skilled movement disciplines in which these
actions are performed. It is hard to think of any martial arts or skilled sporting activities
where participants are not required to bend, stretch, turn, rise, or jump. It
would be expected that the biomechanical principles that enable highly skilled or elite
execution of these movements would be similar whether a dancer, a gymnast, or an
athlete were performing the task. For example, in terms of the trajectory of the whole
body, the biomechanical principles that enable a hurdler to achieve an optimum jump
over a hurdle, are likely to be very similar to those that would enable a ballet dancer to
perform a skilful grande jet (split leap), i.e. principles relating to projectile motion.
What is of interest in this study however, is the identification and measurement of what it
is that distinguishes a balletic jump or a turn, from a jump or turn in another movement
discipline? What is it that a person must do with their body in order to be performing
classical ballet rather than any other movement form? Essentially, it is this defining
characteristic of classical ballet technique that this study aims to investigate, and which
will be narrowed down to a definition of the way in which the limbs and body segments
are oriented with respect to each other while performing the movement. This section of
the literature review will focus on identifying, from a theoretical point of view, how the
limbs and segments of the body must be oriented with respect to one another in order for
a dancer to be executing correct classical ballet technique. The first stage of this
review will be to consider the history of classical ballet, and investigate how this history
has shaped the development of classical ballet technique.
2.1.2 The History and Development of Classical Ballet Technique

Most accounts of the history of classical ballet have placed emphasis on the unique
contributions made by particular individuals and groups as they responded to the social
and political forces of the time. Much attention has been paid to influential personalities
and their creation and/or performance of memorable dance pieces. Dance historians have
highlighted developments and changes in many of the artistic characteristics of classical
- 12 -
ballet, including costumes, music, scenery and style, but few it appears have focused
purely on the evolution and development of the technique itself. If one looks more
narrowly at just the development of ballet technique, one finds a discernable, logical and
rational order in the evolution of the technique over the past four centuries.
Figure 2.1 Influences of selected changes in classical ballet (from Hammond & Hammond, 1989, p21)
Classical ballet technique has developed in a series of stages. The important technical
innovations introduced and absorbed into the ballet tradition during each stage have been
described as being borne out of particular dissatisfactions with the existing conditions
(Hammond and Hammond, 1989). These stages may be related to the periods ordinarily
referred to when discussing the history of the arts (e.g. Middle Ages, Renaissance,
Baroque, Romantic, etc.), however they are not necessarily the same. For this reason, a
history of classical ballet focusing purely on its technical development may produce an
account different from its stylistic history. Recounting the history of classical ballet
technique in this manner, Hammond and Hammond (1989) have identified five stages in
- 13 -
which major technical innovations emerged that are still evident in todays ballet
technique. These stages can be labelled as the development of:
1. Courtesy of grand manners

2. Codification of the five positions of the feet
3. Maximum turnout of the legs
4. Mastery of aerial space
5. Maximum linear extension
This section of the literature review will discuss the history of the development of
classical ballet technique in terms of the five stages presented by Hammond and
Hammond (1979). Although the history of ballet technique will be discussed in relation
to these five stages, other references will also be used to piece together the history of the
technique.
2.1.2.1 The Origins of Ballet: The Emergence of Grand Manners

The early ancestors of classical ballet include various medieval folk practices such as
religious processionals, pagan rituals and village dances (Hammond and Hammond,
1979). Before the eleventh century, the lives lead by the medieval nobility did not consist
of delicacies or refinements that were vastly different from those of the common people
(Lee, 2002). However, at the end of the eleventh century, European cultural centres began
to develop in areas extending from the South of France to Northern Italy and Spain.
During this time, refinement in all things became the objective of medieval court life. To
this end they brought outsiders into their courts to instruct them in etiquette,
comportment, music, dancing and poetry. The troubadours who they employed for this
task were learned poets, songwriters, composers and performers with aristocratic family
background who had travelled widely, absorbing all kinds of information, including much
of the classical learning from scholars of the Eastern Roman Empire. The troubadours
court dances stressed elegance of posture and precision of execution. The early
conventional court dance style of the time thus became one of contrived elegance, with
the dancers often appearing devoid of most of their spontaneous emotion (Lee, 1983).
- 14 -
The troubadours were credited with setting down rigid rules for their dance compositions
and served as educators for close to two centuries during the late Middle Ages, spreading
the presence and instruction of dance amongst the nobility of Western Europe.
By the 1400s instruction in social dancing was the standard for training gentlemen and
women in their social behaviours. All members of the nobility were required to learn
dancing and social etiquette so that they could conduct their lives with an appropriate
manner of grace. By this time the roles of the troubadour had transformed into the
singular profession of the dance master with many achieving high rank as companions
of princes. With the dance master came dance theory (Bland, 1976). Three prominent
Italian dance masters from the 15th century - Domenico da Piacenza, (ca. 1400 - ca.
1475), Antonio Cornazano (ca. 1431 - ca. 1515), and Guglielmo Ebreo (ca. 1440 - ca.
1484) formed a basis of dance theory. Domenicos 1460 treatise De Arte saltandi et
choreas ducendi (On the art of dancing and conducting dances) established that there
was a body of independent ideas and codified dance steps that could be used in various
sequences and contexts, and in this sense Domenico fostered the development of
choreography (Lee 2002). Antonio Cornazano, a student of Domenico, commented
extensively on the execution and relative merits of the steps created by Domenico in a
1465 manuscript entitled Libra dellarte del danzar (Book of the Art of Dancing, yet
another of Domenicos apprentices, Guglielmo Ebreo, wrote a treatise on dance, De
Practica seu arte tripudii (On the Practice or Art of Dancing) (1463), that proved to be
the earliest significant contribution to Western theatrical dance (Bland, 1976). Ebreo
expanded on the ideas published by Domenico and wrote about dance in a way that made
a clear distinction between folk dance and artistic dance with a required manner and
aesthetic (Kassing, 2007). Widely read by dancing masters of the time, Ebreos treatise
called for the strict applications of rules in the execution of dance, and instructed its
readers in the necessary qualifications for dancing that must be rooted in refined
deportment. The dancing masters concepts of the carriage, posture, and movements
necessary for creating court dances for high society were further reinforced in a 1528
commentary on etiquette and social mores written by Count Baldassare Castigline (1478
1529) in Libro del Cortegiano (The Courtier). The Courtier had a profound influence
on the evolution of European sensibilities and was a noted reference work for dance
masters of the time (Lee, 2002).
- 15 -
Figure 2.2 From Practica seu arte tripudii (1463). It was claimed that the dancers
pictured capture the essence of the required skills and decorum for the dances of the
time (from Bland 1976, p33). (Bland, 1976)
The dance styles and technique devised by the early dance masters formed the basis for
much of the dance seen in the splendid Italian court entertainment during the
Renaissance. Lavish productions took place in Italian palaces to celebrate weddings,
birthdays or state visits. The dukes of the Italian city-states, who were greatly concerned
with displaying their wealth and status to one another, competed with each other in giving
these costly spectacles (Lee, 2002). Reasonably skilled members of the court nobility
were selected to perform the exacting floor patterns and steps choreographed for these
productions. The dancers were richly costumed and expected to closely adhere to the
rules prescribed by the dance masters.
In 1547 Catherine de Medici (1519 1589), a member of the ruling family in Florence
married the King of France, Henri II (1519 1559) (Anderson, 1992). She introduced
into the French court the same kind of extravagant entertainment productions she had
known in Italy. Known as Ballet de Cour, many of these productions were staged by
Balthazar de Beaujoyeulx (1535 1587), a gifted musician who came from Italy to be
- 16 -
Catherines chief musician (Anderson, 1992). Although the dance technique did not yet
resemble ballet as we know it today, ballet historians consider one of Beaujoyeulxs
entertainment productions, the Ballet Comique de la Reine (The Dramatic Ballet of the
Queen), to be the first ballet. A description of the production was documented in a
commemorative text, Ballet Comique. Several hundred copies of the printed Ballet
Comique were presented to many European courts and wealthy families, inciting
widespread imitation, thus cementing Frances reputation of superiority in ballet.
With their theoretical contributions, the dance masters of the fifteenth and sixteenth
centuries helped lay down the foundations of a new style of dance that would continue to
find much popularity in the courts of Renaissance Italy and France, eventually leading to
the development of classical ballet as it is known today. Although greatly limited in
technique by todays standards, the style did resemble the classical ballet style of today in
that it was heavily based on the exhibition of a noble demeanor, displaying much
elegance and grace (Homans, 2010). Dancers were required to move with dignity and
danced according to rules carefully laid down by the dance masters of the time. Careless
manners, awkward stance, and imprecise movements were not tolerated. Fundamental to
all was elegant posture straight spine, lifted chest, relaxed shoulders, long neck, erect
head, hands and arms held without tension (Hammond & Hammond, 1979, p594). Such
posture was the trademark of ballet at its beginnings, and remains so without exception in
the ballet classrooms and theatres of today.
2.1.2.2 The Court Ballet and the Five Positions of the Feet
The production of lavish ballets continued in the French courts throughout the sixteenth
and seventeenth centuries. The pinnacle of French court ballet coincided with the lengthy
reign of King Louis XIV (1643 1715) (Anderson, 1992). Louis father, Louis XIII
(1601 1643), had been a fine dancer, thereby instilling Louis XIV with a love of dance
from an early age. By participating in many court ballets Louis XIV gave dance
respectability and encouraged others to work at perfecting the art (Lee, 1983). Although
most dancers in the French court ballets were from aristocratic families, it gradually
became more common to select dancers of less noble class who displayed greater skill.
Louis XIV had great influence in the conversion of dance from amateur court
entertainment to a professional occupation. In 1661 he performed the first formal act to
- 17 -
establish ballet as a professional theatrical art by establishing the Academie Royale de

Danse (The Royal Academy of the Dance), a professional organization to train dancers to
perform for him and his court (Lee, 1983).
The Academie Royale de Danse, consisted of 13 ballet masters who he charged with the
task of re-establishing the art in its perfection. The French academy served to establish
ballet technique and set artistic standards (Hammond and Hammond, 1979, p595), and
later came to be known as the Paris Opera Ballet. It was the king's ballet master Pierre
Beauchamp (1631 1705) who began to work out exactly how each step should be
performed and is credited with establishing the important principle that the feet must
move to and from the five fundamental positions of the feet. He also described how to
dance simple steps such as chasse, glissade and pas de bourree with turned out legs, as is
still done today (Lee, 1983). From 1670 to 1687 Beauchamp, in collaboration with the
talented composer Jean-Baptiste Lully (1632 1687), created dance works of extreme
elegance. However it was not until 1681 that any females performed in the ballets created
by the Royal Academy. This was supposedly a strength issue, as females were considered
to be incapable of carrying the enormous headdresses, full heavy skirts and weighty
corsets that constituted the costumes of the time. Female dancers performed
professionally for the first time in 1681 in Le Triomphe de lAmour (Nordera, 2007).
Figure 2.3 Faune from Lullys Le Triomphe

de lAmour (from Nordera, 2007, p26) (Nordera, 2007)
- 18 -
The most important writing on ballet technique to emerge from the first years of
professional dance was published in 1701 by the dancing master, Roaul Feuillet (c1653
c1709), who had previously served in Beauchamps school (Anderson, 1992). Feuillets
book, Chorgraphie ou lart de decire la danse par caractres, figures et signes
dmonstratifs (Choreography or the Art of Recording the Dance through Types, Figures
and Demonstrative signs), was essentially a technical manual. As written information,
Feuillets work served to stabilise the French ballet terminology and codify existing ballet
steps (Lee, 1983).
It systematically arranged existing steps and reiterated the five
positions of the feet, including the distinction between the five "true" positions (feet
turned outward) and the five "false" positions (feet turned inward).
Figure 2.4 The five positions of the feet (from Homans, 2010, p136).
- 19 -
Figure 2.5 The five positions of the

feet, as depicted by Feuillet in his
Chorgraphie
in
1700
(from
Hammond, 2007, p67). (Hammond,
2007)
Feuillets manual was widely disseminated, being translated and published in 1706 by the
English dancer John Weaver (1673 1760), under the title Orchesography (Art of
Dancing). Eleven years later in 1717, Gottfried Taubert (1679 1746) translated
Feuillets system into German under the title Rechtschaffener Tanzmeister (The Worthy
Dance Master), thus further extending the French influence on European dance. In 1724
another English dance master, Kellom Tomlinson (c.1690 - c.1753), used the various
English translations of Feuillets publication as a reference for his own collection of
dances in the Art of Dancing Explained by Reading and Figures (Hammond, 2007).
Another important shift that took place during the reign of Louis XIV that influenced the
development of ballet technique was the relocation of dance performances from the
ballroom floor to a raised platform at one end of the room, that eventually became the
proscenium stage (Hammond and Hammond, 1979). With ballets now being viewed by
the audience from only one perspective rather than from all sides, the group processions
and geometric floor patterns characteristic of the earlier court ballets gave way to greater
emphasis on the movement of individual performers. Dance became more space oriented
(Kraus et al., 1997), providing new challenges for the choreographers as it was necessary
for the dancer to be able to move skilfully and elegantly sideways so as not to turn away
from their audience. The shift to the proscenium stage, and the consequent need for the
- 20 -
dancers to move sideways, emphasized the usefulness and importance of the turned out
position of the legs. The significance of the correct use of the five turned out positions of
feet thus became entrenched into the classical ballet tradition.
2.1.2.3 Professionalisation and Maximum Turnout of the Legs

Professional ballet persisted as a major form of French artistic expression for decades
after its introduction by Lully and Beuchamp. By the middle of the 18th century ballet
was flourishing as a profession and all large French cities had ballet companies modelled
after Paris (Lee, 2002). The Royal Academy, founded to preserve the dance as a uniquely
French art, held the exclusive privilege of authorising all dance productions in state
owned theatres. As a consequence, it was able to enforce its own traditions and
conventions of opera-ballet in all the most important French theatres (Anderson, 1992).
The tenets of its schooling and its art were guarded by law, and as a result its
choreographers and teachers looked only to the formulas of past successes for ideas for
their productions. Despite its growth in popularity, up until the mid-18th century many
restrictions had been placed on the development of the art of ballet. Skilled dancers were
weighed down by heavy and elaborate costumes, making free-flowing movements, jumps
and lifts extremely difficult (Lee, 2002). By 1760, ballet masters began to question the
restrictions placed on their art, which they saw as emanating from the days when ballet
was part of the rigid protocols of the court. The complaint of the 18th century ballet
masters was that ballet was not artistic enough (Hammond and Hammond, 1979). The
greatest of these ballet masters was the French choreographer Jean Georges Noverre
(1727 - 1810). In his Letters on Dancing and Ballets (1760), Noverre was the first to
propose ideas that became central to the further development of ballet as a serious art
form. Noverre believed that ballet should be a means of expressing a dramatic idea
through the perfect combination of dancing, drama and character. Noverre claimed that
the dancers should be able to illustrate and express the plot, characters and emotions of a
story using only their bodies and faces (Noverre, 1760 - as cited in Lee, 2002, p111).
Speech and singing were therefore discarded, as were restrictive wigs, masks and
cumbersome costumes
. away with those lifeless maskstake off those enormous wigs.discard the use
of those stiff and cumbersome hoops..Renounce that slavish routine which keeps your
- 21 -
art in its infancy.let us gracefully set aside the narrow laws of a schools to follow the
impressions of nature (Noverre, 1760 - as cited in Anderson, 1992, p72).
Noverre was the first ballet master to succinctly classify the types of movements
performed within the ballet repertoire. As stated previously, Noverre identified Seven
Movements of Dance and described them to be;
1. Bend (Pli)
3. Rise (Relev)
4. Jump (Saut)
5. Glide (Gliss)
6. Dart (Elanc)
7. Turn (Tour)
Noverre shared his ideas with many students, dancers, and choreographers of the time,
and developed the ballet daction, the earliest form of dramatic ballet that told the story
completely through movement (Anderson, 1992). In doing so he turned the attention of
ballet masters towards the development of the individual dancer as an artist, thus causing
a major advancement in the dancers technique. For women, who were now receiving
increased recognition as credible dancers, the removal of the long-hooped skirts allowed
a freedom of leg movement not previously seen. High-heeled shoes were replaced with
heelless ballet slippers, which encouraged more steps of elevation (Hammond and
Hammond, 1979). The combination of the five positions of the feet and this new found
freedom meant that dancers were now attempting faster beats, more controlled pirouettes,
and higher leg extensions. They subsequently discovered that even more outward rotation
of the legs better facilitated the execution of these movements. According to Noverre, In
order to dance well, nothing is so important as the turning outwards of the thigh
(Noverre, 1760 as cited in Anderson, 1992, p72).(Noverre, 1760)
At the beginning of the 18th century, the correct use of the five true positions of the feet
required each foot to be rotated outwards by about 45, that is, with the feet positioned at
right angles to each other. However, this degree of rotation was later deemed to be
insufficient for the lighter clad professional dancers of the late 18th century (Hammond
- 22 -
and Hammond, 1979), and by 1800 was increased to 90, which became the new required
technical standard, a permanent feature of classical ballet technique that is still strived for
today.
Thus, by the end of the 18th century the fundamental principles of the classical ballet
technique had expanded to include: the exhibition of grand manners and elegance; correct
use of the five positions of the feet with 90 turnout. By this time many great public
performing institutions had been established, along with their affiliated professional ballet
schools which became the place for specialised ballet instruction (Lee, 2002). The Paris
Opera Theatre, established in 1669, had paved the way for the development of Londons
Haymarket (1705), and later Covent Garden (1732); the Royal Danish Opera (1726), the
Vienna Burgtheater (1748), the Royal Swedish Opera (1773); La Scala in Milan (1778),
and the Bolshoi in Moscow (1776).
2.1.2.4 Ballet Independence and Mastery of Aerial Space

By 1800 ballet was no longer a mere distraction within opera or drama but had gained a
measure of artistic independence and could hold the stage alone (Hammond and
Hammond, 1979). However, the increased focus on the technical qualities of classical
ballet during the 18th century soon started to diminish, and a shift in attention towards the
emotional content of the productions started to occur. The Romantic movement of early
19th century Western Europe brought about a change in the direction of the art form, with
a new focus on the values of human emotion and individualism, in contrast to the rational
and logical ideals of the previous era (Lee, 2002). The significant feature of the Romantic
ballets therefore was their highly emotional content. A new generation of European
artists created dance productions that expressed immense feeling and depicted a strong
desire to transcend the human condition into exotic places inhabited by supernatural
creatures. The creative high point of the Romantic ballet occurred between 1830 and
1845. The two memorable ballets of the period, La Sylphide (1832) and Giselle (1841),
portrayed the realism of rustic village settings against the ethereal world of fairy creatures
known as Sylphs or Wilis. The contrasting themes of the subject matter greatly enriched
the choreographic content and diversity within the Romantic ballets, thereby enhancing
the dancers technical requirements. The unreal or supernatural sections of the ballets
called for a greater development of the aerial aspects of the technique. That is, in order to
- 23 -
successfully interpret these ethereal or otherworldly characteristics, the dancers had to

display a superior quality of lightness, or ballon (Lee, 1983). For this reason, jumping
steps were refined and perfected for the meaning that they could bring to the portrayal of
mystical beings. This notion was suitably captured by the words of famed Italian ballet
teacher and theorist, Carlo Blasis (1795 1878):
Be as light as possibleI would like to see you bound with a suppleness

and agility which gives me the impression you are barely touching the ground and
at any moment may take flight (Blasis, 1820, p46).
The innovation that contributed most significantly to the creation of the ethereal mood of
the Romantic ballets was the inclusion of pointe work. Dancing en pointe, the common
term used to refer to dancing on the tips of the toes, helped dancers achieve the illusion of
defying gravity. The Romantic ballerina Marie Taglioni (1804 1884) is credited with
introducing the modern pointe technique and was seen to dance the full length La
Sylphide en pointe (Kant, 2007). Pointe work became mandatory for all female dancers
during the Romantic period as it provided the ultimate illusion of weightlessness that
characterised the creatures of fantasy depicted in the ballets of the era. The first pointe
shoes were delicate, heel-less, close-fitting, satin slippers which were part of the fashion
of the times (Kant, 2007). They were reinforced around the leather sole and the toes with
extra stitching and featured long ribbons that wound around the ankle to give extra
support. Starched muslin, felt, or cardboard wrapped around the foot also added to the
shoes strength (Lee, 2002). Not until 1880 were the first commercial blocked pointe
shoes introduced, so the female ballet dancer of the Romantic era had to develop her foot
and ankle strength without artificial aids (Lee, 1983).
The thematic material of the Romantic ballets enabled the female dancer to reign supreme
during this period. The ballerinas increasing use of her pointes made her balance more
precarious, requiring highly attentive partnering from her male partner. This contributed
to the development of greater complexity in the pas de deux (dance for two) between the
male and female dancers that served not only to display great tenderness and emotion, but
also the practical need to assist the female dancers balance en pointe. Although there
were some very good male dancers of the time, most were used only to assist the
ballerina and carry her around the stage. Consequently, by the mid 19th century
- 24 -
comparatively few male dancers were enrolled in the professional ballet schools. For the
first time in the history of dance, the female presence superseded that of the male (Lee,
2002).
This emphasis on the female ballet dancer allowed the development of the
ballerinas technique. With more attention paid to their technique, women became
celebrated as performers.
With the rise in importance of the individual female dancer, ballet teachers and masters
set about formulating the rapid growth of pedagogical innovations in classroom
instruction. It was the inspiration of a number of prominent ballet masters, in particular
Blasis, Taglioni, Jules Perrot (1810 1892) and Arthur Saint Leon (1821 1870), that
provided the female dancers with the technical and choreographic challenges on which
their individual fame rested (Lee, 1983).
Although more has been published about Romantic ballet in Paris and London than in
other European cities, a great deal of activity also took place outside these two centres. In
Italy, due to the distinguished reputation of the Italian ballet master, Carlo Blasis (1795
1878), Milan enjoyed a special prominence in the world of 19th century ballet. In the
early part of the century, Blasis published two manuscripts on the analysis of ballet
technique. His 1820 work, Trait lmentaire, thorique, et pratique de l'art de la danse
("Elementary Treaty on the Art of the Dance, theory and practice"), was followed by the
1828 manuscript, The Code of Terpsichore (The Art of Dancing), which became the
standard for European ballet instruction and guided entire generations of dancers (Lee,
2002). Blasis teaching methods perfected the aerial and floating qualities required for
the Romantic ballets. He methodically applied the physical laws of equilibrium to the
human body, an approach that gave rise to the current ideas of balance, placement,
alignment and centredness in classical ballet (Lee, 2002). His geometrical interpretation
of the body with respect to classical ballet positions developed in his students a sense of
the same body placement and alignment that the dancers of today strive to achieve.
Strive to acquire perfect bodily equilibrium and to achieve this end never deviate from
the perpendicular which should extend from the centre of the collar bone down between
the ankles of both feet. (Blasis, 1820, p34).
If a dancer is to remain balanced upon one leg, or for that matter well poised upon both,
he must, besides disposing himself gracefully, establish the correct counterpoise of all
other parts of the body. (Blasis, 1820, p34).
- 25 -
The body should be held erect in a perpendicular line with the legs, except in certain
positions, notable arabesque, where one is obliged to lean backward or forward. The
position notwithstanding, the body must always rest firmly poised upon the hips. Throw
out your chest and hold your waist in, keep your knees springy and brace your back
(Blasis, 1830). (Blasis, 1820, p25).
Blasis manuscripts were the first comprehensive books on ballet technique, as we

understand it today, to appear in print. In 1837 Blasis was appointed as the director of the
Imperial Regia Accademiadi Danze (The Imperial Academy of Ballet and Pantomime) in
Milan. The ballerinas trained there were renowned for their exceptional technique, and
were soon in demand everywhere in Europe. The structure of the todays ballet class
stems from Blasis three-part class. The barre exercises progressively train the body to
form the prescribed classical shapes. The exercises are then repeated in the centre of the
room without the support of the barre. In the final phase of the class, all the elements of
the barre are transformed into dynamic dancing movements with the addition of jumping
and turning movements (Lee, 2002). He reinforced his theories with illustrations
depicting the balanced poses and well-balanced torsos required for the execution of
superior technique. He devised long combinations of steps known as enchainements to
develop lung capacity, leg strength and stability so that the dancers could cope with the
new complexities that choreographers were inventing (Lee, 2002). Thus, from the early
to mid 1800s, the specific theoretical principles of classical ballet technique were refined
and consolidated as the structure of the modern day ballet class developed and countless
ballets replicating La Sylphide, and Giselle were produced. This era therefore saw
classical ballet achieve its modern identity as it acquired many of the characteristics that
are associated with it today: the pointe technique; the tutu; and the desire to create the
illusion of weightlessness and effortlessness. Significantly, most of these characteristics
apply solely to the female dancer, as stated previously, in the course of the 19th century
the male dancer suffered a considerable loss of prestige. Thus, by the mid 1800s classical
ballet had developed into an art form that was clearly based on the features of grand
manners, codified foot positions, 90 turnout, and now the addition of elevation and
pointe work.
- 26 -
2.1.2.5 Decline of Romanticism and Maximum Linear Extension

As the mid-19th century approached, the Romantic ballet began to lose its impact and by
1850 the glory of the Romantic ballets golden age had all but disappeared. The last
quarter of the century saw the abandonment of the Romantic periods artistry and
expressiveness, in favour of a more superficial concentration on technical virtuosity and
visual spectacle. The great European choreographers of the Romantic era had passed
away, and no new talents had risen to take their place. Ballet seemed to have lost its
creative momentum, and the public had ceased to regard it as a serious art form. In
France, where ballet was protected and housed by the Paris Opera, a lack of significant
talent and public interest in the art contributed to its decline. In Italy, art patrons were
beginning to prefer operas to the overly repetitious ballets of the time. Thus, towards the
close of the 19th century, ballet in Western Europe was no longer a mainstream art, as it
had been in the 1830s and 1840s, and appeared to be about to die of exhaustion.
While the Western European public were losing interest in ballet during, the art form was
steadily developing in Russia under the influence of the Russian Empire. Since 1738,
ballet in Russia had received Imperial protection and support because it initially existed
as court entertainment. During the 19th century the Maryinsky Theatre in St. Petersburg
(known as the Kirov Theatre during the Soviet years) and the Bolshoi in Moscow were
state owned and controlled, with the directors of these companies being personally
appointed by the tsar (Anderson, 1992). Although there were some excellent Russian
dancers in St. Petersburg throughout the Romantic period, Western European dancers and
choreographers had been increasingly travelling to Russia, and so contributed to the
development of Russias own successful Romantic age. A new era in Russian ballet
began when a French dancer, Marius Petipa (1818 1910), arrived in St. Petersburg in
1847. While the popularity of ballet was declining in the West, Petipa was initiating a
new force in the development of ballet which saw the synthesis of the most significant
technical advancements from both the Russian and European Romantic era (Lee, 1983).
With the presence and eventual leadership of Petipa, the Maryinsky Theatre in St.
Petersburg became a vital dance centre. For nearly forty of the sixty years he was in
Russia, Petipas concepts of training, choreography and production dominated the dance
at the Imperial School of Ballet and the Maryinsky Theatre, and to a lesser extent, that of
the Bolshoi Theatre in Moscow.
- 27 -
Petipas contribution to the ballet repertoire was significant. Among his masterpieces that
are still performed today, are Don Quixote, 1869; La Bayadere, 1877; Sleeping Beauty,
1890; The Nutcracker, 1892; Swan Lake, 1895; and Raymonda, 1898. However, by
centurys end, the Russian moment in ballet was over, and Petipas generation passed
abruptly from the scene. Petipas legacy however was enormous. Under his leadership,
the entire axis of classical ballet had shifted. For two centuries, the art form had been
quintessentially French, but from this point on, classical ballet would be seen as Russian
(Homans, 2010). It is two great Russians, Serge Diaghilev (1872 1929) and Mikhail
Fokine (1880 1942), who are largely held responsible for the reform and redevelopment
of ballet that soon took place within Europe in the early 20th century.
Diaghilev was born a member of the Russian nobility. He was not a dancer,
choreographer or composer, in fact he studied law, but later became interested in ballet
and opera, and was given a supervisory post at the Maryinsky Theatre. Because of his
independent and forthright character, he found it difficult to work there and resigned after
only a brief period (Kirstein, 1987). He was ambitious and innovative and went on to
produce art exhibitions in St. Petersburg and Paris, and in 1908 he became a theatrical
impresario. It was in this role that he made his significant contribution to dance. In 1909
he assembled a group of leading Russian dancers from the Imperial Ballet to form a new
company, the Ballet Russes, and arranged to present a season of dance in Paris. The
selected dancers, all of whom had graduated from the Imperial Theatre School at or near
the turn of the century, were part of a new generation of Russian dancers that came in the
wake of Petipa. The list of dancers included historically significant names such Mikhail
Fokine, Anna Pavlova (1881 1931), Tamara Karsavina (1885 1978) and Vaslav
Nijinsky (1889 1950). The ballerinas of this generation were physically distinctive
from their voluptuous and sturdy predecessors. They were long and lithe, with smooth
lines, and evenly developed muscles and a soft sensuality (Homans, 2010). This was
primarily due to the training they had received at the Imperial Theatre School under the
guidance of the Italian ballet master Enrico Cecchetti (1850 1928). Cecchetti, who
arrived in St. Petersburg in 1887, belonged to the old Italian school so his teaching
methods were directly rooted in the pedagogy of Carlo Blasis. He emphasised repetition
and tricks, and designed long gruelling enchainements to build strength and endurance. In
doing so, Cecchetti gave his students the tools to redefine classical technique. Rather than
- 28 -
dwelling on the types of steps and bravura stunts, the new generation of dancers used
their technical abilities to sculpt their bodies and develop soft and flowing movements for
which extreme flexibility was highly emphasised. Anna Pavlova, the prototype of the new
Russian ballerina of this time, was described as being thin with highly arched feet,
gangly arms and a long straight neck.the line of her leg was slender and taperingshe
looked wispy and evanescent without a hint of heavy strength or bravura (Homans,
2010, p292). This description is typical of the vast majority of professional female ballet
dancers working today.
In the early 1900s Diaghilev formed a new ballet company, the Ballet Russes, for which
he appointed Michel Fokine as the first choreographer. Fokine, an experienced
choreographer, had previously worked from 1900 with the St. Petersburg company
(Kraus et al., 1997). Together, Diaghilev and Fokine, aimed to create a new and
innovative repertoire. Fokine urged that technique should be the means to express
character and emotion. He felt that a dancers entire body, rather than separate mimed
gestures, should express the story or concept at all times. The choreographic works of
Fokine therefore matched well with the increased physical fluidity and suppleness that the
new generation of graduating dancers of the Imperial Theatre School had developed. The
result for ballet was a significant increase in the breadth of physical movement, especially
in the range and mobility of the limbs - referred to by Hammond and Hammond (1979,
p600) as maximum linear extension. For example, the leg which, in the 19th century,
was raised to a point parallel to the floor, was, in the 20th century, capable of being
extended above the head, almost perpendicular to the floor. This extension was possible
because it built upon its technical predecessors - a capacity for extreme turnout and the
strength and lightness involved in mastering aerial space. That is, it was a movement
style, or element of the technique, that would have been impossible for the ballet dancer
of the 18th century (Hammond and Hammond, 1979).
Displaying new extremes of physicality and flexibility, Diaghilevs Ballet Russes, opened
to great acclaim in Paris in 1909. Until this time, ballet in France had not regained the
highly regarded status it held during the Romantic era. The arrival of the Ballet Russes
was therefore received by Paris audiences with extreme enthusiasm. The male dancers,
among them Vaslav Nijinsky, were particularly admired because elite male dancers had
- 29 -
almost disappeared in Paris. For his time, Fokines choreographic ideas were
revolutionary, and the French patrons were thoroughly excited by the brilliant Russian
dancers disregard for the rigid conventions of the past and their ability to dance in a way
that used their entire body to portray emotion. This event marked the rebirth of ballet in
the West. Following its hugely successful debut in Paris, the Ballet Russes toured
throughout Europe to continued acclaim. Over a period of 20 years (1909 - 1929),
without any permanent base, the company travelled to many European cities, and also to
the Americas. Many new dancers were employed initially from Russia, but World War I
and the Russian Revolution (1917) separated the company from its homeland during the
companys middle years. Fresh new talent was therefore sought from the West, a time
which saw English dancers Alicia Markarova, Antony Dolin and Marie Rambert, and
Irish dancer Ninette de Valois join the ranks. In 1924, towards the end of the companys
existence, Russian born George Balanchine (1904 - 1983) was recruited after defecting
from Soviet Russia. Throughout the lifetime of the Ballet Russes company, Diaghilev
employed a succession of choreographers to produce new and innovative works. In
addition to the founding choreographer, Fokine (Prince Igor, 1909; The Firebird, 1910;
Petrouchka,1911), other great names who choreographed for the Ballet Russes included
Nijinsky (The Afternoon of a Faun, 1912; The Rite of Spring, 1913), Leonide Massine
(1895 1979) (Parade, 1919), Bronislava Nijinska (1890 1942) (Les Noces; 1924), and
in the companys final years, a still very young George Balanchine (Apollo, 1928; The
Prodigal Son, 1929).
Upon Diaghilevs death in 1929, the Ballet Russes disbanded and the companys dancers
and choreographers scattered around the globe to initiate major developments in ballets
worldwide growth. Fokine worked with many companies, including the future American
Ballet Theatre. Massine contributed to the Ballet Russe de Monte Carlo, a company
formed after Diaghilev's death. The Polish-born dancer Dame Marie Rambert (1888
1982) and the British dancer Dame Ninette de Valois (1898 2001), became the founders
of British ballet. Rambert's students included the British choreographers Sir Frederick
Ashton (1906 1988), Antony Tudor (1908 1987), and John Cranko (1927 1973). De
Valois founded the company that became Britain's Royal Ballet, for which celebrated
dancer Margot Fonteyn (1919 1991) danced from 1935 to 1979. Balanchine was invited
by Lincoln Kirstein, a wealthy young American patron of the arts to work in the United
- 30 -
States, from where he went on to form the School of American Ballet and New York City
Ballet. The Russian ballerina Anna Pavlova, who danced in the early seasons, formed her
own company and toured internationally. One of Pavlovas dancers, Edouard Borovansky
(1902 1959) later settled with his wife in Australia where they established the
Borovansky Ballet Company, which in 1962 formed the basis for The Australian Ballet
Company. Before this migration of artists in the 20th century, ballet was only known to
relatively small and elite groups (Lee, 2002), but within a few generations this groups
knowledge and dedication to dance has been disseminated across multiple continents to
millions of people.
Thus, by the early 20th century the fundamental technical requirements of classical ballet
as we know it today had been established, and by the mid 20th century specific detail and
knowledge about the technique had been disseminated around the world. It became
expected that a professional ballet dancer would possess the technical attributes of grand
manners, codified foot positions, 90 turnout, elevation and pointe work, and finally
extreme flexibility displayed through maximum extension of the lower limbs. These
defining qualities of classical ballet technique remain largely unchanged today.
2.1.2.6 Classical Ballet Today

While the defining fundamental technical characteristics of classical ballet technique have
changed little since the early 20th century, the skill and complexity of movements that the
dancers of today are able to execute has steadily increased throughout the 20th and 21st
centuries. This has in large part been due to specific syllabi and training methods that
have been developed throughout different regions of the world. The ballet syllabi most
commonly used to train ballet dancers throughout the world today are the Royal
Academy of Dance Syllabus; the Cecchetti Method, the Vaganova System, and
Balanchine Method.
First established in the United Kingdom in 1920, The Royal Academy of Dance (RAD) is
currently one of the worlds largest dance education and training organisations. Today the
RAD has a presence in 79 countries, with 36 offices and over 13,000 members worldwide
(http://www.rad.org.uk). The RAD offers an internationally recognised portfolio of
examinations and assessments for students of all ages and levels of ability. Classical
- 31 -
ballet technique is taught using a systematic syllabus, for which commencing students
may be as young as 6 years of age. The system consists of Graded (Grades 1-8) and
Vocation levels of study.
Enrico Cecchetti, one of the worlds outstanding teachers of ballet of the late 19th and
early 20th centuries, established a system that passed on the tradition of ballet to future
generations of dancers. This system, the Cecchetti method, was codified and recorded by
Cyril Beaumont, Stanislas Idzikowski, Margaret Craske, and Derra de Moroda (Grant,
2008). The method has a definite program of strict routine and includes a table of
principal daily exercises for each day of the week (Grant, 2008). The Cecchetti Society
was formed in London in 1922 to perpetuate his method of teaching. In 1924, the society
was incorporated in to the Imperial Society of Teachers of Dancing. Entrance to the
Society is by examination and students must pass through a carefully graded system
which has done much to raise standard of dancing throughout the British Empire.
During the 1920s, the Russian ballerina and teacher Agrippina Vaganova developed a
planned instructional system which later became known internationally as the Vaganova
System. Vaganova was personally responsible for revamping the pedagogical system that
produced Russian dancers (Lee, 2002). A new period of dance pedagogy began with the
publication of Vaganovas Basic Principles of Classical Ballet (1934). Her methods
expanded the entire body of technical knowledge so comprehensively that it became the
system on which Russian ballet training is based to this day (Grant, 2008). Vaganovas
exacting demands of her graded syllabus meant that ballet teaching in Russia became
entirely professional. Retiring dancers could no longer claim the right to teach without
first completing a two year training course (Lee, 2002).
The Balanchine Method is a ballet technique and training system devised by the Russian
dancer and choreographer George Balanchine and initially used at the School of
American Ballet, the school of the New York City Ballet, and at many schools of
Balanchine's disciples, such as Miama City Ballet, Ballet Chicago Studio Company and
the Suzanne Farrell Ballet in Washington D.C. The overall illusion of the Balanchine
Method is that dancers are utilising more space in less time: speed, height, length and a
syncopated musicality are created (Schorer, 1999). Dancers must therefore be extremely
fit and flexible.
- 32 -
Technical manuals for each of the aforementioned ballet training syllabi or systems, as
well as other literature sources, will be referred to in the following section, which will
focus on describing and succinctly defining the fundamental principles of classical ballet
technique. In defining the fundamental principles of the technique, it is the similarities
amongst the various training systems that will be focused on, rather than the unique
characteristics of each system. The next section will outline the fundamental principles of
2.1.3 The Fundamental Principles of Classical Ballet Technique

The review of the history of classical ballet technique, shows that the technique we know
today developed through a series of logical progressions and advancements: some due to
social influences (e.g. grand manners); some grew out of physical necessity (e.g. turnout);
and others from the stylistic inspiration of particular individuals or various phases of
European artistic history (e.g. five positions of the feet; mastery of aerial space;
maximum linear extension). It is clear that each stage directly contributes to identifying
and defining a few key principles of the technique. For the purposes of this thesis, four
key principles have been identified:
1) Alignment
2) Placement
3) Turnout
4) Extension
These fundamental principles of classical ballet technique will be clearly defined and
described in the follow sections.
2.1.3.1 Alignment
Grand manners, the first of the technical requirements that had been established by the
beginning of the 20th century, specified deportment and carriage of the upper body. It
outlined a desired aesthetic that remains current, namely extreme elegance and poise.
Classical ballet dancers are expected to maintain this aesthetic while executing various
- 33 -
movements with the lower limbs. For example, a professional ballet dancer is expected to
be able to raise one leg to a very high level at the front, side or back of their body, while
maintaining an upright vertical posture with their torso. This requirement is consistently
identified in many dance literature sources, dating from as far back as the 1820 writings
of Carlo Blasis through to the more contemporary writings of todays major ballet syllabi.
The following quotations and figures support this requirement.
Strive to acquire perfect body equilibrium and to achieve this end never deviate
from the perpendicular which should extend from the center of the collar bone
down to between the ankles of both feet. The body should be held erect in a
perpendicular line. A fine carriage is one of the principal merits in a dancer. The
upper part of the body should always be displayed with elegance. While dancing,
the body must remain quiet and absolutely steady. The dancer who jerks his body,
raises his shoulders in reflex action to his leg or bends his back to facilitate
execution is a ridiculous sight. These mischievous dancers, whose sole
contribution consists in fostering bad taste, should be banished from the theatre
(Blasis, 1820, p24-25 & p34).
Figure 2.6 Correct perpendicular alignment of the body (from Blasis, 1820, p18 & p26)
(Blasis, 1820)
Keep your body poised erect and perpendicular on your legs. In the execution of
your steps, let the body be quiet, firm and free from jerkiness. A common fault of
pupils is throwing out their waist at the same time as throwing out their chest, (ii)
throwing out their posterior when required to keep in their waist (Beaumont and
Idzikowski, 1932, p30). (Beaumont and Idzikowski, 1932)
- 34 -
To master stability in the dance, to gain aplomb, is a matter of primary

importance to every dancer. A body which does not stand straight on the foot, but
inclines towards the barre, will never gain aplomb and balance. The stem of
aplomb is the spine (Vaganova, 1953, p24-25).
Figure 2.7 Correct and incorrect alignment of the body for classical ballet technique
(from Vaganova, 1953, p25).
A ramrod spine is needed. A firm grip in the muscles of the spine is essential. If
the spine and buttocks are relaxed, the trunk will incline forward and the tail will
stick out. (Karsavina, 1962, p10).
Stability is one of the basic elements of classical dance. The basis of stability lies
in the preservation of the vertical axis, which passes through the middle of the
head and body to the ball of the supporting foot when one is standing on demipointe and in front of the heels when one is standing on the whole foot. The body
is in a vertical and pulled-up position in all exercises. (Kostrovitskaya and
Pisarev, 1978, p62). (Kostrovitskaya and Pisarev, 1978)
The tail (coccyx) and pelvis must be pulled downwards and the spine upwards
from the waist. The three natural curves of the spine are thereby straightened. The
head must be held erect so that the crown is directly over the insteps of the feet.
(Lawson, 1979, p13).
The term alignment is used to denote the precise vertical positioning of all the
sections of the torso, in relation to all parts of the legs as it is seen in profile. Good
posture can be thought of as beginning with the spine stretching upwards and
downwards. The feeling in the thoracic area is one of lifting upward in an upward
and vertical pull (Lee, 1983, p207).
- 35 -
Figure 2.8 Correct and incorrect body alignment (from Lee 1983, p207)
The basic stance of classical ballet elongates the spine so that the natural curves
are diminished, though by no means eliminated. The key to good posture is the
correct alignment of the spine. The bony masses of the skeleton head, rib cage,
and pelvis are stacked vertically, like building blocks, over their base, the legs and
feet. This balanced position, with the shoulder girdle hanging easily on top of the
rib cage, causes the line of gravity the imaginary line that exactly bisects the
weight of the body tall fall through the centre of the structure. All parts, on
either side of the line, balance each other. Viewed from the front or the back, the
line bisects the body into two symmetrical halves. (Grieg, 1994, p17).
Central to the basic stance is the concept of aplomb. Mme Vaganova used the
term to refer to vertical stability and alignment. Seen from the side, the line of the
dancers body (from head to toe) will appear to incline almost imperceptibly
forward, so that the shoulders are in vertical alignment over the front of the pelvis,
slightly favouring the balls of the feet. (White, 1996, p90).
The spine forms the central axis of the body, and is composed of two forward
and backward curves. In the dancer these are held in a lengthened position by the
abdominal muscles at the front and the extensor muscles at the back. The sides of
the waist are kept equally lengthened and stabilised by support from the oblique
abdominal muscles. Whether the dancer is performing adage with high extension
or allegro, the length of the back and sides is retained as far as possible. It is the
training, strengthening, and strong control of the spine that advances classical
ballet technique (Royal Academy of Dance, 1997, p8).
- 36 -
Figure 2.9 The correct lengthened

position of the spine (from The Royal
Academy of Dance 1997, p8).
Stand upright like you swallowed a yard stick Mr Balanchine wanted us to

stand straight, with the shoulder square to the barre, held down and open. The
chest was proudly lifted. (Schorer, 1999, p52).
In order to achieve an untrammelled classical line, attention must be given to the

verticality of the body, produced by alignment of the spine. The classically
straight back is achieved by minimising the curvatures of the thoracic and lower
dorsal areas. The practice of alignment starts by standing erect keeping an
elongated spine. This position lengthens the spine, allowing each section of the
torso to occupy the space it needs without impinging on another part. The result of
this alignment is a posture that gives dancers a characteristic super-uprightness. It
is not to be confused with stiffness, in fact, it is its exact opposite; rather this
position can be described as a pliable verticality. Maintaining verticality with ease
means that the line of gravity is shifted to the supporting side; all the muscles of
the torso and pelvis are engaged in that positioning. (Paskeva, 2002, p11 &
Paskeva, 2005, p39).
- 37 -
Figure 2.10 In leg extension to the front or side, the body is expected to remain
close to the vertical (from Paskeva, 2005, p43). (Paskeva, 2005)
Straight spine, relaxed shoulders and neck, arms firmly held, and neck erect is
the first requisite for all exercises. The body is poised and ready to move.
(Walczak and Kai, 2008, p246).
Collectively, the previous quotations indicate the importance of maintaining an elongated
vertical spine while executing all traditional classical ballet movements. For the purposes
of this project, this principle of the classical ballet technique will be referred to as
alignment in all future sections of the thesis. The principle of alignment will thus be
succinctly defined as maintaining verticality of the torso.
2.1.3.2 Placement
To achieve the desired elongated torso, the position of the pelvis cannot be ignored. The
orientation of the pelvic segment contributes to the overall appearance and apparent
length of the spinal column, and therefore plays a crucial role in the execution of correct
classical ballet technique. An incorrectly positioned pelvis, with respect to the thorax,
will influence the shape of the lumbar spine. Thus, in addition to the position of the
thoracic segment, appropriate positioning of the pelvic segment also plays a role
achieving an elongated appearance of the torso, and therefore grand manners. Many
authors on classical ballet technique have provided instruction on the correct position of
the pelvis for classical ballet. It should be noted however that in discussing the pelvic
- 38 -
placement for ballet, some authors refer to the hips rather than pelvis. However, in
relation to this usage of terminology most dance writers are typically referring to the
anterior superior iliac spine (ASIS) of the pelvis, rather than the actual hip joint. The
following quotations from various authors indicate how the pelvis (hips) should be
placed in classical ballet.
He who whilst dancing moves his body by jerks, bends or relaxes his loins to
facilitate the execution of his tems, is unquestionably an object of ridicule, and the
name of a grotesque would suit him much better than that of a dancer (Blasis,
1828, p66).(Blasis, 1828)
Let the body be quiet, firm and free from jerkiness, a common fault is throwing
out the posterior when required to keep in the waist (Beaumont & Idzikoeski,
1932, p30).
When the leg is thrust out to the front, side or back, the upper part of the leg, the
hip, should not be raised too high. It should be done so that the leg works
independently, without involving the body in the movement. (Vaganova, 1953,
p28).
The hips must be square to the position of the feet. The tendency, very frequent,
of throwing out the hip of the working foot must be checked. The shoulders have
to be, at this stage, in line with the hips (Karsavina, 1962, p40).
The basic conditions for the mastery of stability are the correct distribution of the
pulled-up body over both legs or over one, and level hips. (Kostrovitskaya, 1978,
p63)
The pelvis must be balanced over the two legs and held firmly by the so-called
muscular corset with the hips level at all times. The dancer must be in correct
placement with hips and shoulders level and lying parallel to each other, facing
the same plane. In pure classical dance there should be no curving forwards,
sideways or backwards below the waist (Lawson, 1979, p13).
The term placement in ballet refers to the precise horizontal positioning of the
shoulder and pelvis area as viewed from the front or back planes of the body. That
is, the horizontal lines of the shoulder joints and crests of the pelvis are parallel to
each other and the floor when the body is supported by both legs. Correct
placement in ballet technique hinges on two rules of thumb. First when the body
weight is momentarily transferred to one leg, the horizontal and/or parallel lines
of pelvic and shoulder in relation to the parallel floor line, must not be upset.
- 39 -
Second, in large movements using extensions and leaps, the shoulder and pelvis
lines will lose a degree of their parallel relationship with the floor. However, the
balletic concept of placement insists that the horizontal and parallel lines must be
regained as soon as the body weight changes to one or both legs (Lee, 1983,
p209-210).
Figure 2.11 Correct (left image) and incorrect (right image) placement of
the pelvis (from Lee, 1983, p209).
The dancer must find a correct centered placement for the pelvis. The hip bones
are lifted in front, at the back the sacrum is directed strongly downward. The
buttocks are held high on the legs, while the lumbar spine remains elongated.
When one leg is disengaged from a closed position, an important tenet of ballet
training is that the hip crests should remain level and square to the front. In tendus
to the front and side, the pelvis should remain uninvolved, with the crests
absolutely level and facing square to the front (Grieg, 1994, p41).
Figure 2.12 Centred placement of the pelvis

(from Grieg, 1994, p40)
- 40 -
With the abdominal muscles active and the muscles of the buttocks held firmly,
vertically align the pelvis block beneath the lower back. Do not allow the buttocks
to push or tuck under. Shoulders and hips held horizontal the right shoulder
directly above the right hip and the left shoulder directly above the left hip
(White, 1996, p83).
In classical technique the pelvis is kept square. This squareness is controlled by
the oblique abdominal muscles on either side of the waist which hold the pelvis in
relation to the ribcage. The pelvis is also held in balance. This is controlled by
the straight abdominal muscle pulling at the front and the seat muscle pulling
down at the back (Royal Academy of Dance, 1997, p8).
When the dancer moves her leg to the front, her hips and shoulders remain (with
rare exceptions, square to the barre. However, when she moves her leg to the
back she opens the working hip but only as much as needed to reveal a fully
turned-out leg. (Schorer, 1999, p51).
The body must be aligned over the supporting leg and hips and shoulders
squared off to each other. Move the leg without a sympathetic motion of the
pelvis. By engaging the spine and maintaining an upright pelvis, we ensure that
the trapezius and the iliopsoas are optimally positioned to do their work.
Maintaining an elongated spine and an aligned pelvis facilitates access to the
iliopsoas and enables us to utilise its function of lifting the leg. Even a slight
tilting of the pelvis can dramatically affect posture, and most important in our
present context, cause the line of gravity to shift back. In an aligned standing
body, the pelvis is held upright. This position requires that the pubic bone is held
uplifted, creating a connection between the pubis, belly button and the sternum
which allows the ASIS to face straight out like a cars headlights. The pelvis
remains placed and level when legs are lifted (with the exception of arabesque
where a shift forward is allowed. Freedom within the hip joint ensures that the
motion of the working leg is isolated from the pelvis and therefore does not result
in a sympathetic rotation of the pelvis) (Paskeva, 2005, p36).
Make sure that the shoulders remain square and that the leg moves directly
behind (Walczak and Kai, 2008, p44).
It is clear that the literature has placed considerable emphasis on the positioning of the
pelvis. The previous quotations discuss correct pelvic placement with respect to the
sagittal, coronal and/or transverse planes. For the purposes of this study, the term
placement will therefore be used to define the second key principle of classical ballet
technique and will be described as minimal displacement of the pelvis from a centered
position.
- 41 -
2.1.3.3 Turnout
The development and codification of the five positions of the feet made a significant
contribution to the advancement of classical ballet technique. This occurred as far back as
the 1600s. A significant characteristic of these five positions is that they each require the
feet to be placed such that the toes are pointed outwards, i.e. turned towards the lateral
sides of the body. These "turned-out" positions of the feet developed as a result of the
need for ballet dancers to travel sideways on the proscenium stage, without turning their
body away from the audience. While much focus was, and still is, placed on the outward
rotation of the feet, the structure of the normal human lower limb dictates that if the feet
are to be placed in a turned out position then the whole leg must be turned out from the
hips (i.e. external hip rotation). Thus, for correct execution of the five positions of the
feet, a dancer is required to have an ability to externally rotate their hip joint beyond that
typically required for normal activities of daily living. Although in inception, turnout was
devised as a means of facilitating sideways travel of the body, in the later years of the
development of the technique extreme external rotation of the hip joint was found of
benefit in allowing greater elevation of the thigh at the hip joint. Rotating the thigh
outwards at the hip joint changes the relative position of the greater trochanter of the
femur and ilium of the pelvis, thereby allowing greater freedom of the femur at the hip
joint. This increased understanding of the structure of the relevant lower limb bones and
joints, together with the growing expectation that dancers display extreme flexibility,
meant that extreme external rotation of the hip joint became a physical necessity rather
than just an aesthetic requirement for classical ballet. High elevation of the legs, with
minimal disturbance of the pelvis and thorax in an effort to retain an elegant appearance,
is not possible without extreme hip turnout, and for this reason turnout remains a crucial
element of classical ballet technique. The following quotations and figures convey the
importance of turnout in the execution classical ballet technique.
The first essential for the legs is to succeed in turning them fully outward. Strive
after suppleness in the hips in order that the thigh movements shall be free and the
knees well turned out. A dancer whose lower limbs are contracted and who is stiff
in the hips gains little esteem, as these defects are always apparent in his
execution (Blasis, 1820, p11).
- 42 -
Figure 2.13 Turnout (from Blasis, 1828, p12).
In the management of your legs, your chief concern must be to acquire a facility
of turning them well outwards. Therefore your hips must be free so that your
thighs move with ease and your knees turn well outwards. By this means the
openings of your legs are rendered easy and graceful. (Beaumont and
Idzikowski, 1932, p23).
The turnout is an anatomical necessity for every theatrical dance. The turnout is
the faculty of turning out the knee to a much greater extent than is made possible
by nature. The foot turns outward together with the knee. The aim of the turnout is
to turn out the upper part of the leg, the hip bone. The result of the turnout is
freedom of movement in the hip joint. The turnout enlarges the field of action of
the leg. This is the importance of training the legs of a dancer in strict en dehors
(turnout). It is not an aesthetic conception but a professional necessity. The dancer
without a turnout is limited in her movements, while a classical dancer possessing
a turnout is in command of all conceivable richness of dance movements of the
legs. (Vaganova, 1953, p24).
- 43 -
Figure 2.14 Five positions of the feet (from Vaganova, 1953, p17)
The turnout is essential. Trying to turnout from the feet is a mistake which
causes rolling over of the big toes. This is turn brings an undue pressure on the
cartilage of the knee. It is primarily the thigh that controls the turnout of the leg
and when the leg is turned out the feet will follow the lead naturally. (Karsavina
1962, p10).
The foot should be placed with a turned-out heel, and with the whole leg turned
out from the hip joint. (Kostrovitskaya and Pisarev, 1978, p54).
The turnout must take place within the hip-joints; the knees and thighs are
rotated outwards as far as possible so that the feet turn out (Lawson, 1979, p14).
Turnout originates in the hip joint and is characterised by the ball and socket
(enarthrosis) type articulation of its components in which the head of the thigh
bone (femur) fits into the concavity of the lower part of the pelvis (ilium).
Anatomically, turnout is defined as the outward rotation of the entire leg,
beginning at the large rounded head of the femur within the acetabulum cavity of
the ilium (Lee, 1983, p210).
- 44 -
Figure 2.15 Turnout Maximum outward rotation of

the legs in the hip sockets (from Lee, 1983, p211)
Classical ballet is based on the turnout; without it, the technique cannot exist. Far
from being primarily an aesthetic concept, the turnout has a profoundly functional
role. The well turned-out leg makes a fundamental contribution to the stability,
range of motion, mobility and strength of the dancer, as well as to the elongated
shape of the muscles. The rotation of the thigh bone in the hip socket is a
movement, it is an action that the dancer takes, an action that must be learned and
practiced. Visibly the results take place in the thigh just below the pelvis, and
continue into the knee, ankle and foot. The whole leg is rotated as one, and the
relationship of each part to the others remains the same as when the leg is in a
neutral position. The turnout allows the dancer to move with equal ease in any
direction (Grieg, 1994, p53-54).
- 45 -
Figure 2.16 Turnout turnout is a movement

Turn out refers to an outward rotation of the entire leg, not just the feet. Turnout
is the result of combining the basic stance, with a level pelvis block, upward
lifting abdominal muscles, firm buttocks, and an outward rotation of the thigh
muscles at the hip joint. While simultaneously straightening knees, ankles, and
feet, this graceful outward rotation of thigh muscles is designed to facilitate the
multitude of ballet poses and positions and also to maintain stability and poise
during sudden changes of direction while in motion. Finally, turnout helps make
the many poses and positions more aesthetically pleasing. Turnout helps stabilise
the torso laterally. It allows for quick, controlled changes of direction. It is one of
the basic ingredients of the basic stance (White, 1996, p85-86).
The use of turnout is fundamental. The external rotation of the hip is controlled
by the gluteus maximus (the main seat muscle), and the small rotators which are
the deepest muscles in the gluteal area, and most importantly the adductors, or the
inner thigh muscles. With this action, the rest of the leg rotates and the foot is held
in turnout that corresponds with that of the hip (Royal Academy of Dance, 1997,
p9).
In all positions, Balanchine wanted to see the turnout developed from the hip
down. The legs are rotated out from the hips, with inside thighs, calves and heels
brought forward. (Schorer, 1999, p40).
- 46 -
Turnout of the legs contributes to the ballet look. Turnout always begins in the
hip joint, with the rest of the leg in alignment. Turnout came into practice when it
became clear that the outward positioning of the feet, accompanied by the femoral
rotation, provide more stability and was also shown to allow more mobility when
the legs were raised. Turnout will gradually increase until, by the fourth year a
full opening of 90 degrees is possible. When the turnout is firmly held in the hip
joint, a further outward movement at the ankle joint can be expected. Mobility in
the ankle allows the foot to be winged. (Paskeva, 2002, p 14).
The knees and thighs are to be completely turned out and facing side. The heel of
the supporting foot must remain very turned out (Walczak and Kai, 2008, p27 &
p109).
The preceding references show that turnout is a key feature of classical ballet technique.
The literature reveals that turnout has both an aesthetic and a functional purpose in the
execution of classical ballet technique. For this study, turnout has been identified as the
third fundamental principle of the technique, and is defined as maximum external
rotation of lower limbs.
2.1.3.4 Extension
The final phase Hammond and Hammond (1979) identified in the development of
classical ballet technique was described as maximum linear extension. This
development saw ballet dancers acquire extreme flexibility and length of line throughout
their entire body. This aesthetic had already been established very early in the
development of the technique (i.e. grand manners). With the introduction of pointe work
during the Romantic period it meant that straight (extended) knee and ankle joints of
the supporting leg had already been identified as essential requirements to achieve
balance en point. However, it was not until the early 20th century that dancers were
expected to raise their working leg to extreme heights, while also maintaining
maximum extension of the knee and ankle joints. At this time the typical physique of
professional female ballet dancers changed to become more lean and lithe, and the
required aesthetic was one of length and purity of line, as opposed to the more soft and
rounded appearance of dancers in the preceding Romantic era. The following quotations
and figures depict the importance of the full extension of the knee and ankle joint in
achieving maximum length and line in classical ballet technique.
- 47 -
The movement of the knees is inseparable from that of the instep, it is

incomplete unless the leg is extended and the toes forced downward. A person
with knees inclined to be bowed should endeavour to overcome his natural
stiffness by constant practice in stretching to render them more supple. He will
never make a really successfully danseur noble (Blasis, 1820, p12 & p15).
Figure 2.17 Leg extension (from Blasis, 1820, p18).
Pay great attention to your insteps. Do not let them relax either in strength or
elasticity. In forcing your insteps, keep the pointe extended. The movement of the
knee is inseparable from that of the instep, it is perfect only when the knee is
extended. The foot and the knee must remain perfectly straight and be extended to
the utmost (Beaumont & Idzikowski, 1932, p23-24).
The legs must be extended energetically. The legs must be forcefully extended in
the knees, arches and toes. Speaking of the execution of all classic pas, it must be
remembered that they are all done with the toes and arches extended. The leg on
which the dancer stands should be fully extended, like a taut string. (Vaganova,
1953, p29).
- 48 -
Figure 2.18 Leg extension (from Vaganova,

1953, p29-30)
The tension of muscles must be complete and must go through the whole of the
leg: instep, calf and thigh. The knee and thigh of the supporting leg must be very
taut, so as not to give away even slightly under the upward thrust of the working
(lifting) leg. The inability to straighten the knee completely, or having what might
be called a knobbly knee, is a problem. The knee should be tensed while lifting
the leg (Karsavina, 1962, p16).
The arch of the foot is strongly developed and strengthened, and the leg acquires
a beautiful and finished line. The supporting leg must be strongly stretched.
(Kostrovitskaya and Pisarev, 1978, p71).
The dancer must know the feeling of a fully stretched leg. No matter what degree
of turnout is attempted it must be ensured that the centre of the knee, ankle and
middle toes are in the same straight line. (Lawson, 1979, p27).
- 49 -
The straight line of the leg from hip to toe results in the functional and aesthetic
elements necessary to classical ballet. In the various forms of battement, the knee,
ankle and metatarsal joints are trained out of their natural inclination to flex or
relax. The knee and ankle joint must learn to become fully extended to form a taut
leg. (Lee, 1983, p230).
If you view the body from the front, you should be able to draw an imaginary
line through the centre of the hip socket and continue down slightly forward of the
centre of the knee joint (because of the dancers slightly forward stance), and on
into the foot, just in front of the ankle. The dancer must stand on straight legs with
well braced thigh muscles and kneecaps that are lifted by the quadriceps. Correct
alignment through the centre of the ankle to the second toe must be maintained at
all times. It is vital that the joints between the phalanges remain extended, there
must be no hint of clawing the toes. (Grieg, 1994, p88-89 & p103).
Figure 2.19 Leg extension (from Grieg, 1994, p88)
- 50 -
Figure 2.20 Leg extension

Although the legs must be held straight, the knee joints must not be allowed to
lock at their rearward extreme. The knee should help form and control a straight
leg line extending from the ankle to the pelvis. (White, 1996, p83).
In every stretched position, there is a straight line running from the centre of the
knee cap, between the ankle bones, and along the foot between the 2nd and 3rd
toes (Royal Academy of Dance, 1997, p28).
Balanchine wanted to see that stretch from the top of the hip to the tips of the
toes. Every part of the stretched, straight leg lifted off the floor contributed to
making the look he wanted to see; stretched knee and a strong and well shaped
pointe. (Schorer, 1999, p44).
Classical line is characterised by its length, purity and simplicity. The feet are
pointed to create the illusion of a longer leg. Anything that detracts from the
purity of the line can be termed unclassical. (Paskeva, 2002, p12).
Dont look like a dwarf. Pull up and straighten knees completely (Walczak and
Kai, 2008, p64). (Walczak and Kai, 2008).
The ability to completely extend the knee and ankle joints to created a straight line
through the lower limb has been highlighted in many ballet texts as a crucial component
of correct classical ballet technique. The extent to which a dancer can extend their knee
and ankle joints is genetically determined by the bony structure of their joints, and as
such this factor has proven to be one of the deciding physical factors in selection for a
- 51 -
professional career as a ballet dancer. For this project, the final fundamental principle of
classical ballet technique has been identified as extension, and is defined as maximum
elongation of the lower limbs.
2.1.3.5 Summary of the Fundamental Principles of Classical Ballet Technique.

After careful review of the history of the development of classical ballet technique, and
consideration of the extensive dance literature sources spanning many centuries, a
succinct list of fundamental principles of classical ballet technique has been developed.
The fundamental principles of classical ballet technique, and their defining characteristics
have thus been identified as a follows:
1) Alignment - maintaining verticality of the torso

2) Placement - minimal displacement of the pelvis from a centred position
3) Turnout - maximum external rotation of lower limbs
4) Extension - maximum elongation of the lower limbs
In Section 2.1.1 of this literature review, classical ballet technique was defined as the
way in which the limbs and body segments are oriented with respect to each other while
performing the movement. The four key principles of classical ballet technique
identified through this literature review relate well to this definition of technique. Each
principle is based on the position of the limbs and body segments, and can therefore be
described as a kinematic characteristic. Kinematic variables of the human body and
human motion can be quantitatively measured using various form of instrumentation, and
can also be qualitatively analysed through visual observation with the human eye. The
next section will review the academic literature to investigate the extent to which the
biomechanical characteristics of various theatrical dance forms, including classical ballet,
have been assessed through quantitative means. Later sections of the literature review
will focus on the accuracy and reliability of qualitative analysis (i.e. visual observation)
in assessment of various forms of human movement.
- 52 -
2.2 Quantitative Biomechanical Analysis in Dance

2.2.1. Introduction
Investigation of the biomechanical properties of human movement involves
quantification and analytical description of body movement patterns, as well as
measurement of the associated forces and muscle activity that contribute to the
movement. The type of data collected and assessed during quantitative biomechanical
analysis generally fall into one of only a few categories: kinematic data; kinetic data; and
electromyographic data. Kinematic data provide information on individual joint or
segment motion, as well as overall whole body movement, while kinetic data describe the
mechanical forces, torques/moments and powers that cause the resultant motion.
Electromyography (EMG) enables measurement of the electrical activity of individual
muscles, and/or muscle groups.
Over the past 40 to 50 years there has been much scientific research conducted on the
biomechanics of human movement in relation to clinical and sporting aspects of human
motion. Due to the development of more efficient and accurate measurement techniques,
instrumented clinical gait analysis is being used increasingly in hospitals throughout the
world to assist with clinical decision making. A great deal of scientific research is
published in this area every year, with the aim of further improving measurement
methods and data reliability and validity. Similarly, there is much interest in quantitative
sports biomechanics. Throughout the past few decades this area of research has been
given the opportunity to grow and develop within the scientific community through the
interest and support offered by various sporting organisations. In comparison to research
conducted in the clinical setting and sporting arena, quantitative instrumented
biomechanical assessment in dance is in its relative infancy as a field of research. A
review of published scientific literature in this field has revealed that although a few
quantitative studies in dance biomechanics were conducted during the 1970s, 1980s, and
1990s, the vast majority of research in this area has occurred from 2000 onwards, with a
considerable increase in publications since 2005.
- 53 -
References were selected for inclusion in this literature review if they consisted of a full
article published in an academic journal, and if the study protocol involved instrumented
quantitative measurement of biomechanical properties of dance-specific or dance-related
movements performed by trained dancers. A total of 75 scientific references were
selected and examined as part of this literature review.
1970s
4%
1980s
5%
2005-2011
40%
1990s
36%
20002004
15%
Figure 2.21 Summary of publication years of quantitative dance biomechanics research
2.2.2. Instrumentation Used in Quantitative Dance Biomechanics Research

Of the 75 articles selected for inclusion in the review, 3 (4%) were published during the
1970s, 4 (5%) during the 1980s, 27 (36%) during the 1990s, and the majority, 41 (55%),
have been published since 2000. The significant increase in publication of dance
biomechanics literature that has occurred over the past decade has been coincident with a
shift in the methodological processes and instrumentation used by researchers. Early
researchers relied on photography (Hinson et al., 1978; Laws, 1979) and motion picture
film cameras (Ryman, 1978) combined with tracing techniques, to collect and analyse
two-dimensional (2D) kinematic data of selected dance movements. The 1980s saw the
introduction of the use of digitised video data to obtain 2D kinematic data (Laws, 1986;
Laws and Lee, 1989). Use of this method of dance analysis continued into the 1990s
(McNitt-Gray et al., 1992; Dozzi and Winter, 1993; Midgett, 1993; Rasmussen and Hay,
1993; Simpson et al., 1996; Krasnow and Chatfield, 1997; Simpson and Kanter, 1997),
however computerised motion analysis, either 2D (Mouchnino et al., 1992; Mouchnino et
al., 1998; Gamboian et al., 1999) or 3D (Chatfield, 1996), also grew in its usage during
- 54 -
this decade. Since 2000, computerised motion analysis has become the most commonly
used method for obtaining kinematic and kinetic data related to analysis of dance
movements, with a clear trend towards increased use of 3D techniques (Wilmerding et
al., 2003; Masso et al., 2004; Thullier and Moufti, 2004; Wilson et al., 2004; Lin et al.,
2005; Shan, 2005; Lepelley et al., 2006; Chatfield et al., 2007; Hagins et al., 2007; Kwon
et al., 2007; Wilson et al., 2007; Pappas and Hagins, 2008; Golomer et al., 2008;
Golomer et al., 2009a; Golomer et al., 2009b; Alderson et al., 2009; Orishimo et al.,
2009; Bertucco and Cesari, 2010; Mayers et al., 2010; Shippen and May, 2010; Bronner
and Ojofeitimi, 2011; Kulig et al., 2011) compared to 2D techniques (Gamboian et al.,
2000; Bronner et al., 2002; Bronner and Ojofeitimi, 2006). 2D (Couillandre et al., 2008)
and 3D (Beijjani et al., 1990; Feipel et al., 2004) electrogoniometers have also been used
as a means to collect kinematic data by a few researchers in dance biomechanics.
Use of instrumented force plates for analysis of dance movements also increased during
the 1990 and 2000s. Some of these studies were interested in examining dancers balance
and stability by obtaining information on movement of the centre of pressure (Perrin et
al., 2002; Simmons, 2005b), while others (Dozzi and Winter, 1993; Simpson et al., 1996;
Simpson and Pettit, 1997; Lin et al., 2005; Kwon et al., 2007; Hagins et al., 2007; Imura
et al., 2008; Alderson et al., 2009; Orishimo et al., 2009; Shippen and May, 2010; Imura
et al., 2010; Kulig et al., 2011) used force plate data, in combination with kinematic data,
to calculate kinetic variables such as joint reaction forces, torques/moments and/or
powers. A few authors (Miller et al., 1990; McNitt-Gray et al., 1992; Laws and Petrie,
1999; Harley et al., 2002; Dworak, 2005; Hagins et al., 2007; Chockley, 2008; Alderson
et al., 2009; Orishimo et al., 2009; Shippen and May, 2010; Bruyneel et al., 2010; Kulig
et al., 2011) have used force plates to present data on the ground reaction forces
associated with various dance movements.
Electromyography (EMG) has also been used for investigation of some of the motor
control and coordination mechanisms typically used in theatrical dance styles. Some
researchers in dance biomechanics were using EMG as early as the 1970s (Ryman and
Ranney, 1979), but it wasnt until the 1990s (Mouchnino et al., 1992; Chatfield, 1993;
Trepman et al., 1994; Chatfield, 1996; Trepman et al., 1998; Ravn et al., 1999) and 2000s
(Wilmerding et al., 2001; Harley et al., 2002; Masso et al., 2004; Simmons, 2005a;
- 55 -
Lepelley et al., 2006; Hagins et al., 2007; Chatfield et al., 2007; Couillandre et al., 2008;
Bertucco and Cesari, 2010), that EMG was used more widely amongst dance science
researchers.
Although not used widely, other instrumented measurement techniques that have been
utilised for dance biomechanics research include pressure plates, either insole
(Fiolkowski and Bauer, 1997) or external (Miller et al., 1990; Albers et al., 1992), and
accelerometers (Voloshin et al., 1989; Golomer et al., 1999; Guillou et al., 2007;
Couillandre et al., 2008).
2.2.3. Purpose and Aims of Quantitative Dance Biomechanics Research

As indicated in the previous sections, the prevalence of dance biomechanics research has
increased considerably since its early days of the 1970s, with as much as 40% of the
published work being conducted since 2005. Various instrumentation techniques have
been used, with the aim of answering a variety of questions related to understanding the
biomechanical mechanisms associated with the execution of theatrical dance steps.
Review of the literature reveals that, in terms of the specific aims and purpose of each
study, there are a few distinct categories that dance biomechanics academic literature can
be assigned to.
Many studies have focused on providing a biomechanical description of dance

movements. In fact, over one third (36%) of the studies included in the literature review
were placed in this category. The authors of these studies have attempted to provide
detailed quantitative descriptions of the ways in which dance movements are executed in
an effort to inform dancers and dance instructors about the most efficient or skilful way to
perform a movement. It can be clearly noticed that the instrumentation used in each study
relates closely to the aim of the study, and as such the majority of studies (74%) that
aimed to provide a biomechanical description have obtained kinematic data consisting
of displacements, velocities and accelerations. The most commonly presented data types
were relative or local 3D joint angles (Beijjani et al., 1990; Masso et al., 2004; Lin et
al., 2005; Wilson et al., 2007; Bronner and Ojofeitimi, 2011), and local 2D joint angles
(Hinson et al., 1978; Ryman and Ranney, 1979; Trepman et al., 1994; Trepman et al.,
1998; Bronner and Ojofeitimi, 2006; Couillandre et al., 2008). Absolute or global
- 56 -
segment kinematics have also been obtained (Ryman and Ranney, 1979; WoodhullMcNeal et al., 1990; Midgett, 1993; Golomer and Fery, 2001; Wilson et al., 2004;
Bronner and Ojofeitimi, 2006; Wilson et al., 2007; Bronner and Ojofeitimi, 2011), as
well as whole body or centre of gravity kinematic data (Laws, 1979; Laws, 1986; Laws
and Lee, 1989; Laws and Fulkerson, 1992; Rasmussen and Hay, 1993; Midgett, 1993). A
few studies in the biomechanical description category reported kinetic data, such as
joint or segment torques/moments (Lin et al., 2005; Kwon et al., 2007; Imura et al.,
2008), while some presented EMG data, either in conjunction with kinematic data
(Ryman and Ranney, 1979; Trepman et al., 1994; Trepman et al., 1998; Masso et al.,
2004; Couillandre et al., 2008), or as the only data type (Wilmerding et al., 2001). Only
two studies (Laws and Petrie, 1999; Chockley, 2008) used ground reaction force data
obtained from a force plate to provide a biomechanical description of dance
movements.
Dance science and medicine researchers have also been interested in investigating any
links or associations of dance movements with injury. Upon review of the studies
interested in injury, it is also clear that the study aims reflect the instrumentation used and
data type presented. Eighteen studies (24%) were identified as having an aim of looking
at injury link, and of these, 11 (61%) used a force plate to obtain data. The most
commonly presented data type calculated from the force plate measurements were joint
reaction forces (Dozzi and Winter, 1993; Simpson and Kanter, 1997; Simpson and Pettit,
1997; Alderson et al., 2009; Mayers et al., 2010; Shippen and May, 2010), joint reaction
torques/moments (Dozzi and Winter, 1993; Simpson and Kanter, 1997; Simpson and
Pettit, 1997; Orishimo et al., 2009; Mayers et al., 2010; Kulig et al., 2011) and ground
reaction forces (McNitt-Gray et al., 1992; Dworak, 2005; Orishimo et al., 2009; Alderson
et al., 2009; Mayers et al., 2010; Shippen and May, 2010; Kulig et al., 2011). Only a few
studies used kinematic data alone, obtained either from computerised motion analysis
(Bronner et al., 2002; Shan, 2005), digitised video data (Barnes et al., 2000),
accelerometry (Voloshin et al., 1989), or electrogoniometry (Feipel et al., 2004), to
investigate possible associations of dance movements with musculoskeletal injury. It
should also be noted that with respect to the studies interested in identifying any links
between dance and musculoskeletal injury, the majority of studies have analysed high
impact movements (i.e. jumps) or high pressure movements (i.e. en pointe). For example,
- 57 -
the grande jet (split leap) has received attention in the literature (Simpson et al., 1996;
Simpson and Kanter, 1997; Simpson and Pettit, 1997; Shan, 2005; Kulig et al., 2011), as
have other dance jumps involving either springing from two feet (McNitt-Gray et al.,
1992) or from one foot (Dworak, 2005; Orishimo et al., 2009). Concerns have been raised
about the level of pressure and force that dancers feet are subjected to while dancing on
their toes in pointe shoes (en pointe). Teitz et al. (1985) measured the peak pressure on
the first and second toes when en pointe in first position, and Albers (1992) measured the
peak plantar pressure when dancing on a flat foot and when dancing en pointe with the
body weight supported by a single leg or double leg. Dozzi and Winter (1993) measured
the joint reaction forces and torques/moments at the ankle joint when en pointe in second
position. The biomechanical properties of high impact styles of dance such of flamenco
(Voloshin et al., 1989), tap (Mayers et al., 2010) and Irish dance (Shippen and May,
2010) have also been examined with respect to their possible association with
musculoskeletal injury.
A third purpose or aim of the research studies included in this literature review was
identified as that of investigating aspects of motor control. Electromyographic (EMG)
data in combination with kinematic data, were most commonly used for the 19 studies
placed in this category. These studies looked at the muscle activation patterns of either
multiple (Mouchnino et al., 1992; Ravn et al., 1999; Lepelley et al., 2006; Chatfield et
al., 2007) or single (Chatfield, 1996; Harley et al., 2002; Bertucco and Cesari, 2010)
muscles. Some studies used only kinematic data in their investigations of motor control
(Mouchnino et al., 1998; Golomer et al., 1999; Thullier and Moufti, 2004; Guillou et al.,
2007; Golomer et al., 2008; Golomer et al., 2009a; Golomer et al., 2009b), and only two
studies (Chatfield, 1993; Simmons, 2005a) reported EMG data alone. Centre of pressure
(Perrin et al., 2002; Mouchnino et al., 1992; Simmons, 2005b) and ground reaction force
(Harley et al., 2002; Bruyneel et al., 2010) data were also used to study motor control.
Interestingly, most of the studies that looked at motor control analysed movements that
were not actually dance movements (Perrin et al., 2002; Mouchnino et al., 1992;
Mouchnino et al., 1998; Golomer et al., 1999; Ravn et al., 1999; Harley et al., 2002;
Thullier and Moufti, 2004; Simmons, 2005b; Simmons, 2005a; Lepelley et al., 2006;
Chatfield et al., 2007; Guillou et al., 2007; Bruyneel et al., 2010; Bertucco and Cesari,
2010), however these studies have been included in the review because each study
- 58 -
contains a group of trained dancers as subjects. These studies investigated motor control
by comparing the movement and coordination strategies of highly skilled movement
practitioners, such as dancers, and in some cases gymnasts (Thullier and Moufti, 2004;
Guillou et al., 2007) and martial artists (Perrin et al., 2002), with those of untrained
controls. The movements investigated included quiet standing (Perrin et al., 2002;
Simmons, 2005b; Simmons, 2005a), leg raises (Mouchnino et al., 1992; Mouchnino et
al., 1998; Thullier and Moufti, 2004; Lepelley et al., 2006; Bertucco and Cesari, 2010),
jumps (Ravn et al., 1999; Harley et al., 2002) and balancing on a see-saw platform
(Golomer et al., 1999; Guillou et al., 2007). Only a few studies in the motor control
category assessed dance movements. A series of studies by Golomer et al. (2007; 2008;
2009) used 3D kinematic data to compare the motor control abilities of trained dancers
and untrained controls in performing a pirouette (a whole body turn on one leg). The
dancers were instructed to perform a pirouette and the controls a turn on one leg as
they naturally would. Only a few studies looked at the motor control abilities of
dancers alone, and in most of these studies motor control of different groups of dancers
was analysed and compared with respect to the dancers age (Bruyneel et al., 2010), or
skill level (Lepelley et al., 2006; Chatfield et al., 2007). Chatfield (1993) didnt compare
different groups of dancers, but simply presented EMG data for seven advanced college
dancers during their execution of several ballet movements (plis, relevs, developps,
grand battement, sauts).
A few dance biomechanics research studies have attempted to investigate the efficacy or
usefulness of specific dance training programs or regimes. To do this, these studies have
collected quantitative data before and after implementation of a training program.
Kinematic variables were used as the outcome measure for all seven studies in this
category. Specifically, authors obtained before and after data using 2D (Gamboian et
al., 1999; Gamboian et al., 2000) or 3D (Wilmerding et al., 2003) motion analysis data,
2D digitised video data (Krasnow and Chatfield, 1997; Deckert et al., 2007; Holt et al.,
2011) or centre of gravity kinematic data calculated from ground reaction force data
(Poggini et al., 1997). Interestingly, six of the seven articles that aimed to measure the
effects of training programs, analysed ballet movements, and five of the studies were
specifically interested in assessing pelvic anterior/posterior tilt (Gamboian et al., 1999;
Gamboian et al., 2000; Wilmerding et al., 2003; Deckert et al., 2007; Holt et al., 2011).
- 59 -
There was also similarity between the studies in this category with respect to the type of
movements analysed, with five studies (Krasnow and Chatfield, 1997; Gamboian et al.,
1999; Gamboian et al., 2000; Deckert et al., 2007; Holt et al., 2011) measuring the
biomechanical characteristics of the pli (i.e. knee bend).
A less common, yet also very important, study purpose that was identified was that of
effect of dance shoe or surface. Only a few studies (Miller et al., 1990; Fiolkowski and
Bauer, 1997; Hagins et al., 2007; Pappas and Hagins, 2008) investigating this area have
been published. Two of these (Miller et al., 1990; Fiolkowski and Bauer, 1997) measured
peak plantar pressure. Miller (1990) looked at the role that modification of dance shoes
has on peak plantar pressure in professional ballet dancers, and Fiolkowski and Bauer
(1997) studied the effects of different floor surfaces on plantar pressure in student
dancers. Hagins et al. (2007) and Pappas and Hagins (2008) both investigated the effects
of a raked (i.e. inclined) stage on the movement patterns of professional musical theatre
dancers while performing a bilateral drop jump (Hagins et al., 2007), or while standing
still (Pappas and Hagins, 2008). Both these studies used 3D motion analysis to obtain
lower limb joint angle data, and Hagins et al. (2007) also used a force plate to measure
the direction and magnitude of the ground reaction forces associated with landing a jump
onto an inclined surface.
Training
Programs
9%
Injury Link
24%
Dance
Shoes/Surfaces
5%
Movement
Description
36%
Motor Control
26%
Figure 2.22 Summary of study purposes of quantitative dance biomechanics research
- 60 -
2.2.4. Dance Genres and Quantitative Dance Biomechanics Research

There are many different genres or styles of dance, and it is interesting to examine which
of these styles have received the most attention in terms of quantitative dance
biomechanics research.
The academic literature indicates that the biomechanical
characteristics of classical ballet have been analysed most frequently. Of the 75 articles
included in this literature review, 46 (62%) have analysed classical ballet movements.
Perhaps it is the consistency, universality and strong tradition of the technique of classical
ballet lending itself well to systematic research that has brought about this large
proportion of ballet-related biomechanics research. Or indeed it could be that the extreme
and somewhat unnatural physical requirements of the art form create a stronger interest in
the biomechanical properties of classical ballet technique.
Other dance genres that have been researched with respect to quantitative biomechanics
are modern dance (McNitt-Gray et al., 1992; Chatfield, 1996; Simpson et al., 1996;
Simpson and Kanter, 1997; Simpson and Pettit, 1997), flamenco (Voloshin et al., 1989;
Beijjani et al., 1990), tap dance (Mayers et al., 2010) and Irish dance (Shippen and May,
2010). With the exception of Chatfield (1996), all of the studies on modern dance
attempted to address how the particular movements may be associated with potential
injury. 2D digitised video data and force plate data were used to assess this, with authors
reporting joint reaction forces (Simpson et al., 1996; Simpson and Kanter, 1997; Simpson
and Pettit, 1997) or ground reaction forces (McNitt-Gray et al., 1992). Chatfield (1996)
presented kinematic data obtained from 3D motion analysis, together with EMG data, to
examine the potential injury risks associated with advanced college dancers performing a
modern dance sequence. Specifically, their onset of abdominal muscle activity was
measured in relation to their ankle and wrist linear displacements.
The studies that analysed flamenco, tap, and Irish dance movements were all interested in
identifying and measuring biomechanical parameters that may be associated with injury
risk. In relation to flamenco dance, Voloshin et al. (1989) used accelerometers to measure
the accelerations of the anterior superior iliac spines (ASIS) and tibial tuberosity with and
without insoles in the shoes, and Beijjani et al. (1990) used 3D electromagnetic
goniometers to measure the flexion/extension and lateral bending angles of the thoracic
and lumbar spine. Very recently, Mayers et al. (2010) used 3D motion analysis and force
- 61 -
plate data to calculate the hip, knee, and ankle joint reaction forces and torques/moments,
and vertical ground reaction forces associated with professional level tap dancing. Using
similar instrumentation, Shippen and May (2010) reported joint contact forces, muscle
loading estimates (gastrocnemius & soleus), and ground reaction force data, for
professional female Irish dancers performing the Rock step.
Non-specified
dance
1%
Non dance
25%
Tap
1%
Irish
1%
Flamenco
3%
Modern
Ballet
62%
7%
Figure 2.23 Summary of dance genres analysed in quantitative dance biomechanics research
Sixty-two percent (62%) of the studies included in this literature review have analysed
the biomechanical properties or characteristics of classical ballet technique. Since this
research thesis is primarily concerned with analysis of classical ballet technique, the
following subsection will focus in more detail on the academic literature that relates
specifically to quantitative measurement of classical ballet technique.
2.2.5. Quantitative Biomechanics Research in Classical Ballet Technique

As detailed in Section 2.2.4, a large proportion, 46 (62%), of the dance biomechanics
articles included in this literature review have been concerned with quantitative analysis
of classical ballet movements. The year of publication of these studies spans from 1978
through to 2011, with 54% of the research conducted since 2000. In terms of the purpose
of these studies, the majority (54%) of studies aimed to provide a quantitative movement
description, 22% focused on examining possible links with injury, 13% investigated
the effect of training programs, 9% analysed aspects of motor control, and only one
study (2%) was concerned with the biomechanical effects of dance shoes and/or
surfaces.
- 62 -
2.2.5.1 Instrumentation Used in Quantitative Biomechanical Analysis of Ballet Technique

With respect to the instrumentation used, as was the case when reviewing all dance
biomechanics studies, kinematic analysis was also found to be the most commonly used
method for biomechanical assessment when examining classical ballet technique. Seventy
four percent (74%) of studies presented 2D or 3D kinematic data, and of these, 60% used
2D techniques and 40% used 3D techniques. There was more variability in the types of
techniques used for 2D analysis than for 3D analysis. As indicated in Section 2.2.2, the
year of publication has influenced the instrumentation used. Most of the earlier studies
used simplistic, yet time consuming, 2D techniques such as photography and tracing
(Hinson et al., 1978; Laws, 1979), and video and motion photography (Laws, 1986; Laws
and Lee, 1989). Use of video and digitisation of ballet movements increased in the 1990s,
(Woodhull-McNeal et al., 1990; Midgett, 1993; Rasmussen and Hay, 1993; Dozzi and
Winter, 1993; Krasnow and Chatfield, 1997; Barnes et al., 2000), however this 2D
technique has also been used in more recent studies (Deckert et al., 2007; Holt et al.,
2011). Other 2D kinematics techniques that have been used to assess the biomechanics of
classical ballet include goniometers (Trepman et al., 1994; Trepman et al., 1998; Feipel
et al., 2004; Couillandre et al., 2008), and 2D computerised motion analysis (Gamboian
et al., 1999; Gamboian et al., 2000; Bronner et al., 2002; Bronner and Ojofeitimi, 2006).
Since 2004, the majority of the quantitative studies on the kinematics of classical ballet
have used 3D motion analysis (Wilson et al., 2004; Masso et al., 2004; Lin et al., 2005;
Shan, 2005; Wilson et al., 2007; Golomer et al., 2008; Golomer et al., 2009a; Golomer et
al., 2009b; Alderson et al., 2009; Kulig et al., 2011; Bronner and Ojofeitimi, 2011). Prior
to 2004, there were no published articles that used 3D motion analysis to assess classical
ballet movements.
A few studies (Dozzi and Winter, 1993; Lin et al., 2005; Kwon et al., 2007; Imura et al.,
2008; Imura et al., 2010; Kulig et al., 2011) have presented kinetic data, such as joint
reaction forces and/or torques/moments, for the lower limb in classical ballet, however
there is no published data on joint reaction powers. Some studies have reported EMG
data, either as the only data type (Chatfield, 1993; Wilmerding et al., 2001), or in
conjunction with kinematic data (Ryman and Ranney, 1979; Trepman et al., 1994;
Trepman et al., 1998; Masso et al., 2004; Couillandre et al., 2008), and only two studies
- 63 -
(Dworak, 2005; Chockley, 2008) have looked specifically at the ground reaction forces
associated with classical ballet.
2.2.5.2 Quantitative Biomechanical Analysis of the Seven Movements of Classical Ballet

It is also interesting to consider the types of ballet movements or steps that have been
studied in quantitative biomechanics research. The steps that have been analysed will be
considered in terms of the Seven Movements of Dance, as defined by Noverre (1760).
The most widely researched ballet movement to date has been the bend (pli). One of
the most fundamental movements in ballet, the pli is a movement in which the upright
torso, spine, and pelvis are stabilised as they are lowered with coordinated hip and knee
flexion, and then raised back to the starting position with hip and knee extension (Gantz,
1990). Repeated execution of the pli in training class is used to improve strength, timing,
alignment, trunk stability, and coordination of joint movement (Gantz, 1990). The pli is
also often the first component of other movements such as the relev (i.e. rising on the
toes), pirouettes (i.e. turns of the body), and jumps. In total, ten articles have published
quantitative data on the biomechanics of the demi pli (Chatfield, 1993; Trepman et al.,
1994; Gamboian et al., 1999; Gamboian et al., 2000; Deckert et al., 2007; Couillandre et
al., 2008; Holt et al., 2011) and/or grand pli (Krasnow and Chatfield, 1997; Trepman et
al., 1998; Barnes et al., 2000; Deckert et al., 2007). The grand pli, a deep knee bend, has
been identified as a controversial movement (Berardi, 2005) as it has been suggested that
it may be hazardous due to its extreme degree of knee flexion, large patellofemoral forces
and excessive knee joint longitudinal rotation (Clippinger-Robertson et al., 1985). In their
investigation of the lower limb muscle activity associated with the grand pli, Trepman et
al. (1998) reported that EMG activity of the vastus lateralis and vastus medialis was
significantly less in ballet dancers than in modern dancers, despite similar degrees of knee
flexion. Trepman et al. (1998) therefore reported that due to their constant repetition of
the grande pli, ballet dancers may have lower patellofemoral joint reaction force than
modern dancers, and the movement therefore may not necessarily be as potentially
injurious as suggested.
Barnes et al. (2000) used 2D digitised video data to calculate external longitudinal
rotation (ELR) of the knee joint during the grand pli. Their results indicated that ELR
values were highest at the deepest part of the pli, the point of greatest knee flexion, in all
- 64 -
of the five positions of the feet, with the ELR being greater in the 3rd and 4th positions,
compared to 2nd position. On the basis of this research, Barnes et al. (2000) suggested that
dance educators should exercise caution in including the grand pli in the training
process. Krasnow and Chatfield (1997) investigated the alignment of the torso during
the grande pli and Deckert et al. (2007) looked at pelvic anterior/posterior placement
during the grande pli and demi pli. Gamboian et al. (1999; 2000), Couillandre et al.
(2008) and Holt et al. (2011) were also interested in the anterior/posterior placement of
the pelvis during the demi pli. These studies reported that specific training programs can
significantly affect torso alignment (Krasnow and Chatfield, 1997) and pelvic
placement (Gamboian et al., 1999; Gamboian et al., 2000; Deckert et al., 2007;
Couillandre et al., 2008; Holt et al., 2011).
Biomechanical analysis of the rise (relev) in classical ballet has also received some
attention in the academic literature. The relev in classical ballet is a movement in which
the dancer raises themselves up to stand on the balls of the feet (demi-pointe) or all the
way up onto the tips of the toes through use of specially constructed pointe shoes (full
point or sur les pointe) (Grant, 2008). Nine studies on the relev in classical ballet
were identified. Five of these studied the biomechanics of the relev to demi pointe
(Chatfield, 1993; Gamboian et al., 1999; Gamboian et al., 2000; Masso et al., 2004; Holt
et al., 2011) and four investigated the relev to full pointe (Teitz et al., 1985; Albers et
al., 1992; Dozzi and Winter, 1993; Lin et al., 2005). Chatfield (1993) and Masso et al.
(2004) both collected EMG data of dancers performing relevs in an effort to assess their
motor control strategies. Masso et al. (2004) examined the activity of the peroneus
longus, soleus, gastrocnemius, and abductor hallucis muscles in a balletic rise (relev to
demi pointe in first position), compared with a rise in a non-balletic (parallel) position of
the feet. Results indicated decreased abductor hallucis muscle activity in the balletic
position, which the authors suggested may be reflective of deficient stability of the
internal arch and first ray, thereby leading to increased risk of injury. Gamboian et al.
(1999; 2000) and Holt et al. (2011) studied the effects of particular training programs
(somatic training; remote cueing) on pelvic anterior/posterior placement during relev
to demi pointe. Both studies reported improved pelvic placement as a result of the
training programs. In relation to the studies on relevs en pointe, as mentioned in Section
2.2.3 the levels of pressure and force that dancers feet are subjected to when dancing
- 65 -
pointe has been an area of interest for a few researchers (Teitz et al., 1985; Albers et al.,
1992; Dozzi and Winter, 1993). Teitz et al. (1985) measured the peak pressure on the first
and second toes during relev en pointe in first position, and Albers et al. (1992)
measured the peak plantar pressure when performing a single leg (relev retir) or double
leg (relev en first position) relev en pointe. Dozzi and Winter (1993) and Lin et al.
(2005) calculated the joint reaction forces and torques/moments at the ankle joint for the
relev en pointe. Interestingly, Lin et al. (2005) reported that despite high correlations in
bilateral comparisons of ankle joint plantar/dorsiflexion angles, only moderate
correlations were found between the left and right sides in joint reaction
torques/moments. The authors suggested that the two ankle joints may therefore play
different roles in controlling balance and movement.
Classical ballet turns (pirouettes) have also been studied using quantitative biomechanical
techniques. A pirouette is a spinning action of the entire body during which the dancer
rotates with their body weight supported on a single leg. Nine published journal articles
on the biomechanics of the classical ballet pirouette have been identified. The earliest of
these studies were conducted by the research group of Laws and colleagues. These
authors described the whole body angular momentum (Laws, 1979; Laws, 1998), and
whole body angular displacement and velocity (Laws, 1986; Laws and Fulkerson, 1992)
of classical ballet pirouettes. Golomer and colleagues (2008; 2009a; 2009b) presented
data on the relative movements of the shoulder and hips and supporting leg displacements
during whole body rotations (Golomer et al., 2009b), and then went on to examine how
whole body rotations can be affected by mental imagery styles (Golomer et al., 2008) and
vision and motor imagery (Golomer et al., 2009a). Also very recently, Imura and
colleagues described the angular momentum of the shoulder, arm, hip, and leg (Imura et
al., 2008), and the moment/torque of the supporting leg with respect to the friction
coefficient of the floor (Imura et al., 2010), during the fouette turn.
The biomechanical characteristics of jumping (saut) and darting (elanc) movements in

classical ballet have also been investigated. Four of the six studies that analysed ballet
jumps used a force plate to obtain quantitative data (Poggini et al., 1997; Laws and Petrie,
1999; Dworak, 2005; Chockley, 2008). Poggini (1997) compared the centre of gravity
height for sauts in first position before and after implementation of a physio-technique
- 66 -
training program. This author also compared jump elevation when jumping with turned
out legs and with legs in a natural position. Most of the participants achieved a higher
jump using a natural rather than a turned out position of the legs. Laws and Petrie
(1999), and Chockley (2008) measured the ground reaction forces (GRF) associated with
the sauts in first position. Laws and Petrie (1999) assessed the influence of arm position
on GRF during take-off and on jump height. GRF was higher and jump elevation
increased when subjects raised their arms over their head during take-off, compared to
using a static arm position. Chockley (2008) examined the vertical jumps of ballet
dancers landing en pointe and landing on the full foot, landing en pointe only required
72% of the GRF that landing on the full foot required. Dworak (2005) measured and
compared the GRF during landing from a selection of standard ballet jumps (grand pas
de chat, grand pas assembl, entrelac, saut basque, double tour), GRF was greatest
when landing from the grand pas de chat.
With respect to quantitative analysis of darting (elanc) movements in ballet, seven

articles were located, each of these investigated the grand jet (split leap). Hinson et al.
(1978) described the segment and joint angles of the torso and lower limb during the
grand jet en tournant entrelac (turning split leap), and the other six authors (Laws and
Lee, 1989; Miller et al., 1990; Midgett, 1993; Rasmussen and Hay, 1993; Shan, 2005;
Kulig et al., 2011) investigated the grand jet en avant (forward split leap). Most of these
authors (Laws and Lee, 1989; Midgett, 1993; Rasmussen and Hay, 1993) reported on the
trajectory of the whole body centre of gravity during the airborne phase of the movement,
and Miller et al. (1990) and Kulig et al. (2011) measured the GRF associated with
landing from the grand jet. In addition to GRF, Miller et al. (1990) also measured
plantar pressure at the first and second toes, the first and second metatarsal heads, the
medial and lateral arches, and heel while wearing different types of dance shoes. Results
indicated that the GRF was not affected by shoe type, but the modifications in shoe
design resulted in a more even distribution of pressure over the medial arch, away from
the toes, metatarsal heads and heel. Kulig et al. (2011) analysed knee joint mechanics as
well as GRF for the landing and take-off of the grand jet en avant. Knee joint angles and
torques/moments were calculated, and it was reported that landing was accomplished
with less knee angular stiffness than take-off, despite greater peak GRF, and therefore
landing from the jet was considered to be more injurious to the knee soft tissue than
- 67 -
take-off. Shan (2005) measured the hip, knee and ankle joint angles and estimated muscle
lengths (biceps femoris, adductor magnus, gracillis, semimembranosus) for ballet dancers
and tae kwon do martial artists performing the grand jet en avant. Results showed that
the most over lengthened muscle was the small adductor magnus muscle, and it was
therefore suggested that small muscles are more likely to be injured in repetitive motion.
The stretch (battement) movement in classical ballet has also been analysed through
quantitative means. Five journal articles on the battement were identified in the published
literature. This list included the earliest identified quantitative study on the biomechanics
of classical ballet technique (Ryman and Ranney, 1979) and two of the most recent
studies (Bronner and Ojofeitimi, 2011; Holt et al., 2011). Most of these studies (Ryman
and Ranney, 1979; Deckert et al., 2007; Bronner and Ojofeitimi, 2011; Holt et al., 2011)
were interested in analysing pelvic movement during the battement movement. As
mentioned in Section 2.1.3.2, pelvic placement was identified as one of the fundamental
principles of classical ballet technique. Specifically, correct pelvic placement involves
minimal movement or disturbance of the pelvis from its centred or neutral position when
raising the leg to the front, side, or back of the body. Deckert et al. (2007) and Holt et al.
(2011) reported on the effects of individualised training approaches on sagittal plane
pelvic placement for the battement. Both these studies used 2D video and digitisation
techniques, and both reported that anterior pelvic tilt decreased after individual tutoring
for durations of 3 weeks and 8 weeks, respectively. Ryman and Ranney (1979) and
Bronner and Ojofeitimi (2011) compared the relative positions of the pelvis and thigh
segments during execution of the grand battement. Ryman and Ranney (1979) used 2D
techniques (cinecamera and tracings) and only examined the grand battement devant
(front kick), while Bronner and Ojofeitimi (2011) used 3D computerised motion analysis
and examined the grand battement to the front, side and back of the body. Despite the
traditional manuals stressing that the working leg alone accomplishes the kick, and pelvic
and spinal movements must be kept to a minimum for aesthetic reason (Ryman and
Ranney, 1979), both studies reported that there is actually significant involvement of the
pelvis during the grand battement devant. Ryman and Ranney (1979) concluded that as
the leg is kicked to the front, the pelvis goes into increased posterior tilt, causing the
lumbar curve to flatten. Bronner and Ojofeitimi (2011) also reported that in the battement
devant, posterior pelvic tilt occurs, and that this exceeded the degree of posterior pelvic
- 68 -
tilt during grand battement to the side. Pelvic obliquity toward the stance limb was also
demonstrated during the battement in each direction, as was transverse plane pelvic
rotation (Bronner and Ojofeitimi, 2011). The grand battement has also been examined
though EMG analysis (Ryman and Ranney, 1979; Chatfield, 1993). Such EMG data has
illustrated considerable variations in patterns of muscle activity, despite similarities in
dancer training and kinematic execution of the movement (Ryman and Ranney, 1979).
The last of Noverres Seven Movements of Dance (1760) that has not yet been
discussed is the glide (gliss) movement category. It was not possible to identify any
published journal articles that used quantitative biomechanics technique to analyse the
gliss, this movement category will therefore not be discussed any further in this
literature review.
Although not included in Noverres Seven Movements of Dance, another movement

type that has received some attention in the academic literature, and that is often
performed in classical ballet repetoire, is the lift. Lifts in classical ballet typical involve
a male dancer lifting a female dancer off the floor and supporting her entire body weight
in various positions against his body or above his head. Alderson et al. (2009) used 3D
motion analysis to measure lumbar extension velocity, horizontal hand to feet distance,
and to estimate the forces generated at the L5/S1 joint in professional and preprofessional male dancers performing two different lifts (full pressage lift; arabesque
lift). Ground reaction force data were also presented.
Interestingly, the estimated peak
lumbar anterior shear force was found to occur at the beginning of each lift,
approximately 0.01 seconds prior to the ballerina moving vertically into the lift. The
compression forces were found to be greater than the National Institute of Occupational
Safety and Health, Back Compression Design Limit (3400N), suggesting a requirement
for administrative controls to be placed on the number of lifts performed by male dancers
daily, in order to reduce the risk of lower back injury (Alderson et al., 2009).
In consideration of Noverres Seven Movements of Dance (1760), review of the

published literature reveals that the majority (87%) of studies that have conducted a
quantitative analysis of classical ballet movement have analysed the biomechanical
characteristics of only a single step or single movement category. Only a few studies have
- 69 -
attempted to examine more than one movement category. For example, some researchers
have analysed two of the seven movements, such as the bend (pli) and rise (relev)
(Gamboian et al., 1999; Gamboian et al., 2000; Holt et al., 2011), bend (pli) and stretch
(battement) (Deckert et al., 2007), or bend (pli) and jump (saut) (Couillandre et al.,
2008). Only two studies (Chatfield, 1993; Holt et al., 2011) attempted to assess steps
from more than two movement categories. Holt et al. (2011) measured pelvic tilt for the
bend (pli), rise (relev), and stretch (battement); and Chatfield (1993) presented EMG
data for steps from the bend (pli), rise (relev), stretch (battement), and jump (saut)
movement categories. No published studies were identified that analysed the
biomechanical characteristics of all seven movement categories.
2 Categories
9%
3 Categories
2%
Other
17%
4 Categories Stretch
2%
(Battement)
4%
Bend
(Plie)
9%
Rise
(Releve)
11%
Jump (Saute)
9%
Lift
2%
Turn (Tour)
20%
Dart (Elence)
15%
Figure 2.24 Summary of dance movement categories analysed in quantitative

ballet biomechanics research
2.2.5.3 Quantitative Biomechanical Analysis of the Fundamental Principles of

Classical Ballet
In Section 2.1.3, four fundamental principles of classical ballet technique were identified
and described. These were defined as alignment, placement, turnout, and
extension. The biomechanical analyses that will be conducted as part of the current
research project will involve assessment of classical ballet movements in relation to these
four principles. It is interesting therefore to consider how, or indeed if, the existing
research on the biomechanics of classical ballet technique relates to these principles.
- 70 -
The review of the literature indicates that while it is possible to relate some of the existing
research to the four fundamental principles, the majority (52%) of the quantitative
biomechanics studies on classical ballet technique cannot easily be associated with one of
these fundamental principles. This is largely due to the fact that most of these studies
collected and presented data types other than kinematic data. For example, joint reaction
torques/moments (Kwon et al., 2007; Imura et al., 2008; Imura et al., 2010; Kulig et al.,
2011), ground reaction forces (Miller et al., 1990; Laws and Petrie, 1999; Dworak, 2005;
Chockley, 2008; Kulig et al., 2011), EMG data (Chatfield, 1993; Wilmerding et al., 2001;
Couillandre et al., 2008) and pressure data (Teitz et al., 1985; Miller et al., 1990; Albers
et al., 1992) have provided interesting and useful information to assist in understanding
some of the biomechanical aspects of ballet steps, however these data types do not reveal
any detail about the resulting movement pattern displayed by the dancer, i.e. the visually
perceivable elements of the movement. Other studies that did report kinematic data, but
that still could not be easily related to the four fundamental principles of ballet technique,
were mostly concerned with whole body angular momentum (Laws, 1979), angular
displacement (Laws, 1986; Laws and Fulkerson, 1992), angular velocity (Laws, 1986),
and linear centre of mass displacement (Laws and Lee, 1989; Midgett, 1993; Rasmussen
and Hay, 1993; Poggini et al., 1997), rather than inter-segmental interactions.
Amongst those studies that could be linked with one or more of the four fundamental
principles of classical ballet technique, kinematic characteristics related to alignment
and placement have received the most attention. Nine studies were identified that
reported kinematic variables relating to the principle of alignment. In one of the earlier
studies, Woodhull-McNeal (1990) assessed
the linearity of dancers posture by
measuring the horizontal distances of bony land marks (ear, shoulder, hip, knee, ankle)
from the line of the centre of the gravity. In a similar study, Krasnow and Chatfield
(1997) calculated the horizontal anterior-posterior displacement of ear, shoulder and
greater trochanter markers from the foot marker to obtain a measure of vertical central
alignment. Trepman and colleagues (1994; 1998) measured anterior-posterior body
sway; and lumbar flexion/extension was analysed by Gamboians research group (1999;
2000), Feipel et al. (2004), Wilson et al. (2004) and Alderson et al. (2009).
- 71 -
With respect to the principle of placement, orientation of the pelvic segment is the
kinematic variable of relevance. Of the 10 studies whose results can be related to
placement, those that used 2D kinematic techniques presented only sagittal plane pelvic
tilt data (Ryman and Ranney, 1979; Gamboian et al., 1999; Gamboian et al., 2000;
Deckert et al., 2007; Holt et al., 2011), while those that used 3D techniques were also
able to present coronal plane pelvic data (Wilson et al., 2004; Wilson et al., 2007;
Bronner and Ojofeitimi, 2011), and/or transverse plane pelvic data (Golomer et al., 2008;
Golomer et al., 2009b; Bronner and Ojofeitimi, 2011). Some of these studies investigated
the placement of the pelvis in relation to thigh segment movement (Ryman and Ranney,
1979; Wilson et al., 2004; Wilson et al., 2007; Bronner and Ojofeitimi, 2011), or thoracic
segment movement (Golomer et al., 2008; Golomer et al., 2009b), while others looked at
pelvic placement in isolation from the movements of other segments (Gamboian et al.,
1999; Gamboian et al., 2000; Deckert et al., 2007; Holt et al., 2011).
The results of quantitative research on classical ballet can also be related to the principle
of extension. This principle is concerned with the elongation of the lower limb (i.e.
maximum knee extension and maximum ankle plantarflexion). Hinson et al. (1978)
reported that the knees and ankles were well extended during the airborne phase of a
tour jet (turning leap), and Shan (2005) measured knee and ankle angles during a grand
jet en avant (forward leap). Lin et al. (2005) and Dozzi and Winter (1993) calculated
ankle plantarflexion angles during the relev en pointe in first and second positions
respectively, and Masso et al. (2004) reported ankle plantarflexion angles for the relev
to demi-pointe in first position.
Classical ballet requires maximum external rotation, turnout, of the lower limb joints.
Interestingly, despite the fact that turnout is such a crucial element of classical ballet
technique, there is relatively little published quantitative research, involving instrumented
biomechanical analysis, that can be related to this principle. Only a few studies were
identified that had some relevance to turnout. Barnes et al. (2000) measured
longitudinal rotation (ELR) of the knee joint during the grand pli (deep knee bend). The
relatively high ELR measured at the base of the grand pli was suggested to be indicative
of an increased injury risk of repetitive performance of the grand pli (Barnes et al.,
2000). In relation to hip rotation, Shan (2005), Wilson et al. (2007) and Bronner and
- 72 -
Ojofeitimi (2011) all used 3D motion analysis to measure hip rotation during the grand
jet en avants, the grand rond de jambe en lair en dehors, and the grand battement,
respectively.
2.2.5.4 Subjects Included in Quantitative Biomechanical Analysis of Ballet

The numbers and types of subjects recruited for biomechanical assessment of ballet
technique is another interesting point to consider. Review of the literature indicates that
the majority (80%) of studies have recruited only one group of subjects. The number of
subjects recruited into a single group has ranged from 1 participant (i.e. a case study) up
to 61 participants. The mean number of participants across all 37 studies that analysed
only one group of subjects was 10, while the most frequently occurring number of
participants recruited to a single group was 1. Figure 2.25 depicts the subject group sizes
for single group quantitative ballet studies. As displayed in the graph, the majority of
single group studies had less than 16 subjects recruited to the group.
Number of Studies
7
6
5
4
3
2
1
0
1
11
16
21
26
31
36
41
46
51
56
61
Number of Subjects per Group

Figure 2.25 Sample size for single group studies involving quantitative instrumented
biomechanical analysis of classical ballet technique
With respect to the selection criteria used for recruitment of dancers to the single group
studies, most studies (40%) have been interested in analysing the biomechanical
characteristics of professional ballet dancers, while some (24%) recruited elite preprofessional ballet dancers. Eight studies (23%) described their subjects as dance
students, categorised under various headings, e.g. advanced (Teitz et al., 1985;
- 73 -
Rasmussen and Hay, 1993) and highly trained (Shan, 2005), or according to a specific
number of years of training (Woodhull-McNeal et al., 1990; Albers et al., 1992; Poggini
et al., 1997; Lin et al., 2005; Imura et al., 2008; Imura et al., 2010). Seven studies (19%)
recruited university or college dancers with varying levels of training (Chatfield, 1993;
Midgett, 1993; Krasnow and Chatfield, 1997; Gamboian et al., 1999; Gamboian et al.,
2000; Chockley, 2008; Bronner and Ojofeitimi, 2011), and the remaining studies (Laws
and Lee, 1989; Laws and Petrie, 1999) did not provide any indication of the level of
training of the recruited dancers.
Some research studies have compared biomechanical parameters between different
groups of dancers performing classical ballet steps. Comparisons have been made
between professional dancers of different genres, e.g. professional ballet dancers versus
professional modern dancers (Trepman et al., 1994; Trepman et al., 1998); or between
professional ballet dancers and practitioners of other movement disciplines, such as tai
kwon do (Shan, 2005). One study was interested in the differences in lower limb
kinematic patterns between males and females in their execution of classical ballet
movements (Bronner
and Ojofeitimi,
2006),
while
others
have
investigated
biomechanical differences between groups in relation to their level of dance training

(Wilson et al., 2004; Kwon et al., 2007; Golomer et al., 2008; Golomer et al., 2009b).
Wilson et al. (2004) and Kwon et al. (2007) looked at the differences between skilled and
novice college dancers in lower limb kinematics and kinetics, respectively. Golomer and
colleagues (2008; 2009b) analysed and compared supporting leg displacements and
relative positions of the shoulders and hips, in expert ballet dancers and untrained dancers
in their execution of a single leg spin (pirouette). Dancers have also been recruited to
quantitative studies on ballet biomechanics based on the type of treatment they have
received for injury. Bronner et al. (2002) investigated the effects of the type of graft
selected for anterior cruciate ligament (ACL) reconstruction on lower limb biomechanics.
Hip and knee joint displacements and velocities for subjects who had received a
semitendinosus - gracilis graft were compared with those who had received a patella
tendon - bone graft. The movement patterns of both groups were compared with skillmatched controls. Although dancers with ACL reconstruction had returned to full dancing
and performance at the time of analysis, results showed that kinematic differences still
- 74 -
existed in both the involved and uninvolved lower limbs of those with ACL
reconstruction compared to controls.
Number of Studies
5
4
3
2
1
0
1
10 11 12 13 14 15
Number of Subjects per Group

Figure 2.26 Sample size for multiple group studies involving quantitative instrumented
biomechanical analysis of classical ballet technique
The numbers of subjects recruited to each group in the inter-group comparison studies
has ranged from groups of 3 up to groups of 15. The mean and standard deviation of the
number of subjects per group, across the 9 multiple group studies indentified, was 6.6 and
2.5, respectively. As indicated in Figure 2.26, most inter-group comparison studies had
group sizes of between 5 and 8 subjects.
2.2.5.5 Three-dimensional Motion Analysis of Classical Ballet Technique.

Since the methodology of this thesis will include 3D motion analysis of classical ballet
movements, the research involving 3D motion analysis of classical ballet technique will
be reviewed in more detail. Only 11 such journal articles were identified, and Table 2.1
provides a summary of each of the different aspects of these studies.
- 75 -
Author
Year
Study
Purpose
Movement
Category
Subjects
Kinematic
Variables
Measured
Ankle plantarflexion
Relevant
Ballet
Principle
Extension
Masso et al.
2004
Movement
description
Relev
Professional ballet dancers (N=18)
Wilson et al.
2004
Movement
description
Other
Skilled college dancers (N=5),

Novice college dancers (N=5)
Trunk & pelvis

global angles
Alignment
Placement
Shan
2005
Injury link
Elanc
Highly trained ballet dancers (N=6),

Highly trained Tai Kwon Do artists (N=5)
Hip, knee & ankle

3D joint ROM
Extension
Turnout
Lin et al.
2005
Movement
description
Relev
Dancers with over 5 years dance training

and over 3 years pointe experience
(N=13)
Extension
Wilson et al.
2007
Movement
description
Other
Experienced female ballet dancers (N=8)
Placement
Turnout
Golomer et al.
2008
Motor
control
Tour
Professional ballet dancers (N=8),

Untrained controls (N=7)
Alderson et al.
2009
Injury link
Other - Lift
Professional and pre-professional male

ballet dancers (N=8)
Pelvis global
angles, hip 3D
angles of
"working" leg
Horizontal
shoulder-pelvis
angles;
Supporting leg
displacements
Trunk flexion,
lateral flexion and
rotation
Golomer et al.
2009a
Motor
control
Tour
Professional ballet dancers (N=10)
Supporting foot
displacement
n/a
Golomer et al.
2009b
Motor
control
Tour
Professional ballet dancers (N=8),

Untrained controls (N=7)
Placement
Kulig et al.
2011
Injury link
Elanc
Elite pre-professional dancers (N=12)
Horizontal
shoulder- pelvis
angles;
Supporting leg
displacements
Knee flexion
n/a
Bronner &
Ojofeitimi
2011
Movement
description
Battement
Advanced college dancers (N=17)
Pelvis and hip

angles
Placement
Turnout
Placement
n/a
Table 2.1 Summary of quantitative biomechanics research studies that used 3D motion analysis and
reported kinematic data.
As has been previously mentioned, the methodology used for kinematic biomechanical
analysis closely reflects the year of publication, and as such, all of the studies that have
used 3D techniques to assess classical ballet technique have been published since 2004.
In terms of the purpose or aim of these studies, most aimed to provide a biomechanical
description of the movements (Wilson et al., 2004; Masso et al., 2004; Lin et al., 2005;
Wilson et al., 2007; Bronner and Ojofeitimi, 2011), however work by Golomer and
colleagues (2008; 2009a; 2009b) focused on motor control, while Shan (2005),
Alderson et al. (2009) and Kulig et al. (2011) investigated ways in which classical ballet
movements may be associated with injury. None of the 3D motion analysis studies looked
at the effects of training programs or dance shoes or surfaces on classical ballet
technique.
- 76 -
The most commonly analysed movement category was the turn (tour), (Golomer et al.,
2008; Golomer et al., 2009a; Golomer et al., 2009b), followed by the rise (relev) (Masso
et al., 2004; Lin et al., 2005) and dart (elanc) (Shan, 2005; Kulig et al., 2011). Only one
study (Bronner and Ojofeitimi, 2011) used 3D motion analysis to analyse the stretch
(battement), and the bend (pli), jump (saut), and glide (gliss) have not been examined
in any published studies involving 3D motion analysis of ballet technique. No study
attempted to analyse more than one of the Seven Movements of Dance (Noverre 1760).
With respect to subjects, nearly half of the studies (Masso et al., 2004; Golomer et al.,
2008; Golomer et al., 2009a; Golomer et al., 2009b; Alderson et al., 2009) recruited
professional ballet dancers, while other studies described their participants as novice
(Wilson et al., 2004), skilled (Wilson et al., 2004) or advanced (Bronner and
Ojofeitimi, 2011) college dancers. Shan (2005) described recruited subjects as being
highly trained, without reference to the dancers being college students, professional
dancers, or having any other such affiliation. Some studies compared kinematic variables
between different groups of dancers (Wilson et al., 2004); between dancers and martial
artists (Shan, 2005); or between dancers and untrained controls (Golomer et al., 2008;
Golomer et al., 2009b); however most studies analysed data for only a single group of
dancers (Masso et al., 2004; Lin et al., 2005; Wilson et al., 2004; Alderson et al., 2009;
Golomer et al., 2009a; Kulig et al., 2011; Bronner and Ojofeitimi, 2011). Across all 11
studies that used 3D motion analysis for analysis of ballet technique, the mean and
standard deviation for the number of participants per group was 9.1 and 4.1, respectively.
Placement was the fundamental principle of classical ballet technique that most studies
could be related to (Wilson et al., 2004; Wilson et al., 2007; Golomer et al., 2008;
Golomer et al., 2009b; Bronner and Ojofeitimi, 2011). Turnout (Shan, 2005; Wilson et
al., 2007; Bronner and Ojofeitimi, 2011) and extension (Masso et al., 2004; Lin et al.,
2005; Shan, 2005) have each been discussed in three studies, while data relating to
alignment has only been reported in one study (Wilson et al., 2004). Also with respect
to the four fundamental principles, most single studies could only be linked with one
(Masso et al., 2004; Lin et al., 2005; Golomer et al., 2009a; Golomer et al., 2009b) or
two (Wilson et al., 2004; Shan, 2005; Wilson et al., 2007; Bronner and Ojofeitimi, 2011)
of the fundamental principles.
- 77 -
In completing this section of the literature review, it has become clear that interest in the
field of quantitative dance biomechanics has increased in recent years. Forty percent
(40%) of the quantitative research identified in this review has been produced during the
past six years, with all of the 3D motion analysis research being conducted during this
time. The studies that have been published to date have had a variety of aims and have
analysed a few different dance genres, however classical ballet has been the most
commonly researched area. With respect to the studies that have investigated classical
ballet technique, what is of extreme relevance to the current research thesis is the fact that
only 11 published studies have used 3D motion analysis to report kinematic data. Of even
more importance, is the finding that amongst these 11 studies, no single published study
has used 3D motion analysis to analyse classical ballet technique in terms of a set of
fundamental biomechanical principles of the technique, nor in terms of all Seven
Movements of Dance, as described by Noverre (1760). For this reason, the aims and
methodologies proposed in the current research thesis are unique amongst existing
literature, and as such it is anticipated that this thesis will make a distinct and original
contribution, not only to the field of dance science research, but also to the broader area
of quantitative analysis of human movement.
- 78 -
2.3 Qualitative Analysis of Human Movement

2.3.1. Introduction
The ability to accurately and reliably observe movement patterns is a key factor in all
fields endeavouring to enhance and alter human movement. Dance teachers and sports
coaches work to assist their students to become highly skilled in their chosen discipline,
integral to which is observing their students movement patterns and providing immediate
useful and effective feedback. Often this process must happen over durations as short as a
fraction of a second (McPherson, 1996). This process is not limited to instruction of
skilled or expert movement. In clinical practice, visual observation of human movement
is often used to identify movement disorders in activities of daily living, and to evaluate
treatment or intervention outcomes. It is a skill used by physiotherapists to improve the
movement abilities of patients undergoing physical rehabilitation, and to estimate work
postures and movements in the fields of ergonomics and occupational health and safety.
This process of visually analysing human movement has been labelled as qualitative
analysis and
has been defined as the systematic observation and introspective
judgment of the quality of human movement for the purpose of providing the most
appropriate intervention to improve performance (Knudson and Morrison, 1997, p4).
Qualitative analysis is a complex process that instructors must be able to employ using
their innate perceptual abilities, combined with their knowledge of the skill or movement
discipline being instructed. Thus, most sports coaches, dance teachers, clinicians and
ergonomists regularly use qualitative analysis in their everyday practice to diagnose and
correct movement errors.
The accuracy and reliability of qualitative movement analysis is vitally important to the
teaching and learning of physical skills. Accurate qualitative analysis helps students to
improve their motor skills and gain confidence in the analyst and themselves. Inaccurate
or inconsistent qualitative analysis however, can just as easily lead to frustration and poor
performance (Knudson, 2000). In clinical practice, in ergonomics and in some sporting
disciplines, specific tools have been developed to help ensure that qualitative analysis is
conducted with as much reliability and accuracy as possible. The following sections will
firstly outline some of the tools typically used in the application of qualitative analysis of
- 79 -
human movement, and will then define the terms reliability and accuracy in relation to
movement assessment. Lastly a review of the academic literature that has investigated the
accuracy of qualitative analysis of human movement in clinical practice, in occupational
health and safety, and in various sporting disciplines will be presented.
2.3.2 Application of Qualitative Analysis of Human Movement

2.3.2.1 Qualitative Analysis of Human Movement in Clinical Practice
Gait analysis constitutes a major aspect of physiotherapy practice and is frequently
employed to assess patients movement abilities. Due to the ease of use of visual gait
analysis compared to 3D computerised gait analysis, many clinicians routinely rely on
this method to assess their patients. Over the past few decades, a number of observational
gait assessment tools have been developed for use in clinical practice. The assessment
tools usually consist of rating scales for numerous gait parameters for which
abnormalities commonly exist in pathological populations. A few observational gait
analysis scales have been developed for general analysis of all clinical populations
(Winter, 1985; Perry, 1992), however the need for more population-specific scales has
led to the development of various visual gait assessment scales specific to particular
clinical populations. Scales have been developed for patients with lower limb amputation
(Saleh and Murdoch, 1985) rheumatoid arthritis (Eastlack et al., 1991), podiatry patients
(Keenan and Bach, 1996), hemiplegia following stroke (Goodkin and Diller, 1973;
Miyazaki and Kubota, 1984; Hughes and Bell, 1994; Lord et al., 1998; McGinley et al.,
2003; Watelain et al., 2005; McGinley et al., 2006), orthopaedic disorders (Brunnekreef
et al., 2005), and for cerebral palsy (Corry et al., 1998; Krebs et al., 1985; Koman et al.,
1994; Boyd and Graham, 1999; Dickens and Smith, 2006; Kawamura et al., 2007; Read
et al., 2003; Piripis et al., 2001; Flett et al., 1999).
When using observational gait analysis tools observers are typically required to estimate
joint angles at particular phases or points in the gait cycle. The joint angles are often
expressed in terms of levels of deviation from normal gait. These estimations are made
using a variety of techniques. In one of the earlier scales developed, the Waterloo Gait
Profile Form (Winter, 1985), observers record their ratings by circling one of a selection
of stick diagrams and/or symbols that they believe most closely represent the trunk, knee
and foot movement patterns they observed. The Rancho Los Amigos System (Olsson,
- 80 -
1990; Perry, 1992) is based on a form that is laid out in a tick box format. The user
identifies deviations from normal gait by ticking a box for the frontal, sagittal and
transverse plane rotations at the trunk, pelvis, hip, knee, ankle and toes. Deviations from
normal gait are categorised as minor or major, but no specific guidelines or joint angle
ranges are provided to assist the user in establishing what constitutes major or minor
deviation. The Rivermead Visual Gait Assessment (Lord et al., 1998), a tool developed
for assessment of neurologically impaired patients, comprises a 4-point scale (0 3) to
grade joint or segment positions as either normal (0), mild (1), moderate (2), or severe
(3), and where appropriate indicate the direction of deviation. However, as for the
Rancho Los Amigos System, there is some ambiguity about the criteria for selection of
each grade in this system. The Physician Rating Scale (Koman et al., 1994) is a tool that
is commonly used to assess the gait of children with cerebral palsy. It examines the knee,
ankle and equinus foot in the sagittal plane, and uses either 3- or 4-point scales to identify
deviations from normal. For example, the extent to which a patient exhibits crouch gait
(i.e. increased knee flexion) is recorded using a 4-point scale (i.e. severe (0), moderate
(1), mild (2), or none (3)), however in this scale users are also presented with associated
joint angle ranges to assist in selecting the appropriate grade or category.
Modified versions of the Physicians Rating Scale (PRS) (Corry et al., 1998; Flett et al.,
1999; Boyd and Graham, 1999; Mackey et al., 2003; Wren et al., 2005; Dickens and
Smith, 2006; Brown et al., 2008) have also been developed. These modified versions all
still focus on knee, ankle and foot parameters, but alterations have been made to the
number and type of ratings required for each joint. While most modified versions of the
PRS provide joint angle ranges as well as references to normality, some other versions of
the PRS (Wren et al., 2005) provide joint angle ranges with no indication as to how the
joint range categories relate to normal. In contrast, other newer scales (Kawamura et al.,
2007) have been developed in which raters are asked to assess joint movements in
multiple planes only in terms of normality. For example, in Kawamura et al. (2007),
observers are required to use a 3-point scale to record if the joint movement is normal,
decreased, or increased. All the above mentioned gait scales are typically conducted
using live visual observation, rather than video-based observation. However, there are
some scales that have been designed specifically for video-based observation. A well
known gait assessment tool typically conducted via video analysis, is the Edinburgh
- 81 -
Visual Gait Score (Read et al., 2003). It requires raters to make observations of 17
different joint parameters for each lower limb of patients with cerebral palsy. Joint and
segment angles for the trunk, pelvis, hip, knee and ankle are estimated using a 5-point
scale, for which joint angle ranges and references to normality are provided. As suggested
by the large number of parameters contained in the Edinburgh Visual Gait Score, use of
video analysis allows for observations of more parameters than live observations, as the
video can be replayed multiple times to obtain all the required information.
Observation of human movement in clinical practice is not limited to observational gait

analysis. In the clinical setting therapists often use other functional performance tests
based on visual observation to evaluate the movement abilities of patients (Harrison et
al., 1994; Bernhardt et al., 1998; Haupstein and Goldie, 2000; Bernhardt et al., 2002;
Bjrklund et al., 2006; von Porat et al., 2008). The one-legged standing balance test
(Harrison et al., 1994; Haupstein and Goldie, 2000) is one protocol which is used
frequently in the fields of neurology, paediatrics, geriatrics, orthopaedics and sports
medicine. In this test subjects are asked to stand as still as possible over the narrow base
of support provided by one foot, with eyes open or closed. Inferences are made about
balance on the basis of how steadily this challenging posture can be held. Clinicians
evaluate the movement strategy that the person uses to maintain balance to judge whether
a rehabilitation program is required, or has been effective in correction of any observed
deficiency in balance. Other functional performance tests used in clinical practice may
involve evaluation of running, one- and two-legged squatting , sit-to-stand, distance
hoping, cross-over hoping, and vertical jumping (Bjrklund et al., 2006; von Porat et al.,
2008). Upper limb function (Bernhardt et al., 1998; Bernhardt et al., 2002), and head and
neck posture (Youdas et al., 1991; Passier et al., 2010) are also often assessed by visual
observation in clinical practice.
2.3.2.2 Qualitative Analysis of Human Movement in Human Factors and Ergonomics

Due to legal and other requirements, various methods have been developed for assessing
exposure to risks linked with work-related musculoskeletal disorders, and for
documenting changes in exposure associated with introducing workplace interventions.
These include instrumentation-based techniques, subjective self reports and other
psychophysiological and visual observational methods (Li and Buckle, 1999).
- 82 -
It is
presumed that the most accurate exposure data are obtained by instrumentation-based
technical measurements (Genaidy et al., 1993). However, as per 3D gait analysis, these
techniques may be too expensive and time consuming when attempting to analyse large
numbers of subjects. Thus, systematic visual observation applied by experienced
ergonomists may offer a compromise. Observational techniques, specially developed for
the evaluation of occupational activities, have therefore become popular as they are
widely applicable, cost effective and often easy to use (Burdorf et al., 1992). In general,
almost all of the observation methods published to date concentrate on working postures,
with some methods adding assessment of other factors such as force, duration, frequency,
perceived exertion, or type of work activity. Most postural observation systems
concentrate on assessment of the back, neck, shoulders and arms. This is understandable
as the prevalence of work-related musculoskeletal problems of the lower extremity is not
as frequently reported as those of the upper body segments (Li and Buckle, 1999). Some
observational methods have been developed for use in specific industries, such as
construction (OWAS), transportation (TRAC) or healthcare (REBA), while others have a
more general use in various industries (HARBO).
Most workplace qualitative analysis methods are based upon repeated observations of the
worker throughout either a number of representative work cycles or a specified period of
time. Two general methodologies for visual observation techniques can be distinguished,
(1) continuous recording of events as they occur in real time (e.g. PEO, TRAC, VIRA
HARBO), and (2) time-sampled or intermittent observations in which data are collected
at pre-specified time intervals (e.g. OWAS, RULA, PATH, ARBAN, 3DMatch).
Continuous real time observations are typically used to assess work activities that can
change rapidly and frequently and do not always follow a consistent pattern from one
cycle to another. Continuous observations thus provide information about the sequence,
duration and frequency of the observed postures, while time-sampled observations only
provide an estimate of this information. In time-sampled observations, the percentage of
time spent in a given posture is computed as the ratio of the number of observations
recorded for this posture and the total number of observations (Genaidy et al., 1994),
therefore it is assumed that the higher the sampling frequency, the more accurate the
time-based records obtained, however very few reports give any advice on appropriate
sampling rates when using this method. Some observation tools such as ROTA (Ridd et
- 83 -
al., 1989), TRAC (van der Beek et al., 1992), and RULA (McAtamney and Corlett, 1993)
facilitate optional time sampling or continuous real time monitoring. It has been reported
(Bao et al., 2007) that there is poor agreement between RULA scores obtained with
event-based (continuous observations) and time-based (intermittent observations)
postural analysis methods. Therefore, postural analysis data based on one postural
analysis method would not necessarily be equivalent to those generated by the other
method. (Karhu et al., 1977; Heinsalmi et al., 1986; Keyserling, 1986)
In addition to the sampling strategy of observation, visual observation tools can also be
categorised as being based on either macropostural classifications or micropostural
classifications (Genaidy et al., 1994). Macropostural classification systems group multiplanar non-neutral positions around a joint into one category. For example, in one of the
earliest macropostural systems, the Ovako Working Postures Analysing System (OWAS),
Karhu et al. (1977) classified the flexion, extension and lateral bending of the lower back
as bending. Arm movements are classified in the OWAS as elevation above or below
the shoulder joint, with elevation including flexion, extension, abduction and adduction.
Heinsalmi (1986) expanded the OWAS system to include head positions around the neck
for which postures were described as bending and rotation. The micropostural
classifications are more detailed than the macropostural classification procedures
(Genaidy et al., 1994), and usually contain more than one category or level of non-neutral
uni-planar joint posture. For example, Keyserling (1986) modelled lower back flexion
into two levels; moderate flexion (16 - 45) and severe flexion (>45), and shoulder
flexion has been classified into three categories (Kilbom et al., 1986), and four categories
(Armstrong et al., 1982). Genaidy et al. (1993) and Armstrong et al. (1982) both
modelled wrist flexion/extension into five angle categories.
2.3.2.3 Qualitative Analysis of Human Movement in Sports

Due to the huge variety of sports, and the vast array of tasks, aims, and measures of
competency that are associated with each different sporting discipline, it is not feasible to
review specific assessment scales that have been devised for individual sports. Instead,
this review will briefly discuss some of the comprehensive models of qualitative analysis
of skilled movement that have been proposed by renowned scholars in sports pedagogy
- 84 -
(Arend and Higgins, 1976; Hay and Reid, 1982; McPherson, 1990; Knudson and
Morrison, 1997).
Arend and Higgins (1976) made a significant contribution to the study of qualitative
analysis by presenting one of the first systematic methods for assessing skilled human
movement performance. The system consisted of three major phases; (1) pre-observation,
(2) observation, and (3) post-observation. Pre-observation involves multilevel breakdown
or decomposition of the movement (Arend and Higgins, 1976) in order to obtain
information about the prerequisites essential for successful performance of the skill. This
includes identification of the critical features of the skill, as well as specific
characteristics of the performer and their environment, in order to determine what it is
that one expects to see. Critical features were defined as parts of the movement that
could be least modified to be successful (Arend and Higgins, 1976, p45). Some authors
have since advocated that critical features be customised for the age or developmental
level of the performer (Abendroth-Smith et al., 1996). Secondly, in the observation
phase it is necessary to make systematic observations over repeated trials with each trial
used to view a particular aspect or phase of the movement, so that by the last trial a clear
record of the entire performance has been obtained. This information should be clearly
documented by answering a series of carefully devised questions.
Lastly, Arend and Higgins (1976) state that by utilising the information gained in the first
two stages, comparisons and evaluations can be made during the post-observation phase,
between what the observer expected to see happen and what actually happened. From this
information feedback is provided to the performer. The Arend and Higgins (1976) model
was very significant in the field of sports pedagogy as it was one of the first models to
provide an integrated analysis of human movement incorporating aspects of
biomechanics, pedagogy and motor development.
- 85 -
Figure 2.27 A strategy for the qualitative analysis of human movement (from Arend and Higgins 1976,
p38)
Another widely cited model is the deterministic model of qualitative analysis, first
presented by Hay and Reid (1982). This model consists of four steps, (1) develop a model
of the skill, (2) observe performance and identify faults, (3) rank the priority of faults, and
(4) instruct the performer. The first task of developing a model of the skill requires the
observer to identify the primary goal, result, or outcome of the skill to be investigated,
and to identify those factors that produce this result.
The outcome of a performance can
be an objective measure (e.g. distance, height, time, etc), or a subjective measure (e.g.
points awarded in gymnastics and diving competitions) (Chow and Knudson, 2011), and
the factors that produce this should normally be mechanical quantities. The correct
identification of these factors relies upon the observer having thorough background
knowledge of the skill being assessed. In the second step of their model, Hay and Reid
(1982) state that it is important to use all sensory information: visual, aural, tactile, and
kinaesthetic to observe the performance. They also state that the position of the observer
should be at right angles to the direction of the performers body motion, and similar to
Arend and Higgins (1976), Hay and Reid (1982) recommend observation of multiple
- 86 -
trials to systematically observe separate parts of the movement. In identifying faults, it is

suggested that the mechanical method is used. This method is based upon identifying
faults in the most fundamental biomechanical characteristics that contribute to execution
of the movement. The third step in the model involves evaluating and ranking the faults.
Hay and Reid (1982) suggest that faults should be prioritised according to the order that
generates the most improvement in the time available to correct them. Hay and Reid
(1982) recommend that in the final step of their model, instructions to the performer
should be limited to one fault at a time. They also advocate initially providing direct
corrections or literal descriptions of what the performer should do, rather than more
figurative feedback.
McPherson (1990) provided yet another well known model of qualitative analysis of
skilled movement. This model also consists of multiple phases, similar in structure and
purpose to the phases described by Arend and Higgins (1976) and Hay and Reid (1982).
McPhersons phases are; (1) pre-observation, (2) observation, (3) diagnosis, and (4)
remediation. McPherson (1990) states that during pre-observation, careful development
of an observation plan and a recording form are essential, and that during observation an
optimal viewing location must be identified and used. Unique to this model, is the
inclusion of the consideration of direct and indirect constraints. Direct constraints
relate to mechanical factors that directly affect how a movement is performed, while
indirect constraints refer to factors other than mechanical elements that may affect the
performance of the skill, this might include motivational or environmental factors. Figure
2.28 provides a schematic representation of the phases of the model developed by
McPherson (1990).
- 87 -
Figure 2.28 Model of qualitative analysis of sports movement (from McPherson 1996, p s86).
Lastly, Knudson and Morrison (1997) presented another multi-phase model consisting of;
(1) preparation, (2) observation, (3) evaluation and diagnosis, and (4) intervention. Like
the other models, Knudson and Morrison (1997) recommend careful preparation,
including identification of critical features, and use of an appropriate and systematic
observation strategy, in which multiple observations (at least 5 8) are performed, using
all perceptual senses. The distinguishing factor about this model is that the authors
emphasise the importance of applying a more comprehensive and interdisciplinary
approach during the final phases of the model. Rather than applying the traditional error
detection and correction strategy and overemphasising negative elements of the
performance, Knudson and Morrison (1997) suggest that the analyst should strive to
identify both the strengths and weaknesses of the performance and use all this
information in applying useful intervention in the form of conditioning, cueing and
modified practice (Knudson, 2000).
- 88 -
The four models discussed present many similarities in the phases and tasks that are seen
as being necessary for a comprehensive approach to qualitative analysis. The multiphase
approach to qualitative analysis in sports is thus a widely accepted model in the academic
literature, however it is difficult to know whether the techniques presented are used
regularly in everyday practice in sports pedagogy. It is possible that some sports coaches
do not apply such systematic methods in analysing their students movements, but rather
rely on ad hoc methods developed over years of trial and error.
2.3.2.4 Qualitative Analysis of Human Movement in Dance

Students of dance and classical ballet usually undergo formal assessments or evaluations
throughout their dance training. These assessments often take the form of examinations in
which specific exercises or dances, developed as part of a syllabus, are evaluated by an
external examiner who is registered with an examining organisation (e.g. Royal Academy
of Dance (RAD), or Cecchetti). In these examinations, dancers are assessed against
specific criteria, amongst which technique is usually only one component. Other
components often include elements such as musicality and performance.
The
examinations held by organisations such as the RAD and Cecchetti are conducted
regularly in dance centres throughout the world and provide a consistency of standard,
within the parameters of the syllabus, against which dancers and dance teachers can be
evaluated. Some dance organisations have devised their own assessment criteria for inhouse assessment of their dancers. Again, these assessment tools usually list a number of
different components, of which technique is only one element. Whether conducted by
large external examining bodies, such as the RAD or Cecchetti, or through in-house
assessment, it appears that the technique component in most dance assessments does not
consist of detailed analyses of specific joint movements and/or segment orientations. For
example, in the examinations conducted by the RAD, the technical elements that are
assessed are listed as:
x Correct posture and weight placement
x Co-ordination
x Control
x Line
x Spatial awareness
- 89 -
x Dynamic values
(Royal Academy of Dance, 2011, p19) (Royal Academy of Dance, 2011)
For the Cecchetti syllabus the technical aspects of ballet are assessed according to these
criteria:
x
Stance and poise
Use of technique
An understanding of correct stance and placement with an application of classical

technique
(Imperial Society of Teachers of Dancing, 2011)
(Imperial Society of Teachers of Dancing, 2011, p12)
These values do not relate specifically to particular body regions, or to specific aspects of
the technique, and are therefore somewhat difficult to interpret. A few in-house technique
assessment tools used in tertiary dance institutions were also identified (Krasnow and
Chatfield, 1997; Krasnow and Chatfield, 2009), and these proved to be just as ambiguous
and non-specific in relation to assessing dancers movement patterns at individual joints.
Although these assessment methods typically give a numerical score, all judgments on
the merits of a dancers competence are based on subjective opinion through qualitative
assessment. Dance assessment criteria used by Krasnow and Chatfield (2009) have
included:
x
Skill
Space
Time
Energy
Phrasing
Presence
Krasnow and Chatfield (1997) assessed dancers according to the criteria of:
x
Full body involvement
Body Integration and Connectedness
Articulation of Body Segments
Movement Skills
- 90 -
There are a few dance research groups and organisations (Liederbach, 1997; Molnar and
Esterson, 1997; Solomon, 1997; Siev-Ner et al., 1997; Plastino, 2005; Hamilton, 2006;
Potter and Baas, 2011; Gibbs et al., 2006) who have devised tools for identification of
potential pre-cursors to dance-related injury. Although often based largely on assessment
and measurement of static postures, strength, and joint ranges of motion, some of these
tools also include a section involving qualitative analysis of dynamic performance of core
dance steps (Liederbach, 1997; Molnar and Esterson, 1997; Gibbs et al., 2006; Potter and
Baas, 2011). Molnar and Esterson (1997) developed a screening for students in a preprofessional ballet school. In addition to many static assessment techniques, this
screening process included qualitative analysis of performance of the pli, relev, pass,
developp, battement tendu and ports be bras.
Elements such as alignment and
stability were assessed through qualitative analysis of these movements. Liederbach

(1997) described the contents of a Screening for Functional Capacity in Dancers
developed at the Harkness Centre for Dance Injuries, New York. Although also based
largely on static postural and range of motion tests, the tool also included qualitative
assessment of a few dynamic movements such as the pli, saut, jet, and single
pirouette. Gibbs et al. (2006) presented guidelines for the Post-Hire Health Screen for
Professional Dancers developed by the Dance/USA Taskforce on Dancer Health. This
screening tool contains a small section in which dance-specific movements (i.e. pli and
relev) are analysed qualitatively, with the aim of assessing lumbo/pelvic stability, malalignment of the knee with respect to the feet, and the position of the ankle joint (i.e.
pronation/supination). Potter and Baas (2011) reported on the basic requirements for
effective dance screening, and highlighted the work of the Dancer Wellness Project
(DWP), a consortium of dance organisations, that promote dancer health. The screen
developed by the DWP contains a dynamic functional assessment component, which
includes qualitative analysis of the battement tendu, developp, and ports be bras. Potter
and Baas (2011) state that in assessing these movements, it is essential to look for
attributes of faulty movement patterning. Thus, it appears that greater awareness of the
importance of assessment of dancers dynamic functional capacity is developing
throughout the dance medicine and science community.
No matter which type of movement is being observed, or in which setting or field

qualitative assessment is being conducted, the efficacy of any method or test used to
- 91 -
assess human movement relies heavily on its reliability and validity (Spielholz et al.,
2001). The definitions and characteristics of test reliability and validity will be discussed
in the next section.
2.3.3 Reliability and Accuracy of Qualitative Analysis of Human Movement

2.3.3.1 Definition of Test Reliability
The reliability of a test or measurement is the degree to which repeated measurements of
the same trait are reproducible under the same condition (Morrow et al., 2005). It relates
to the consistency or repeatability of a measurement or observation. A test is said to be
reliable if the same score or result is obtained each time the test is administered to the
same individual by the same rater (intra-rater reliability) or by different raters (inter-rater
reliability). The greater the dispersion of repeated scores of the same event, the greater
the variance and the lower the reliability or repeatability (Toro et al., 2003). A measure
can be reliable but not valid, whereas an unreliable measure cannot be valid. In order to
determine if an assessment has adequate reliability for the purposes of clinical or
ergonomic use, one must understand what measures are used to determine reliability,
what are the strengths and limitations of these measures, and what the results actually
mean in the context of the field or discipline in which they are obtained (Innes and
Straker, 1999).
2.3.3.2 Definition of Test Accuracy

The terms test accuracy and test validity are often used interchangeably in scientific
research. In general, validity refers to the degree of truthfulness of a test score or result
(Morrow et al., 2005). That is, a test score or result, once found to be reliable, is valid if it
accurately measures what it reports to measure. There are various levels of validity. The
two broad categories of validity are construct validity and criterion-related validity.
Construct validity is comprised of the two sub-categories of face validity and content
validity. Face validity is concerned with how a measure or procedure appears. That is, it
pertains to whether the test "looks valid" in that it seems like a reasonable way to gain the
information the researchers are attempting to obtain. Face validity does not depend on
established theories for support, it only requires an intuitive judgment. Content validity
requires more rigorous statistical tests than face validity, and gives an indication of the
degree to which the particular items in a test reflect the knowledge actually required for a
- 92 -
given area of research (Toro et al., 2003). In clinical practice for example, content
validity refers to the relevance between the test items included on a performance rating
scale and the common symptoms or pathologies of a particular condition. For example,
numerous studies (Patla et al., 1987; Hughes and Bell, 1994; Lord et al., 1998; Schutte et
al., 2000; Romei et al., 2004; Watelain et al., 2005; Hillman et al., 2007; Novacheck et
al., 2000) have investigated the content validity of visual gait analysis rating scales.
The criterion-related validity of test instruments, is a measure of agreement between the

results obtained by the given test instrument and more "objective" results for the same
population (Toro et al., 2003). The "objective" results are obtained either by a well
established instrument ("the gold standard") or by direct measurement. In terms of
criterion-related validity in visual observation of human movement, data obtained from
computerised 3D motion capture have come to be considered as the gold standard" for
criterion-related validity tests (Toro et al., 2003), however other instrumented
measurement techniques, such as inclinometry (Burdorf et al., 1992; Teschke et al., 2009;
Village et al., 2009), and electrogoniometry (Yen and Radwin, 2000; Ketola et al., 2001;
Spielholz et al., 2001; Lowe, 2004a; Lowe, 2004b; Andrews et al., 2008), have also been
used to assess criterion-related validity. Assessment tools that can demonstrate criterionrelated validity have the greatest justification for their use since they instil high
confidence in their ability to genuinely reflect what is actually occurring. Thus, for the
purposes of this study it is the criterion-related validity of qualitative analysis that will be
used as a measure of accuracy. That is, qualitative analysis will be considered to be
highly accurate if the data are in close agreement with criterion-related quantitative data.
2.3.3.3 Measurement of Test Accuracy and Reliability

Measures of reliability and validity are usually reported as coefficient values. Many
reliability or validity coefficients are based on measures of correlation (e.g. Pearsons
Product-Moment Correlation Coefficient (r); Spearman Rank Correlation Coefficient (r);
Intra-class Correlation Coefficient (ICC)). Correlation reflects the degree of association
between data sets, and gives an indication of variance rather than agreement between
data. For example, the greater the dispersion of repeated scores of the same event the
greater the variance and the lower the reliability, and the smaller the variance between
repeated scores the higher the reliability (Toro et al., 2003). For many applications,
- 93 -
however, it is necessary to determine agreement between measures (i.e. two sets of

measures are the same). When the unit of measurement in an assessment is categorical
(e.g. poor/ fair/good/ excellent; normal/abnormal; correct/incorrect), reliability or validity
is more appropriately determined as a measure of agreement. In its simplest form this
would be measured as percentage agreement, however this value often gives an overestimation of true reliability (Innes and Straker, 1999). Another measure of agreement is
the kappa () statistic, which is used to correct for chance agreement. This statistic
represents the average rate of agreement for an entire set of scores, and is an analysis of
exact agreement with no room for close agreement, making it suitable for nominal and
ordinal data, but not continuous data (Innes and Straker, 1999). The weighted kappa takes
into account the distance between categories. It penalises more for disagreement between
extreme categories than for adjacent ones (Denis et al., 2002).
Measure of Reliability or Validity
Range of Values
Interpretation of Values
Correlation Coefficients (r)
0.00 - 0.25
Little or no relationship
0.26 - 0.50
Poor to fair
0.51 - 0.75
Moderate to good
> 0.75
Good to excellent
Required for clinical application to

ensure valid interpretation of findings
< 0.40
Poor reliability
0.40 - 0.75
Fair to moderate reliability
> 0.75
Good to excellent reliability
Required for clinical application to

ensure valid interpretation of findings
0.00
Chance agreement
0.01 - 0.20
Poor agreement
0.21 - 0.40
Slight agreement
0.41 - 0.60
Moderate agreement
0.61 - 0.80
Substantial agreement
Intra-class Correlation Coefficient (ICC)
Kappa (k) value
0.81 - 1.00
Excellent to perfect agreement
Table 2.2 Interpretation of measures of reliability or validity (Landis and Koch, 1977; Fleiss, 1986;
Portney and Watkins, 1993)
It appears that there is little consensus about which statistical methods are best to analyse
reliability and validity (Bao et al., 2009), however research has shown that percentage
agreement values are generally higher than kappa values (van der Beek et al., 1992; de
Looze et al., 1994; de Bruijn et al., 1998; Burt and Punnett, 1999; Ketola et al., 2001;
- 94 -
Denis et al., 2002; Kociolek and Keir, 2010) and ICC values (Ebersole and Armstrong,
2006; Bao et al., 2009), and perhaps overestimate the actual reliability and validity. Once
a numerical measure of reliability or validity has been calculated, in the form of a
coefficient or kappa value, typically a rating or interpretation is applied to this value.
Correlation coefficients and kappa values can theoretically vary between 0 and 1.0, where
0 indicates no reliability and 1 indicates perfect reliability (Weir, 2005). Landis and Koch
(1977) devised some guidelines for interpretation of particular ranges of kappa values,
while other guidelines for interpretation of correlation coefficients and ICC coefficients
and have also been developed, these are presented in Table 2.2.
Studies that have investigated the criterion-related validity or accuracy of qualitative

analysis in human movement have used a variety of the above mentioned statistical
methods to compare qualitative responses with quantitative data. Kappa values have
been the most commonly used statistic in these studies (de Looze et al., 1994; Ketola et
al., 2001; Paquet et al., 2001; Nicholls et al., 2003; Mackey et al., 2003; Lowe, 2004a;
Lowe, 2004b; Stott et al., 2005; Dickens and Smith, 2006; Krosshaug et al., 2007;
Kawamura et al., 2007; Dallas et al., 2011), followed by Pearsons correlation coefficient
(Yen and Radwin, 2000; Haupstein and Goldie, 2000; Spielholz et al., 2001; Neumann et
al., 2001; McGinley et al., 2003; McGinley et al., 2006; Moseley et al., 2008; Village et
al., 2009; Teschke et al., 2009), and Spearmans rank correlation coefficient (Burdorf et
al., 1992; Harrison et al., 1994; Lowe, 2004a; Lowe, 2004b; von Porat et al., 2008). It
appears that researchers are aware of the limitations of the measure of percentage
agreement (Po), as this statistic has only been reported in a few studies (de Looze et al.,
1994; Leskinen et al., 1997; Paquet et al., 2001; Piripis et al., 2001; Read et al., 2003;
Brown et al., 2008), with very few of these being conducted in recent years. As an
alternative to the techniques listed above, some studies have obtained measures of
accuracy based on calculation of mean error (Genaidy et al., 1993; Hillman et al., 1998;
Morrison et al., 2005; Plessner and Schallies, 2005; Krosshaug et al., 2007; van Wyk et
al., 2009; Marques-Bruna et al., 2008; Lau and Armstrong, 2011), or the number or
percentage of accurate responses (Osborne and Gordon, 1972; Patla and Clouse, 1988;
Baluyut et al., 1995; Wren et al., 2005; Andrews et al., 2008; Passier et al., 2010). Little
evidence was found of the use of ICC for analysis of the accuracy of qualitative analysis
in human movement.
- 95 -
The following section will review the findings of the studies on criterion-related validity
or accuracy of qualitative analysis of human movement. These studies have been
conducted in the clinical setting, in the field of ergonomics and human factors, and in
various sporting disciplines. Considerable research has also been conducted on the interrater and intra-rater reliability of qualitative analysis of human movement, particularly in
clinical practice, however this work will not be reviewed in this thesis. By definition, if
an assessment technique has high criterion-related validity it will also have high
reliability, whereas a highly reliable assessment tool may not necessarily be valid or
accurate. For this reason, this review will focus only on studies that have assessed the
criterion-related validity, hereafter referred to as accuracy, of qualitative assessment of
human movement. Articles have been selected for inclusion in the review if they contain
results on the accuracy of visual observation methods in which kinematic variables (e.g.
body segment and joint orientations) have been observed and compared with quantitative
data obtained through instrumented measurement techniques.
2.3.4 Investigation of the Accuracy of Qualitative Analysis of Human Movement

Given the widespread use of qualitative analysis in routine clinical practice, in assessment
of workplace postures, and in sports coaching, it is imperative that a high level of
confidence is held in the accuracy of the techniques used. Studies investigating the
accuracy of visual observation of human movement are therefore extremely important.
The increasing availability of computerised motion analysis and other instrumented
measurement techniques, has created the opportunity for more detailed study of the
accuracy of observational assessment of human movement.
A number of studies have been conducted by clinical research groups to investigate the
accuracy of the use of observational motion analysis scales. These studies have
encompassed a wide diversity in both the type and number of variables selected for
rating. Several studies (Hillman et al., 1998; Piripis et al., 2001; Read et al., 2003;
McGinley et al., 2003; Mackey et al., 2003; Wren et al., 2005; Dickens and Smith, 2006;
McGinley et al., 2006; Kawamura et al., 2007; Brown et al., 2008) have used 3D
quantitative motion analysis techniques to assess the accuracy of the observational gait
analysis skills of clinicians working in musculoskeletal rehabilitation. These studies have
investigated the accuracy with which raters are able to estimate specific joint angles
- 96 -
(Hillman et al., 1998; Piripis et al., 2001; Read et al., 2003; Mackey et al., 2003; Wren et
al., 2005; Dickens and Smith, 2006; Brown et al., 2008), spatio-temporal properties
(Mackey et al., 2003) and kinetic parameters (McGinley et al., 2003; McGinley et al.,
2006). Accuracy of visual observation of clinical functional performance tests has also
been investigated (Harrison et al., 1994; Bernhardt et al., 1998; Haupstein and Goldie,
2000; Bernhardt et al., 2002; von Porat et al., 2008). Some of these studies have
compared qualitative responses to 3D kinematic motion analysis data (Bernhardt et al.,
1998; Bernhardt et al., 2002; von Porat et al., 2008), while others have used force plate
data (Harrison et al., 1994; Haupstein and Goldie, 2000) or specifically designed
equipment, such as an instrumented foot plate (Moseley et al., 2008), to determine
accuracy.
This section will also review the published research on the criterion-related validity of
visual observation of work-related postures and movements. The accuracy of
observational-based methods for assessing working posture has been investigated in
several studies (Burdorf et al., 1992; Genaidy et al., 1993; de Looze et al., 1994; Baluyut
et al., 1995; Leskinen et al., 1997; Yen and Radwin, 2000; Paquet et al., 2001; Spielholz
et al., 2001; Ketola et al., 2001; Neumann et al., 2001; Lowe, 2004b; Lowe, 2004a;
Andrews et al., 2008; van Wyk et al., 2009; Teschke et al., 2009; Village et al., 2009; Xu
et al., 2011; Lau and Armstrong, 2011).
These studies have been conducted by
comparing analysts observational estimates of working postures with direct

measurements of relevant joint angles. The methods of direct measurement have included
3D motional analysis (de Looze et al., 1994; Leskinen et al., 1997; Neumann et al., 2001;
Lowe, 2004b; Lowe, 2004a; Andrews et al., 2008; van Wyk et al., 2009; Xu et al., 2011),
electrogoniometer measurements (Yen and Radwin, 2000; Paquet et al., 2001; Spielholz
et al., 2001; Ketola et al., 2001; Baluyut et al., 1995; Genaidy et al., 1993), inclinometer
measurements (Burdorf et al., 1992; Teschke et al., 2009; Village et al., 2009), and more
simplistic methods such as use of a plexiglass calibration board (Lau and Armstrong,
2011).
It appears that less research has been conducted on the accuracy of qualitative analysis of
sporting activities than in clinical practice and occupational health and safety. Due to the
highly varied nature of sports-related movements and the complexity of the movements
- 97 -
involved, it may be more difficult to perform detailed studies on the accuracy of

quantitative analysis of sports movement than for clinical gait analysis, functional
performance tests, or work-related tasks. It is perhaps for this reason that it was difficult
to find an abundance of studies that have used quantitative techniques to assess accuracy
of qualitative analysis of sporting activities. However a few studies of this type were
identified, which include investigation of qualitative analysis in gymnastics (Imwold and
Hoffman, 1983; Plessner and Schallies, 2005; Dallas et al., 2011), handball (Krosshaug et
al., 2007; Stensrud et al., 2011), baseball (Nicholls et al., 2003), soccer (Marques-Bruna
et al., 2008), and general physical education activities such the vertical jump (Knudson,
1999). The majority of these studies have obtained quantitative data through video
analysis and 2D digitisation (Knudson, 1999; Plessner and Schallies, 2005; MarquesBruna et al., 2008; Stensrud et al., 2011; Dallas et al., 2011), while few studies have used
3D motion analysis (Nicholls et al., 2003; Krosshaug et al., 2007). Other techniques have
included traces and measurements made from photographs (Imwold and Hoffman, 1983).
Investigation of the accuracy of qualitative analysis in dance, measured by comparison of

qualitative data with quantitative data, has received no attention in the academic
literature. A thorough review of the published academic literature on quantitative
biomechanical assessment in dance, presented in Section 2.2, did not reveal any studies
that specifically compared quantitative biomechanical data with qualitative responses
based on observational assessment. In addition, literature searches in which accuracy or
validity of qualitative analysis were key search criteria did not return any studies that
focused on qualitative analysis in dance.
In total, 47 published studies were identified that have used instrumented quantitative
methods to obtain criterion-related measures of validity (accuracy) of qualitative
analysis of human movement. Figure 2.29 depicts the various percentages of the studies
that have been conducted in relation to clinical practice, occupational health and safety,
sports, and dance. Overall, the majority of these studies (41%) have been conducted in
the clinical setting, with most of these clinical studies being related to qualitative gait
analysis.
- 98 -
Dance
0%
Sport
20%
OHS
39%
Clinical Gait
26%
Clinical Other
15%
Figure 2.29 Summary of percentages of studies conducted

on the accuracy of qualitative analysis in clinical practice,
occupational health and safety (OHS), sport, and dance.
There is a great deal of diversity in the results for studies that have assessed accuracy of
observation by comparing quantitative measurements with qualitative responses. Kappa
(k) values have ranged from 0.01 to 0.94 (poor to excellent agreement), Pearsons
coefficients (r) have ranged from 0.24 to 0.98 (little relationship to excellent
relationship), Spearmans coefficients (r) have ranged from 0.37 to 0.62 (fair
relationship to moderate relationship), percentage agreement (Po) values have ranged
from 38% to 83%, and percentage of accurate responses have ranged from 20% to 95%.
A variety of different protocols were used in the studies to obtain accuracy data for
observation of various body regions, for subjects performing a multitude of different
activities. Given the vast array of conditions under which these data were obtained, there
are a number of factors that could have contributed to the diversity of accuracy results.
The following sections will review the literature in relation to a number of key factors
that may impact upon the accuracy results obtained. These factors include the mode of
observation (i.e. live vs. video), the type of assessment scales used, the number of
variables simultaneously observed, the body region observed, and the level of experience
of the person conducting the qualitative analysis.
- 99 -
2.3.4.1 Effects of Observation Mode on Accuracy of Qualitative Analysis

The accuracy of observation of either video (Baluyut et al., 1995; Yen and Radwin, 2000;
Spielholz et al., 2001; Neumann et al., 2001; Lowe, 2004b; Lowe, 2004a; Andrews et al.,
2008; van Wyk et al., 2009; Xu et al., 2011) or live on-site (Burdorf et al., 1992; de
Looze et al., 1994; Leskinen et al., 1997; Ketola et al., 2001; Paquet et al., 2001; Teschke
et al., 2009) postural observation in the work place has been studied. It was not possible
to identify any studies that directly compared the accuracy of observation between the
two modes. From the data published to date, it appears that the only suitable comparison
that can be made across studies to attempt to address this question is between Spielholz et
al. (2001) and Ketola et al. (2001). Both these studies reported data on the accuracy of
visual observation of wrist flexion/extension and radial/ulnar deviation, with a very
similar number of joint angle categories. The difference between the studies is that
Spielholz et al. (2001) used video observations, while Ketola et al. (2001) used live
observations. For both studies only poor correlation or agreement between observed and
measured wrist joint angles was achieved. Spielholz et al. (2001) reported an average
correlation coefficient for the wrist of 0.26, and Ketola et al. (2001) reported an average
kappa value of 0.20. Thus, from this single comparison it could be suggested that the
accuracy of visual observation of postures is similar whether observations are made via
video recordings or live observations.
Most of the accuracy studies conducted in clinical practice to date have analysed ratings
based on observations of subjects on video (Patla and Clouse, 1988; Harrison et al., 1994;
Bernhardt et al., 1998; Hillman et al., 1998; Haupstein and Goldie, 2000; Piripis et al.,
2001; Bernhardt et al., 2002; McGinley et al., 2003; Mackey et al., 2003; Read et al.,
2003; Stott et al., 2005; Wren et al., 2005; Dickens and Smith, 2006; Kawamura et al.,
2007; Brown et al., 2008; von Porat et al., 2008; Moseley et al., 2008). Only a few
studies have investigated the accuracy of live observations (McGinley et al., 2006;
Passier et al., 2010). Given the scarcity of studies that have investigated accuracy of live
observations, and the diversity of protocols and assessment scales used, it is difficult to
make many direct comparisons across live and video-based studies. Thus, only a few
studies are helpful in making comparisons across these modes of observation. McGinley
et al. (2003) asked experienced physiotherapists to watch video recordings of post-stroke
hemiplegics and rate their level of ankle push-off during terminal stance of gait. The
- 100 -
observers ratings were correlated with ankle kinetic data (peak ankle power) obtained
from 3D motion analysis. High correlations between the observational ratings and the
criterion measure were reported (mean Pearson r = 0.84), representing fair to excellent
values. McGinley et al. (2006) also investigated the accuracy of observation of ankle
push-off in post stoke hemiplegics, however in this study raters viewed the subjects under
live real-time conditions, rather than observations of videos. The same protocol, rating
scale and criterion-related measure were used as in McGinley et al. (2003). A strong
positive linear relationship between the observed data and 3D kinetic data (mean Pearson
r = 0.91) was also obtained for the live observations. Comparison of validity results
obtained from McGinley et al. (2003) with those from McGinley et al. (2006)
demonstrates that under comparable conditions, the accuracy of rating ankle push-off in
terminal stance is similar for videotaped and live observations.
Wren et al. (2005) appears to be the only other research group that has investigated and
compared accuracy results for visual gait observations obtained from live and videobased observation techniques. Results showed that accuracy of joint angle estimates was
similar for ratings made from normal speed videos and live observation. An additional
element of this study was to compare the effects of video playback speed on accuracy of
observation. For items where accuracy was especially poor at normal speed (e.g. ankle
dorsiflexion of less than 5), greater accuracy was obtained when viewing the slow speed
video. Moseley et al. (2008) also investigated the effects of video play back speed on
accuracy of qualitative analysis. For this study of the accuracy of observation of the ankle
joint movement of able bodied subjects during stair descent, results also indicated that
greater accuracy of observation was obtained when viewing the videos in slow motion
(mean r = 0.40) compared to normal speed (mean r = 0.30).
In relation to investigation of accuracy of qualitative analysis in sporting activities, most

studies (Imwold and Hoffman, 1983; Knudson, 1999; Nicholls et al., 2003; Morrison et
al., 2005; Plessner and Schallies, 2005; Krosshaug et al., 2007; Marques-Bruna et al.,
2008; Dallas et al., 2011) have investigated accuracy of video analysis, while only one
study (Stensrud et al., 2011) looked into the accuracy of live observational analysis. It is
therefore difficult to make valid comparisons of the accuracy of live and video-based
qualitative analysis in sport.
- 101 -
Given such wide use of video based assessments in qualitative analysis, it is important to
understand the effects that the position of the camera in relation to the joint angles of
interest have on the accuracy of visual observation. Some very recent studies in
workplace posture analysis (Lau and Armstrong, 2011; Xu et al., 2011) have attempted to
study this.
Lau and Armstrong (2011) aimed to investigate whether wrist flexion/extension and
radial/ulnar deviation postures were estimated differently when different viewing angles
were used. A single digital video camera was used and placed in three different camera
angle positions, (1) the ideal camera angle, placed orthogonal to the plane of motion, (2)
the in-line view, placed directly in the plane of joint motion and through the axis of
rotation, and (3) the off-axis view, placed neither orthogonal nor in-line with the plane
of motion of the observed joint.
For measures of accuracy, observation data were
compared with joint angle measurements obtained using a plexiglass calibration board.
Results showed that the viewing angle of the camera did affect the accuracy of visual
observations. Observations from the orthogonal camera angle produced the most accurate
observations, while the in-plane views produced the least accurate observations. The
authors therefore confirmed that the orthogonal view is preferable and in-plane views
should be avoided, however, if possible, use of multiple cameras for video-based
observation methods is preferred. Xu et al. (2011) also provided evidence to indicate that
the viewing angle of the camera does affect accuracy of visual observation of video-based
posture assessment tools. In this study one digital video camera was placed such that a
side-view of the subject was obtained. Participating raters were asked to observe trunk
flexion/extension and side flexion, as well as flexion/extension of the shoulder, elbow,
knee and ankle joints. Visual observation data were compared with 3D motion analysis
data to obtain a measure of observation accuracy. The estimated joint angles were
strongly correlated with the measured segment angles for all the observed angles (r >
0.8), except the trunk side flexion angle (r = 0.21), the only variable for which the camera
was not placed orthogonal to the plane of motion. In agreement with Lau and Armstrong
(2011), Xu et al. (2011) also suggested that use of multiple cameras placed at different
camera angles would optimise accuracy of visual observation data obtained using videobased posture assessment tools.
- 102 -
The effects of video camera angle on the accuracy of qualitative analysis has also been
studied in relation to sporting activities, in particular to judging gymnastics (Plessner and
Schallies, 2005; Dallas et al., 2011). Plessner and Schallies (2005) examined the
influence of camera angle on the accuracy of judgements of the angle of deviation from
the horizontal of the abducted arms for male gymnasts performing the cross on the
rings. Results showed that even experienced judges were significantly influenced by their
viewing position, with their error rate increasing with increasing deviation of the camera
angle from a frontal view. Dallas et al. (2011) also asked judges to estimate the degree of
deviation from the horizontal of the abducted arms, however it was for the inverted
cross. Three different camera positions were used to study the accuracy of observation,
and in agreement with Plessner and Schallies (2005), results indicated that perceptual
errors in the judges evaluations were systematic with regard to angle of observation, with
unacceptable accuracy being obtained when using an observation angle of 45 to the
frontal plane.
Thus, review of the studies that have investigated the effects of the mode of observation
on the accuracy of qualitative analysis of human movement, reveals little convincing
evidence to suggest there are significant differences in accuracy when observation is
conducted via live or video-based methods. However, if video-based methods are used,
results indicate that better accuracy of observation is obtained using slow playback speed,
rather than normal playback speed. Furthermore, there is strong evidence to suggest that
the viewing angle does have an effect on accuracy of observation, indicating that the
position of the camera or live viewer are of utmost importance in obtaining accurate
evaluations.
2.3.4.2 Effects of Scale Type on Accuracy of Qualitative Analysis

There is much diversity in the type of assessment scales or techniques used in qualitative
analysis of human movement. When considering all the studies identified from the fields
of clinical practice, occupational health and safety (OHS), and sport that have
investigated the accuracy of qualitative analysis, the majority of studies (65%) have used
nominal or categorical scales. Interval scales, visual analogue scales (VAS) and specific
numerical estimates have also been used, however much less frequently. Figure 2.30
- 103 -
displays the percentages of the studies that have used each of these different assessment
scale types or techniques.
In terms of the use of particular scales in studies conducted in each setting or field, all
scales types were used in studies conducted in clinical practice, while only categorical
scales and numerical estimates were used in OHS studies, and all scales except interval
scales were used in studies of accuracy of qualitative analysis in sport. Table 2.3
summarises the types of scales studied in each of the different fields.
VAS
11%
Numerical
estimate
13%
Interval
scale
11%
Categorical
scale
65%
Figure 2.30 Summary of percentages on studies that have used

categorical scales, interval scales, visual analogue scales (VAS), or
numerical estimates for qualitative analysis of human movement.
Clinical
OHS
Sport
Categorical scale
10
15
Interval scale
Visual Analogue Scale
Numerical estimate
Total
19
18
Table 2.3 Summary of the numbers of studies that have used categorical
scales, interval scales, visual analogue scales (VAS), or numerical
estimates for qualitative analysis of human movement in various fields.
- 104 -
It is possible that the type of scale used in a visual observation tool may affect the
accuracy of the observations obtained. When choosing an appropriate method for visual
observation of human movement, it is therefore important to consider the type of data
recorded and which type of scale or data recording technique can capture this information
most accurately. Thus, review of results obtained from studies using different rating
scales may offer some insight into how the scale or number of categories used may affect
accuracy.
In relation to nominal or categorical scales, it is evident that various numbers of

categories are typically used, ranging from binary nominal scales through to scales with
as many as nine categories. Central to the observational methods that use joint angle
categories to record postural information, is the effect that the number or size of
categories has on the accuracy of the observation data recorded. A variety of studies have
investigated the effects on accuracy of the number of categories in nominal scales
(Leskinen et al., 1997; Piripis et al., 2001; Lowe, 2004a; Lowe, 2004b; Village et al.,
2009; van Wyk et al., 2009).
In the clinical setting, Piripis et al. (2001) and Lowe and colleagues (2004a; 2004b) have
investigated the question of how the number of posture categories may affect accuracy.
Piripis et al. (2001) studied the accuracy of observational gait analysis using the Hugh
Williamson Gait Analysis Scale (HWGAS). It was reported that accuracy of observation
of foot position during stance was better when using a 5 category scale (Po = 0.67)
compared to a 7 category scale (Po = 0.46). Lowe et al. (2004a; 2004b) studied the effects
of the number of posture categories on the accuracy of visual observation of shoulder,
elbow, forearm and wrist postures. All joint postures were assessed with a 3-category
scale and a 6-category scale. Observations of video data were compared with data from
electrogoniometer measurement, and results indicated that, with the exception of shoulder
elevation, accuracy of visual observation was better when using a 3-category scale (r =
0.44 0.67), compared to the 6-category scale (r = 0.07 0.27).
The effects of the types of labelling included or omitted from scales has also proven to be
influential in clinical practice. Accuracy of observational gait analysis has been studied
using various categorical ratings scales (Mackey et al., 2003; Wren et al., 2005; Dickens
- 105 -
and Smith, 2006; Kawamura et al., 2007; Brown et al., 2008). For each of these scales the
observers gave their qualitative response by selecting one of a small number of joint
range categories. In modified versions of the Physicians Rating Scale (PRS), used by
Mackey et al. (2003), Dickens and Smith (2006) and Brown et al. (2008), as well as
providing joint angle ranges the category definitions also contained labels indicating
which was the normal range and how the other ranges deviated from normal (e.g.
severe flexion, mild flexion). However, in Wren et al. (2003), joint angle categories
were given without any indication as to how each category related to normal. In contrast,
Kawamura et al. (2007) used a rating scale in which observers were not provided with
any specific joint angle categories, but instead were only provided with references to
normality. On comparison of the results of Mackey et al. (2003), Read et al. (2003),
Wren et al. (2005), Dickens and Smith (2006), Kawamura et al. (2007) and Brown et al.
(2008), it appears that greater accuracy was obtained for observational gait analysis when
raters were provided with rating scales that contained specific joint angle categories as
well as labels to indicate the level of normality associated with each joint range category.
Neither the joint range categories alone, nor the references to normality alone seemed to
assist raters in achieving accurate results.
Overall, in considering data from a variety of studies that have investigated the effects of
different scale types, it appears that better accuracy of qualitative analysis is obtained
when fewer categories are used, and when categories are adequately labelled with
appropriate references to normality or levels of competency.
2.3.4.3 Effects of Number of Parameters Observed on Accuracy of Qualitative Analysis

The number of posture or movement parameters that raters are asked to observe at one
time may also have an effect on the accuracy of visual observations of human movement.
A few studies have attempted to investigate the effect of the total number of parameters
observed on the accuracy of postural assessment in the workplace. Leskinen et al. (1997)
conducted a study in which visual observations of neck and trunk flexion of workers
performing manual handling tasks were compared with 3D motion analysis data. A total
of six different parameters were observed in the study, however four different trial types
were conducted, consisting of different combinations of parameters. In one trial type,
subjects were asked to observe all six parameters. In addition to neck and trunk flexion,
- 106 -
these parameters included the size of the load lifted, the type of lift performed, whether
the hands were elevated above or below the shoulder, and whether the subject was
kneeling or squatting. In two trial types the number of parameters observed was limited to
four, and in one trial type subjects were asked to observe only two parameters. Results
showed that percentage agreement between visual observation data and 3D motion
analysis data for neck flexion was better when only two parameters were observed (Po =
0.53) than when six parameters were observed (Po = 0.39). However for trunk flexion, the
effect of the number of observed variables was not so clear. For the two trial types that
contained four observation parameters, proportion of agreement for trunk flexion was
0.59 and 0.68, while for the trial type containing six parameters, proportion of agreement
was 0.61. Perhaps these data suggest that once a certain number of observation
parameters are added to an observation tool, there is limited effect on the accuracy of
observation, but keeping the number of observation parameters very small (i.e. only two),
does enable improved accuracy of visual observation.
Paquett et al. (2001) investigated the validity of visual observations when using the
Posture, Activities, Tools and Handling (PATH) tool (Buchholz et al., 1996), and also a
simplified version of PATH. For the full PATH method, raters were required to record
nine different parameters, but for the simplified version of PATH, the observers used the
same template but were required to record data for only three postural parameters. A
measure of accuracy of the visual observation data for the tools was obtained by
comparing it with inclinometer and electrogoniometer data. In general, agreement with
the reference measurements was slightly higher for the simplified PATH tool (k = 0.60 0.87) than with the full PATH method (k = 0.51 - 0.74). However, it should be noted that
each observer was assigned to perform either the simplified PATH or full PATH for the
duration of the study. It is therefore possible that the differences found may be due to
inter-observer differences rather than to the decreased cognitive demands of the
simplified PATH method. Nevertheless, the authors were confident in suggesting that the
accuracy of posture observation may improve when the observer is required to code
fewer parameters.
In sports analysis, the effect of the number of parameters observed on accuracy of

observation has been studied in gymnastics (Imwold and Hoffman, 1983; Plessner and
- 107 -
Schallies, 2005). Imwold and Hoffman (1983) asked raters to observe either two, three or
four different parameters related to the running front handspring. Decreasing accuracy
was obtained with increasing number of observation targets. Plessner and Schallies
(2005) study required both experienced and inexperienced gymnastics judges to estimate
arm angles during performance of the cross. In some trials, estimation of arms angles
was the only observation required, while in others, observers were also required to
estimate the duration that the position was held. Data indicate that there was no
statistically significant main effect of the number of tasks on the accuracy of observation.
Thus, data suggests that when asked to estimate more than two parameters, accuracy of
observation decreases as the number of observation targets increases, but there is little
difference in accuracy of observation between one and two observation targets.
2.3.4.4 Effects of Body Region Observed on Accuracy of Qualitative Analysis

A variety of studies have reported data to suggest that the level of accuracy of
observation of human movement is dependent upon which joints or body regions are
being observed. In relation to workplace posture, Baluyut et al. (1995) compared videobased visual observation data with goniometric measurements. Data for the percentage of
correct responses (Pc) indicate that it was harder for subjects to accurately evaluate
postures of the distal joints, such as the elbow (Pc = 53 82%) and wrist (Pc = 65 80%),
than for the shoulder (Pc = 54 92%), neck (Pc = 76 95%) and lower back (Pc = 71
98%). Lowe and colleagues (2004a; 2004b) also reported that estimates of shoulder
elevation (r = 0.67) were generally more accurate than estimates of the more distal
movements of elbow flexion (r = 0.45), forearm supination (r = 0.46), and wrist flexion (r
= 0.44). These data were obtained by comparison of visual observation data with
electrogoniometer data. A few accuracy studies (Spielholz et al., 2001; Ketola et al.,
2001) have focused primarily on observation of the distal upper limb joints. Spielholz et
al. (2001) compared wrist and forearm posture estimates obtained from observation of
video images with data from electrogoniometric measurements. Pearsons correlation
coefficients were low for all variables, but were worse for forearm pronation/supination (r
= 0.07), compared to wrist flexion/extension (r = 0.33) and wrist radial/ulnar deviation (r
= 0.21). Ketola et al. (2001) also investigated the validity of visual observation of wrist
postures by comparison with electrogoniometric measurements. Results were consistent
- 108 -
with those of Spielholz et al. (2001) with only poor to fair agreement obtained for all
wrist postures (k = 0.04 0.38).
Other workplace studies (Burdorf et al., 1992; Leskinen et al., 1997; Paquet et al., 2001;
Neumann et al., 2001; Xu et al., 2011) have looked mainly at the accuracy of observation
of the trunk, neck and shoulder joints. Burdorf et al. (1992) compared visual observations
of trunk bending with inclinometric measurements, and calculated moderate
correlations between the two methods for sedentary (r = 0.62 and dynamic (r = 0.57)
tasks. Leskinen et al. (1995) reported that estimations of trunk flexion angles (Po = 0.63)
were more accurate than estimations of neck flexion angles (Po = 0.46), as judged by
comparison of visual observation data with 3D motion analysis data obtained by Selspot
II equipment. Paquet et al. (2001) obtained higher accuracy for visual observation of
shoulder (k = 0.75 0.80) postures compared to trunk postures (k = 0.51 0.60). These
data were obtained by comparison of live visual observation data with electrogoniometer
and inclinometer data. Very recently, Xu et al. (2011) also investigated the accuracy of
visual observation of trunk and shoulder angles, however in this study accuracy results
were similar for observation of these different body regions. By comparing visual
observation data with 3D motion analysis data, Pearsons correlation coefficients (r >
0.83) indicated good to excellent validity for observations of sagittal plane trunk and
shoulder postures. Validity of observation of trunk flexion/extension was studied by
Neumann et al. (2001) through comparison of observation of static video images with 3D
motion analysis data. The observational data showed very high correlation with data from
the 3D motion analysis system (r = 0.92). Collectively, the results of studies investigating
the validity of visual observation of work task postures indicate that accuracy of
observation is weakest for distal upper limb joints, such as the wrist, forearm and elbow,
and is most accurate for observation of trunk or lower back postures. Results have varied
with respect to the relative accuracy of shoulder, trunk and neck observations.
Studies have also been conducted in clinical practice in relation to the accuracy of
observations of movements at different joints of the body. Read et al. (2003) reported
that ankle dorsiflexion in swing showed the best percentage agreement (83%) between
the visual score and the quantitative motion analysis data. The proximal recordings of
peak hip extension in stance, peak hip flexion in swing, and pelvic rotation in mid-stance
- 109 -
were also all well correlated (75%, 68%, and 69%, respectively), however the knee
observations in the sagittal plane showed lower correlations of 60% for peak knee flexion
in swing, 55% for knee extension in terminal swing, and 47% for peak knee extension in
stance. These results indicate that it was most difficult for observers to estimate joint
angle magnitude at knee level during gait. Mackey et al. (2003), showed a moderate to
substantial level of agreement between data from 3D gait analysis and observational data
for knee position in mid-stance (average k = 0.57), initial foot contact (average k = 0.59),
foot position in mid-stance (average k = 0.64), and timing of heel rise (average k = 0.65).
Like Read et al. (2003), Mackey et al. (2003), also reported higher correlations for the
items relating to foot position than for knee position. However, both Wren et al. (2005)
and Kawamura et al. (2007), reported that contrary to previous studies (Read et al., 2003;
Mackey et al., 2003), accuracy was greatest for knee position and lowest for hip and
ankle positions. In Wren et al. (2005), accuracy was particularly poor when ankle
dorsiflexion was less than 5. Thus, although overall accuracy was better for the knee
than for the ankle or hip, the level of accuracy did depend on the degree of knee flexion
demonstrated by the subjects. For the knee, raters assigned appropriate ratings in 73%83% of cases when flexion was greater than 20, in 61%-75% of cases between 5 and 20
of flexion, and only 11% - 20% when the knee was close to neutral. That is, raters had
more difficulty categorising knee position when it was close to neutral.
A few workplace studies (Baluyut et al., 1995; Yen and Radwin, 2000; Village et al.,
2009; Teschke et al., 2009) have investigated accuracy of observation of neutral versus
non-neutral joint postures. Results from a study by Baluyut et al. (1995), in which visual
estimates of joint angles were compared with goniometric measurements, indicate that
joint angle observations are more accurate for larger deviations from the neutral position.
This was found to be the case for trunk flexion, shoulder flexion, wrist flexion and elbow
flexion especially. Percentage correct response data for neutral and extreme joint
positions were 71% and 98% for trunk flexion, 54% and 77% for shoulder elevation, 55%
and 82% for elbow flexion, and 65% and 85% for wrist flexion. Yen and Radwin (2000)
compared visual observation data with data obtained from electrogoniometers to assess
the criterion-related validity of visual observations of work task postures of the shoulder,
elbow, forearm and wrist. Subjects observed static frames of video recordings of workers
performing six industrial jobs. Overall, the Pearson correlation coefficient for comparison
- 110 -
between visual observation and electrogoniometric measurement was 0.75. When

considering only larger joint motions that exceeded the posture classification angle
resolution, the correlation between postural classification and electrogoniometer data
increased to 0.82. Village et al. (2009) also presented data to indicate that visual
observation is more accurate the further the joint postures deviate from the neutral
position. By comparing data obtained from observation of static video images with data
from an inclinometer, these authors found that Pearson correlation coefficients for
validity of trunk flexion observations increased as the range of motion increased (0 - 10
flexion, r = 0.13; 10 - 20 flexion, r = 0.16; 20 - 45 flexion, r = 0.19; 45 - 60 flexion, r
= 0.42; > 60 flexion, r = 0.90). Teschke et al. (2009) also reported that accuracy of
visual observation of trunk flexion increased with increasing displacement from the
neutral position. This study also compared observation data with inclinometer data, and
correlation coefficients increased as flexion angle categories increased from 10 - 20 (r =
-0.052), to 20 - 45 (r = 0.12) to >60 (r = 0.90).
With regards to sporting activities, for mens gymnastics Plessner and Schallies (2005)
and Dallas et al. (2011) both investigated the accuracy of observation of the deviation of
the arm position from the horizontal during performance of the upright (Plessner and
Schallies, 2005) and inverted (Dallas et al., 2011) cross. Plessner and Schallies (2005)
reported similar levels of accuracy for arm positions with true deviations of 0 - 15,16 30, 31 - 45 and <45 from the horizontal, however Dallas et al. (2011) found that the
degree of angular deviation did affect accuracy. Dallas et al. (2011) reported that
percentage accuracy was best when the true angular deviation was 31 - 45 (74%),
followed by 16 - 30 (65%) and least accurate for 1 - 15 deviation (56%).
Overall, data from studies investigating workplace activities indicate that distal joint
movements of the upper limb are more difficult to accurately observe than proximal joint
movements, however in clinically related studies of the lower limb, results to not provide
strong evidence to suggest that a particular joint or region of the lower limb is usually
more accurately observed than another. There is however, convincing evidence to
indicate that the degree of deviation of a joint from neutral does affect accuracy of
observation, with accuracy increasing with increasing deviation from neutral.
- 111 -
2.3.4.5 Effects of Observer Experience on Accuracy of Qualitative Analysis

The effect of observer experience on accuracy of qualitative analysis has also received
some attention in the academic literature. A few authors (Bernhardt et al., 2002;
McGinley et al., 2003; Brown et al., 2008) have investigated this relationship with
respect to clinical assessment tools. Bernhardt et al. (2002) investigated the influence of
clinical experience on the accuracy of observational kinematic assessment of upper limb
function and found that all groups of observers (highly experienced, less experienced
and novice) made moderate to highly accurate judgments with no significant
differences between groups. Similarly, McGinley et al. (2003) found no evidence of a
relationship between clinical experience and accuracy of observational analysis in their
investigation of the accuracy of assessment of ankle push-off during terminal stance in
gait. Most recently, Brown et al. (2008) reported that in the hands of experienced
observers observational gait analysis showed better percentage agreement with 3D
motion analysis data (57.75%) than when assessed by inexperienced observers (49.5%).
However, the difference between the two groups was only 8.25%. Having had no
previous experience in observational gait analysis, the inexperienced observers improved
their accuracy in a second observation in all categories except for knee peak flexion in
swing. In contrast, Moseley et al. (2008) did report a difference in accuracy of
observation due to the level of experience of the observer. In this study, in which
participants observed the ankle joint movement of able bodied subjects during stair
descent, results indicated that greater accuracy of observation was achieved by
neurological physiotherapists who had experience in observational gait analysis
(Pearsons r = 0.39) compared to cardiothoracic physiotherapists (Pearsons r = 0.26) and
undergraduate physiotherapy students (Pearsons r = 0.23) who had little or no experience
in observational gait analysis.
As was the case for the clinical setting, only a few studies (Lowe, 2004a; Andrews et al.,
2008) were identified that specifically investigated the effect of observer experience on
accuracy of visual observations of work-related postures. Lowe et al. (2004a) reported
that analysts self-reported years of experience in the field of ergonomics exhibited no
meaningful correlation with posture estimation accuracy for the elbow and shoulder.
Andrews et al. (2008) examined the effect of rater experience or expertise level on the
effect of classification of trunk postures by visual observation. Three groups of
- 112 -
participants were recruited, two novice groups and one expert group. The difference
between the novice groups was that one group had completed courses in functional
human anatomy, while the other group had not. Trunk flexion/extension and lateral side
flexion postures were classified by observation of static video frames. To determine the
level of accuracy of posture classifications, observation data were compared with the
actual joint angles, as measured by 3D motion analysis (IsotrakTM). The authors
hypothesised that more posture classification errors would be made by novice operators
compared to experts. Results showed that there were significant differences between
groups in the percentage of misclassification errors (p < 0.001) for trunk
flexion/extension. Both novice groups obtained greater percentages of misclassification
than the expert group, but interestingly, it was the Novice 2 group (those who had
completed an anatomy course) rather than the Novice 1 group (those with no anatomy
training) who obtained the highest percentage of misclassification of all groups. For
observation of lateral trunk flexion all groups had similar percentages of misclassification
errors. Results of the few studies that have investigated the effects of rater experience on
visual observation of work task postures are mixed and therefore appear to be
inconclusive. Lowe et al. (2004a) reported that experience had no effect on accuracy of
visual observations, while data from Andrews et al. (2008) suggest that the effect of rater
experience may vary, depending upon which joints of planes of motion are being
observed.
In relation to sporting activities, only three articles were found that specifically looked at
the effect of experience on accuracy of qualitative analysis (Imwold and Hoffman, 1983;
Knudson, 1999; Plessner and Schallies, 2005). Results of these studies have varied.
Imwold and Hoffman (1983) reported that experience does improve the observational
analysis of a gymnastics skill, the running front handspring, with specialist gymnastics
coaches only achieving 54% accuracy, compared to 46% accuracy by generalist physical
education teachers and undergraduate physical education students. Also in relation to
assessment in gymnastics, Plessner and Schallies (2005) found that experienced judges
(mean experience = 15.4 years) were more accurate than novices in their estimates of
arms angles when observing male gymnasts perform the cross. In contrast, Knudson
(1999) revealed that 60% of students were effective in rating joint range of motion during
the vertical jump, compared to only 20% of highly experienced professionals. So, it
- 113 -
seems that collectively, studies conducted in multiple fields have not provided convincing
data to definitively indicate that experience is or is not advantageous for increasing the
accuracy of qualitative analysis of human movement.
Table 2.4 summarises the conclusions drawn from the literature review on the criterionrelated validity or accuracy of qualitative analysis of human movement.
Factor
Mode of observation
Conclusions from Review of Literature

Data are inconclusive as to whether better accuracy is achieved
when using live or video-based observation techniques.
Video playback speed
Accuracy is improved when slower video playback speeds are used.
Camera or viewing angle
Camera angle or position of the observer in relation to the plane of

joint movement significantly affects accuracy of observation.
Better accuracy is obtained when the camera angle and plane of
movement are orthogonal to each other.
Number of scale categories
Better accuracy is obtained when fewer scale categories are used.

Better accuracy is obtained when categories are adequately
labelled, for example with appropriate references to normality or
levels of competency.
Number of parameters observed
Accuracy decreases as the number of simultaneous observation

targets increases.
Body region observed
Distal upper limb joint movements are more difficult to accurately

observe than proximal upper limb joint movements.
Data do not strongly suggest that a particular joint or region of the
lower limb is more accurately observed than another.
Accuracy increases with increasing deviation of a joint angle from
its neutral position.
Observer experience
Results are inconclusive as to the effect of observer experience on

accuracy of qualitative analysis.
Table 2.4 Summary of conclusions from literature review of accuracy of qualitative analysis of human
movement.
- 114 -
2.4 Summary of Literature Review and Project Rationale

2.4.1 Summary of Literature Review
This chapter has presented a review of the literature on: the development of classical
ballet technique; the fundamental biomechanical principles that characterise classical
ballet technique; instrumented quantitative biomechanical analysis of theatrical dance
styles; and the characteristics, application, and accuracy of qualitative analysis of human
movement. The following is a summary of the key findings:
x The fundamental principles of classical ballet technique have their origin in
historical context and have evolved over a period of nearly five centuries. Since its
inception in 15th century Europe, development and refinement of classical ballet
technique occurred in response to stylistic and functional necessity.
x According to the historic and current dance literature, professional ballet dancers
must possess: the technical attributes of grand manners; codified foot positions;
extreme turnout of the lower limb; elevation and pointe work; and extreme
flexibility displayed through maximum extension of the lower limbs. These
characteristics can be summarised into four fundamental principles of classical
ballet technique.
x The four fundamental principles of classical ballet technique, and their defining
characteristics were identified as:
x Quantitative instrumented analysis of dance movement is a relatively new field of
academic research. The vast majority of research has occurred from 2000 onwards,
with a considerable increase in publications since 2005. Use of 3D motion analysis
of dance has increased substantially since 2005.
- 115 -
x The majority of quantitative dance analysis studies have investigated the dance
genre of classical ballet. However, of these, only 11 published studies were
identified that used 3D motion analysis to report kinematic data for classical ballet.
Amongst these 11 studies, no single study used 3D motion analysis to analyse
classical ballet technique across multiple steps in terms of a set of fundamental
biomechanical principles of the technique.
x A quantitative tool, based on 3D motion analysis, for assessment of classical ballet
technique was not identified in the academic literature.
x Qualitative analysis is a form of movement assessment typically applied in clinical
practice, in the fields of occupational health and safety and ergonomics, in sports
analysis and coaching, and in dance assessment and instruction.
x Studies investigating the accuracy of qualitative analysis of human movement have
been conducted in the clinical setting, in relation to workplace tasks, and for some
sporting activities. Studies on the accuracy of qualitative analysis of dance
movements were not identified in the academic literature.
x A number of factors affect the accuracy of qualitative analysis of human movement.
These include: camera or viewing angle; video playback speed; number of scale
categories; number of parameters observed; and body region observed.
x Data are inconclusive as to the effect of mode of observation (i.e. video vs. live),
and observer experience on the accuracy of qualitative analysis of human
movement.
2.4.2 Project Rationale
It is evident from the literature that the requirements for execution of correct classical
ballet technique can be succinctly defined by a few key theoretical principles. These
theoretical principles can be translated into four fundamental biomechanical principles,
each of which can be defined in terms of specific kinematic variables. The number of
scientific studies involving 3D quantitative motion analysis of classical ballet technique
- 116 -
has steadily increased since 2005, however no studies were identified that used 3D
motion analysis to assess the technique in relation to a number of fundamental principles
across multiple ballet steps. Moreover, the existing published literature has not attempted
to compare professional level practical execution of classical ballet technique with the
key theoretical principles outlined in the dance-related literature. Thus, the current study
will redress this anomaly in the scientific literature. To do this, 3D motion analysis will
be conducted on professional and non-professional ballet dancers performing seven
categories of ballet steps. The aim will be to compare the practical kinematic data relating
to each of the fundamental biomechanical principles of the classical ballet technique with
the theoretical ideals presented in the literature. It is proposed that the findings of such a
study could have implications for the teaching practices currently used in instruction of
Review of the literature also highlights that qualitative analysis of human movement is
used routinely in clinical practice, occupational health and safety, and for instruction of
sport and dance, and is therefore a very important skill that should be performed with a
high degree of accuracy. Studies investigating the accuracy of qualitative analysis were
identified in relation to clinical practice, occupational health and safety, and sport, but not
for qualitative analysis of dance. This finding thus identifies another area in which
scientific investigation is warranted, but is currently lacking. An additional aim of this
research project therefore is to investigate the accuracy with which ballet teachers are
able to use qualitative analysis to identify correct and incorrect classical ballet
technique, and to investigate the effect of teaching experience on this skill. The results
obtained can provide important information to assist with the development of curricula
and teaching programs for dance teachers.
Finally, although use of 3D motion analysis has increased in dance science research over
recent years, it is not currently used as a tool for routine assessment of dance technique.
Such a tool is therefore proposed as a useful method to systematically and accurately
assess classical ballet technique. The final analysis section of this thesis therefore
presents the methods used, results obtained and conclusions reached during development
of a 3D quantitative assessment tool for evaluation of classical ballet technique.
- 117 -
Chapter 3 Comparison of Theoretical and Practical Execution of Classical Ballet Technique
CHAPTER 3
COMPARISON OF THEORETICAL AND

PRACTICAL EXECUTION OF CLASSICAL
BALLET TECHNIQUE
- 118 -
CHAPTER 3
COMPARISON OF THEORETICAL AND PRACTICAL
EXECUTION OF CLASSICAL BALLET TECHNIQUE
3.1 Introduction
Classical ballet has been studied and performed throughout the world for many centuries.
Its earliest forms were performed in the 1500s as entertainment for the nobility of the
French royal courts. Over time, specific principles, guidelines and steps of the classical
ballet repertoire were established. While different regions/countries have developed and
established their own styles and characteristics (e.g. French, Russian, Italian, Danish),
there are some basic principles of the classical ballet technique that have been adopted
universally. As described in Chapter 2, these include principles such: as maintaining the
verticality of the torso and minimising the natural curvatures of the spine during
execution of ballet movements; maintaining a centred pelvic position and minimising
pelvic displacement when executing movements of the lower limb; achieving maximum
turnout (i.e. external rotation) of the lower limbs; and achieving maximal extension (i.e.
knee extension and ankle plantarflexion) of the lower limbs.
Through the process of conducting the literature review, the above mentioned principles
were categorised into four main fundamental principles, which have been defined as
follows:
3) Turnout - maximum external rotation of the lower limbs
This section of the study aims to use quantitative biomechanics techniques to investigate
the level of agreement between what is considered to be theoretically correct execution of
ballet technique as outlined in the dance literature, and the practical execution of classical
ballet technique demonstrated by elite level ballet dancers in professional companies.
- 119 -
That is, the study aims to examine empirically if professional classical ballet dancers
actually do execute classical ballet technique more closely to the theoretical principles
outlined in the texts than non-professional dancers. To achieve this aim, a set of standard
classical ballet steps will be selected for 3D motion analysis. A sample of professional
ballet dancers (N=15) and a sample of non-professional ballet dancers (N=15) will be
recruited. They will be asked to perform the basic classical ballet movements while
undergoing full body 3D motion analysis and video analysis.
Data analysis will involve assessing the kinematic differences between the two groups of
dancers in their execution of the steps. For example, thorax verticality, pelvic thoracic
rotation, pelvic tilt, pelvic obliquity, external hip rotation, knee extension, and maximum
ankle plantarflexion will be compared to determine whether or not on average
professional dancers execute classical ballet technique that is more like the theoretically
correct characteristics than non-professional dancers.
It is intended that the findings of this study will inform the dance community about the
level of consistency between the theoretical principles of classical ballet technique, and
the practical execution of classical ballet technique as demonstrated by elite level ballet
dancers in professional companies. Such information is relevant to the dance community,
to ballet teachers in particular, as it will inform them of the degree to which previously
established principles of ballet technique are actually executed at an elite level, and may
therefore provide insight into ways in which teaching methods or instructional cues can
be modified or enhanced for optimal teaching practice.
- 120 -
3.2 Methods
3.2.1 Participant Recruitment
Ethics approval was obtained for this study through the University of New South Wales
Human Research Ethics Advisory (HREA) Panel (Appendix A). The target number of
subjects for recruitment to the study was 15 non-professional ballet dancers and 15
professional ballet dancers. The sample sizes were not determined to provide a specific
power, rather, the sample sizes were determined in accordance with standard practice in
biomechanics (Mullineaux et al., 2001), and according to the practical constraints of
completing the planned program of research.
Criteria for inclusion as a non-professional ballet dancer were;

x
being at least 18 years of age at the time of data collection.
never having been employed as a ballet dancer in a professional ballet company.
having previously passed the Royal Academy of Dance (RAD) Intermediate

(formerly known as Elementary) examination, or equivalent level in an alternative
syllabus.
injury-free at time of data collection, and within the 6 months preceding the time
of data collection.
For recruitment of the non-professional dancers, dancers enrolled in the University of

New South Wales Bachelor of Arts (Dance)/Bachelor of Education program were
provided with information about the study through their attendance at group information
sessions on the university campus. During these sessions, potential participants were
informed of the inclusion criteria for participation in the non-professional group, and
were provided with information on what they would be required to do as study
participants and how their data would be used for the study. Each potential participant
was given an information and consent form during the information session.
Inclusion in the professional group was based on having achieved employment as a

dancer in a professional ballet company, indicating that through the company audition
- 121 -
process they had achieved an exemplary level of skill in classical ballet technique.
Specific criteria for inclusion as a professional ballet dancer were;
x
employed as a dancer in a professional ballet company at the time of data

collection, or within the 6 months preceding the time of data collection.
injury-free at time of testing, and within the 6 months preceding time of testing.
The professional dancers were recruited from various professional ballet companies and
classes within Australia. Methods for recruitment of professional ballet dancers included
posting online advertisements on dance-related websites (e.g. Ausdance website), phone
calls and emails to Australian dance companies (e.g. The Australian Ballet Company,
Mod Dance Company), and through visiting a number of dance studios in the Sydney
metropolitan area and speaking directly with dancers and dance teachers. Upon
identifying potentially suitable subjects for the professional group, each potential
participant was informed of the inclusion criteria for participation in the professional
group, and was provided with information on what they would be required to do as study
participants and how their data would be used for the study. Each potential participant
was given an information and consent form.
Non-professional and professional dancers who were willing to participate in the study
were contacted directly via email or phone to arrange the date and time of their testing
sessions. One 3 hour testing session was scheduled for each participating dancer at a time
of their convenience. A copy of the information and consent form was emailed to all
participating subjects at the time of scheduling their testing session. All subjects provided
written informed consent, in the form of a signed and dated consent form, prior to
commencing any data collection for the study. All dancers were offered a pair of dance
foot thongs (Bloch Foot Thong III S0675) for participation in the study.
3.2.2 Instrumentation
3.2.2.1 Motion Analysis System Description
3D kinematic data were collected with an 8-camera (M2Cam) motion analysis system
(Vicon 612, Oxford Metrics Ltd, Oxford, UK). VICON 612 is a 3D motion measurement
system comprising specialised hardware, and application software. The hardware
- 122 -
components of the system comprise two main parts: a datastation and a workstation. The
datastation contains the 8-input video converter (vi-con), a 64-channel analogue to
digital converter, timing and control circuits, and communication ports. The cameras and
analogue devices are connected to the datastation. The workstation is a personal computer
running Vicon Workstation Version 5.2.4.
Vicon 612 software is divided between the datastation and workstation. A single
interactive program (Vicon Workstation Version 5.2.4.) on the workstation controls
system calibration, data capture, 3D data reconstruction, and display of results in the
appropriate windows. The measured quantities are processed by the software, producing
an output file (C3D file), which contains 3D motion data ready for further processing
specific to the users requirements.
The system tracks the trajectories of a number of retro-reflective markers in the field of
multiple video cameras. The video cameras emit and then detect infra-red light reflected
back from the optical markers which are placed on specific anatomical locations on a
subject. 3D reconstruction of optical data starts from a set of marker images in the 2D
view of each camera. In order to generate 3D coordinates from 2D images the Vicon 612
system must first have reliable and accurate calibration parameters for each camera.
3.2.2.2 System Calibration.

The positions of all eight cameras were checked before each data collection session, and
were repositioned, if necessary, so as to give the best possible view of all optical markers.
Once all the cameras were positioned ready to perform a trial, they were calibrated to
determine their exact position and orientation with respect to a laboratory set of
coordinate axes. This process, known as photogrammetric calibration, was done by
collecting video data from each of two calibration reference objects, whose markers
were at known 3D positions. The system was calibrated using static and dynamic
calibration methods. The performance of the Vicon 612 system was dependent upon on
the accuracy with which this process was conducted.
Firstly, a static calibration was conducted using an L-Frame marker device, and then a
dynamic calibration was conducted using a calibration wand (240mm Space Bar Wand).
- 123 -
The L-Frame device, consisting of four retro-reflective markers attached at specific

distances from each other (Figure 3.1), was used to determine the location and orientation
of the laboratory coordinate axes and origin, and the direction of each camera relative to
the axes and origin. The L-Frame was placed in the same precisely known position for the
static calibration for each data collection session. Alignment plates attached to the LFrame device enabled accurate positioning of the L-frame for each static calibration. At
least 20 frames of data were captured for each static calibration with the L-Frame.
Figure 3.1 L-Frame device for static calibration
Following the static data capture, the L-Frame device was removed from the space and a
second longer period of dynamic data was captured. To obtain this dynamic calibration
data an operator moved around the entire measurement space, waving the calibration
wand (Figure 3.2) on which three markers were mounted at a known separation. The
accuracy of the calibration depended critically upon the distance between the two markers
on the calibration wand. This distance completely determined the size of the units of
distance. An error in this distance would have led to the units of distance (normally
millimetres) being incorrectly scaled.
- 124 -
Figure 3.2 240mm Space Bar Wand device for dynamic

calibration
At the end of the dynamic calibration, the calibration calculations were performed using
an optimisation process run by Vicon Workstation software. In this process, the direction
to the origin and the orientation of the vertical were determined for each camera using the
static calibration data. Then, using the dynamic calibration data, calculations of
increasing accuracy were made of the positions of the markers on the calibration wand,
while the camera locations and orientations were also calculated with increasing
accuracy, until the best fit to the data was obtained. During these calculations, the average
calibration residual was updated, continually reducing with each iteration. The residual is
the average distance by which each direction measurement, from the camera concerned,
deviates from the location of the markers used in the calibration. Calibrations for which
the residual values were below 1.3% of the reconstruction volume for each of the eight
cameras were considered to be acceptable. The Vicon 612 system stored the calibration
parameters for each calibration in a CP file, which was later used during the 3D
reconstruction of optical data.
3.2.2.4 Marker Set

A full body marker set comprised of 38 reflective, spherical markers was used to create a
15-segment model (Figures 3.3 & 3.4). The reflective markers had a diameter of 15mm
and were placed according to the Vicon Plug-in-Gait model. Markers were placed at the
following anatomical sites:
- 125 -
Right and left front of head (RFHD, LFHD)
Right and left back of head (RBHD, LBHD)
7th cervical vertebra (C7)
10th thoracic vertebra (T10)
Jugular notch between clavicles (CLAV)
Xiphoid process of sternum (STRN)
Right scapula (RBAK)
Right and left acromio-clavicular joint (RSHO, LSHO)
Right and left upper arm (RUPA, LUPA)
Right and left lateral epicondyle of elbow (RELB, LELB)
Right and left forearm (RFRA, LFRA)
Right and left styloid process of radius (RWRA, LWRA)
Right and left styloid process of ulna (RWRB, LWRB)
Right and left head of the second metacarpal (RFIN, LFIN)
Right and left anterior superior iliac spine (RASI, LASI)
In-between right and left posterior superior iliac spines of sacrum (SACR)
Right and left thigh (RTHI, LTHI)
Right and left lateral femoral epicondyle (RKNE, LKNE)
Right and left tibia (RTIB, LTIB)
Right and left lateral malleolus (RANK, LANK)
Right and left head of the second metatarsal (RTOE, LTOE)
Right and left calcaneus (RHEE, LHEE)
- 126 -
Figure 3.3 Vicon Plug-in-Gait reflective marker placement for full body analysis (from Vicon Plug-in-Gait
Manual).
- 127 -
Figure 3.4 Reflective marker placement on participant
Knee alignment devices (KADs) were used to assist with the knee marker placement. The
KAD (Figure 3.5) is a lightweight, spring-loaded G-clamp with adjustable jaws that span
the knee. It was used during static trials to determine the flexion/extension axis of the
knee joint, and thus to correctly place the knee markers (RKNE and LKNE). The stem is
aligned with the knee flexion/extension axis, one 15mm marker is attached to the tip of
this stem and two 15mm markers are attached to the ends of two additional rods fixed to
the device. The three markers are exactly equidistant from the point where the stem meets
the jaws of the clamp, allowing the 3D position of this point, known as the virtual knee
marker, to be calculated. The knee markers (RKNE and LKNE) were placed directly
over this point upon removal of the KADs.
- 128 -
Figure 3.5 Knee Alignment Device (KAD)
The KAD was also used to define the alignment of the ankle flexion/extension
(dorsiflexion/plantarflexion) axis. That is, during the static trial, the knee and ankle
flexion/extension axes are assumed to be parallel. If for any reason this assumption is
invalid, the Shank Rotation Offset can be manually reset, by entering a measurement of
the subjects tibial torsion into the model. Tibial torsion was therefore measured for each
subject and entered into the model for accurate calculation of the ankle flexion/extension
axis. Tibial torsion was defined as the angle between the bi-malleolar axis and the knee
flexion/extension axis projected onto the transverse plane (Vicon Plug-in gait Manual). It
was measured according to a method commonly used in clinical practice (Staheli and
Engel, 1972), whereby the patient is prone, their knee is flexed to 90, the ankle is held
neutral, a line is drawn on the sole of the foot between the malleoli, and a manual
goniometer is used to measure the angle between this line and the estimated knee
flexion/extension axis, assumed to be 90 to the long axis of the thigh.
- 129 -
3.2.2.5 Optical Data Capture and Reconstruction

All the optically derived kinematic data were sampled at 100 Hz. Kinematic data were
reconstructed using Vicon Workstation software (Oxford Metrics Ltd, Oxford, UK).
Reconstruction is the calculation of the 3D position of a marker and the linking of the
reconstructed points in each frame into trajectories. Reconstruction was conducted
according to the following process. Immediately after data capture, the Vicon
Workstation software automatically saved the marker coordinate data in a TVD file, a
binary file containing unprocessed data created whenever video data are captured. TVD
files, together with the associated CP file obtained during system calibration, provided the
input for 3D reconstruction which occurs through a process called direct linear
transformation (DLT). Reconstruction parameters were saved by Vicon Workstation in a
Capture, Analogue and Reconstruction (CAR) file, the parameters of which could be
edited if necessary to improve the reconstruction accuracy.
After the reconstruction, in order to uniquely identify the trajectories in the each frame,
all the trajectories were labelled, according to the abbreviations specified above in
Section 3.2.2.4. Once reconstruction and marker labelling was complete, the data were
saved as a C3D file, a binary file created whenever video data are reconstructed, labelled
and saved.
At the same time as recording 3D kinematic data, digital video data were obtained from
the coronal and sagittal view for all subjects for all steps performed using two Sony
digital camcorders.
3.2.2.6 Processing and Biomechanical Model

It was likely that the 3D kinematic data obtained contained imperfections that are
inherent in all 3D motion capture data. Therefore, interpolation and filtering processes
were applied to the data before any joint angle calculations were made. An interpolation
(i.e. gap filling)
procedure was conducted if gaps of up to 10 frames occurred for a
single marker trajectory. This process was achieved using cubic spline curve
interpolation, a form of interpolation where the interpolant is a special type of piecewise
polynomial called a spline (Wood & Jennings, 1979; Wood, 1982). The interpolation
- 130 -
procedure was run using the Vicon Workstation software (Additional information on the
cubic spline curve interpolation is presented in Appendix E).
Since non-filtered data can appear noisy, all data were filtered to provide smooth and
meaningful signals. Woltring's general cross-validatory (GCV) quintic spline routine
(Woltring, 1985; Woltring, 1986) was used for the filtering process. The GCV filter
offers the facility of selectively filtering the data signal more where it is required, and less
where it is not required. The Woltring GCV filter was run using the Vicon Workstation
software. (Additional information on the Woltring GCV filter is presented in Appendix
F).
3D kinematic analysis of the human body is typically conducted by modelling the body as
a series of rigid body segments linked at joint centres, i.e. locations at which the segment
rotations occur. For each rigid body segment, a 3D coordinate axis system must be
defined, and it is measurement of the relative movement between the 3D coordinate axis
systems of neighbouring segments that facilitates calculation of joint rotation angles in
each anatomical plane. A minimum of three non-collinear markers are required to
establish a 3D coordinate axis system for each rigid body segment. For this study, the
segment 3D coordinate axis systems and joint centre locations were defined according to
the Vicon Plug-in-Gait model. The process through which the 3D coordinate axis systems
were established for each rigid body segment is described in the following paragraphs.
The thorax formed a rigid segment defined by the markers on the seventh cervical
vertebra (C7), the tenth thoracic vertebra (T10), the jugular notch (CLAV) and the
xiphoid process of the sternum (STRN) (Figure 3.6). The 3D coordinate axis system for
the thorax was defined as follows. The orientation of the thorax was defined before the
origin. The first axis (Z) was defined as the line from the midpoint between the clavicle
marker (CLAV) and cervical marker (C7) to the midpoint between the sternum (STRN)
and thoracic markers (T10). The second axis (X) was formed between the midpoint of C7
and T10 to the midpoint of CLAV and STRN. The third axis (Y) was perpendicular to
both the first (Z) and second (X) axes, and was therefore directed in the medial-lateral
direction. The thorax origin was then calculated from the CLAV marker, with an offset of
half a marker diameter backwards along the X axis.
- 131 -
2 (X)
Figure 3.6 Marker placement and three-dimensional coordinate axis system for the thorax segment. The
number on each axis indicates the order in which the axes were established, the letter indicates the
assigned axis direction definition (adapted from http://medical-dictionary.thefreedictionary.com and Vicon
Plug-in-gait Manual).
The pelvis formed a rigid segment created by the two anterior superior iliac spine markers
(LASI, RASI) and the sacrum marker (SACR) (Figure 3.7). The origin of the pelvis
segment was taken as the midpoint between the left and right ASIS markers. The first
axis (Y) was the direction from the right ASIS marker (RASI) to the left ASIS (LASI)
marker. The second axis (X) was taken perpendicular to the first axis in the plane formed
by all three pelvis markers (LASI, RASI, SACR). The third axis (Z) was directed
upwards, perpendicular to the first and second axes.
Figure 3.7 Marker placement and three-dimensional coordinate axis system for the pelvis segment. The
assigned axis direction definition (adapted from Vicon Plug-in-Gait Manual).
- 132 -
The thigh segment was defined by the hip joint centre located in the pelvis segment, the
knee joint centre, the lateral thigh marker (LTHI/RTHI) and the lateral knee marker
(LKNE/RKNE) (Figure 3.8). The segment origin was taken as the knee joint centre. The
first axis (Z) was taken from the knee joint centre (KJC) to the hip joint centre (HJC). The
second axis (Y) was taken parallel to the line from the knee joint centre to the virtual
knee marker (established using the KAD). The third axis (X) for the thigh segment was
perpendicular to the first and second axes.
Figure 3.8 Marker placement and three-dimensional coordinate axis system for the thigh segment. The
The lower leg or shank segment was defined using the ankle joint centre, the knee joint
centre, the lateral tibia marker (LTIB/RTIB) and the lateral ankle marker (LANK/RANK)
(Figure 3.9). The first axis (Z) joined the ankle and knee joint centres. The second axis
(Y) passed through the lateral ankle marker (LANK/RANK) and the ankle joint centre,
which was located at a distance equal to half the ankle width plus half a marker diameter
from LKNE. The first and second axes both lay in the plane formed by the knee joint
centre and the markers LTIB and LANK. The third axis (X) was perpendicular to the first
and second axes.
- 133 -
Figure 3.9 Marker placement and three-dimensional coordinate axis system for the shank segment. The
The foot segment was defined using the ankle joint centre, the lateral ankle marker
(LANK/RANK), the toe marker (LTOE/RTOE) and the heel marker (LHEE/RHEE)
(Figure 3.10). The segment coordinate system was constructed using the line joining the
toe marker (TOE) and the ankle joint centre (AJC) as the first axis (Z). The direction of
the second axis (Y) from the shank segment was used to define the second axis (Y) of the
foot segment. The third axis (X) of the foot was perpendicular to the first and second
axes.
Figure 3.10 Marker placement and three-dimensional coordinate axis system for the foot segment. The
- 134 -
All joint angular displacements were calculated using the Vicon Plug-in-Gait full body
model, which uses the following methods for segment and joint angle calculations.
Movements of the thorax and pelvic segments were measured with respect to the global
reference frame (X, Y, and Z laboratory coordinate axes established by system
calibration). Sagittal plane thorax movement (forward/backward tilt) was defined as
anterior/posterior movement around the Y-axis, coronal plane thorax movement (side tilt)
was defined as right and left tilt around the X-axis, and transverse plane thorax rotation
(forward/backward rotation) was defined as rotation around the longitudinal Z-axis.
Sagittal plane pelvic movement (anterior/posterior pelvic tilt) was defined as anteriorposterior motion around the Y-axis, coronal plane pelvic movement (pelvic obliquity)
was defined as sideways movement around the X-axis, and transverse plane pelvis
(forward/backward rotation) was defined as rotation around the longitudinal Z-axis.
Hip, knee, and ankle angular displacements were all calculated with respect to relative
movement of the local 3D coordinate axis systems embedded within the segments.
Sagittal plane hip movement (flexion/extension) was defined as anterior-posterior motion
around the Y-axis, coronal plane hip movement (adduction/abduction) was defined as
sideways
movement
around
the
X-axis,
and
transverse
plane
hip
rotation
(internal/external rotation) was defined as rotation around the longitudinal Z-axis of the
thigh segment. Sagittal plane knee movement (flexion/extension) was defined as anteriorposterior motion around the Y-axis, coronal plane knee movement (valgus/varus) was
defined as sideways movement around the X-axis, and transverse plane knee rotation
(internal/external rotation) was defined as rotation around the longitudinal Z-axis of the
shank segment. Sagittal plane ankle movement (dorsi/plantarflexion) was defined as
anterior-posterior motion around the Y-axis, coronal plane ankle movement
(inversion/eversion) was defined as sideways movement around the X-axis, and
transverse plane ankle rotation (internal/external rotation) was defined as rotation around
the longitudinal Z-axis of the foot segment.
Cardan angle conventions were used for the calculation of all joint angle rotations.
Cardan angle rotations are one of the commonly used approaches to determine anatomical
joint angles, and are especially relevant and useful for biomechanics applications as they
- 135 -
can be easily used to represent the anatomical joint actions of flexion/extension,

adduction/abduction, and longitudinal (internal /external) rotation. Using the Cardan
angle approach, 3D joint angles are defined as a sequence of three ordered rotations from
the initial position of one coordinate system. For example, a common Cardan rotation
sequence used in biomechanics involves the following order of rotations.
1. Rotation about the medial-lateral directed axis

2. Rotation about the anterior-posterior directed axis
3. Rotation about the vertical/longitudinal axis
Using this order of rotations, for the segment and axis definitions employed in the current
study, sagittal plane rotations (i.e. flexion/extension; dorsi/plantarflexion) would be
calculated first about the Y-axis of the proximal segment; followed by coronal plane
rotations (i.e. lumbar side flexion, hip adduction/abduction, knee varus/valgus and ankle
inversion/eversion) about the X-axis; followed by internal/external rotation about the Zaxis (longitudinal axis) of the distal segment (Wu and Cavanagh, 1995). This sequence is
equivalent to a Y, X, Z order of rotation.
It should be noted that as Cardan angles are calculated, each rotation causes the axis for
the subsequent rotation to be shifted. For example, in the second rotation, the axis has
already been acted upon and shifted by one previous rotation, and in the third rotation, the
axis has already been acted upon and shifted by the two previous rotations.
Figure 3.11 Example of starting position (a) and end position (b) for a series of 3 Cardan angle rotations.
DQGUHSUHVHQWWKHUHVXOtant angles in each plane of motion at completion of all 3 rotations (from
mathworks.com)
- 136 -
Given this definition of Cardan angles, it would be expected that in 3D kinematic

analyses, the calculated segment rotations would be highly dependent on the Cardan
angle rotation sequence used (Cappozzo et al., 2005). However, Lees et al. (2010)
demonstrated that pelvic data and sagittal plane data for the hip, knee and ankle are robust
with regard to the rotation sequence used. For coronal and transverse plane data for the
hip, knee and ankle it was reported that the influence of rotation sequence was evident in
terms of the data offset and profile (Lees et al., 2010). For the hip joint, the rotation
sequence influenced transverse plane data when hip flexion was greater than 50, and for
the ankle joint transverse plane data offset varied according to rotation sequence across
the full range of sagittal joint motion. For these reasons, the kinematic data relating to the
principle of turnout will need to be interpreted with caution.
Figure 3.12 Directions and signs for kinematic measurement variables for thorax, pelvis, hip, knee and
ankle rotations (adapted from Vicon Plug-in-gait Manual).
- 137 -
Figure 3.12 depicts the signs (i.e. positive or negative) applied to the directions of the
rotations occurring in each plane at each joint for the current study. In accordance with
the conventions typically used in clinical biomechanical analyses, thorax posterior tilt,
lumbar spine extension, pelvic posterior tilt, hip and knee extension, and ankle plantar
flexion in the sagittal plane were denoted as negative in all tables and graphs. Coronal
plane movements of downward pelvic obliquity, hip abduction, knee varus, and ankle
inversion, and all external rotations in the transverse plane were also denoted as negative.
It should also be noted that the kinematic model used in this study calculates 3D
movement of the ankle joint complex rather than that of a single anatomical joint. That
is, movement occurring at this structure is due to combined motions of the talo-crural
joint and the sub-talar joint. Due to the orientation of the talo-crural joint axis, the
dominant movement of this joint can be considered to be dorsi/plantarflexion. However,
the sub-talar joint axis lies oblique to the three anatomical planes of motion with the
majority of the joint movement occurring simultaneously within the transverse and
coronal planes. Consequently, the kinematic measurements obtained for the coronal and
transverse plane are not directly analogous to specific anatomical joint function, but
instead reflect the projection of motion of the ankle joint complex into each of these
planes.
3.2.3 Kinematic Measurement Variables

As described in Section 3.1, through reading a number of texts on classical ballet
technique, four fundamental principles of the technique were identified. These principles
were defined as:
1) Alignment
2) Placement
3) Turnout
4) Extension
In order to use quantitative techniques to analyse these principles, and determine the
extent to which professional and non-professional ballet dancers adhere to these
principles in their execution of classical ballet technique, it was necessary to a identify a
- 138 -
number of biomechanical parameters that could be used as indicators of these principles.

Based on the defining characteristics on which these principles are based, a collection of
appropriate kinematic measurement variables were selected for analysis for the study.
Each of the four fundamental principles, their defining characteristics, and their
associated kinematic measurement variables are outlined in Table 3.1.
Fundamental Principle
Defining Characteristics of
of Classical Ballet
Fundamental Principle of Classical
Technique
Ballet Technique
1. Alignment
Maintenance of verticality and
maximum elongation of the torso
Biomechanical Quantitative
Measurement Variables
x Thoracic anterior/posterior tilt
x Thoracic side tilt
2. Placement
Minimal displacement of the pelvis

from its centred position
x Pelvic anterior/posterior tilt

x Pelvic obliquity
x Pelvic-thoracic transverse rotation
3. Turnout
Maximum external rotation of the

lower limbs
x Hip external rotation

x Ankle external rotation
4. Extension
Maximum elongation of the lower

limbs
x Knee extension
x Ankle plantarflexion
Table 3.1 Fundamental principles of classical ballet technique and associated kinematic variables
3.2.3.1 Potential Measurement Error

As indicated in the preceding section, for 3D biomechanical analyses, reflective markers
placed on the body surface are used to define the embedded coordinate axes, and
therefore the movement, of underlying bone segments. However, it is well known that
error in estimation of bone segment motion exists due to movement of the markers on the
surface of the body moving with respect to the underlying bones. This discrepancy
between the movement of the markers and underlying bone is based on two different
sources of error: anatomical landmark mislocation (i.e. marker placement error) and soft
tissue artifact (Stagni et al., 2005). Various studies have investigated the effects that
marker placement error and soft tissue artefact have on the accuracy of 3D joint angle
calculations.
Kadaba et al. (1989) conducted a repeatability study of joint angles computed from a
lower limb marker set very similar to the one used in the current study, the only
difference being the use of wand markers by Kadaba et al. (1989) to mount the sacrum
- 139 -
and lateral thigh markers. Advances in the accuracy and resolution of motion analysis
cameras since 1989 means that use of wand markers is no longer a methodological
necessity, therefore in the current study these markers were attached directly to the skin.
Results of Kadaba et al. (1989), presented as coefficient of multiple correlation (CMC),
indicate that sagittal plane repeatability of hip (CMC = 0.978-0.996), knee (CMC =
0.981-0.994) and ankle joint (CMC = 0.933-0.978) data were excellent both within a
test day as well as between test days. Repeatability for the frontal (CMC = 0.783-0.962)
and transverse plane (CMC = 0.582-0.893) joint angles within-days and between-days
were lower than for sagittal plane data. The fact that the current study will not use wand
markers, which have a tendency to wobble, suggests that increased measurement
reliability will be obtained in the current study compared to Kadaba et al. (1989). Another
study by this group (Kadaba et al., 1990) reported that for a 1cm variation in marker
placement, a maximum constant offset of 2 in the sagittal plane angle patterns was
obtained, however the ranges of joint rotations were not affected. Della Croce et al.
(1999) also investigated the effect of marker placement error on the calculation of 3D hip,
knee and ankles joint angles. Of the variables that are relevant to the current study, the
precision of the intra-examiner data were 1.0, 1.6, 3.9 and 5.3 for knee
flexion/extension, ankle dorsi/plantarflexion, ankle internal/external rotation, and hip
internal/external rotation, respectively. These results from previous studies indicate that
with repeated use of the same operator for reflective marker placement, excellent
repeatability measures can be obtained for sagittal plane hip, knee and ankle joint angles,
with slightly lower repeatability occurring for transverse plane variables. In addition,
larger marker placement errors (1cm) than would be expected in the current study have
been found to produce only relatively minor discrepancies in joint angle data.
Several studies have been published on the amplitude and implications of soft tissue
artefact in 3D kinematic analysis. For reflective markers located on the thigh, Camomilla
et al. (2009) obtained root mean square soft tissue displacements ranging from 2.523.0mm. Stagni et al. (2005) reported that displacement of skin markers with respect to
the corresponding embedded coordinate axis system, was generally larger on the thigh
segment (up to 31mm) than on the shank (up to 21 mm). In addition, as was the case for
marker placement error, Stagni et al. (2005) also reported that coronal plane and
transverse plane data were more affected by soft tissue artefact than sagittal plane data.
- 140 -
The increased likelihood of soft tissue displacement on the thigh, together with the higher
influence of soft tissue displacement on transverse rotation data, indicates that hip
transverse plane data (i.e. internal/external rotation) may contain multiple sources of error
and should therefore be assessed with some caution.
3.2.4 Experimental Protocol

3.2.4.1 Subject Anthropometric Measurements and Marker Placement
The experiments were conducted at the University of New South Wales in the
Biomechanics Laboratory in the School of Risk and Safety Sciences. Upon arrival at the
laboratory, participants were familiarised with the testing environment and were informed
again of what to expect and what was required of them. They were assured of the
anonymity of their motion analysis data, but were reminded that by giving consent to
participate in the study they were agreeing to their video data being viewed by
participating dance teachers in a later part of the study. A signed and dated information
and consent form (Appendices B and C) was collected from each subject at the beginning
of each test session, prior to any data being collected.
Before collecting any motion analysis data, subjects were also asked to complete a
questionnaire outlining their date of birth, dance training and performance history, study
background, and dance injury history. This questionnaire consisted of 40 questions which
were a mixture of short answer, yes/no and multiple choice questions (Appendix D).
Subject height, mass, and anthropometric data were measured upon commencement of
each motion analysis session. Body mass was measured using electronic scales.
Anthropometric data obtained included leg length, inter anterior superior iliac spine
(ASIS) distance, knee width, ankle width, shoulder anterior-posterior girth, elbow width,
wrist width, foot thickness, hand thickness, and tibial torsion. During the data collection
session, subjects were required to wear tight black short bike-pants, a black crop-top and
foot thongs (Bloch Foot Thong III S0675).
All subjects were asked to watch a demonstration video detailing each of the dance
movements that they would be required to perform during the data collection session.
Non-professional and professional ballet dancers were required to perform the same steps
and therefore all participants were shown the same demonstration video. Immediately
- 141 -
after watching the demonstration video all subjects were given the opportunity to ask
questions about the movement requirements and to opt out of the study if they felt that
any of the movements displayed on the video could be potentially injurious to them. If
willing to proceed with the test session, subjects were asked to complete a self-directed
warm-up that they felt would sufficiently prepare them to perform the required steps to
the best of their ability. Each dancer was allowed as much time as required for this, most
dancers took 10-15 minutes to complete this process. It should be noted that no dancers
opted out of the study after viewing the demonstration video.
Upon completion of their warm-up, the reflective markers were attached to the subject by
palpation of the appropriate bony landmarks. Marker attachment was by way of double
sided tape, and reinforcement with 3M Micropore medical tape. After attachment of all
reflective markers, except the right and left knee markers (RKNE, LKNE), three static
trials using the KADs were obtained. For each of these static trials the KAD was aligned
in a slightly different position, with the medial side of the devise being placed in middle,
anterior or posterior positions with respect to the estimated knee flexion/extension axis
(estimated by visual observation and palpation of the femoral condyles). A final static
trial was taken without the KADs. For this trial the right and left knee markers (RKNE,
LKNE) were placed on the location of the virtual knee marker as defined by use of the
KADs. Anthropometric measurement, reflective marker placement and KAD application
was conducted by the same operator (the author) for all subjects that participated in the
study.
The first dynamic trials conducted were walking trials. Subjects were instructed to walk
down the middle of the laboratory, between designated points, at a self-selected speed
using their usual walking pattern. Six walking trials were captured for each subject.
Walking trials were obtained to ensure the accuracy of the reflective marker placement.
Gait cycle data were processed using the Vicon Plug-in Gait Lower Body model in Vicon
Workstation (Oxford Metrics Ltd, Oxford, UK). Graphical outputs for the gait cycle data
were created and displayed in Vicon Polygon Authoring Tool (see Appendix G). Given
that all subjects appeared to exhibit a normal gait pattern, evaluation of the gait analysis
graphical outputs enabled assessment of the validity of the joint displacement data. All
kinematic parameters were compared to the normal data set (mean 1 standard deviation
- 142 -
for normal adults) provided with Polygon Authoring Tool, but in particular, knee
varus/valgus data were used as a means of assessing the accuracy of the location of the
knee flexion/extension axis. The knee flexion/extension axis, along with the long axis of
the femur, defines the coronal plane of the thigh segment and therefore has a strong
influence on the accuracy of hip transverse rotation data.
If the knee flexion/extension axis is misplaced it can convert knee flexion movement into
apparent knee varus or valgus movement (crosstalk), appearing as increased knee
varus/valgus movement during the gait cycle. Knee varus/valgus range of motion (ROM)
data are therefore a very useful indicator of marker placement accuracy and thus of the
quality of the kinematic data. A knee varus/valgus ROM of less than or equal to 10 for
the entire gait cycle was used as an indication of sufficient accuracy of the location of the
knee joint flexion/extension axis, and thus sufficient accuracy of the optical marker
placement. If the knee varus/valgus ROM was greater than 10 for the entire gait cycle,
then the subjects static rotation offsets (obtained from processing a static trial) were
adjusted by processing a different static trial that is, a static trial with a different position
(middle, anterior, or posterior) of the medial side of the KAD. Upon selecting a static trial
that resulted in a knee varus/valgus ROM less than or equal to 10 for the entire gait
cycle, this static trial was subsequently used to process all dance movement trials.
3.2.4.2 Dance Movement Descriptions

The ballet movements the subjects were asked to perform during the data collection
session were of an elementary/intermediate level and are typical of those performed in the
standard classical ballet classes that all subjects would have previously experienced.
Dancers were advised that if at any time during the testing session they experienced pain
or discomfort, or for any other reason did not want to continue performing the dance
movements, they were free to immediately cease participation and withdraw from the
study. It should be noted that all participating dancers who started the data collection
session completed all requirements of the experimental protocol. No dancers withdrew
from the study or reported that their execution of the movements was hindered or altered
in any way by pain or discomfort.
- 143 -
In order to ensure the consistency between dancers of the speed of execution of each step,
the same musical accompaniment, played from a CD player, was used for each dancer for
each type of movement performed. The music used for data collection was the same
music used in the demonstration video. Dancers were reminded of what was required for
execution of each different movement before performing it (e.g. sequence of movements
and correct musical timing), but were given no instruction on how to perform each
movement, that is, no technical instruction or coaching was provided at any stage
throughout the testing session.
The dance movements selected for analysis were each from one of the categories of the
Seven Movements of Dance, as defined by Jean Georges Noverre (1727 1810) in
1760. Noverre, one of the great ballet masters of the 18th century, has been described as
the person who laid down the scientific foundation of ballet movement (Conyn, 1948,
p7). It is commonly stated that the basic principles and categories of classical ballet
movement that were outlined by Noverre have remained unaltered through the centuries
and still guide every ballet dancer today. For this reason, the movement categories
defined by Noverre played a pivotal role in the selection of ballet steps for analysis in this
study. Noverres Seven Movements of Dance are characterised in terms of the quality
of the step, and are defined as: (Conyn, 1948).
1. Bend (Pli)
3. Rise (Relev)
4. Jump (Saut)
5. Glide (Gliss)
6. Dart (Elanc)
7. Turn (Tour)
At least two steps from each of these seven movement categories were selected for
analysis in this study. The steps are described below.
- 144 -
3.2.4.2.1 Movement Category 1 Bend (Pli)

x Step 1 Grand Pli in First Position
x Step 2 - Grand Pli in Second Position
The grand pli movement is a deep bend of the knees while the feet are held in one of the
five positions of the feet. All subjects were asked to perform the grand pli movement
with their feet in first position (Figure 3.13), and then in second position (Figure 3.14).
The subject begins the movement with both legs straight (in either first or second
positions) and with the arms held in second position (Figure 3.13a & Figure 3.14a), Upon
commencement of the pli, the arms are lowered from second position, reaching bras bas
(i.e. arms form an oval shape in front of the dancer with arms hanging down) by the
maximum depth of the pli (Figure 3.13c & Figure 3.14c). The arms are then brought
through first position while ascending from the grand pli (Figure 3.13d and Figure
3.14d), and then back to second position upon completion of the whole movement (i.e. on
returning to straight legs) (Figure 3.13e & Figure 3.14e).
Figure 3.13 Grand pli in first position. (a) start of pli; (b) descent of pli (c) peak of pli; (d) ascent of
pli; (e) end of pli.
Figure 3.14 Grand pli in second position. (a) start of pli; (b) descent of pli (c) peak of pli; (d) ascent of
pli; (e) end of pli.
- 145 -
3.2.4.2.2 Movement Category 2 Stretch (Battement)

x Step 1 - Battement Devant
x Step 2 - Battement Derriere
The battement movement is defined as a kicking movement of the working leg (i.e. the
leg that is performing the movement). All subjects were asked to perform the battement
movement to the front (devant) and back (derriere) of their body, at four different levels
of leg elevation. The subject begins the movement with knees straight, with their feet in
first position and with the arms held in second position (Figure 3.15a & Figure 3.16a), the
arms remain in second position for the duration of the movement. Upon commencement
of the battement, the subject moves their leg in one of the two directions, front (Figure
3.15) or back (Figure 3.16), while attempting to keep both knees straight. The four
different levels of battement leg elevation performed by each subject in each of the two
directions were; a battement tendu (kicking the working leg while keeping the toes in
contact with the floor) (Figure 3.15b & Figure 3.16b), a battement gliss (kicking the
working leg to a height just above the floor) (Figure 3.15c & Figure 3.16c), a battement
jet (kicking the working leg to an angle of 45 to the vertical) (Figure 3.15d and Figure
3.16d), and a grand battement (kicking the working leg to an angle of 90 to the vertical)
(Figure 3.15e & Figure 3.16e).
Figure 3.15 Battement devant (front). (a) starting position for all battements devant; (b) peak position for
battement tendu devant; (c) peak position for battement gliss devant; (d) peak position for battement jet
devant to 45; (e) peak position for grand battement devant to 90.
- 146 -
Figure 3.16 Battement derriere (back). (a) starting position for all battements derriere; (b) peak position
for battement tendu derriere; (c) peak position for battement gliss derriere; (d) peak position for
battement jet derriere to 45; (e) peak position for grand battement derriere to 90.
3.2.4.2.3 Movement Category 3 Rise (Relev)

x Step 1 - Rise in First Position
x Step 2 - Rise in Second Position
The rise movement is a controlled and gradual increase in ankle plantarflexion while
remaining on the balls of the feet. This movement can be performed with the feet held in
any one of the five positions of the feet. All subjects were asked to perform the rise
movement with their feet in first position (Figure 3.17), and then in second position
(Figure 3.18). The subject begins the movement with both legs straight (in either first or
second positions) and with the arms held in bras bas (Figure 3.17a & Figure 3.18a), the
arms remain in the bras bas position for the entire duration of the movement. The peak of
the rise movement is reached when the dancer has achieved their maximum level of ankle
plantarflexion (Figure 3.17b & 3.18b), this position of the foot and ankle is referred to as
demi-pointe (i.e. half-point) in classical ballet terminology.
Figure 3.17 Rise in first position. (a) start of rise; (b) peak of rise; (c) end of rise
- 147 -
Figure 3.18 Rise in second position. (a) start of rise; (b) peak of rise; (c) end of rise
3.2.4.2.4 Movement Category 4 Jump (Saut)

x Step 1 - Saut in First Position
x Step 2 - Temps Lev Ordinaire
The saut jump is a vertical jump with an initial countermovement (i.e. a preparatory
bend of the knees (demi-pli)) to assist with the achievement of greater elevation. It is
typical for this movement to be performed with the feet in either first or second position.
For this study all subjects were asked to perform the saut in first position (Figure 3.19).
In the saut, the dancer pushes off from two feet with both knees bent (Figure 3.19a),
then while airborne the dancer must try to extend both knees and plantarflex both ankles
as much as possible (Figure 3.19b). The dancer then lands back onto two feet in the same
foot position from which they commenced the jump (Figure 3.19c). Arms were held in
the bras bas position for the entire movement (Figure 3.19). In addition, the subjects were
also required to perform a single leg hop, known as a temps lev in classical ballet
terminology. All subjects performed the temps lev with their working leg (leg
contralateral to the leg on which they were jumping) held with maximal external rotation
of the hip, knee flexed to approximately 60, and the medial side of the ankle of the
working leg touching the calf of the supporting leg (Figure 3.20). Similar to the saut, the
temps lev is also performed with a countermovement (i.e. bend of the knees) prior the
jump (Figure 3.20a), then while airborne the dancer is required to achieve maximum knee
extension and plantarflexion of the supporting leg (Figure 3.20b), before landing back
onto this leg with a bent supporting knee (Figure 3.20c).
- 148 -
Figure 3.19 Saut in first position. (a) start of saut; (b) peak of saut; (c) end of saut
Figure 3.20 Temps lev. (a) start of temps lev; (b) peak of temps lev; (c) end of temps lev.
3.2.4.2.5 Movement Category 5 Glide (Gliss)

x Step 1 - Glissade Dessus
x Step 2 - Glissade Dessous
The glissade is a gliding travelling movement along the floor that can be performed in
various directions (forwards, backwards or sideways). It is typical for this movement to
be performed with the feet starting in fifth position. In the glissade, the dancer starts by
bending both knees (demi pli) (Figure 3.21a & Figure 3.22a), then sliding the leading
foot along the floor in the direction of intended movement while gradually extending the
knee of the leading side until it reaches full extension (Figure 3.21b & Figure 3.22b). The
- 149 -
dancer then pushes off the following leg, reaching a position where both knees are in full
extension (Figure 3.21c & Figure 3.22c), before smoothly transferring their weight onto a
bent leading leg while extending the following leg (Figure 3.21d & Figure 3.22d). The
following leg then slides in to meet the leading leg in fifth position to complete the
movement (Figure 3.21e & Figure 3.22e). Various forms of the glissade movement exist,
they are categorised according to the direction of travel, forward (en avants), backwards
(en arriere), and sideways (de cote). Glissade de cote can be further categorised
according to which foot leads the movement, front foot (devants) or back foot (derriere),
and whether or not the feet change position during the movement. When the front foot
leads the movement and changes to the back by the end of the movement the step is
called glissade under (dessous) (Figure 3.21), when the back foot leads the movement
and changes to the front by the end of the movement the step is called glissade over
(dessus) (Figure 3.22). For this study, all subjects were asked to perform four consecutive
glissades de cote, in the sequence of under, over, under, over (dessus, dessous, dessus,
dessous). All subjects performed the glissades dessus and dessous starting with their
feet in fifth position. Subjects were asked to hold their arms in the bras bas position for
the duration of the movement sequence. The 2nd (dessus) and 3rd (dessous) glissades of
the sequence were selected for analysis.
Figure 3.21 Glissade dessous (under). (a) starting position for glissade; (b) push-off from following leg; (c)
peak of glissade; (d) weight transference onto leading leg (e) ending position for glissade.
- 150 -
Figure 3.22 Glissade dessus (over). (a) starting position for glissade; (b) push-off from following leg; (c)
peak of glissade; (d) weight transference onto leading leg (e) ending position for glissade.
3.2.4.2.6 Movement Category 6 Throw or dart (Elanc)

x Step 1 - Jet Ordinaire
x Step 2 - Grand Jet Elanc en Avants
An elanc movement requires the dancer to make a strong push away from, but also just
over, the floor (Lawson, 1979). The jet movement possesses the elanc quality and was
considered to be a suitable step for analysis. Jet is a generic term used for numerous
jumping steps whereby the dancer jumps from one leg to the other while one leg is
thrown outward in the process of execution (Lee, 1983). For this study, all subjects
were asked to perform a jet ordinaire derriere (Figure 3.23) and a grand jet elanc en
avants (Figure 3.23). For the jet ordinaire derriere, the dancer starts with their feet in
fifth position (Figure 3.23a). Upon commencing the movement the back leg is thrust out
towards the side of the body, while the knee is straightened to full extension (Figure
3.23b). As the leg is thrust out the knee of the other leg (the supporting leg) bends (demipli) and then the dancer push forcefully off the supporting leg jumping into the air
(Figure 3.23c). The dancer lands back on the floor onto the leg that was initially thrust out
(Figure 3.23d).
The grand jet elanc en avants movement (Figure 3.24) is also a throwing jumping
movement in which the dancer pushes off one leg and lands on the other. It is essentially
a leap from one leg (Figure 3.24a) to the other during which the dancer tries to travel as
far as possible in the forward direction, while splitting their legs to their maximum ability
at the peak of the movement. At the peak of the jet, the dancer should have their leading
- 151 -
hip flexed as much as possible in front of their body with their knee at full extension,
while their back leg is simultaneously extended as much as possible behind their body,
also with the knee in full extension (Figure 3.24b).
a
b
c
d
Figure 3.23 Jet ordinaire. (a) starting position for jet ordinaire; (b) push-off from following leg; (c)
peak of jet ordinaire.; (d) ending position for jet ordinaire.
a
b
c
Figure 3.24 Grand jet elanc en avants. (a) push-off position for grand jet; (b) peak of jet grand jet;
(d) ending position for grand jet.
3.2.4.2.7 Movement Category 7 Turn (Tour)

x Step 1 Double Pirouette En Dehors
x Step 2 - Double Pirouette En Dedans
The pirouette is a turning movement in which the dancer spins on the spot while standing
on one leg with the heel raised to so that the base of support consists only of the ball of
the foot (Figure 3.25 & Figure 3.26). The supporting leg on which the dancer stands
- 152 -
must be as extended as possible (i.e. maximum knee extension and maximum ankle
plantarflexion) for the duration of the movement. The other leg, the working leg is held
in a retir position for the duration of the movement. For correct execution of the retir
position, the working knee is flexed such that the toe of the working leg is placed at knee
height of the supporting leg. The working ankle should be held in maximum
plantarflexion, and the working and supporting hips should both be held in maximum
external rotation for the duration of the movement (Figure 3.25b & c, Figure 3.26c & d).
For this study all subjects were asked to perform two different versions of the double
pirouette, a double pirouette en dehors (Figure 3.25) and a double pirouette en dedans
(Figure 3.26). Both these movements are performed with the supporting and working legs
placed in the positions described above, however the difference between these two steps
is the direction in which the dancer spins. For the double pirouette en dehors the dancer
performs two complete revolutions spinning towards the direction of the working leg,
while for the double pirouette en dedans, the dancer performs two complete revolutions
spinning towards the direction of the supporting leg. For both these movements the
subjects were asked to initiate the pirouette with their feet in fourth position, and their
arms in third position (Figure 3.25a and Figure 3.26a). Subjects were asked to hold their
arms in first position for the duration of each version of the double pirouette (Figure
3.25b & c, Figure 3.26c & d).
Figure 3.25 Pirouette en dehors. (a) preparatory position for pirouette; (b) start of pirouette; (c) end of
pirouette; (d) finishing position for pirouette
- 153 -
Figure 3.26 Pirouette en dedans. (a) preparatory position for pirouette; (b) initiating movement for
pirouette (i.e. fouette movement) (c)start of pirouette; (d) end of pirouette; (d) finishing position for
pirouette.
3.2.5 Data Extraction

After calculation of the kinematic variables, each trial was exported from Vicon
Workstation and saved as an ASCI file. In order to analyse the kinematic data and make
comparisons between groups, it was necessary to identify specific time points within each
trial that corresponded with the start, peak and end of each movement. This was achieved
through the use of Visual Basic code. For each movement category, specific criteria were
established according to which the relevant time points were identified. A different
Visual Basic macro was written by the author for each different category of movement.
The criteria used for identification of the relevant time points for each movement
category are described below.
3.2.5.1 Movement Category 1 Bend (Pli)

Pli commencement (0% of the movement) (Figure 3.13a & Figure 3.14a) was defined as
the instant of the first of three consecutive increases in knee flexion (Knee-X) of at least
2%. The peak of the grand pli (Figure 3.13b & Figure 3.14b), the instant of maximum
depth of the grand pli, was defined as the point at which maximum knee flexion
(maximum Knee-X) was achieved The end of the grand pli (100% of the movement)
was defined as the instant of the last of three consecutive decrease in knee flexion (KneeX) of less than 2%.
- 154 -
3.2.5.2 Movement Category 2 Stretch (Battement)

The sagittal plane thigh angle variable was used to identify significant time points for
battements towards the front (devant), or back (derriere) of the body. Specifically, for
battements devant, battement commencement (0% of the movement) (Figure 3.15a) was
defined as the first of three consecutive increases of at least 2% in sagittal plane thigh
angle. The peak of the battement devant (Figures 3.15b, 3.15c, 3.15d, 3.15e), the instant
of maximum height of the battement devant, was defined as the point at which maximum
sagittal plane thigh angle was achieved. The end of the battements devant (100% of the
movement) (Figure 3.15a) was defined as the last of three consecutive decreases of less
than 2% in sagittal plane thigh angle.
For battements derriere, battement commencement (0% of the movement) (Figure 3.16a)
was defined as the first of three consecutive decreases of at least 2% in sagittal plane
thigh angle. The peak of the battements derriere, (Figures 3.16b, 3.16c, 3.16d, 3.16e), the
instant of maximum height of the battement derriere, was defined as the point at which
minimum sagittal plane thigh angle was achieved. The end of the battements derriere
(100% of the movement) (Figure 3.16a) was defined as the last of three consecutive
increases of less than 2% in sagittal plane thigh angle.
3.2.5.3 Movement Category 3 Rise (Relev)

Rise commencement (0% of the movement) (Figure 3.17a & Figure 3.18a) was defined as
the first of three consecutive increases of at least 2% in ankle plantarflexion (i.e. three
consecutive decreases in Ankle-X) The peak of the rise (Figure 3.17b & Figure 3.18b),
instant of maximum height of the rise, was defined as the point at which maximum
plantarflexion (i.e. minimum Ankle-X) was achieved. The end of the rise (100% of the
movement) (Figure 3.17c & Figure 3.18c) was defined as the last of three consecutive
decreases of less than 2% in ankle plantarflexion (i.e. three consecutive increases in
Ankle-X).
3.2.5.4 Movement Category 4 Jump (Saut)

Jump commencement (0% of the movement) (Figure 3.19a & Figure 3.20a) was defined
as the point at which maximum knee flexion (i.e. maximum Knee-X) was achieved. The
peak of the jump (Figure 3.19b & Figure 3.20b), the instant of maximum height of the
- 155 -
jump, was defined as the point at which maximum vertical displacement of the whole
body was achieved (i.e. maximum COM-Z). The end of the jump (100% of the
movement) (Figure 3.19c & Figure 3.20c) was defined as the point at which maximum
knee flexion (i.e. maximum Knee-X) was achieved again.
3.2.5.5 Movement Category 5 Glide (Gliss)

Glissade commencement (0% of the movement) (Figure 3.21a & Figure 3.22a) was
defined as the point at which maximum knee flexion (i.e. maximum Knee-X) was
achieved on the leading leg at the beginning of the movement. The peak of the glissade
(Figure 3.21c & Figure 3.22c) was defined as the point at which maximum vertical
displacement of the whole body was achieved (i.e. maximum COM-Z). The end of the
glissade (100% of the movement) (Figure 3.21e & Figure 3.22e) was defined as the point
at which maximum knee flexion (i.e. maximum Knee-X) was achieved on the following
leg.
3.2.5.6 Movement Category 6 Throw or dart (Elanc)

For both the jet ordinaire derriere and the grand jet elanc en avants commencement
(0% of the movement) (Figure 3.23b & Figure 3.24a) was defined as the point at which
maximum knee flexion (i.e. maximum Knee-X) was achieved on the supporting (pushoff) leg at the beginning of the movement. The peak of the jet (Figure 3.23c & Figure
3.24b) was defined as the point at which maximum vertical displacement of the whole
body was achieved (i.e. maximum COM-Z). The end of the movement (100% of the
movement) (Figure 3.23d & Figure 3.24c) was defined as the point at which maximum
knee flexion (i.e. maximum Knee-X) was achieved on the leading/landing leg.
3.2.5.7 Movement Category 7 Turn (Tour)

For the turning steps (pirouettes), Visual Basic code was written to identify four rather
than three time points in each trial. These were the points at which the subject
commenced the pirouette, reached the pirouette retir position, finished the pirouette
retir position, and then placed their working foot back on the ground again. For both the
pirouette en dehors and the pirouette en dedans commencement (0% of the movement)
(Figure 3.25a & Figure 3.26a) was defined as the point at which the working foot left the
ground. This time point was defined as the first of three consecutive increases of at least
- 156 -
2% in the ankle marker vertical displacement, (i.e. the first of three consecutive increases
in LANK-Z or RANK-Z). The instant of reaching the pirouette retir position (Figure
3.25b & Figure 3.26c) was defined as the last of three consecutive increases of at least
2% in knee flexion (i.e. Knee-X). The instant of finishing the pirouette retir position
(Figure 3.25c & Figure 3.26d) was defined as the first of three consecutive decreases of at
least 2% in knee flexion (i.e. Knee-X). The end of the pirouette (100% of the movement)
(Figure 3.25d & Figure 3.26e) was defined as the last of three consecutive decreases of
less than 2% in the ankle marker vertical displacement, (i.e. the last of three consecutive
decreases in LANK-Z or RANK-Z). The instances of the movement start and peak
used for the inter-group analysis were the instant of reaching the pirouette retir position
and the instant of finishing the pirouette retire, respectively.
For comparison of kinematic data between the non-professional and professional groups,
the relevant kinematic measurement variables at the start and peak of each step for each
subject were extracted and tabulated using a Visual Basic macro written by the author.
3.2.6 Statistical Analysis

SPSSTM version 18.0 software was used for all statistical analysis. Descriptive statistics
for all relevant kinematic measurement variables at the start and peak of each step were
calculated for each group.
Independent-samples t-tests were used to make comparison between groups of some of

the specific kinematic variables used for identification of the start and peak of time points
for each step in each movement category. That is, to ensure the validity of making
comparisons between groups in all the kinematic variables that relate to the four
fundamental principles of classical ballet technique (listed in Table 3.1), it was important
to check the level of consistency or similarity between groups in the time point defining
variables at each time point. This was particularly important for the battement steps. That
is, as mentioned in Section 3.2.5.1, the start and peak time points for the battements
(kicks) to the front (devant) and back (derriere) of the body were identified according to
global sagittal plane thigh angle data, (i.e. thigh elevation). During the data collection
sessions, participants were instructed to kick their legs to specific degrees of elevation for
each different battement step, so that comparison of the nine kinematic measurement
- 157 -
variables could be made at a consistent peak of the movement. The elevation of the
thigh could have a direct effect on the position of the thorax and pelvis, the external
rotation of the hip, and the extension of the knee (i.e. the kinematic variables associated
with the fundamental principles of classical ballet), it was therefore important to ensure
that kinematic variables were being compared at similar levels of thigh elevation at the
peak of the battement movements. Based on these analyses, only two battement steps
were selected for inclusion in future analyses.
After identification of the battements for which there was adequate consistency between
groups in the time point defining variables, independent-samples t-tests were conducted
to test for significant differences between groups at the start and peak of each step in the
nine kinematic measurement variables. Each step included in this analysis was
investigated with respect to the four fundamental principles of alignment, placement,
turnout and extension. Because repeated measures of the same variable were made for
each participant, e.g. each parameter was measured for each participant performing
multiple steps, a repeated measures general linear model analysis was conducted for each
variable to adjust for multiple comparisons. Within-subject factors used in this analysis
were step, and time point, and the between-subject factor was group. For all
statistical analyses, results were considered to be significant if a p-value of less than 0.05
was obtained.
- 158 -
3.3 Results
3.3.1 Subject Demographics
Within the time available for completion of the study, twenty-six healthy female adult
ballet dancers volunteered to participate. Fourteen participants satisfied the criteria for
inclusion in the non-professional group, and twelve participants satisfied the criteria for
inclusion in the professional group. The non-professional dancers were all students
enrolled in the University of New South Wales Bachelor of Arts (Dance)/Bachelor of
Education program. The professional dancers were recruited from various professional
ballet companies and classes within Australia.
Professional ballet dancers were older (24.7 4.4 years) than the non-professional
dancers (19.6 2.0 years). There was no significant difference in the mean total years of
ballet training between groups, however the professional group had a mean duration of
full-time (vocational) ballet training of 4.1 2.0 years, while none of the non-professional
dancers had received any full-time training in classical ballet throughout their dance
training history. There were no statistically significant differences between groups in
mean height or mass, however there was a difference in body mass index (BMI). The
lower mean BMI (19.61.3) recorded for the professional dancers is consistent with the
slender body aesthetic typically required for success as a professional ballet dancer.
Group
Non-professional
Professional
Age (years)
19.6 2.0
24.7 4.4*
Height (cm)
163.7 6.6
165.4 4.2
Mass (kg)
57.7 7.2
53.7 4.7
Body Mass Index (BMI)**
21.5 2.0
19.6 1.3*
Total years ballet training

11.6 4.3
Years full-time (vocational) ballet
0.0 0.0
training
Table 3.2: Participant demographics (Mean SD)
*t-test (p<0.05): Age, BMI, and Years full-time ballet training.
** BMI = mass(kg)/(height(m))2
- 159 -
14.3 4.3
4.1 2.0*
3.3.2 Comparison of Variables Used to Define Start and Peak Time Points for
Battements
All dancers were instructed to perform multiple battement movements at varying degrees
of thigh elevation. Despite being given specific instructions, it was important to ensure
that, on average, both groups of dancers performed the battements subsequently selected
for analysis to a statistically similar height, thereby ensuring that valid comparisons
between groups could be made. Table 3.3 shows the results of the independent-samples ttests for the time point (start and peak) defining variables for the battement steps.
Step
Time
point
Time Point
Defining
Variable
Mean
Difference
(Nonprofessional
Professional)
p value
(2-tailed
t-test)
Std. Error
Difference
95% Confidence
Interval of the
Difference
Lower
Upper
Global Thigh
Start
2.762*
0.003
0.836
1.032
4.492
Sagittal Angle
Battement
Tendu Front
Global Thigh
Peak
-1.190
0.274
1.062
-3.389
1.008
Sagittal Angle
Global Thigh
Start
2.024*
0.026
0.854
0.261
3.787
Sagittal Angle
Battement
Gliss Front
Global Thigh
Peak
-3.702
0.143
2.432
-8.765
1.360
Sagittal Angle
Global Thigh
Start
2.833*
0.036
1.274
0.200
5.467
Battement
Sagittal Angle
Jet 45
Global Thigh
Front
Peak
-5.440
0.193
4.061
-13.842
2.961
Sagittal Angle
Global Thigh
Start
2.690
0.043
1.240
0.091
5.290
Grand
Sagittal Angle
Battement
Global Thigh
90 Front
Peak
-11.488*
0.001
2.981
-17.691
-5.286
Sagittal Angle
Global Thigh
Start
1.179
0.295
1.095
-1.112
3.469
Sagittal Angle
Battement
Tendu Back
Global Thigh
Peak
1.071
0.225
0.852
-0.722
2.865
Sagittal Angle
Global Thigh
Start
1.083
0.290
0.999
-0.985
3.152
Sagittal Angle
Battement
Gliss Back
Global Thigh
Peak
4.298*
0.048
2.069
-0.031
8.626
Sagittal Angle
Global Thigh
Start
1.607
0.150
1.081
-.626
3.841
Sagittal Angle
Battement
Jet 45 Back
Global Thigh
Peak
5.845*
0.037
2.647
0.380
11.310
Sagittal Angle
Global Thigh
Start
0.690
0.518
1.503
-1.485
2.866
Grand
Sagittal Angle
Battement
Global Thigh
90 Back
Peak
11.786*
0.000
2.048
7.465
16.106
Sagittal Angle
Table 3.3 Results of the independent-samples t-tests for global thigh sagittal angle (thigh elevation) at the
start and peak of the battement movements. *t-test (p<0.05)
- 160 -
For the battements devant (kicks forwards), the battement jet to 45 was the highest
battement for which there was not a statistically significant difference between groups in
mean global thigh sagittal angle (thigh elevation) at the peak of the step. For this reason,
the battement jet to 45 was chosen for further analysis. For the battements derriere
(kicks backwards), a statistically significant difference between groups in mean sagittal
thigh elevation at the peak of the step was obtained for all steps except the battement
tendu derriere (kick backward along the floor), therefore this step was selected as the
battements derriere step that would undergo further analysis. Thus, the battement jet
devant to 45 and battement tendu devant were the two steps chosen to represent the
battement movement category.
Fourteen steps, two from each movement category, were included in the between group
analysis of the nine kinematic measurement variables that represent the fundamental
principles of classical ballet technique. These inter-group comparisons are presented in
the next section.
3.3.3 Between Group Analysis of Fundamental Principles with Respect to Joint and
Segment Kinematics
Results for the comparison of kinematic variables between the non-professional and
professional groups will be presented with respect to the four fundamental principles of
classical ballet technique that have been identified by literature review. Group means for
each of the nine kinematic measurement variables selected as indicators of the four
fundamental principles, and results for independent-samples t-tests between group means,
are presented in Figures 3.27 3.44. Significant differences between group means are
indicated by asterisks (*). Descriptive statistics (means, standard deviations, and pvalues) for each of the nine kinematic measurement variables are presented in Appendix
H. Results of the general linear model analyses are presented in Tables 3.4 3.15.
It should be noted that although all participants completed all steps during the data
collection sessions, there is some variability between kinematic variables in the sample
sizes presented in the results. This is because for some travelling steps (i.e. some jets and
glissades) the subjects were on the edge of the capture volume as the step commenced
- 161 -
and it was therefore not possible to resolve all marker placements for all participants at
this time point. Since the general linear model includes data from multiple steps, a
reduction in data points for one step will decrease the resulting sample size for the whole
analysis.
3.3.3.1 Alignment
The principle of alignment relates to the orientation and shape of the upper body.
Specifically, this principle is defined by the characteristics of maintaining the verticality
of the thorax and minimising the natural curvatures of the spine during execution of ballet
movements. The kinematic measurement variables selected as biomechanical indicators
of this principle are thorax anterior/posterior tilt and thorax side tilt. Group means for
these parameters at the start and peak of each step, and results for independent-samples ttests between group means are presented in Figures 3.27 - 3.30 Results of the repeated
measures general linear model analysis, conducted to adjust for multiple comparisons, are
presented in Table 3.4 and Table 3.5.
3.3.3.1.1 Thorax Anterior/Posterior Tilt

Figures 3.27 and 3.28 indicate that for most steps and time points, non-professional and
professional dancers held their torso in a position of posterior thoracic tilt. Significant
differences between groups in thorax anterior/posterior tilt were only obtained for the
start and peak of the grand pli in first position, the start of the grand pli in second
position, and the peak of the temps lev. There is no observable trend when comparing
mean thorax anterior/posterior tilt between groups at the start of each step. That is, greater
mean thorax posterior tilt was demonstrated by the professionals for half of the steps,
with this relationship being reversed for the other half of steps. Greater posterior thoracic
tilt was demonstrated by the professionals at 71% of step peaks (10 out of 14), however
statistical significance was only achieved for two of these steps.
- 162 -
Thorax Anterior/Posterior Tilt at Start of Step (degrees)
20
Non-professional (N=11)
Anterior
15
Professional (N=11)
10
5
Posterior
0
-5
-10
Double Pirouette En Dedans
Double Pirouette En Dehors
Step
Grand Jete Elance En Avant
Jete Ordinaire Derriere
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Battement Jete 45 Front
Grand Plie 2nd Position
Grand Plie 1st Position
Battement Tendu Back
-20
Rise 1st Position
-15
Figure 3.27 Thorax anterior/posterior tilt at start of step.

* Significant difference between groups (t-test: p<0.05). Error bars = 1 standard deviation
Thorax Anterior/Posterior Tilt at Peak Step (degrees)
20
Anterior
15
Professional (N=11)
10
5
Posterior
0
-5
-10
Figure 3.28 Thorax anterior/posterior tilt at peak of step.

- 163 -
Jete Ordinaire Derriere
Glissade Over
Step
Glissade Under
Temps Levee
Saute 1st Position
Rise 2nd Position
Rise 1st Position

n
-20
-15
Time point
Mean Difference () (Non- Standard

professional Error
Professional)
p-value
95% Confidence Interval for

Difference
Lower Bound
Upper Bound
Start
-0.468
0.808
0.569
-2.153
1.218
Peak
-1.313
0.937
0.175
-3.252
0.627
Start & Peak

1.133
0.838
0.192
0.616
2.882
Table 3.4 Pairwise comparison of thorax anterior/posterior tilt for all steps and start, peak, and start and
peak combined. Based on estimated marginal means.
*Mean difference is significant at the 0.05 level
Adjustment for multiple comparisons.
Table 3.4 shows the results of the repeated measures general linear model analyses,
conducted to adjust for multiple comparisons of thorax anterior/posterior tilt across
multiple steps at multiple time points (i.e. start and peak). These results indicate that
when examining the effect of group on thorax anterior/posterior tilt for the start and
peak for all 14 steps, a significant difference between the non-professional and
professional groups (i.e. p > 0.05) was not obtained for the start, the peak, or when
combining the start and peak of steps.
3.3.3.1.2 Thorax Side Tilt

Figures 3.29 and 3.30 indicate that for thoracic side tilt, the only significant difference
between non-professional and professional dancers was found for the peak of the grand
jet elanc en avants. In terms of the group mean values, all dancers exhibited minimal
deviation of the thorax from vertical (0) in the coronal plane, with relatively small
standard deviations, for the plis, battements, rises, and sauts. For these steps, dancers
held their thorax within 1 or 2 of the vertical for the start and peak of the movements.
Overall, there is no observable trend when comparing values for mean thoracic side tilt
between groups at the start and peak of each step. For the few steps where discrepancies
between groups are indicated there is no consistent direction of this deviation.
- 164 -
Thorax Side Tilt at Start of Step (degrees)

15
Professional (N=11)
Right
10
5
Left
0
-5
-10
Step
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-15
Figure 3.29 Thorax side tilt at start of step.

Thorax Side Tilt at Peak of Step (degrees)
Non-professionals (N=11)
Right
10
Professional (N=11)
Left
0
-5
-10
Figure 3.30 Thorax side tilt at peak of step.

- 165 -
Jete Ordinaire
Glissade Over
Glissade Under
Step
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-20
-15
Time point
Mean Difference ()
(Non-professional Professional)
Standard
Error

p-value
Difference
Lower Bound
Upper Bound
Start
0.682
0.542
0.223
-0.449
1.812
Peak
-0.360
0.451
0.433
-1.293
0.573
Start & Peak

0.081
0.383
0.834
-0.717
0.880
Table 3.5 Pairwise comparison of thorax side tilt for all steps and start, peak, and start and peak
combined. Based on estimated marginal means.
conducted to adjust for multiple comparisons of thorax side tilt across multiple steps at
multiple time points (i.e. start and peak). These results indicate that when examining the
effect of group on thorax side tilt for all 14 steps, there is no significant difference (i.e.
p > 0.05) for this variable between the non-professional and professional dancers for the
start, the peak, or when combining the start and the peak of the steps.
3.3.3.2 Placement
The principle of placement relates to the orientation and range of motion of the pelvis.
Specifically, this principle is defined by the characteristics of maintaining a centred or
neutral pelvic position and minimising pelvic displacement when executing movements
of the lower limb. The kinematic measurement variables selected as biomechanical
indicators of this principle are pelvic anterior/posterior tilt, pelvic obliquity, and pelvic thoracic transverse rotation. Group means for these parameters at the start and peak of
each step, and results for independent-samples t-tests between groups, are presented in
Figures 3.31 3.36. Results of the repeated measures general linear model analysis,
conducted to adjust for multiple comparisons, are presented in Tables 3.6 -3.8.
3.3.3.2.1 Pelvic Anterior/Posterior Tilt

Figures 3.31 and 3.32 indicate that for the majority of steps and time points, both groups
of dancers held their pelvis in a position of anterior pelvic tilt. Significant differences
between the non-professional and professional groups in mean pelvic anterior/posterior
tilt were only obtained for the start of the glissade under and the peak of the rise in first
position. Despite significant differences only being obtained for two parameters, a
- 166 -
notable trend shown in figures 3.31 and 3.32 is one of increased anterior pelvic tilt for
the non-professional group compared to the professional group for the majority of steps at
both the start and peak time points.
Pelvic Anterior/Posterior Tilt at Start of Step (degrees)

30
Professional (N=12)
Anterior
20
10
-10
Step
Figure 3.31 Pelvic anterior/posterior tilt at start of step.
- 167 -
Jete Ordinaire
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-30
Glissade Over
-20
Glissade Under
Posterior
Pelvic Anterior/Posterior Tilt at Peak of Step (degrees)
Professional (N=12)
Anterior
40
30
20
10
-10
Step
Figure 3.32 Pelvic anterior/posterior tilt at peak of step.
Time point
Mean Difference ()
Standard
p-value
Error
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
-30
Rise 2nd Position
-20
Rise 1st Position
Posterior

Difference
Lower Bound
Upper Bound
Start
-0.591
0.894
0.515
-2.440
1.259
Peak
-2.260*
0.926
0.023
-4.175
-0.344
Start & Peak

-1.425
0.775
0.079
-3.028
0.178
Table 3.6 Pairwise comparison of pelvic anterior/posterior tilt for all steps and start, peak, and start and
*Mean difference is significant at the 0.05 level
conducted to adjust for multiple comparisons of pelvic anterior/posterior tilt across
multiple steps at multiple time points (i.e. start and peak). These results indicate that
when examining the effect of group on pelvic anterior/posterior tilt for all 14 steps,
there is no significant difference (i.e. p > 0.05) for this variable between the nonprofessional and professional dancers for the start, or when consider the start and the peak
together, however there is a statistically significant difference when only considering the
peak of the movements (p = 0.023).
- 168 -
3.3.3.2.2 Pelvic Obliquity

Figures 3.33 and 3.34 indicate that for both groups of dancers least pelvic obliquity was
demonstrated at the start of the plis, battements, rises, and sauts. Significant
differences between the non-professional and professional groups in mean pelvic
obliquity were only obtained for the start of the jet ordinaire. For this step and time
point, the non-professional group demonstrated increased lift of the pelvis on the
supporting side of the body compared to the professional group. Overall, the mean
values for pelvic obliquity are similar for each time point for each group for the majority
of the 14 steps. However, despite not reaching statistical significance, there are a few
steps and time points (start and peak pirouette en dehors, start and peak pirouette en
dedans, peak temps lev, peak gliassade under, peak glissade over) where discrepancies
between group means are evident, and for each of these the observable trend is that the
professional group held their pelvis closer to neutral coronal plane placement than the
non-professional group.
Pelvic Obliquity at Start of Step (degrees)

Professional (N=12)
30
Up
20
10
Down
0
-10
Figure 3.33 Pelvic obliquity at start of step.

- 169 -
Jete Ordinaire
Glissade Over
Glissade Under
Step
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-30
-20
Pelvic Obliquity at Peak of Step (degrees)
Professional (N=12)
30
Up
20
10
Down
0
-10
-20
Step
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-30
Figure 3.34Pelvic obliquity at peak of step.

Time point
Mean Difference ()
Standard
p-value
Error

Difference
Lower Bound
Upper Bound
Start
-0.005
0.660
0.995
-1.370
1.361
Peak
-0.159
0.846
0.852
-1.909
1.590
Start & Peak

-0.082
0.623
0.896
-1.370
1.206
Table 3.7 Pairwise comparison of pelvic obliquity for all steps and start, peak, and start and peak
conducted to adjust for multiple comparisons of pelvic obliquity across multiple steps at
multiple time points (i.e. start and peak). These results indicate that when examining the
effect of group on pelvic obliquity for all 14 steps, there is no significant difference (i.e.
p > 0.05) for this variable between the non-professional and professional dancers for the
start, the peak, or when combining the start and the peak.
- 170 -
3.3.3.2.3 Pelvic Thoracic Transverse Rotation

Figures 3.35 and 3.36 indicate that significant differences between the non-professional
and professional groups in mean pelvic thoracic transverse rotation were only obtained
for the start and peak of the grand jet elanc en avant, and the peak of the battement jet
front. There is no notable trend when comparing the direction of discrepancy in mean
values for these statistically significant different steps and time points. For all other steps,
the mean values for pelvic thoracic transverse rotation are very similar for each time
point for each group. For the few steps where discrepancies between groups are indicated
there is no consistent direction of this deviation. Minimal pelvic thoracic rotation was
demonstrated for the start and peak of the plis, rises and sauts.
Pelvic-Thoracic Transverse Rotation at Start of Step (degrees)

15
Professional (N=12)
Forward
10
5
0
-10
Jete Ordinaire
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-20
Glissade Over
-15
Backward
-5
Step
Figure 3.35 Pelvic- thoracic transverse rotation at start of step.
- 171 -
Pelvic-Thoracic Transverse Rotation at Peak of Step (degrees)
50
Professional (N=12)
40
Forward
30
20
10
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
-20
Rise 1st Position
-10
Backward
Step
Figure 3.36 Pelvic thoracic transverse rotation at peak of step.
Time point
Mean Difference ()
Standard
p-value
Error
Difference
Lower Bound
Upper Bound
Start
-0.032
0.828
0.969
-1.759
1.694
Peak
-0.281
0.714
0.698
-1.758
1.196
Start & Peak

-0.244
0.702
0.732
-1.707
1.220
Table 3.8 Pairwise comparison of pelvic - thoracic transverse rotation obliquity for all steps and start,
peak, and start and peak combined. Based on estimated marginal means.
comparisons.
Adjustment for multiple
conducted to adjust for multiple comparisons of pelvic - thoracic transverse rotation
across multiple steps at multiple time points (i.e. start and peak). These results indicate
that when examining the effect of group on pelvic - thoracic transverse rotation for all
14 steps, there is no significant difference (i.e. p > 0.05) for this variable between the
non-professional and professional dancers for the start, the peak, or when combining the
start and the peak.
- 172 -
3.3.3.3 Turnout
The principle of turnout relates to the rotation of the lower limb in the transverse plane.
Specifically, this principle is defined by the characteristics of the dancer trying to achieve
and maintain maximal external transverse plane rotation of the lower limb so as to
increase the external foot progression angle. The kinematic measurement variables
selected as biomechanical indicators of this principle are hip transverse plane rotation and
ankle transverse plane rotation. Group means for these parameters at the start and peak of
each step, and results for independent-samples t-tests between groups, are presented in
Figures 3.37 3.40. Results of the repeated measures general linear model analysis,
conducted to adjust for multiple comparisons, are presented in Tables 3.9 and 3.10.
3.3.3.3.1 Hip Transverse Rotation
Figures 3.37 and 3.38 depict that both groups of dancers exhibit external hip rotation for
the start and peak of all steps except the grand jet elanc en avants. There is a clear
trend of increased external hip rotation demonstrated by the professional dancers
compared to the non-professional dancers at the start and peak of the majority of steps.
The only steps and time points for which this was not the case was the start of the jet
ordinaire, the peak of the grand pli in second position, and the peak of the grand jet
elanc en avants. Significant differences between groups in hip external rotation were
obtained for 8 of the 14 steps at the start time point, and for 7 of the 14 steps at the peak
time point.
- 173 -
Hip Transverse Rotation at Start of Step (degrees)
10
0
External
-10
-20
-30
-40
Step
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-50
Professional (N=12)
Figure 3.37 Hip transverse rotation at start of step.

Hip Transverse Rotation at Peak of Step (degrees)
20
10
External
0
-10
-20
-30
-40
Figure 3.38 Hip transverse rotation at peak of step.

- 174 -
Jete Ordinaire
Glissade Over
Step
Glissade Under
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-50
Temps Leve
Professional (N=12)
Time point
Mean Difference ()
Standard
p-value
Error

Difference
Lower Bound
Upper Bound
Start
-4.704*
1.209
0.001
-7.205
-2.204
Peak
-4.137*
1.799
0.031
-7.859
-0.416
Start & Peak

-4.421*
1.438
0.005
-7.395
-1.447
Table 3.9 Pairwise comparison of hip transverse rotation for all steps and start, peak, and start and peak
conducted to adjust for multiple comparisons of hip transverse rotation across multiple
steps at multiple time points (i.e. start and peak). These results indicate that when
examining the effect of group on hip transverse rotation for all 14 steps, there is a
significant difference (i.e. p < 0.05) for this variable between the non-professional and
professional dancers. When considering all steps and time points, the professional
dancers exhibited a mean value of 4.4 more external hip rotation than the nonprofessional dancers for the start, the peak, and when combining the start and the peak.
3.3.3.3.2 Ankle Transverse Rotation

Figures 3.39 and 3.40 depict a trend of increased external ankle rotation demonstrated by
the professional dancers compared to the non-professional dancers at the start and peak of
the majority of steps. The few steps and time points for which this was not the case were
the start of the grand pli in first position, the peak of the grand pli in second position,
the peak of the saut in first position, the peak of the jet ordinaire, and the start and peak
of the grand jet elanc en avants. Despite an observable trend of increased external
ankle rotation demonstrated by the professional dancers compared to the nonprofessionals, significant differences between groups in ankle external rotation were only
obtained for 3 of the 14 steps at the start time point, and for 2 of the 14 steps at the peak
time point.
- 175 -
Ankle Transverse Rotation at Start of Step (degrees)

Professional (N=7)
Internal
30
20
10
0
External
-10
-20
-30
-40
Step
Figure 3.39 Ankle transverse rotation at start of step.
Ankle Transverse Rotation at Peak of Step (degrees)
Double Pirouette En Dehorss
Jete Ordinaire
Glissade Over
Glissade Underr
Temps Leve
n
Rise 2nd Position
Rise 1st Position
-60
Saute 1st Position
-50
30
Professional (N=7)
Internal
20
10
0
-10
External
-20
-30
-40
Step
Figure 3.40 Ankle transverse rotation at peak of step.
- 176 -
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Rise 2nd Position

n
n
Rise 1st Position
-60
Saute 1st Position
-50
Time point
Mean Difference ()
Standard
p-value
Error

Difference
Lower Bound
Upper Bound
Start
-5.292*
2.393
0.046
-10.461
-0.123
Peak
-1.640
1.650
0.331
-5.053
1.774
Start & Peak

-3.871
1.891
0.061
-7.957
0.215
Table 3.10 Pairwise comparison of ankle transverse rotation for all steps and start, peak, and start and
conducted to adjust for multiple comparisons of ankle transverse rotation across multiple
examining the effect of group on ankle transverse rotation for all 14 steps, on average
the professionals exhibited 5.3 more external ankle rotation than the non-professional
dancers for the start, but for the peak, and when combining the start and the peak, there
are not a significant differences (i.e. p > 0.05) for this variable between the nonprofessional and professional dancers.
3.3.3.4 Extension
The principle of extension relates to the extension of the lower limb in the sagittal plane.
Specifically, this principle is defined by the characteristics of the dancer trying to achieve
maximal extension at the knee and ankle joints in order to create an elongated and
lengthened appearance of the limb in accordance with the aesthetic requirements of the
classical ballet art form. The kinematic measurement variables selected as biomechanical
indicators of this principle are knee flexion/extension and ankle dorsi/plantarflexion.
Group means for these parameters at the start and peak of each step, and results for
independent-samples t-tests between groups, are presented in Figures 3.41 3.44. Results
of the repeated measures general linear model analysis, conducted to adjust for multiple
comparisons, are presented in Tables 3.11 and 3.12.
3.3.3.4.1 Knee Flexion/Extension

Figures 3.41 and 3.42 show group means for knee flexion/extension at the start and peak
of the steps for which maximum knee extension is required at these time points. Data are
- 177 -
omitted for the start of the steps in the jump, glissade and jet movement categories, and
the peak of the steps in the pli movement category, because maximum knee extension is
not required at these time points for these steps.
Figures 3.41 and 3.42 show that for
most steps, both groups of dancers achieve a position of knee hyperextension, but there is
a clear trend of increased knee extension demonstrated by the professional dancers
compared to the non-professional dancers at the start and peak of all 14 steps. Significant
differences between groups in knee flexion/extension were obtained at the start time point
for 5 of the 8 steps for which maximum knee extension was required at this time point.
Significant differences were obtained at the peak time point for 5 of the 12 steps for
which maximum knee extension was required at this time point.
Knee Flexion/Extension at Start of Step (degrees)

20
15
Professional (N=12)
Extension
10
5
0
-5
-10
Figure 3.41 Knee flexion/extension at start of step.

- 178 -
Double Pirouette En Dedanss
Jete Ordinaire
Glissade Over
Glissade Under
Step
Temps Leve
Saute 1st Position
Rise 2nd Position
Battement Jete 45 Frontt
Battement Tendu Backk
n

n
-20
Rise 1st Position
-15
Knee Flexion/Extension at Peak of Step (degrees)

20
Professional (N=12)
Flexion
15
10
5
Extension
0
-5
-10
Step
Figure 3.42 Knee flexion/extension at peak of step.
Time point
Mean Difference ()
Standard
p-value
Error
Jete Ordinaire
Glissade Overr
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Battement Tendu Backk
-20
Rise 1st Position
-15

Difference
Lower Bound
Upper Bound
Start
-2.979*
1.227
0.023
-5.511
-0.447
Peak
-2.881*
1.281
0.034
-5.532
-0.230
Start & Peak

-3.097*
1.201
0.017
-5.582
-0.612
Table 3.11 Pairwise comparison of knee flexion/extension for all steps and start, peak, and start and peak
conducted to adjust for multiple comparisons of knee flexion/extension across multiple
examining the effect of group on knee flexion/extension for all 14 steps, there is a
significant difference (i.e. p < 0.05) for this variable between the non-professional and
professional dancers. When considering all steps and time points, the professional
dancers exhibited a mean value of 3.0 more knee extension than the non-professional
dancers for the start, the peak, and when combining the start and peak timepoints.
- 179 -
3.3.3.4.2 Ankle Dorsi/Plantarflexion

Figures 3.43 and 3.44 show group means for ankle dorsi/plantarflexion at the start and
peak of the steps for which maximum ankle plantarflexion is required at these time
points. Data are omitted for the start of all steps except the pirouette en dehors and
pirouette en dedans, and for the peak of the steps in the pli category of movement. This
is because maximum ankle plantarflexion is not required at these time points for these
steps.
Figures 3.36 and 3.37 show that for the majority of steps and time points, both
groups of dancers achieved a mean of at least 40 plantarflexion, and a trend of increased

ankle plantarflexion demonstrated by the professionals compared to the non-professional
is observed for the start of the pirouette en dehors, and for the peak of all steps.
Significant differences between groups in ankle dorsi/plantarflexion were not obtained at
the start of the pirouettes en dehors or pirouette en dedans. Significant differences were
obtained at the peak time point for 4 of the 14 steps for which maximum ankle
plantarflexion was required at this time point.
Ankle Dorsi/Plantarflexion at Start of Step (degrees)

0
-10
Plantarflexion
-20
-30
-40
-50
-60
-70
Professional (N=12)
Step
Figure 3.43 Ankle dorsi/plantarflexion at start of step.
- 180 -
Jete Ordinaire
Glissade Over
Glissade Under
Temps Leve
Saute 1st Position
Rise 2nd Position
Rise 1st Position
-80
Ankle Dorsi/Plantarflexion at Peak of Step (degrees)

0
-10
Plantarflexion
-20
-30
-40
-50
-60
-70
Grand Jete Elance En Avantt
Jete Ordinaire
Glissade Overr
Temps Leve
Glissade Under
Saute 1st Position
Rise 2nd Position
Rise 1st Position
Professional (N=12)
Profes
-80
Step
Figure 3.44 Ankle dorsi/plantarflexion at peak of step.
Time point
Mean Difference ()
Standard
p-value
Error

Difference
Lower Bound
Upper Bound
Start
-1.202
2.044
0.562
-5.421
3.016
Peak
-3.802*
1.621
0.028
-7.156
-0.449
Start & Peak

-3.503*
1.564
0.035
-6.737
-0.268
Table 3.12 Pairwise comparison of ankle dorsi/plantarflexion for all steps and start, peak, and start and
conducted to adjust for multiple comparisons of ankle dorsi/plantarflexion across multiple
examining the effect of group on ankle dorsi/plantarflexion for all 14 steps, there is a
significant difference (i.e. p > 0.05) for this variable between the non-professional and
professional dancers. When considering all steps, compared to the non-professional
dancers the professional dancers exhibited 3.8 more plantarflexion at the peak of the
steps, and 3.5 more plantarflexion for the start and peak time points combined.
- 181 -
Summaries of the results for the repeated measures general linear model analyses for each
of the nine kinematic measurement variables are presented in Tables 3.13 3.15. These
tables show that using this form of statistical analysis, in which adjustments were made
for multiple comparisons of the same variable across multiple steps, of all of the 9
measurement variables, statistically significant differences in group means were only
obtained for 3 variables at the start of the steps, for 4 variables at the peak of the steps,
and for 3 variables when considering both the start and peak together. Most notably,
significant differences between groups were found for hip transverse rotation and knee
extension when considering start and peak data seperately, and also when combining start
and peak time points. These parameters fall into 2 of the 4 fundamental principle
categories that were indentied for classical ballet technique, i.e. turnout and extension.
Biomechanical
Principle
Alignment
Placement
Turnout
95% Confidence Interval

a
for Difference
Kinematic Variable
Mean
Difference
(Nonprofessional
Professional)
Std.
Error
p-value
Thorax Ant/Post Tilt
-0.468
0.808
Thorax Side Tilt
0.682
Pelvic Ant/Post Tilt
Lower
Bound
Upper
Bound
0.569
-2.153
1.218
0.542
0.223
-0.449
1.812
-0.591
0.894
0.515
-2.440
1.259
Pelvic Obliquity
-0.005
0.660
0.995
-1.370
1.361
Pelvic - Thoracic
Rotation
-0.032
0.828
0.969
-1.759
1.694
Hip Rotation
-4.704*
1.209
0.001
-7.205
-2.204
Ankle Rotation
-5.292*
2.393
0.046
-10.461
-0.123
Knee Extension
-2.979*
1.227
0.023
-5.511
-0.447
Ankle
-1.202
2.044
0.562
-5.421
3.016
Plantarflexion
Table 3.13 Summary of pairwise comparisons between groups at start of step, using repeated measures
general linear model. Based on estimated marginal means.
* The mean difference is significant at the 0.05 level.
a
Extension
- 182 -
Biomechanical
Principle
Std.
Error
p-value
0.937
Thorax Side Tilt
-0.360
Kinematic Variable
Placement
Turnout
Lower
Bound
Upper
Bound
0.175
-3.252
0.627
0.451
0.433
-1.293
0.573
-2.260*
0.926
0.023
-4.175
-0.344
Pelvic Obliquity
-0.159
0.846
0.852
-1.909
1.590
Pelvic - Thoracic
Rotation
-0.281
0.714
0.698
-1.758
1.196
Hip Rotation
-4.137*
1.799
0.031
-7.859
-0.416
Ankle Rotation
-1.640
1.650
0.331
-5.053
1.774

Alignment

a
for Difference
Mean
Difference
(Nonprofessional
Professional)
-1.313
-2.881*
1.281
0.034
-5.532
-0.230
Knee Extension
Extension
Ankle
-3.802*
1.621
0.028
-7.156
-0.449
Plantarflexion
Table 3.14 Summary of pairwise comparisons between groups at peak of step, using repeated measures
general linear model. Based on estimated marginal means.
a
Biomechanical
Principle
Alignment
Placement
Turnout

a
for Difference
Kinematic Variable
Mean
Difference
(Nonprofessional
Professional)
Std.
Error
p-value
1.133
0.838
Thorax Side Tilt
0.081
Lower
Bound
Upper
Bound
0.192
0.616
2.882
0.383
0.834
-0.717
0.880
-1.425
0.775
0.079
-3.028
0.178
Pelvic Obliquity
-0.082
0.623
0.896
-1.370
1.206
Pelvic - Thoracic
Rotation
-0.244
0.702
0.732
-1.707
1.220
Hip Rotation
-4.421*
1.438
0.005
-7.395
-1.447
Ankle Rotation
-3.871
1.891
0.061
-7.957
0.215
Knee Extension
-3.097*
1.201
0.017
-5.582
-0.612
Extension
Ankle
-3.503*
1.564
0.035
-6.737
-0.268
Plantarflexion
Table 3.15 Summary of pairwise comparisons between groups for start and peak time points combined,
using repeated measures general linear model. Based on estimated marginal means.
a
- 183 -
3.3.4 Analysis of Fundamental Principles with Respect to Movement Categories and

Steps
To further analyse the kinematic data with respect to the four fundamental principles of
classical ballet technique, results are also displayed in a test matrix in which each
movement category and step is displayed against each fundamental biomechanical
principle. The test matrix, presented in Table 3.16, summarises the result of the
independent-samples t-tests between the professional and non-professional groups for
each kinematic variable.
Table 3.16 shows that with respect to the Seven Movements of Dance (Noverre, 1760),
the plis were most effective in detecting deviations in alignment, specifically in
thoracic anterior/posterior tilt. Significant differences between groups were obtained for
the start and peak of the grand pli in first position, and for the start of the grand pli in
second position.
The battement steps were able to detect differences between groups in two of the
fundamental principles, turnout and extension. Statistically significant differences
were obtained for all inter-group comparisons of hip rotation (i.e. turnout) and knee
extension (i.e. extension) for the start and peak of the battement tendu derriere and the
battement jet devant. Deviations in turnout were also highlighted well by the relev
and pirouette movement categories. For the relevs in first and second position,
differences between groups were observed for hip rotation and ankle rotation, while for
the pirouettes, differences in hip rotation were detected. None of the movement
categories were effective in detecting differences in placement. That is, significant
differences between groups were not obtained for pelvic variables for more than one step
per movement category. Thus, with respect to the four fundamental principles of classical
ballet technique, the movement categories of bend (pli), stretch (battement), rise
(relev), and turn (tour) were best able to detect kinematic differences between groups
across the multiple time points.
- 184 -
BIOMECHANICAL PRINCIPLE
ALIGNMENT
MOVEMENT
CATEGORY
1. BEND
(PLI)
2. STRETCH
(BATTEMENT)
3. RISE
(RELEV)
4. JUMP
(SAUT)
5. GLIDE
(GLISS)
6. DART
(ELANC)
7. TURN
(TOUR)
STEP
PHASE
Thorax
Ant/
Post
Tilt
Grand
Pli 1st
Position
Start
9
Peak
9
Grand
Pli 2nd
Position
Start
9
Peak
8
Battement
Tendu
Back
Start
Peak
Battement
Jet 45
Front
Thorax
Side
Tilt
8
8
8
8
PLACEMENT
Pelvic
Ant/
Post
Tilt
8
Pelvic
Obliq
PelvicThorax
Rot
8 8
8
8
TURNOUT
EXTENSION
Hip
Ext
Rot
Knee
Ext
9
n/a
8 n/a
n/a
9
8 9
n/a
n/a
n/a
8
8
8
8
9
9
n/a
8
8
8
8
9
8
9
9
Start
8
8
8
8
9
8
9
n/a
Peak
8
8
8
9
9
8
9
Rise 1st
Position
Start
8 8 8 8
9
n/a
Peak
8
9
9
Rise 2nd
Position
Start
8
8 8 8 8
9
n/a
Peak
8
8
8
9
9
Saut 1st
Position
Start
8
8 8 8
8 n/a
Peak
8
8
8 8
Temps
Lev
Start
8
8 8 8
8 n/a
n/a
Peak
9
8
8 8 8
8 9
9
Glissade
Under
Start
8
8
9
n/a
Peak
8
8 8 8
8 8
Glissade
Over
Start
8
8
8
8
8 n/a
n/a
Peak
8
8
8
8
8 9
9
Jet
Ordinaire
Start
8
9
8
8
8 n/a
n/a
Peak
8
Grand
Jet
Elanc En
Avants
Double
Pirouette
En Dehors
Double
Pirouette
En
Dedans
Start
Peak
9
Start
8
End
8
Start
8
End
8
8
9
8
Ankle
Plantflex
8 9
8 8
8
Ankl
e Ext
Rot
8 8
8
8
8
9
9
9
8 8 8 8
8 8 8
9
9
8 8 9

8
n/a
8 8
8
n/a
8
8
8
n/a
8 9
9
8
9
8
8
9
8 8
8
9
8
9
8
9
8
8
Table 3.16 Kinematic Variable Test Matrix
9 = Significant difference between professional and non-professional groups (t-test: p<0.05)

8 = No significant difference between professional and non-professional groups (t-test: p0.05)
n/a = Not applicable (Joint movement not executed at time point, (e.g. maximum knee extension and
plantarflexion not required at peak of grand plis in first and second positions).
Movement categories for which a significant difference between professional and non-professional groups
occurs for at least 3 of the 4 time points analysed. Steps for which statistically significant differences in group
means occurred for at least 3 of the 4 four fundamental principles
- 185 -
n/a
The test matrix in Table 3.16 also enables comparison of the extent to which statistically
significant differences in group means occur for each measurement variable in each
individual step. Results show that there are only 3 steps out of 16 for which statistically
significant differences in start and peak group means occur for at least 3 of the 4 four
fundamental principles of classical ballet technique. These three steps are the grand pli
in first position, the battement tendu back, and the grand jet elanc en avants.
Based on the data depicted in Table 3.16, further results for the three steps for which
statistically significant differences in group means were obtained for at least 3 of the 4
four fundamental principles are presented in Figures 3.45 to 3.55. These figures display
the time-normalised group means from the start (0%) to finish (100%) of each step for the
measurement variables for which statistically significant differences in group means were
obtained. For each figure, the vertical line represents the instant at which the peak of the
movement occurs. A red vertical line at the start and/or peak of the movement indicates
that a statistically significant difference in group means was obtained at that time point.
Significant differences in group means at the start and/or peak of the grand pli in first
position occurred for thorax anterior/posterior tilt, hip transverse rotation and knee
flexion/extension, these kinematic variables fall into the Alignment, Turnout and
Extension principle categories, respectively. Figures 3.45 and 3.46 show that for the
grand pli in first position, the professional group demonstrated increased thorax
extension and increased external hip rotation not only at the start and peak of this step,
but at all time points throughout the step. Figure 3.47 indicates that although a
statistically significant difference in knee flexion/extension was obtained at the start of
the grand pli in first position, the degree of knee flexion throughout the rest of the
movement was very similar between groups.
- 186 -
Thorax Tilt (Anterior / Posterior)

5
0
Degrees
10
20
30
40
50
60
70
80
90
100
-5
-10
-15
-20
Percentage Complete Movement (%)
PR ave (R)
NP ave (R)
Figure 3.45 Thorax anterior/posterior tilt for grand pli in first position.
Blue line = mean for professional dancers; green line = mean for nonprofessional dancers. Vertical lines indicate step start and peak (Red line
= significant difference at time point; Black line = no significant
difference at time point)
Hip Rotation (Internal / External)
0
-5
10
20
30
40
50
60
70
80
90
100
Degrees
-10
-15
-20
-25
-30
-35
-40
PR ave (R)
NP ave (R)
Figure 3.46 Hip transverse rotation for grand pli in first position.
Blue line = mean for professional dancers; green line = mean for nonprofessional dancers. Vertical lines indicate step start and peak (Red
line = significant difference at time point; Black line = no significant
Knee Flexion / Extension
160
140
Degrees
120
100
80
60
40
20
0
-20
10
20
30
40
50
60
70
80
90
100

PR ave (R)
NP ave (R)
Figure 3.47 Knee flexion/extension for grand pli in first position.

- 187 -
Significant differences in group means at the start and/or peak of the battement jet front
occurred for pelvic-thoracic transverse rotation, hip transverse rotation, and knee
flexion/extension, these kinematic variables fall into the Placement, Turnout and
Extension principle categories, respectively. Figure 3.48 indicates that although the
professional group started and finished the battement jet front with their pelvis-thoracic
rotation in a similar position to the non-professionals (i.e. close to neutral), between 20%
and 80% of the movement the professional group demonstrated increased forward pelvic
rotation with respect to the thorax compared to the amateur group. For the other variables
for which statistically significant differences were obtained at the start and/or peak of the
battement jet front, the direction of difference between group means at the start and/or
peak time points was consistent throughout the movement. That is, the professional group
demonstrated increased external hip rotation (Figure 3.49) and increased knee extension
(Figure 3.50) throughout the entire movement compared to the non-professional group.
Pelvic-Thoracic Transverse Rotation (Forward / Backward)

30
Degrees
20
10
0
0
10
20
30
40
50
60
70
80
90
100
-10
-20
-30
PR ave (R)
NP ave (R)
Figure 3.48: Pelvic-thoracic transverse rotation for battement jet front.

- 188 -

0
0
10
20
30
40
50
60
70
80
90
100
-5
Degrees
-10
-15
-20
-25
-30
-35
PR ave (R)
NP ave (R)
Figure 3.49: Hip transverse rotation for battement jet front.

Blue line = mean for professional dancers; green line = mean for
non-professional dancers. Vertical lines indicate step start and peak
(Red line = significant difference at time point; Black line = no significant

0
-2 0
10
20
30
40
50
60
70
80
90
100
-4
Degrees
-6
-8
-10
-12
-14
-16
-18
-20
PR ave (R)
NP ave (R)
Figure 3.50 Knee flexion/extension for battement jet front.

Significant differences in group means at start and/or peak of the grand jet elanc en
avants occurred for thorax side tilt, pelvic-thoracic transverse rotation, hip transverse
rotation, knee flexion/extension and ankle dorsi/plantarflexion. These kinematic variables
fall into each of the four principle categories of Alignment, Placement, Turnout and
Extension, respectively. Figure 3.52 shows that the professional group exhibited
decreased forward rotation of the pelvis relative to the thorax compared to the amateurs
for the entire grand jet elanc en avants. For all other variables for which statistically
significant differences were obtained at the start and/or peak of the grand jet elanc en
avants, there is variability in the direction of difference between group means for other
- 189 -
time points within the movement cycle. For example, as shown in Figures 3.54 and 3.55,
the professionals show increased knee extension and ankle plantarflexion from the start to
approximately 70% of the step (i.e. the airborne phase), but increased knee flexion and
ankle dorsiflexion from about 75% -100% (i.e. the landing phase) of the movement. The
direction of difference between group means for thorax side tilt and hip transverse
rotation also fluctuates during the movement cycle, as indicated in Figures 3.51 and 3.53
respectively.
Thorax Side Tilt (Leading/Following Side)

20
15
Degrees
10
5
0
-5
10
20
30
40
50
60
70
80
90
100
-10
-15
-20

PR ave (R)
NP ave (R)
Figure 3.51 Thorax side tilt for grand jet elanc en avants. Blue line =
mean for professional dancers; green line = mean for non-professional
dancers. Vertical lines indicate step start and peak (Red line =
significant difference at time point; Black line = no significant
Pelvic-Thoracic Transverse Rotation (Forward/ Backward)

40
Degrees
30
20
10
0
0
10
20
30
40
50
60
70
80
90
100
-10
-20
PR ave (R)
NP ave (R)
Figure 3.52 Pelvic-thoracic transverse rotation for grand jet elanc en

avants. Blue line = mean for professional dancers; green line = mean
for non-professional dancers. Vertical lines indicate step start and
peak (Red line = significant difference at time point; Black line = no
significant difference at time point)
- 190 -

10
5
Degrees
0
-5
10
20
30
40
50
60
70
80
90
100
-10
-15
-20
-25
-30

PR ave (R)
NP ave (R)
Figure 3.53 Hip transverse rotation for grand jet elanc en avants.
140
120
Degrees
100
80
60
40
20
0
0
10
20
30
40
50
60
70
80
90
100
-20

PR ave (R)
NP ave (R)
Figure 3.54 Knee flexion/extension for grand jet elanc en avants.

Ankle Dorsi / Plantar Flexion

60
Degrees
40
20
0
0
10
20
30
40
50
60
70
80
90
100
-20
-40
-60

PR ave (R)
NP ave (R)
Figure 3.55 Ankle dorsi/plantarflexion for grand jet elanc en avants.

difference at time point).
- 191 -
3.4 Discussion
A primary aim of this project was to determine the degree to which professional and nonprofessional ballet dancers execute classical ballet technique relative to the theoretical
principles, thereby identifying the level of agreement between the theoretical concepts
and practical execution of the technique at an elite level. To achieve this, 3D joint and
segment angular displacement data were obtained for professional and non-professional
ballet dancers performing 14 different ballet steps. The steps could each be assigned to
one of seven different movement categories identified in Noverres Seven Movements of
Dance (Noverre, 1760). Kinematic data were obtained for the thorax, pelvis, hips, knees
and ankles, and comparisons of mean kinematic variables were made between the
professional and non-professional dancers. Data were analysed with respect to the four
fundamental principles of classical ballet technique, as defined and described in Section
2.1.2. The four fundamental principles are alignment, placement, turnout, and
extension, and results will be discussed in relation to each of these principles
3.4.1 Alignment
In the literature review, the principle of alignment was defined as maintaining
verticality of the torso.
To measure this biomechanical principle, the kinematic
variables of thoracic anterior/posterior tilt and thoracic side tilt were selected as the
relevant parameters. In accordance with the theoretical principles of classical ballet, it
could be expected that professional dancers would hold their thorax in a more vertically
neutral position within the sagittal and coronal planes than non-professionals, and that
this would be revealed by smaller deviations of the thorax from vertical (i.e. from a
measurement value of zero) at the start and peak of each step. When considering all steps
collectively, there were no significant differences between groups in thoracic
anterior/posterior tilt or thoracic side tilt at either the start or peak of the movements. This
finding is in agreement with that of Wilson et al. (2004), who found no significant intergroup difference in sagittal and coronal plane trunk motion between novice and skilled
dancers, when using 3D motion analysis to investigate the grand rond de jambe en lair.
- 192 -
Interestingly, although in the dance literature much emphasis has been placed on
maintaining a verticality of the torso, it is evident from the current data (Figures 3.27 &
Figure 3.28) that both non-professional and professional dancers hold their torso in a
position of posterior thoracic tilt, rather than a truly vertical position. In support of this
finding, Wilson et al. (2004) reported mean anterior/posterior trunk angles of 18
posterior for skilled ballet dancers performing the grand rond de jambe en lair. When
comparing the data obtained for thoracic anterior/posterior tilt during execution of
classical ballet steps with that for normal adult gait, it is evident that the thoracic position
adopted by ballet dancers is extreme in relation to normal human locomotion. Sartor et
al. (1999) and Chung et al. (2010) reported mean global sagittal plane thoracic positions
of approximately 0 (i.e. approximately vertical), throughout the normal adult human gait
cycle. Thus, it seems than on a practical level, the desired aesthetic of grand manners
and extreme elegance and poise is actually achieved by taking ones thorax beyond a
position of verticality, into a position of posterior thoracic tilt. This was demonstrated to a
moderate extent by non-professional dancers, and in some instances to an even greater
degree by highly skilled professional dancers.
(Sartor et al., 1999; Chung et al., 2010)
With respect to the range of thoracic motion exhibited throughout the ballet steps, it is
evident that for stationary steps that do not require knee flexion or elevation off the floor
(i.e. battements; rises) there was little range of thorax anterior/posterior motion between
the step start and peak for either group of dancers. That is, at the start and peak of the
battements, and rises, all mean values for thoracic anterior/posterior tilt for both groups
of dancers were held within the range of 5 - 9 posterior thoracic tilt. Thus, despite some
discrepancy with the dance literature in relation to the thorax position being posterior
rather than vertical, thoracic range of motion data are consistent with the literature in the
sense that while executing these steps the dancers held their thorax very still, as indicted
by Blasis (1820, p24) while dancing, the body must remain quiet and absolutely steady,
and more recently by the Royal Academy of Dance (1997, p8), the position of the back
and sides is retained as far as possible. For the steps that required deep (i.e. grand plis)
or moderate (i.e. sauts; temps levs; jets) knee flexion the range of thoracic
anterior/posterior motion throughout the steps was not so limited. For example, at the
point of maximum knee flexion during the grand plis, both groups of dancers moved
their thorax approximately 12 anterior of their starting position, and during the
- 193 -
countermovement knee bend required for push-off into the airborne sauts, temps levs
and jets, all dancers moved their thorax to a more anterior position than that displayed
during most other steps. Also of note, is the observation that it was at the peak of the
steps requiring vertical elevation off the floor (i.e. sauts; temps levs) for which the
professional dancers demonstrated the greatest posterior tilt of the thorax. Therefore, this
group exhibited a higher range of trunk anterior/posterior motion between the start and
peak of the jumps (10 posterior) compared to the non-professionals (5 posterior). Such
a result contradicts the theoretical principle that ballet dancers are supposed to keep their
back still while performing jumps (the dancer is required to jump with lightness and
control, and to mask the effort by keeping the body erect, Royal Academy of Dance
1997, p79), in that practical data from highly skilled professionals indicate that some
movement of the trunk is in fact functionally necessary.
From a purely biomechanical perspective, increased forward lean of the trunk during a
grand pli, and during the preparatory phase of jumps, facilitates greater torque and
hence greater power generation for the hip extension required for ascent out of the grand
pli and for push-off into the air for jumps (Lees et al., 2004; Vanrenterghem et al.,
2008). Vanrenterghem et al. (2008) studied the effect of forward inclination on the power
output in vertical jumping. Subjects (non-dancers) performed maximal countermovement
jumps as they naturally would (i.e. with some forward thoracic flexion during
preparation), and also while holding the trunk as upright as possible. The more upright
position of the thorax during the countermovement preparation phase resulted in a 27%
reduction in hip joint flexion, and a consequent statistically significant decrease of 37% in
hip extension torque and power during push-off. The maximum jump height reached was
also significantly decreased by 10%. So, in the context of classical ballet jumps, the
theoretically desired aesthetic of minimal anterior/posterior motion of the trunk, yet
maximum elevation off the floor, are biomechanically opposing goals. The increased
range of trunk motion demonstrated by the professionals compared to the nonprofessionals indicates therefore that the more highly skilled group may be sacrificing
some of the theoretically ideal aesthetics required of the trunk, in order to access maximal
hip extension power during push-off, thus potentially enabling an aesthetically impressive
increase in jump height.
- 194 -
In relation to thoracic side tilt, for the symmetrical steps where the right and left sides of
the body performed the same movement (i.e. plis, rises, sauts), all dancers exhibited
very little thoracic movement in the coronal plane, and held their thorax within 1 or 2 of
the vertical for the start and peak of the movements. This result is therefore in strong
agreement with the definitions of the principle of alignment. It was the steps which
commenced while primarily balanced on a single leg (i.e. temps lev, glissades, jets,
pirouettes), that deviations from neutral in thoracic coronal movement were noted in both
groups of dancers. For example, when executing the temps lev (i.e. single leg hop), all
dancers displayed thoracic side tilt towards the supporting leg, and for the glissades
(gliding step), thoracic side tilt was towards the direction of travel. Thus, as was the case
for sagittal plane thoracic motion, larger ranges of motion in coronal plane thoracic
movement are associated with the functional requirements of the step.
3.4.2 Placement
The principle of placement has been described in the literature review as being
minimal displacement of the pelvis from a centred position (Section 2.1.3.2). To assess
this principle, pelvic positions with respect to the sagittal, coronal and transverse planes
were measured. Analysis of the kinematic data relating to pelvic anterior/posterior tilt
reveals that at the start and peak of the majority of steps, both groups of dancers held their
pelvis in a position of at least 10 anterior tilt. At the start of the steps that commenced
from a basic ballet stance, that is with straight knees, externally rotated hips, and upright
posture (i.e. plis, battements, rises), mean values for pelvic position ranged from 15 to
23 anterior for the non-professional dancers, and from 14 to 28 anterior for the
professional dancers. Thus, it is evident that if one considers a biomechanically neutral or
centred position to be 0, i.e. the convention typically applied in quantitative
biomechanical analysis, then the pelvis is very rarely actually held in a truly neutral
position while executing classical ballet movements. This finding is supported by the
studies of Gamboain et al. (1999; 2000), Deckert et al. (2007) and Holt et al. (2011) who
all used 2D video analysis to measure pelvic tilt. For all of these studies the steps
analysed were basic ballet movements in which neither leg was elevated off the floor.
Pelvic tilt ranges for each of these studies were 7 - 23 anterior (Gamboian et al., 1999;
Gamboian et al., 2000), 2 - 14 anterior (Holt et al., 2011), and 9 - 19 anterior (Deckert
et al., 2007). The sagittal plane position of the pelvis typically adopted in normal human
- 195 -
locomotion is approximately 10 anterior, with minimal excursion from this position

occurring throughout the gait cycle (Perry, 1992; Schwartz, 2004). The results of the
current study, as well as those of previous researchers, therefore indicate that on a
practical level, the highly desired neutral position of the pelvis, referred to in the dance
literature, is actually a position of anterior pelvic tilt, similar to that typically adopted
during normal gait. Interestingly, even highly skilled professional ballet dancers do not
achieve correct technique and placement of the pelvis by altering their sagittal pelvic
position from that typically held during normal locomotion.
Also of note in relation to the pelvic anterior/posterior tilt, is the observation that for steps
in which one leg was moved to the front (i.e. battement jet devant to 45) or back (i.e.
battement tendu derriere) of the body, the pelvis did in fact deviate from its starting
position. This was observed for both groups of dancers. When moving the leg to the front
in the battement jet devant, the pelvis moved from a starting position of 15 anterior for
both groups of dancers, to 8 and 5 anterior for professional and non-professional
dancers, respectively. Such a result contradicts the dance literature, in that Grieg (1994,
p41) for example, states that in battement to the front, the pelvis should remain
uninvolved, and White (1996, p83) writes do not allow the buttocks to push or tuck
under. For the step involving posterior leg movement, pelvic tilt increased to at least 25
anterior for both groups of dancers by the peak of the movements. Results showing
involvement of the pelvis during movement of the leg to the front or back of the body in
classical ballet (grand rond de jambe en lair and grand battements) have also been
reported by Wilson et al. (2004; 2007) and Bronner and Ojofeitimi (2011). Thus, despite
considerable theoretical information indicating that ideally there should be limited
movement of the pelvis in the sagittal plane when moving one leg to either the front or
back of the body, research data show that the pelvis is involved, even when the foot of the
working leg remains in contact with the floor, as was the case for the battement tendu
derriere. It should also be pointed out however, that when considering data from all steps
combined, there was a significant difference in pelvic anterior/posterior tilt between the
non-professional and professional dancers at the peak of the steps, with the professional
dancers demonstrating a mean value of 2.2 less anterior tilt than the non-professionals.
Given that there was not a significant difference between groups in this variable at the
start of the movements, this finding suggests that although some pelvic movement did
- 196 -
occur from start to peak, the more highly skilled ballet dancers were in fact able to
perform the ballet steps with a decreased range of sagittal plane pelvic motion, compared
to non-professional dancers.
With respect to pelvic obliquity, there was only one step and time point (i.e. start of jet
ordinaire) for which a significant difference was found between the professional and nonprofessional dancers. In this instance, the professional dancers exhibited raised pelvic
obliquity of 4 on the supporting side, compared to 7 demonstrated by the nonprofessionals. This difference is consistent with many of the theoretical references to
correct placement, (e.g. the horizontal and/or parallel lines of pelvis and shoulder in
relation to the parallel floor line, must not be upset Lee, 1983, p209), in that it indicates
that professional dancers are able to maintain their pelvis more closely to the horizontal,
even when performing single-leg support movements such as the start of the jet
ordinaire. Despite this result, when all steps were considered collectively, there were no
significant differences between groups in pelvic obliquity at the start, the peak, or for the
start and peak combined. As was the case for thoracic side tilt, in relation to pelvic
obliquity it was the symmetrical steps, where the right and left sides of the body
performed the same movement (i.e. plis, rises, sauts), that all dancers exhibited very
little pelvic movement in the coronal plane. For these steps, all dancers held their pelvis
within 1 of the horizontal for the start and peak of the movements, a result that also
supports the definitions of the principle of placement. However, pelvic obliquity for
both groups deviated up to as much as 18 for asymmetrical steps that required sideways
travel (e.g. glissade under), indicating that if a step involves movement from a single
spot, then functionally it is necessary to involve pelvic coronal plane motion.
Also related to the principle of placement, is the observation that for the battement jet
to 45 front, both groups of dancers raised the pelvis on the working side (i.e. the side of
leg movement) to 9 above the horizontal at the movement peak. A recurring theme in
dance literature is that the pelvis should not be lifted when elevating the leg to the front
(Vaganova, 1953; Karsavina, 1962; Grieg, 1994). Given that the dancers were only asked
to elevate their leg to an angle of 45, and that at best the literature relating to ballet
technique only concedes that there is displacement of the pelvis when the working leg
moves above 90 (Paskeva, 2002), this degree of pelvic movement for a relatively low
- 197 -
leg elevation from highly skilled professional dancers is somewhat surprising. Such a
result suggests that more coronal plane pelvis movement is demonstrated on a practical
level than would be expected based on theoretical information alone.
Lastly in relation to placement, the kinematic variable of pelvic thoracic transverse

rotation will be discussed. For this variable, when considering all steps collectively, no
significant differences between groups were obtained at the start, peak, or start and peak
combined. The only individual steps or time points for which significant differences were
found were the start of the glissade under, the peak of the battement jet front to 45 and
grand jet elanc en avants. The result for the battement jet is particularly interesting as
it indicates that, in contradiction to the dance literature, when one leg is elevated in front
of the body the pelvis and thorax do not necessarily remain facing the same plane. Many
dance authors (Karsavina, 1962; Lawson, 1983; White, 1996; Schorer, 1999) have
indicated that when one leg is elevated off the ground, the pelvis and thorax should
remain parallel to each other and facing the same plane. However, the results obtained in
the current study reveal that this was in fact the case for the non-professional dancers but
not for the professional dancers. Although lifting the leg to a statistically similar height
(Table 3.3), the professionals performed this movement with a pelvic - thoracic transverse
rotation value of 6 at the peak of the movement, compared to only 1 for the nonprofessionals. Additionally, in the battement tendu to the back, although a significant
difference was not obtained between groups for pelvic-thoracic transverse rotation, the
data indicate that for the professional dancers up to 9 of rotation took place between the
pelvis and thorax. This result is also in conflict with the theoretical suggestion that the
pelvis and thorax should remain parallel or square to each other. Thus, the dance theory
has again been challenged in practise as shown by the practical results obtained from
highly skilled professional ballet dancers.
3.4.3 Turnout
The results relating to the principle of turnout clearly indicate that professional dancers
perform classical ballet steps with more external rotation of the hips than nonprofessional dancers. Whether considering the start, the peak, or the start and peak
combined, significant differences between groups were found for this variable, with the
professional dancers exhibiting an average of 4.4 more external hip rotation than the
- 198 -
non-professionals. In terms of the theoretical dance literature, which clearly states that
extreme external rotation of the hip joints is an essential characteristic of classical ballet
technique, the current data are in strong agreement with this theoretical information, and
add convincing evidence to support this theoretical concept.
It can be seen that at the start of the steps, the professionals hip external rotation data
ranged from 4 through to 29 external, while values at the peaks ranged from 3 internal
to 41 external. On a practical level, there was therefore a considerable range in the
degree of hip turnout actively used by professional dancers. In discussing these values
of hip external rotation data however, it is important to consider the possible effects of the
Cardan angle rotation sequence on the values calculated. Lees et al. (2010) reported that
pelvic data and sagittal hip, knee and ankle data are robust with regard to Cardan angle
rotation sequence, but that hip transverse plane data are influenced by rotation sequence.
In particular, hip internal/external rotation values varied across a range of 60 when six
different rotation sequences were used, however this was only the case when hip flexion
moved beyond 50 (Lees et al., 2010). Below 50 hip flexion, hip internal/external
rotation data were not affected by the rotation sequence. Based on this finding, since the
majority of steps analysed in the current study did not involve hip flexion beyond 50, it
can be concluded that for most steps and time points the hip external rotation data can be
considered to be a reasonable representation of the actual movement occurring at the
joint. The step and time points for which hip transverse plane data may be inaccurate are
the start and peak of the grand jet elanc en avants (i.e. split leap). In this step, the
working leg is thrust forward into hip flexion to facilitate the split position.
Accordingly, the hip rotation data calculated for the start and peak of this step are
considerably different to those for all other steps, with values reaching into the internal
rotation range (Figure 3.38).
If the data for the grand jet elanc en avants are not considered, then hip rotation data
range from 10 to 40 external for the professional dancers across the start and peak of all
steps. It has been stated (Hardaker et al., 1984; Paskeva, 2002) that ideally dancers should
strive to dance with as close as possible to 90 of external rotation on each lower limb,
thus giving an overall appearance of 180 between the external progression line of each
foot. Some dance texts state that close to all of this rotation should occur in the hip
- 199 -
(Paskeva, 2002) while others have said that only 60 - 70 of the external rotation
actually occurs in the hip joint, with the rest of the rotation being facilitated primarily by
the ankle joint (Hardaker et al., 1984). Whether the ideal external hip rotation is
considered to be 60 or 90, in comparison to these values, only moderate values of hip
external rotation were achieved by the professional dancers in the current study. Thus,
although the professionals do exhibit increased external hip rotation compared to the nonprofessionals, it is interesting that the actual values for active external hip rotation
demonstrated by the professionals were relatively low compared to the so called
theoretical ideals. Moreover, the review of existing literature reveals that relatively
moderate values of external hip rotation have also been reported in previous studies that
have used 3D motion analysis to measure hip rotation in classical ballet (Shan, 2005;
Wilson et al., 2007; Bronner and Ojofeitimi, 2011). Bronner and Ojofeitimi (2011)
analysed the battement to the front, side and back of the body, and reported hip rotations
at the movement peak of 20 external, 10 internal and 20 external, respectively. Wilson
et al. (2007) reported external hip rotation angles between 30 and 45 during execution
of the grand rond de jambe en lair en dehors, and Shan (2005) obtained external hip
rotation angles ranging from 0 to approximately 30 during the grand jet en avants. The
results for these studies are therefore similar to the current study, in that the hip rotation
values are well below the theoretical ideal of 60 - 90 external rotation.
Data for ankle transverse rotation are partially in agreement with the theoretical literature.
For the start of the steps, professional dancers exhibited an average of 5.3 more external
ankle rotation than the non-professionals. This however was not the case when
considering the step peaks, or start and peaks combined. For these analyses, mean
differences between the groups reduced to only 1.6 and 3.9 degrees, respectively. It was
not possible to identify any previously published studies, that used comparable joint
movement definitions, to measure ankle external rotation in classical ballet dancers, and
therefore comparisons with existing data could not be made.
It is evident that the extent of external ankle rotation achieved is associated with or
determined by the sagittal plane motion of the ankle joint at each time point. For
example, in all plis, battements and rises, the ankle is in an approximately neutral
sagittal position as the steps commence, and for each of these steps ankle external
- 200 -
rotation ranged from 10 to 20. For the jumping or gliding steps (i.e. glissades), in which
the dancers commenced (i.e. pushed-off) from a position of ankle dorsiflexion, ankle
external rotation ranged from 35 to 40. At the other extreme, steps for which the
analysis limb commenced in a position of plantarflexion (i.e. jets and pirouettes),
external ankle rotation was very low with values even moving into low ranges of internal
rotation. A similar relationship was observed for the step peaks, at which all steps (with
the exception of the plis) require maximum plantarflexion, as this occurred in
association with ankle rotation values ranging from 4 external to 10 internal. Although
reference to the actual values for ankle internal/external rotation has been made, it should
be noted that it is the relationship of ankle sagittal plane and transverse plane data that is
the point of interest. As indicated in Section 3.2.2.6, the Cardan angle rotation sequence
has an effect on the offset of ankle transverse plane rotation data. Consequently, there is
uncertainty as to the accuracy of the actual values calculated. Despite this, it is still
possible to draw confident conclusions about the difference in ankle external rotation data
between groups, and to comment on the relative plantar/dorsiflexion and internal/external
rotation that occur concurrently for each ballet step. Although, due to different
methodologies, it is not possible to compare the actual values of ankle rotation obtained
with those from previous studies, Shan (2005) did report a similar relationship between
ankle joint rotation and ankle plantar/dorsiflexion, as that observed in the current study.
Thus, the forward or winged position, i.e. external ankle joint rotation, that has been
referred to in the dance literature (e.g. for the preservation of the turnout, the heel is held
forward and the toe kept back (Kostrovitskaya and Pisarev, 1978, p71)) does appear to
be better achieved by professional dancers than non-professionals in some instances, and
if total turnout is considered to be determined by inclusion of the foot progression
angle, then ankle rotation does in fact contribute to turnout to a greater extent in
professional dancers than non-professionals. However, the degree to which external
rotation of the ankle joint is biomechanically possible, even for the most highly skilled
dancer, is determined by the extent of concurrent plantarflexion or dorsiflexion, and is
considerably restricted during positions of extreme plantarflexion.
- 201 -
3.4.4 Extension
Knee joint extension and ankle plantarflexion were the kinematic variables selected to
investigate the principle of extension, i.e. maximum elongation of the lower limb.
Data indicate that for all instances when maximum knee extension was required, on
average the professional dancers exhibited 3.0 more knee extension than the nonprofessionals. Analysis of the actual knee flexion/extension values reveals that both
groups of dancers demonstrated a sagittal plane knee position beyond neutral, reaching
hyperextension for the majority of steps and time points. For the professional dancers,
the average knee hyperextension reached 13 in some instances (e.g. start of grand pli in
second position). Comparisons with previous studies is limited by the fact that the
majority of other researchers who have measured sagittal plane knee joint motion during
ballet movements have been interested in flexion rather than extension (Bronner et al.,
2002; Bronner and Ojofeitimi, 2006; Kulig et al., 2011). Shan (2005) was the only study
identified that did report on knee joint extension during execution of classical ballet,
however differences in the joint angle definitions make it difficult to directly compare
data. Thus, comparisons with previous research will not be discussed. However, it is
evident that in relation to knee extension, the current data are in agreement with the
theoretical principle that highly skilled ballet dancers display extreme elongation
(extension) of their lower limb.
The results for ankle plantarflexion are also consistent with the theoretical principle of
extension, in that for all steps and time points, professional dancers exhibited more
plantarflexion than non-professionals. The general linear model analysis produced
significant differences in plantarflexion when considering the step peaks, and starts and
peaks combined, but for the step start a significant difference between groups was not
found. It should be pointed out, however, that only two steps were included in the
analysis of the start, as most steps did not commence with a requirement of maximum
plantarflexion. This may have affected the overall result for the analysis of the start time
point.
For the professional dancers, mean plantarflexion values ranged from 30 through to 60.
Plantarflexion values were lower when the foot was in a position of demi-pointe (i.e.
standing on the ball of the foot) as in the rises and pirouettes (30 - 52), compared to a
- 202 -
non weight bearing position with a fully pointed foot (53 - 60). The current data are
comparable to results from the limited number of published articles that have used similar
techniques to investigate active plantarflexion during execution of classical ballet.
Specifically, Lin (2005) used 3D motion analysis to measure maximum ankle
plantarflexion during the relev en point, with an average plantarflexion angle of 53
being reported. Thus, as was the case for knee extension, the practical data for ankle
plantarflexion are also consistent with the theoretical concept that extreme elongation
(extension) of the lower limb is a characteristic of elite ability in classical ballet
technique.
3.4.5 Analysis of Fundamental Principles with Respect to Movement Categories and

Steps
Four of the seven movement categories proved most successful in detecting differences
between the professional and non-professional dancers in the fundamental principles of
classical ballet technique. Specifically, it was the bend (pli), stretch (battement),
rise (relev), and turn (tour) categories for which inter-group differences of specific
variables were detected. The most commonly detected deviation across these movement
categories was for transverse hip rotation, a result that is consistent with the significant
difference obtained for this variable through the repeated measures general linear model
analyses. The turn (tour) category was only sensitive to transverse hip rotation (turnout)
deviations, while the rise (relev) detected differences in both kinematic variables
relating to turnout (i.e. transverse hip rotation and transverse ankle rotation). It appears
that the reduction in contact area between the foot and the floor that occurs during the
rise, introduces instabilities in stance that provide increased challenge to the maintenance
of hip and ankle joint rotation. The battements also highlighted differences between
groups in transverse hip rotation (turnout). This result suggests that movement of a
single leg to the front or back of the body can perhaps compromise the ability of the
external hip joint rotators to achieve optimal function. In addition, the battements also
captured differences between groups in knee extension (extension). The working leg
of the professional dancers reached greater knee hyperextension than that of the nonprofessionals. The discrepancy between groups in knee extension observed during
execution of these steps, but not other steps, suggests that the battements are particularly
effective in highlighting weaknesses, or lack of use, of the knee extensor musculature.
- 203 -
Finally, the plis, detected differences between groups in thoracic anterior/posterior tilt
(alignment). That is, the non-professionals demonstrated approximately 5 more
anterior thoracic tilt than the professionals at the peak of the movement, i.e. at maximum
knee flexion. In line with the finding that increased forward trunk lean facilitates
increased hip extensor power (Lees et al., 2004; Vanrenterghem et al., 2008), it is
possible that as a group the non-professional dancers have weaker hip extensors and are
able to access more power for pli ascent by increasing hip and trunk flexion during pli
descent. It has also been shown that increased forward trunk flexion during knee flexion,
decreases the torque at the knee joint (Lees et al., 2004), essentially by decreasing the
horizontal displacement between the bodys centre of gravity and the knee
flexion/extension axis. This has the effect of decreasing the load on the knee extensors
(i.e. quadriceps), a compensation which may be beneficial for the less experienced
dancers who may have weaker quadriceps. Therefore, the discrepancies in alignment,
observed during execution of the plis, may in fact be a reflection of deficiencies in areas
other than the trunk itself.
Thus, it appears that there are particular ballet steps or movement categories that are more
effective than others in highlighting deviations of certain kinematic variables from the
ideal. These deviations may be indicative of biomechanical problems directly related to
the area being assessed, or could be compensatory mechanisms due to deficiencies in
other areas of the body.
3.4.6 Limitations
3.4.6.1 Limitations of the Biomechanical Model and Measurement Techniques
It is acknowledged that the biomechanical model used in the current study is associated
with certain errors and assumptions. For example, as discussed in Section 3.2.3.1,
attachment of reflective markers has been shown to be associated with soft tissue artefact
(Stagni et al., 2005), and the Vicon Plug-in-Gait model is based on the assumption that
the joint structure and body segment proportions of the subjects assessed are in agreement
with the anthropometric reference data on which the model is based (Davis et al., 1991).
In addition the model was designed specifically for assessment of human gait, rather than
for analysis of dance-related movements, which involve much larger ranges of joint
motion. In particular, the hip and ankle joint movements required for execution of
- 204 -
classical ballet technique are typically beyond that used in normal human gait. However,
with respect to the steps that were included in the analyses, examination of the data
obtained from performance of multiple ballet steps, indicate that there were no instances
of data flipping due to gimbal lock issues. Gimbal lock describes the discontinuity in
data as the second orientation angle reaches +90 or -90 for a Cardan rotation sequence
(Lees et al., 2010). In addition, the 3D kinematic values obtained for most of the hip and
ankle joint data were reasonable with respect to the movements performed. The one
kinematic variable for which data seemed different to that expected was transverse hip
rotation. It was noted in Section 3.4.3, that the external hip rotation measures obtained in
this study were lower than those that could be expected based on theoretical information.
Dance literature states that external hip rotation values ranging from 60-90 are typical
for experienced dancers, however the highest mean hip rotation value obtained in this
study was only 41. However, in accordance with the findings of Lees et al., (2010), the
Cardan angle rotation sequence was not considered to be influential on the hip rotation
data for the steps analysed, and the values obtained were similar to data reported in
multiple previous studies that used the same techniques. It is reasonable to suggest
therefore, that the data are in fact an accurate reflection of what is actually happening at
the hip joint.
It is also acknowledged that the accuracy of marker placement is crucial in determining

the accuracy of data obtained from 3D motion analysis. Therefore, to ensure the accuracy
and consistency of optimal marker placement across all participants, marker placement
was conducted by the same person (the author) for all participants in strict accordance
with the Vicon Plug-in-Gait guidelines. Attachment of all knee markers was completed
with use of a Knee Alignment Device (KAD), thereby adding to the consistency and
reliability of the marker placement process. Once all markers were in place, additional
accuracy checks were conducted through collection of gait data for each participant, and
use of the data accuracy assessment methods typically applied in clinical gait analysis.
This included examination of knee varus/valgus data, and comparison with mean data for
normal human gait. The fact that this process was conducted adds additional support for
the accuracy of the hip rotation data.
- 205 -
In summary, maximum care and effort was taken to ensure optimal accuracy of the
kinematic data obtained. In conducting this study, all potential technical limitations were
mitigated as much as was practically possible, and therefore any additional issues relating
to the equipment or biomechanical model can only be considered to be common to all
users of this form of movement analysis, and are therefore not limitations that are unique
to this study.
3.4.6.2 Limitations Due to Participant Sample Size

Fourteen non-professional and 12 professional dancers participated in this study. It could
be argued that in terms of scientific research studies these are relatively small sample
sizes, thus potentially imposing limitations on the conclusions reached. A review of
existing literature reporting on biomechanical analysis of dance-related movements
(Section 2.2.5.4) revealed that a mean sample size of 10 was obtained across 37 studies.
The majority of these studies had less than 15 subjects, with only six studies recruiting
between 16 and 25 subjects. In relation to ballet-specific studies, for which recruitment of
subjects possessing a highly specialised skill is required, sample sizes ranged from 1 to
16. A review article by Mullineaux et al. (2001, p739) states that in biomechanics
research the sample sizes used are often small. They reported a mean sample size of 14.5
participants for research papers published in the Journal of Biomechanics in 1998
(Mullineaux et al., 2001). Thus, although the participant sample sizes obtained in the
current study were constrained by identification and availability of suitable subjects, and
also by practical limitations imposed by restrictions on laboratory and equipment
availability, the final sample sizes obtained are well within the ranges of those typically
used in biomechanics analysis. Moreover, the relatively low standard deviations, obtained
for each group across most steps, indicate that larger sample sizes may not have resulted
in vastly different results than those obtained with the current numbers of participants. It
should be noted that the few steps for which quite large standard deviations were obtained
(i.e. pirouettes and grand jet elanc en avant), are some of the more difficult steps in the
classical ballet repertoire, and large standard deviations would be expected for any
sample size.
- 206 -
3.4.6.3 Limitations Due to Participant Motivation

Another possible limitation that may have affected the data could be the level of personal
motivation of the dancers during execution of the steps in the data collection sessions.
The technician who conducted all data collection sessions (the author) was not personally
familiar with any of the participating dancers, and had not seen them dance prior to their
attendance at the biomechanics laboratory. It was therefore not possible to truly know if
all dancers exerted maximum effort, and attempted to perform all movements to the best
of their ability. All dancers were instructed to perform each step as best they could, and
one would expect that the unusual situation of being inside a laboratory, covered in
markers, surrounded by various measurement devices, multiple cameras, and computers
would elicit maximum motivation for optimal performance from all dancers. Based on
the authors personal observation of the mannerisms, demeanour and body language of all
dancers during their motion analysis session, it is believed that all dancers were highly
motivated, and did in fact perform to the best of their ability. In addition, the fact that a
high degree of consistency of kinematic data between dancers was observed in overlays
of time-normalised graphs, indicates that the dancers were executing the steps in
accordance with standard classical ballet practice.
3.4.7 Conclusions
Professional ballet dancers perform the fundamental biomechanical principles of
turnout and extension in closer agreement with the theoretical principles of classical
ballet technique than non-professionals. These theoretical principles therefore accurately
reflect what occurs in practice during highly skilled execution of classical ballet
technique, and are therefore appropriate guidelines on which to base current teaching
practice. However, this is not necessarily the case for the principles of alignment and
placement. Interestingly, in relation to alignment, ballet dancers adopt posterior
thoracic alignment, rather than the vertical alignment referred to in the theoretical
literature. In addition, when functionally biomechanically advantageous (e.g. when
jumping), the alignment of professional dancers deviates from the theoretical ideal more
than that of the non-professionals. With respect to placement, the pelvic position
adopted by ballet dancers is not neutral and is not dissimilar to the pelvic position
typically used during normal human gait. Although a decreased sagittal plane pelvic
range of motion (i.e. superior control of sagittal placement) shown by professionals is in
- 207 -
agreement with the dance literature, in contrast, these more experienced dancers utilise
increased coronal and transverse plane pelvic range of motion while performing certain
movements. This finding is contradictory to that expected based on theoretical
information, and has implications for the instruction of classical ballet technique. The
specific implications of these conclusions with respect to classical ballet teaching practice
will be discussed in Chapter 6 of this thesis.
This study has also determined that there are particular ballet steps or movement
categories that are more effective than others in highlighting kinematic deviations related
to the fundamental principles of classical ballet technique. These are the bend (pli),
stretch (battement), rise (relev), and turn (tour). In Chapter 5, all 14 ballet steps will
be further investigated with respect to their ability to detect deviations in kinematic
variables, and to rank the technical competency of the group of non-professional dancers
against that of the professional dancers (i.e. the gold standard). To do this, the rank
order of the non-professional dancers obtained using different numbers and combinations
of ballet steps will be compared. The results from such an analysis could contribute to the
development of a quantitative assessment tool to evaluate classical ballet technique.
Whether or not such a tool would in fact be necessary ultimately depends on the abilities
of dance teachers to qualitatively analysis dance movements. Quantitative biomechanical
assessment techniques are typically used (e.g. clinical gait analysis) because it is assumed
that the accuracy of visual perception is inferior to the accuracy of quantitative data
obtained from instrumented techniques (Toro et al., 2003). The ability of experienced and
inexperienced ballet teachers to accurately observe, i.e. qualitatively analyse, classical
ballet movement will therefore be investigated in the next chapter. Ballet teachers
qualitative responses, obtained through observation of videos of dancers who participated
in the current study, will be compared to the quantitative data obtained in the current
study.
- 208 -
Chapter 4 Accuracy of Qualitative Analysis of Classical Ballet Technique
CHAPTER 4
ACCURACY OF QUALITATIVE ANALYSIS OF

CLASSICAL BALLET TECHNIQUE
- 209 -
CHAPTER 4
ACCURACY OF QUALITATIVE ANALYSIS OF
CLASSICAL BALLET TECHNIQUE
4.1 Introduction
Movement instructors, such as sports coaches and dance teachers, are faced with many
challenges in their efforts to facilitate skill acquisition. The ability to accurately observe
learner responses as a precursor to corrective feedback stands as one of the most
important operations in motor skill instruction (Imwold and Hoffman, 1983). Without
obvious external measures of skill competency, dance students must rely on the
observation abilities of their teachers in order to receive useful feedback and correction
on the execution of their technique. The effectiveness of a ballet teachers analysis of the
technical abilities of their students is based firstly upon having a thorough knowledge of
the fundamental principles upon which the movement technique is based, and secondly
on the accuracy with which they can see the movement patterns executed by their
students. Having a clear and detailed understanding of the ideal or desired movement
pattern for each skill enables the instructor to develop an internally held criterion or
mental image of the desired execution of the movement (Whiting, 1972).
When analysing the movement patterns or responses of their students, motor skill
instructors need to be able to form a mental image of the desired response and compare
that image with the learners response during immediate observation. Once the nature
and extent of any deviation of the learners response from the desired response is
identified, teachers can then provide appropriate feedback and instruction to their
students. This whole process has been defined as qualitative analysis and has been
described as the systematic observation and introspective judgment of the quality of
human movement for the purpose of providing the most appropriate intervention to
improve performance (Knudson and Morrison, 1997, p4).
Other terms such as
movement analysis, clinical diagnosis, skill analysis, error detection, observational

assessment, systematic observation, and introspective judgement have also been used in
- 210 -
the same context (Knudson and Morrison, 1997), but for the purposes of this study the
term qualitative analysis will be used.
Not only is the skill of qualitative analysis of human movement important for motor skill
acquisition, but it also plays a vital role in injury prevention. Poor technique is an
important risk factor for the development of overuse injuries in classical ballet, as well as
many other forms of skilled movement. If deficiencies in technique can be identified
immediately and corrected, then the risk of injury can be reduced or prevented. The
accuracy with which an instructor can see the motor response of their students, and thus
determine how it deviates from the desired or ideal response, is therefore crucial for many
aspects of successful motor skill instruction.
It has been suggested that humans may have special perceptual abilities relating to the
analysis of human motion, which may extend beyond the kind of data obtained from pure
quantitative analysis (Morrison, 2000). The implicit assumption is that people who
regularly use the skill of qualitative analysis, such as dance teachers, sports coaches,
clinicians and ergonomists, have the ability to observe motor skill performances
systematically and critically and to identify flaws accurately. Whether this is actually the
case has been the subject of studies in the clinical setting, by some in the fields of
ergonomics and workplace safety, and by a few in the sporting arena (Section 2.3.4).
However, the accuracy with which dance teachers can visually observe their students
movement patterns, and identify correct or incorrect technique, has received no attention
in the academic literature to date. It has been stated (Imwold and Hoffman, 1983) that
neglect of this topic by researchers may retard the design and implementation of training
programs for movement teachers who, without access to such programs, will be forced to
train themselves through the random events of day-to-day teaching. Left to their own
devices, and given sufficient time and experience, some movement teachers and coaches
may develop competency in qualitative analysis, however it has been proposed (Arend
and Higgins, 1976) that the teacher who is a trained analyser of human movement will be
better equipped to enhance the process of skill acquisition than the teacher who has not
received specific training in the process.
- 211 -
Many professional dancers retire from their performing career and immediately
commence teaching dance. It seems to be an accepted belief that a highly accomplished
dancer will instantly have the skills required to be a highly effective teacher. However,
this may not necessarily be the case in that an expert level of practical ability in a
movement discipline may not necessarily be accompanied by a high level of cognitive
ability in identifying and articulating the theoretical principles and corrections that are
crucial for successful execution of skilled movement. Thus, it may be that some ballet
teachers are deficient in their qualitative analysis skills and could benefit from specific
training in this area. This study therefore aims to investigate the accuracy with which
ballet teachers are able to qualitatively assess the technique of classical ballet dancers.
The ability of ballet teachers to identify correct and incorrect classical ballet
technique will be assessed, and analysed with respect to their level of experience in
classical ballet teaching, their ballet training history, and their professional performance
experience. Measures of accuracy will be obtained by comparing qualitative assessments
with quantitative data obtained from 3D motion analysis data.
It has been stated (Runeson, 1984) that if perceptual observation can be identified as a
skill, then it might be anticipated that more experienced teachers should have an
improved ability to detect the significant features in technical execution of a movement.
Thus, the primary hypothesis is that experienced ballet teachers will more accurately
recognise the occurrence of correct and incorrect classical ballet technique, and more
accurately determine the type of correction required to improve incorrect technique. A
secondary hypothesis is that experienced ballet teachers will be more accurate in
determining which body region contributes most significantly to the execution of
incorrect ballet technique.
- 212 -
4.2 Methods
4.2.1 Participant Recruitment
Professional and student ballet teachers were recruited to take part in the study. Prior to
commencing the recruitment process, ethics approval was obtained through the
University of New South Wales Human Research Ethics Advisory (HREA) Panel
(Appendix J). The target number of subjects for recruitment to the study was 20 student
ballet teachers and 20 professional ballet teachers.
Criteria for inclusion as a student ballet teacher were,

x
being at least 18 years of age at the time of data collection
never having been employed as a professional ballet teacher.
being currently enrolled in a dance teacher training program.
For recruitment of the student ballet teachers, second year students enrolled in the
University of New South Wales Bachelor of Arts (Dance)/Bachelor of Education
program were invited to attend an information session held on the university campus,
during which information was provided about the project. This session was facilitated
with the assistance of the program authority. During this session, potential participants
were informed of the inclusion criteria for participation in the student teacher group, and
were provided with information on what they would be required to do as study
participants, and how their data would be used for the study. Interested potential
participants provided their name and contact details at the end of the session, and were
subsequently emailed an information and consent form prior to attending a data collection
session.
Inclusion criteria for participation as a professional ballet teacher were,

x
being employed as a professional ballet teacher at the time of data collection, or

within the 12 months preceding time of data collection.
- 213 -
The professional ballet teachers were recruited from various ballet schools around the
Sydney metropolitan area. Recruitment of professional ballet teachers occurred via
emailing and calling administrative staff at local dance schools, by distributing
information flyers throughout dance schools, by giving brief presentations at dance
teacher seminars, and by contacting dance teachers on an individual basis as identified by
word-of-mouth within the Sydney dance community.
Upon identification of potentially suitable and interested participants for the professional
group of teachers, all potential participants were informed of the inclusion criteria for
participation in the professional group, and were emailed detailed information on what
they would be required to do as study participants and how their data would be used for
the study. Each participating professional teacher was given an information and consent
form prior to their data collection session.
All subjects provided written informed consent, in the form of a signed and dated consent
form (Appendices K and L), prior to commencing any data collection for the study. No
form of incentive or compensation was offered or provided to any student or professional
teachers for their participation in the study.
4.2.2 Development of the Qualitative Assessment Questionnaire

The qualitative assessment questionnaire was comprised of 40 multiple choice questions.
It consisted of eight questions for each of the five steps selected for qualitative analysis.
The same eight questions were asked for each step. The questionnaire was presented to
the participants in the form of a Microsoft PowerPoint presentation.
The questionnaire was developed through careful consideration of the existing academic
literature that has investigated the accuracy of qualitative analysis of human movement.
That is, through conducting the literature review, described in Section 2.3, a number of
findings and conclusions from previous studies provided useful information that assisted
in designing the type of questions, and the number and type of scale categories.
The majority of questions in the Qualitative Assessment Questionnaire related to

identification of correct or incorrect classical ballet technique. The first question for
- 214 -
each step required the teachers to observe the whole body and to indicate which body
region, if any, was most in need of correction. This style of question was chosen for
inclusion as it did not direct the teachers to focus on any one specific body area, and thus
provided an opportunity to assess and compare which technical aspects of classical ballet
teachers are most concerned with when allowed to perform non-directed qualitative
analysis. The remaining questions for each step provided more specific direction as to
which body region to observe. Data from the academic literature indicate that better
accuracy of observation is achieved when the number of parameters an analysts is
required to observe is kept to a minimum, ideally limited to only one or two. Thus, each
remaining question for each step directed the participating teachers to only observe
movement at one body segment or joint and in only one plane of movement. The four
different segments or joints chosen for inclusion were thoracic anterior/posterior tilt,
pelvic anterior/posterior tilt, hip external rotation, and knee extension. Each of these
regions related to one of the four fundamental principle of classical ballet technique (i.e.
alignment, placement, turnout, extension), as outlined in Section 2.1.
As was indicated through review of the existing academic literature, better accuracy of
qualitative analysis is obtained when categorical scales containing fewer categories are
used, and when categories are adequately labelled with appropriate references to the skill
level or normality. For these reasons, binary categorical scales were used for each of the
questions relating to a specific body region. Although the academic literature did indicate
that in clinical practice better accuracy of observation is achieved when joint range
categories as well as appropriate labels are included on the scale, it was decided to only
use labels, and these were in the form of references to correctness of technique. This
decision was based on the fact that, unlike physiotherapists, dance teachers do not
typically observe and refer to joint ranges of motion in specific numerical terms, so this
form of categorising would not have been familiar to the majority of participating
teachers. Therefore, for each of the four individual parameters chosen for inclusion in the
questionnaire, teachers were asked to identify if movement at the joint/segment was
correct or incorrect with respect to their own knowledge of classical ballet technique.
It should be noted that teachers were not instructed as to what constituted correct or
incorrect technique before or during any of the data collection sessions.
- 215 -
In addition to identifying correct or incorrect technique for each individual parameter

for each step, participants were also asked to indicate which type of correction might be
required to correct incorrect technique. That is, if a participant selected the incorrect
category for a particular movement parameter, they were then presented with an
additional question, which they were required to answer through use of another binary
scale containing two categories of possible corrections.
A copy of the Microsoft PowerPoint presentation format used to present the Qualitative
Assessment Questionnaire, and an example of the questions for a single step are
presented in Appendix N.
4.2.3 Video Data Selection and Preparation

After selecting the type and number of questions that would be included in the Qualitative
Assessment Questionnaire, a selection of dance videos were chosen for inclusion in the
data collection sessions. The same videos were viewed by all participating teachers in all
sessions. Five videos, consisting of five different ballet dancers performing five different
ballet steps were selected. Each of these dancers had participated in the motion analysis
section of the study (described in Chapter 3) in which simultaneous video and 3D motion
analysis data were obtained for each dancer performing steps from each of the Seven
Movements of Dance (Noverre, 1760). All dancers who participated in this section of the
study provided written consent for their video data to be watched by ballet teachers
participating in the current section of the study.
As indicated in the academic literature, when conducting video-based qualitative analysis

the position of the camera in relation to the plane of joint motion has a significant effect
on the accuracy of observation. Specifically, better accuracy is achieved when the
direction of the camera and the plane of joint motion are orthogonal. Authors have
suggested that use of multiple cameras placed at various perspectives is therefore
advantageous in optimising accuracy of observation. Each video therefore consisted of a
dual perspective view of the dancer (Figure 4.1), in which frontal and sagittal plane
aspects were simultaneously played to the viewer. This setup was considered to provide
the observer with maximum opportunity to accurately view all relevant parameters.
- 216 -
Figure 4.1 Example of sagittal and frontal views played simultaneously for all videos.
All video footage was obtained in the same laboratory with the same recording equipment
(two Sony digital camcorders), and all dancers wore the same dance clothes in all videos
(i.e. black bike shorts, black crop top, and Bloch foot thongs). The accompanying music,
to which all dancers performed the steps during the motion analysis sessions, was not
audible to the viewers during the qualitative observation sessions. Each video was edited
so that the start and finish of the step corresponded to the start and finish as defined by
the Visual Basic macros described in Chapter 3 of the thesis. All edited videos were
inserted into a Microsoft PowerPoint presentation that was presented to all participating
teachers during the qualitative analysis data collection sessions.
The five videos selected for inclusion in the qualitative analysis data collection sessions
were chosen such that each video contained footage of a different dancer performing a
step from one of five different movement categories. These five categories were:
- 217 -
Bend (pli)
Rise (relev)
Stretch (battement)
Turn (pirouette)
Jumps (saut)
Videos of steps in the glide (gliss) and dart (elanc) movement categories were not
included in the qualitative analysis data collection sessions. This was due to the fact that
the travelling nature of these movements meant that the dancers moved in and out of view
of the stationary cameras throughout the execution of the movement. It was therefore not
possible to obtain a clear, full body frontal and sagittal view of these steps, and they were
consequently omitted from the analysis. The specific steps chosen for inclusion in the
qualitative analysis are listed in Table 4.1.
In assembling the final list of videos included in the qualitative analysis sessions, it was
not only necessary to choose a combination of five videos that represented each of the
five different movement categories, as listed above, performed by five different dancers,
but it was also important that the selection of videos contained approximately equal
numbers or correct and incorrect kinematic variables. It was important to choose five
different dancers to ensure that participating teachers would not become familiar with the
individual movement patterns or habits of any one dancer throughout the analysis session,
thereby reducing the likelihood of any learning effect occurring as the session progressed.
An equal spread of correct and incorrect kinematic variables across the five videos
was necessary so that the incidence of any Type I and Type II Errors could be analysed.
Type I Errors occur when a null hypothesis is rejected when it is really true, while a
Type II Errors occur when a null hypothesis is accepted when it is really false (Vincent,
1995, p248). In the context of the current study, a Type I Error would be a situation
where correct ballet technique is rated as incorrect, and a Type II Error would occur if
incorrect ballet technique is rated as correct.
In terms of the kinematic variables that participants were asked to observe during the
analysis sessions, these were selected so as to align with the four fundamental
biomechanical principles of classical ballet technique, as identified in Chapter 2 and
- 218 -
analysed in Chapter 3. In order to maintain consistency in the way in which the steps
were observed, the same four kinematic variables were selected for analysis for each step.
Thus, in observing each of the five steps, participants were required to rate the dancers
technique in terms of each of the four fundamental biomechanical principles of classical
ballet technique, with each of these principles relating to a different body region.
4.2.3.1 Measurement of Response Accuracy: Calculation of Quantitative Reference
Criteria
As described in Section 4.2.2, participants were asked a number of questions which
required them to focus on individual body regions, rather than on the whole body. For
each separate body region, in answering the question does the body region need
correction?, participants were required to select one of two options, yes or no. In
order to identify whether or not respondents were accurate in answering this question, it
was necessary to establish objective quantitative reference criteria against which all
qualitative responses could be compared, and thus identified as either accurate or
inaccurate. The videos that were chosen for inclusion in the qualitative observation
sessions were therefore selected according to their accompanying kinematic data,
obtained from simultaneous 3D quantitative motion analysis. That is, in order to make a
final selection of the videos for inclusion in the study, it was necessary to assess how the
kinematic data for individual dancers compared to the mean data for the professional
group of dancers. To make this comparison, z-scores were calculated to compare joint
and segment movement patterns of individual dancers with mean values of the gold
standard data, i.e. mean joint angle data for the groups of professional ballet dancers.
The z-score is a standard score expressed in terms of standard deviation units on a normal
curve (Hastad and Lacy, 1994). This value indicates how far above or below the group
mean the raw score is located. The further above the mean the raw score is, the higher the
z-score will be. If the raw score is below the mean the z-score will be negative, and will
increase in magnitude the further below the mean the raw score is located. If a raw score
is exactly the same as the group mean it will have a z-score of zero. The z-score for an
individual value is calculated using the group mean and standard deviation, according to
the following formula:
- 219 -
z=
where = group mean, = raw score , and s = standard deviation.
For the purposes of systematically and objectively identifying whether or not each
kinematic variable for each step for each dancer required correction, an absolute z-score
of 1 was chosen as the cut-off point or threshold above which the variable should be rated
as incorrect, and thus in need of correction. That is, a z-score of 1.0 indicates that a raw
score is 1 standard deviation above the mean, while a z-score of -1.0 indicates that a raw
score is 1 standard deviation below the mean. In any normal distribution curve, 68.26%
of all data falls between the mean and 1 standard deviation from the mean (Vincent,
1995). Therefore, a cut-off value of 1 standard deviation (i.e. a z-score within the range
-1 to 1) allows a large range within which the majority of data for the population in
question falls. Thus, in using the 3D kinematic data to rate the technical performance of
each dancer, an objective system was put into place in which any kinematic variable with
a z-score greater than 1.0 or less than -1.0 was deemed to be incorrect, and therefore in
need of correction. Any variable whose z-score fell in the range -1.0 < 0 <1.0, was
labelled as correct, and therefore not in need of correction. Previous studies (Nicholls et
al., 2003; Moseley et al., 2008) have reported the same method for determining proper
or improper sports technique (Nicholls et al., 2003) or flexible or inflexible ankle
movement (Moseley et al., 2008). The magnitudes of the absolute value of the z-scores
were therefore used for determination of whether a kinematic variable was correct or
incorrect, and therefore provided the reference criteria used to determine whether
respondents were accurate or inaccurate in assessing the dancers technique. Table
4.1 lists the ballet steps that were chosen for qualitative analysis, the movement category
each step falls into, the kinematic variables chosen for observation and their associated
fundamental principle, the mean z-score for each kinematic variable for each step, and
whether the variable was determined to be correct or incorrect. The total number of
correct and incorrect kinematic variables are also listed.
- 220 -
Kinematic Variable
Subject ID
Thorax anterior/posterior tilt
(Principle = Alignment)
Pelvis anterior/posterior tilt
(Principle = Placement)
Hip rotation
(Principle = Turnout)
Knee flexion/extension
(Principle = Extension)
Grand
Pli in
First
Position
Rise in
Second
Position
Step
Grand
Battement
Derriere
90
Double
Pirouette
En Dedans
Temps
Lev
(Bend)
(Rise)
(Stretch)
(Turn)
(Jump)
PR007
PR010
PR001
PR005
PR002
-1.17 ()
-1.35 ()
-1.20 ()
-0.21 (9)
1.53 ()
1.01 ()
1.53 ()
-1.35 ()
0.15 (9)
-1.31 ()
-0.17 (9)
1.38 ()
-0.49 (9)
-0.59 (9)
0.36 (9)
-0.23 (9)
-1.37 ()
1.33 ()
0.22 (9)
0.64 (9)
Total
Total Correct Technique
Total Incorrect Technique
11
4
4
4
4
4
20
Total
Table 4.1 Mean z-score for each step for selected kinematic variables observed during qualitative analysis
data collection sessions. () = Incorrect technique; (9) = Correct technique
If participants responded with an answer of yes to the question does the body region
need correction?, they were then asked which type of correction is required? An
objective reference criterion for this question was determined according to the sign of the
z-score. For a kinematic variable with a z-score of less than -1.0, the required correction
would be towards the positive direction, and for a z-score of more than 1.0, the required
correction would be towards the negative direction. For example, as shown in Table 4.1,
in her performance of the grand pli in first position, subject PR007 obtained a z-score of
-1.17 for thoracic anterior/posterior tilt. For this variable, a negative value for the raw
data indicates that the dancers thorax is in posterior tilt, and a positive value indicates
anterior tilt. Therefore, a z-score of -1.17 indicates that the dancers thorax was posterior
compared to the professional group mean by more than one standard deviation, and this
dancer should therefore be corrected to shift her thorax into a more anterior position.
Another question on the qualitative assessment questionnaire required the participants to

observe the dancers whole body during performance of the ballet movement, rather than
focusing on individual body segments. Specifically, participants were asked to answer the
- 221 -
question if you could give this dancer only one correction on their execution of this step,
which body region would you concentrate on correcting?. Participants were required to
give their answer by selecting one of seven multiple choice options, as outlined below;
1. Thorax (upper torso)

2. Pelvis
3. Hips
4. Knees
5. Ankles/Feet
6. Other
7. None No correction required
The correct answer for each step, based on a quantitative reference criterion, was
determined by calculation and comparison of the mean z-score across each step, for each
plane of motion (sagittal, frontal, transverse) for the thorax, pelvis, hip, knee, and ankle.
The kinematic variable with the largest absolute value for the z-score was identified as
the variable most in need of correction, and was therefore determined to be the correct
answer. The respondents answer selection was compared to this variable in order to
determine whether they were accurate or inaccurate in answering this question.
Table 4.2 displays the mean z-scores for each kinematic variable for each step, and
indicates which variable was identified quantitatively as being most in need of
correction.
Participants were not asked to provide so much detail as to indicate which plane of joint
or segment motion was most in need of correction, but only which body region. Answers
were therefore considered to be accurate if the selected body region matched the
quantitative reference criteria, regardless of the plane of motion.
- 222 -
Kinematic Variable
Step
Grand Pli
in First
Position
Rise in Second
Position
Grand
Battement
Derriere 90
Double
Pirouette En
Dedans
Temps
Lev
(Bend)
(Rise)
(Stretch)
(Turn)
(Jump)
Subject ID
PR007
PR001
PR010
PR005
PR002
Thorax Anterior/Posterior Tilt
-1.17*
-1.20
-1.35
-0.21
1.53
Thorax Side Tilt
0.58
-0.62
-0.34
0.39
-0.71
Thorax Rotation
0.23
0.05
0.48
-0.53
0.01
Pelvic Anterior/Posterior Tilt
1.01
-1.35
1.53*
0.15
-1.31
Pelvic Obliquity
0.01
1.27
-0.08
-0.35
-1.86*
Pelvic Rotation
0.87
1.20
-0.83
-0.55
-1.23
Hip Flexion/Extension
0.11
-0.73
1.18
0.00
-0.52
Hip Abduction
0.08
-0.77
-0.07
-0.72
-1.64
Hip Rotation
-0.17
-0.49
1.38
-0.59
0.36
Knee Flexion/Extension
-0.23
1.33
-1.37
0.22
0.64
Knee Rotation
0.17
-0.24
-0.65
0.31
0.27
Knee Varus/Valgus
0.31
1.38
-1.22
0.89
0.02
Ankle Dorsi/Plantarflexion
-0.35
1.52*
-0.97
0.44
0.44
Ankle Inversion/Eversion
0.33
1.40
1.40
-0.19
-0.77
Ankle Rotation
0.24
-1.07
1.37
-0.54
0.43
Largest absolute z-score
1.17
1.52
1.53
0.72
1.86
Thorax
Pelvic
*Kinematic variable with
None
Anterior/
Ankle Dorsi/
Anterior/
Pelvic
largest z-score (i.e. variable
Posterior
Plantarflexion
Posterior
Obliquity
most in need of correction)
(all < |1|)
Tilt
Tilt
Table 4.2 Mean z-scores for all kinematic variables for all steps observed during qualitative data collection
sessions.
4.2.4 Experimental Protocol

All participating student teachers attended a single qualitative analysis data collection
session. This was conducted on the UNSW campus. Professional teachers were more
restricted in their availability, so multiple data collection sessions, consisting of one to
three participants, were scheduled across a number of weeks in order to accommodate the
limited availability of these participants. All sessions for professional teachers were also
conducted on the UNSW campus.
- 223 -
At the beginning of each session, before collecting any data, a signed and dated consent
form was obtained from each participant (Appendices K and L). Participants were then
asked to complete a Ballet Teacher Questionnaire. This form consisted of 40 questions
asking information about the participants date of birth, dance training history, dance
performance history, teacher training history, teacher employment history, and anatomy
and biomechanics training history. Participants spent 10-15 minutes completing this
questionnaire. A copy of the questionnaire is presented in Appendix M.
After collecting all completed consent forms and questionnaires the qualitative analysis
data collection began. This session consisted of an introductory speech, given by the
author, in which the terminology that would be used in subsequent questions was
clarified, and participants were given instructions on the procedure that would take place.
All instructions were displayed in writing as part of a Microsoft PowerPoint
presentation, with accompanying verbal instruction also being provided. Subjects were
informed that they would be shown five videos, and would be asked to answer the same
eight questions for each of the videos. It was emphasised to all participants that they were
only required to focus on technical aspects of the dancers execution of the steps, as this
study was not addressing overall performance elements such as artistry and presentation.
Subjects were also told that they were free to withdraw from the study at any time during
the data collection session if they so desired. Participants were given an opportunity to
ask any questions before presentation of the videos began.
All videos, questions and answer options were presented to the participants in the form of
a Microsoft PowerPoint presentation (Appendix N) projected onto a large screen at the
front of a university tutorial room. Subjects sat comfortably at desks at a self selected
distance from the screen. For each of the eight questions for all five ballet steps, the
participants were firstly shown the question and possible answer options, and were then
shown the video. After watching the video, the subjects were shown the question and
answer options again, and were asked to select the answer option they felt was most
appropriate. Participants responded to questions by selecting the desired answer number
through use of a hand held remote control device (ResponseCard keypads), and
associated hardware (TurningPoint USB receiver device) and software (TurningPoint
2008 v4.1).
- 224 -
Each qualitative analysis data collection session took approximately 45 minutes to

complete. Upon completion of each session, the captured data were exported to a
Microsoft Excel spreadsheet, through use of TurningPoint software (TurningPoint
2008 v4.1), and saved for subsequent analysis.
4.2.5 Data Analysis

Measures of accuracy of qualitative assessment of classical ballet technique were
obtained by comparison of subjects qualitative responses with quantitative 3D kinematic
data, as described in Section 4.2.2. Measures for accuracy of qualitative assessment for
each type of question in the qualitative assessment questionnaire were calculated as a
value of percentage accuracy.
4.2.5.1 Accuracy of Identification of the Body Region most in need of Correction

All participants responded to five questions, one for each step, in which they were asked
to identify which body region, if any, was most in need of correction. Participants gave
their responses by selecting one of seven possible answer options. A figure for overall
percentage accuracy for this question was therefore calculated by combining the
response to this question for all five step. Since participants were asked the same question
for all five steps, the accuracy of responses for was also analysed with respect to the step
observed.
4.2.5.2 Accuracy of Identification of Correct or Incorrect Technique

Each participant responded to a total of 20 questions in which they were asked to rate the
dancers technique as correct or incorrect, one question for each of the four different
body regions for each of the five steps. The figure for overall percentage accuracy for
this type of question was therefore calculated by combining responses for all body
regions for all steps. In addition, the proportions of Type I and Type II errors, and
measures of sensitivity and specificity were calculated for each group. Sensitivity and
specificity are closely related to the concepts of Type I and Type II errors, and are
measures of the performance of a binary classification test (Loong, 2003). Sensitivity
measures the proportion of positives results (i.e. for this study incorrect technique)
- 225 -
which are accurately identified, while specificity measures the proportion of negative
results (i.e. correct technique) which are accurately identified.
4.2.5.3 Accuracy of Identification of Type of Correction Required

If a participant identified that the technique for a particular body region for a particular
step was incorrect, they were then asked to identify which type of correction is
required to correct the technique. Subjects answered this question by selecting one of
two answer options, e.g. segment (thorax/pelvis) needs to move anterior (forwards) or
posterior (backwards). A percentage accuracy score was calculated for each subject,
based upon the number of variables they had identified as incorrect, and the proportion
of these that were answered accurately. The total number of applicable responses to this
type of question could therefore potentially be different for each subject, depending upon
the number of kinematic elements they considered to be incorrect.
4.2.6 Statistical Analysis

The effect of professional ballet teaching experience on the accuracy of qualitative
assessment of classical ballet technique, data were analysed in terms of two categories of
subjects.
x
Group 1:
Group 2: Professional ballet teachers
Student ballet teachers
4.2.6.1 Accuracy of Identification of the Body Region most in need of Correction

Binary logistic regression was used to statistically assess the accuracy of identification of
which body region was most in need of correction. Since in answering this question the
participants were required to observe the whole body, rather than being directed to
observe individual body segments, these data were analysed separately from the data
relating to single body regions. The two categories for the dependent binary variable of
accuracy were yes and no. These data were analysed in terms of professional
teaching experience using the two categories of the independent variable teaching
experience, as listed above.
- 226 -
4.2.6.2 Reliability of Identification of the Body Region most in need of Correction

The inter-rater reliability of identification of which body region was most in need of
correction was determined by calculating percentage agreement (Po) and Randolphs
kappa (k) values (Randolph, 2005; Randolph, 2008). Unlike Cohens kappa, which only
allows comparison between two raters, Randolphs kappa enables calculation of
reliability of observations between multiple raters (Warrens, 2010). Kappa data for each
group were interpreted and assigned a reliability rating based on the guidelines provided
by Landis and Koch (1977).
(Landis and Koch, 1977)
K value
Interpretation
<0
Poor agreement
0.01 0.20
Slight agreement
0.21 0.40
Fair agreement
0.41 0.60
Moderate agreement
0.61 0.80
Substantial agreement
0.81 1.00
Almost perfect agreement
Table 4.3 Kappa value ranges and interpretation

(from Landis and Koch 1977, p 165)
4.2.6.3 Accuracy of Identification of Correct or Incorrect Technique

To compare the accuracy of qualitative assessment of correct or incorrect technique
for individual body regions, binary logistic regression was also used. The two categories
for the dependent binary variable of accuracy were yes and no. A result of yes
was achieved for accuracy if a correct kinematic variable was rated as correct, or if
an incorrect kinematic variable was rated as incorrect. If either a Type I Error (correct
technique rated as incorrect), or a Type II Error (incorrect technique rated as correct)
occurred, then a result of no was obtained for accuracy. The independent variables
examined via binary logistic regression were the two categories of professional ballet
teaching experience, as listed above.
In addition, two separate regression models were run to test for the effect of teaching
experience on the occurrence of Type I and Type II errors. One model contained accuracy
data for variables for which no correction was required (i.e. variables for which a Type I
Error could be obtained), and the other model contained variables for which a correction
- 227 -
was required (i.e. variables for which a Type II Error could be obtained). Both regression
models contained one independent categorical variable, years teaching experience, and a
binary dependent variable of accuracy, with two possible options of yes or no.
To further assess the effect of the number of years of professional teaching experience, on
the accuracy of qualitative assessment, the percentage of accurate responses was
calculated for each professional teacher, and then linear regression was performed against
the number of years of professional teaching experience.
For the professional teachers who were at least 24 years of age (i.e. those of an age at
which they could have completed at least 6 years of professional performance experience
as a classical ballet dancer), independent sample t-tests were conducted to check for
differences in mean percentage accuracy for the following sub groups:
x
Ex-professional ballet dancer versus non ex-professional ballet dancer
Completed teacher training course versus did not complete teacher training course
4.2.6.4 Accuracy of Identification of Type of correction required

Binary logistic regression was also used to test for a statistically significant effect of the
independent variable, ballet teaching experience, on the accuracy of response to which
type of correction is required. The binary dependent variable was accuracy with two
possible options of yes or no, and again the categorical independent variable
examined was the two categories of professional ballet teaching experience, as listed
above.
- 228 -
4.3 Results
4.3.1 Participant Demographics
A total of thirty five professional and student dance teachers volunteered to participate in
the study, all 35 participants completed all data collection requirements. Of these
recruited participants, 19 were student teachers enrolled in the University of New South
Wales Bachelor of Arts (Dance)/Bachelor of Education program, and 16 were
professional ballet teachers recruited from local dance schools. Upon review of the
participant information questionnaires, data from 2 participants were excluded from the
study. One student teacher from UNSW had received significantly less ballet training (1.5
years) compared to all other participants (mean 14.0 years: SD 2.6 years), and was
excluded on these grounds, while one professional teacher with 20 years experience
described herself primarily as a Pilates teacher rather than a ballet or dance teacher.
Pilates and ballet are different movement disciplines, with differing criteria on which
technical skill and competency are based, the data from this participant were therefore
also excluded from the study.
On synthesising and reviewing the subject demographic data, it became apparent that
rather than assigning participants to one of two groups (i.e. student ballet teachers and
professional ballet teachers) as originally intended and outlined in the methods section, it
would actually be more appropriate to assign participants into three different groups. Six
of the student teachers listed on their questionnaire that they had already had some
professional experience in classical ballet teaching (1 4 years), and the range of
experience listed by the professional ballet teachers was very broad (2 33 years). It was
therefore decided to divide the participant data into three groups based on years of
professional ballet teaching experience.
Therefore, data obtained from each of the 33
recruited teachers included in the analysis were placed into one of three groups,
categorised as follows:
x
Group 1:
Group 2: 1 to 9 years ballet teaching experience
Group 3: Greater than 10 years ballet teaching experience
0 years ballet teaching experience
- 229 -
Group 1 consisted of 12 participants, all of whom were student dance teachers enrolled at
UNSW who were yet to work professionally as classical ballet teachers. Group 2
contained a total of 11 participants, 6 of whom were student dance teachers enrolled at
UNSW who listed that they had worked for between 1 and 4 years as professional ballet
teachers, and 5 participants were professional ballet teachers, with ballet teaching
experience ranging from 2 to 9 years. Ten professional ballet teachers were assigned to
Group 3, their professional ballet teaching experience ranged from 10 to 33 years.
0 Years Ballet
Teaching
Experience
(N=12)
1-9 Years Ballet

Teaching
Experience
(N=11)
z
Teaching
Experience
(N=10)
Years professional ballet teaching

experience (mean SD) *
0.0 0.0
3.9 2.8
18.1 8.2
Total years ballet training (mean SD)
14.0 2.8
15.3 1.8
12.7 2.8
Years employed as professional ballet

dancer (mean SD) *
0.0 0.0
1.1 2.4
4.6 6.0
Age in years (mean SD) *
19.7 1.1
23.5 5.5
44.9 10.0
Group
Table 4.4 Participant demographics. *ANOVA (p<0.05)
A oneway ANOVA was conducted to test for statistical differences in demographic data
between groups.
Statistical differences were observed between groups for years
professional teaching experience (F = 42.70; p =0.000), years employed as a professional

ballet dancer (F = 4.804; p = 0.015), and age (F = 48.17; p =0.000). Statistically
significant differences were not observed between groups for total years ballet training (F
= 2.610; p = 0.090).
Since the total years of ballet training was not significantly different between groups, the
effect of this variable on accuracy of qualitative analysis of classical ballet technique
cannot be investigated in this study, and will therefore not be included in any statistical
analyses.
- 230 -
4.3.2 Accuracy and Reliability of Identification of the Body Region Most in Need of
Correction
On their initial viewing of each step, participants were asked to answer the following
question:
If you could only give this dancer one correction on the execution of this step, which
body region would you concentrate on correcting?
Frequency of responses for each body region for each step for Groups 1-3 are presented
in Tables 4.5 4.7, respectively. The percentage of accurate responses for each step and
the overall percentage accuracy for all steps combined is also presented.
Step
(Accurate Response)
Answer Options
Total
%
Accurate
Responses
Thorax
Pelvis
Hips
Knees
Ankles/
Feet
Grand Pli in First

(Thorax)
0*
12
0%
Rise Second
(Pelvis)
2*
12
17%
Grand Battement
Derriere
(Ankles/Feet)
1*
12
8%
Double Pirouette En
Dedans
(None)
4*
12
33%
Temps Lev
(Pelvis)
2*
12
17%
60
(100%)
15%
Other
None
14
13
3
4
12
4
10
(23%)
(22%) (5%)
(6%)
(20%)
(6%)
(17%)
Table 4.5 Response counts for Group 1 (Teaching Experience = 0 years) (N = 12)
*Accurate response.
Total for All Steps
- 231 -
Step
(Accurate Response)
Answer Options
Total
%
Accurate
Responses
Thorax
Pelvis
Hips
Knees
Ankles/
Feet
Other
None
Grand Pli in First

(Thorax)
2*
11
18%
Rise Second
(Pelvis)
2*
11
18%
Grand Battement
Derriere
(Ankles/Feet)
1*
11
9%
Double Pirouette En
Dedans
(None)
5*
11
45%
Temps Lev
(Pelvis)
1*
11
9%
Total for All Steps
18
(33%)
10
(18%)
4
(7%)
9
(16%)
4
(7%)
1
(2%)
9
(16%)
55
(100%)
20%
Total
%
Accurate
Responses
Table 4.6 Response counts for Group 2 (Teaching Experience = 1-9 years) (N = 11)
*Accurate response.
Step
(Accurate Response)
Answer Options
Thorax
Pelvis
Hips
Knees
Ankles/
Feet
Grand Pli in First

(Thorax)
1*
10
10%
Rise Second
(Pelvis)
7*
10
70%
Grand Battement
Derriere
(Ankles/Feet)
0*
10
0%
Double Pirouette En
Dedans
(None)
2*
10
20%
Temps Lev
(Pelvis)
4*
10
40%
Total for All Steps
16
(32%)
23
(46%)
0
(0%)
3
(6%)
4
(8%)
2
(4%)
2
(4%)
50
(100%)
28%
Other
None
Table 4.7 Response counts for Group 3 (Teaching Experience \HDUV1
*Accurate response.
- 232 -
When considering total responses for all steps, the most experienced teachers showed an
overall tendency to focus most of their corrections on the pelvis (46%), followed by the
thorax (32%) (Table 4.7). A more even distribution of responses was given within the
two groups of less experienced teachers (Table 4.5 & Table 4.6). When combining all
subjects, the thorax (29%) and pelvis (28%) were the two body regions most frequently
selected for correction (Table 4.8).
Step
(Accurate Response)
Answer Options
Total
%
Accurate
Responses
Thorax
Pelvis
Hips
Knees
Ankles/
Feet
Grand Pli in First

(Thorax)
3*
10
33
9%
Rise Second
(Pelvis)
12
11*
33
33%
Grand Battement
Derriere
(Ankles/Feet)
15
2*
33
6%
Double Pirouette En
Dedans
(None)
13
11*
33
33%
Temps Lev
(Pelvis)
16
7*
33
21%
21
(13%)
165
(100%)
21%
Other
None
48
46
7
16
20
7
(29%)
(28%) (4%)
(10%)
(12%)
(4%)
Table 4.8 Response counts for all participants (N = 33)
*Accurate response.
Total for All Steps
Reliability data (Table 4.9) indicate greater interrater reliability of responses between
teachers with at least 10 years experience (k = 0.269), compared to teachers with less than
10 years experience (k = 0.070 - 0.075). Although higher than the less experienced
groups, the reliability of accurate responses amongst the most experienced group was still
very modest, with a reliability rating of only fair being achieved.
0 Years Ballet
Teaching Experience
1-9 Years Ballet

Teaching Experience
z
Teaching Experience
Percentage Agreement (Po)
0.203
0.207
0.373
Kappa (k)
0.070
0.075
0.269
Reliability Rating
Slight
Slight
Fair
Group
Table 4.9 Inter-rater reliability data for each group of teachers
- 233 -
In relation to the accuracy of responses, comparison of the teachers qualitative responses

with quantitative data obtained from 3D motion capture reveals that there is only a very
low level of response accuracy for all three groups of subjects. The highest mean
percentage accuracy for any group is only 28%, this was achieved by the group of
teachers with at least 10 years of professional experience (Figure 4.2).
100%
Percentage Responses
90%
80%
70%
60%
50%
Accurate
40%
Innacurate
30%
20%
10%
0%
0 years experience
(N=12)
1-9 years experience

(N=11)
(N=10)
Professional Ballet Teaching Experience

Figure 4.2 Percentage accurate and inaccurate responses by group for identification of which
body region requires the most correction. Data include all responses for all steps. The range
for accurate responses was 0-40% for all three groups. Error bars = +1 standard deviation.
120%
Percentage Responses
100%
80%
60%
Accurate
40%
Innacurate
20%
0%
Grand Plie in
First Position
* in
Rise
Second
Position
* En Temps Leve
Grand
Pirouette
Battement
Dedans
Derriere
Step Observed
Figure 4.3 Percentage accurate and inaccurate responses by step observed for identification of
which body region requires the most correction. Data include all responses for all subjects. (*p
< 0.05). Error bars = +1 standard deviation.
- 234 -
Despite an apparent trend of increasing accuracy with increasing experience in classical

ballet teaching (Figure 4.2), statistically, results of the binary logistic regression indicate
that there is no significant main effect for teaching experience (p = 0.147) (Table 4.10).
A significant main effect for step observed was demonstrated (p = 0.028), with greatest
accuracy being obtained for the rise (p = 0.037) and pirouette (p = 0.021) steps.
B
Experience
Step*
Constant
S.E.
0 years
df
Sig. (p)
0.147
Exp(B)
Odds
Ratio
95% C.I. for Exp(B)

Lower
Upper
1-9 years
0.519
0.525
0.323
1.681
0.6
4.705
1.005
0.514
0.051
2.732
0.997
7.487
0.028
Rise*
1.505
0.722
0.037
4.504
1.094
18.55
Battement
-0.443
0.953
0.642
0.642
0.099
4.159
Pirouette*
1.65
0.717
0.021
5.207
1.276
21.246
Temps Lev
1.011
0.747
0.176
2.747
0.635
11.884
-2.848
0.705
0.058
Pli
Table 4.10 Output (variables in the equation) for binary logistic regression of dependent variable
accuracy of identification of body region most in need of correction. Each category for each independent
variable (Experience; Step) is compared to the reference category (i.e. first category listed). (*p < 0.05)
4.3.3 Accuracy of Diagnosis of Correct or Incorrect Technique

Measures of accuracy for both correct and incorrect technique are presented in terms
of an overall percentage accuracy score, calculated using responses to all 20 questions.
Percentage accuracy data for each group are presented in Figure 4.4.
Because there was a non-uniform spread of correct and incorrect kinematic variables
across each step and each body region, accuracy of qualitative analysis was not assessed
in relation these parameters. That is, any apparent results indicating significant effects of
step or body region could potentially be artefacts of whether or not correction was
required. For this reason, the effects of step and body region on accuracy of qualitative
analysis were not examined. The regression model therefore only contained one
independent categorical variable: years teaching experience (0 years, 1-9 years,
years). The binary dependent variable was accuracy with two possible options of yes
or no.
- 235 -
100%
Percentage of Responses
90%
80%
70%
60%
50%
Accurate
40%
Inaccurate
30%
20%
10%
0%
*
1-9 years

experience
experience
0 years experience
Figure 4.4 Percentage accurate and inaccurate responses by group for identification of correct
and incorrect technique. Data include all responses for all steps. Error bars = +1 standard
deviation.
B
Experience*
Constant
S.E.
0 years
df
Sig. (p)
0.013
95% C.I. for

Exp(B)
Exp(B)
(Odds
Ratio)
Lower
Upper
1-9 years
0.077
0.187
0.679
1.081
0.749
1.559
ars*
0.547
0.196
0.005
1.728
1.176
2.539
0.05
0.129
0.699
1.051
accuracy of diagnosis of correct or incorrect technique. Each category for the independent variable
(Experience) is compared to the reference category (i.e. first category listed). (*p < 0.05)
Results of the regression model show that there is a significant main effect of ballet
teaching experience on the accuracy of qualitative analysis of ballet technique (p =
0.013). Specifically, teachers with 10 or more years of teaching experience (Group 3)
were more accurate in identifying correct or incorrect ballet technique (p = 0.005) than
teachers with no experience (Group 1), or with 1 to 9 years experience (Group 2).
4.3.3.1 Further Investigation of the Effects of Teaching Experience on Accuracy of
Diagnosis of Correct or Incorrect Technique
Having established, using categorical data, that there is a statistically significant main
effect of teaching experience on the accuracy of diagnosis of correct or incorrect
- 236 -
ballet technique, further analysis was conducted on the data from all subjects that had a
minimum of one year of teaching experience. This analysis took the form of a linear
regression of years teaching experience against percentage accuracy.
Overall, using data from all subjects with professional teaching experience (N = 21),
results of the regression model again indicate that there is a significant effect of years
teaching experience (p = 0.023) on the accuracy of identification of correct or
incorrect ballet technique.
Percentage Correct Responses
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 21
R = 0.2951
p = 0.023
10
15
20
25
30
35
Professional Ballet Teaching Experiene (years)

Figure 4.5 Linear regression of years teaching experience against percentage
accuracy for identification of correct or incorrect ballet technique.
Separate linear regression analyses were also conducted on data for subjects with 1-9
years teaching experience (N = 11), and subjects with 10 or more years of teaching
experience (N = 10). Again, regressions were conducted for years teaching experience
against percentage accuracy. Results of the linear regression models show that there is
a significant relationship, between years teaching experience and percentage
accuracy for teachers with 1 to 9 years teaching experience (p = 0.011), but for 10 years
of experience and above there is a no significant relationship between years teaching
experience and percentage accuracy (p = 0.961). Figures 4.6 and 4.7 graphically
display the results of these linear regressions.
- 237 -
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 11
R = 0.5311
p = 0.011
4
6
8
10
Professional Ballet Teaching Experience (years)

accuracy for identification of correct or incorrect ballet technique for all
teachers with 1-9 years experience
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 10
R = 0.0003
p = 0.961
0
10
15
20
25
30
35

accuracy for identification of correct or incorrect ballet technique for all
teachers with 10 or more years experience.
Having established that ballet teachers in the earlier period of their professional
experience (i.e. 1-9 years) demonstrate improvement in accuracy with increasing
experience, it is interesting to investigate how it is that their accuracy improves. That is,
do they become more competent in accurately diagnosing correct technique or incorrect
technique, or both? Separate linear regressions were therefore conducted to test for a
relationship between accuracy of qualitative analysis of correct and/or incorrect ballet
- 238 -
technique. Figures 4.8 and 4.9 depict the results of these analyses for teachers with 1-9
years experience.
Perentage Accurate
Incorrect Variables Only (11 variables)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 11
R = 0.4247
p = 0.027
0
10

accuracy for identification of incorrect ballet technique for all teachers with 1-9
years experience
Perentage Accurate
Correct Variables Only (9 variables)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 11
R = 0.0172
p = 0.728
0
10

accuracy for identification of correct ballet technique for all teachers with 1-9
years experience
Results of the separate linear regressions of percentage accuracy of diagnosis of correct

and incorrect technique, against years of teaching experience indicate that with increasing
experience early career ballet teachers (1-9 years experience) become more competent at
accurately identifying incorrect technique (p = 0.027), but do not significantly change in
their ability to accurately identify correct technique (p = 0.728).
- 239 -
Perentage Accurate
Incorrect Variables Only (11 variables)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 10
R = 0.0493
p = 0.537
10
15
20
25
30
35

accuracy for identification of incorrect ballet technique for all teachers with 10 or
more years experience.
Perentage Accurate
Correct Variables Only (9 variables)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
N = 10
R = 0.0423
p = 0.573
10
15
20
25
30
35

accuracy for identification of correct ballet technique for all teachers with 10 or
more years experience.
A similar linear regression analysis was conducted on teachers with 10 or more years
experience. These data reveal that for this group, the overall lack of increase in accuracy
with increasing experience (Figure 4.7), was due to a lack of significant improvement in
accurate diagnosis of either incorrect (Figure 4.10) or correct (Figure 4.11) ballet
technique.
- 240 -
4.3.3.2 Effect of Teaching Experience on Type of Diagnostic Error

Despite achieving a statistically significantly higher percentage accuracy than all
participants with less than 10 years teaching experience, the teachers with 10 or more
years experience only obtained a mean percentage accuracy of 65%. Since a relatively
large proportion of inaccurate response were returned by all groups, it is therefore
interesting to investigate the types of errors made by each group. Figure 4.12 presents the
proportions of Type I Errors (correct technique rated as incorrect) and Type II Errors
(incorrect technique rated as correct) with respect to professional ballet teaching
experience.
100%
90%
80%
70%
60%
50%
Accurate
40%
Type I Error
30%
Type II Error
20%
10%
0%
0 years experience
1-9 years
experience
**
experience

Figure 4.12 Percentages of Accurate responses, and Type I and Type II Errors by group.
(*p < 0.05). Error bars = +1 standard deviation.
The student teachers with no professional ballet teaching experience achieved the lowest
percentage (15%) of Type I Errors, in which they rated correct technique as incorrect,
and the highest percentage (34%) of Type II Errors, in which incorrect technique was rate
as correct (Table 4.12). The proportions of error type were the opposite for the most
experienced group of teachers. For the group with 10 or more years of experience, Type I
Errors (21%) were more frequent than Type II Errors (14%).
Results of the binary logistic regression models show that there is not a significant main
effect of ballet teaching experience on occurrence of Type I Errors (i.e. rating correct
- 241 -
technique as incorrect) (p = 0.109) (Table 4.12), but there is a significant main effect
for Type II Errors (i.e. rating incorrect technique as correct), (p = 0.000).
Specifically, teachers with at least 10 years experience obtained significantly fewer Type
II errors than both groups of less experienced teachers (p = 0.000) (Table 4.13).
Experience
95% C.I. for

Exp(B)
Exp(B)
(Odds
Ratio)
Lower
Upper
S.E.
df
2
Sig. (p)
0.109
1-9 years
-0.177
0.291
0.544
0.838
0.473
1.483
ears
-0.604
0.294
0.040
0.547
0.307
0.972
0.693
0.204
0.001
2.00
0 years
Constant
occurrence of Type I Error. Each category for the independent variable (Experience) is compared to
the reference category (i.e. first category listed).
Experience*
95% C.I. for

Exp(B)
Exp(B)
(Odds
Ratio)
Lower
Upper
S.E.
df
2
Sig. (p)
0.000
1-9 years
0.280
0.256
0.273
1.324
0.802
2.184
ears *
1.537
0.283
0.000
4.651
2.673
8.093
-0.463
0.179
0.010
0.630
0 years
Constant
occurrence of Type II Error. Each category for the independent variable (Experience) is compared to
the reference category (i.e. first category listed). (*p < 0.05)
The significantly fewer occurrences of teachers with at least 10 years experience

obtaining Type II Errors is consistent with the sensitivity data (Table 4.14). This group
achieved the highest sensitivity value of 75%, indicating their increased ability to
accurately identify incorrect technique. In contrast, this group achieved the lowest
specificity (i.e. accurate identification of correct technique), this is consistent with their
low percentage for Type I Errors.
Teaching
Experience
Sensitivity
Specificity
0 years
39%
67%
1-9 years
45%
63%
75%
52%
Table 4.14 Sensitivity and specificity by teaching experience
- 242 -
It is also interesting to assess the size of the deviations in joint angles that were
considered, through quantitative means, to be incorrect. That is, what size difference in
joint angle results in incorrect technique for each joint, and how well can ballet
teachers visually detect these differences? Table 4.15 provides a summary of the
magnitude of the deviations in joint angles that were identified through quantitative
means as being incorrect. Data outlining the percentage of teachers in each category of
teaching experience that were able to use visual observation to detect joint angle
differences of these magnitudes are also presented in Table 4.15. These data show that
mean differences of 5 in thoracic anterior/posterior tilt and 6 for pelvic
anterior/posterior tilt were detected by 78% and 88% of experienced teachers (10
experience teachers), respectively. For knee flexion/extension, mean differences of 5
were less well detected by experienced dance teachers (60%), as were 7 differences in
external hip rotation (40%). The high percentages of detection of deviations by
experienced teachers indicate that for the thorax and pelvis, differences of 5 and 6
would constitute a meaningful functional difference to the dance community. However
for knee flexion/extension and external hip rotation, based on the qualitative analysis
abilities of experienced ballet teachers participating in the current study, deviations of 5
and 7 may not constitute a meaningful functional difference in the dance community.
Thorax
Ant/Post
()
Pelvic
Ant/Post
()
Hip Int/Ext
Rot
()
Knee
Flex/Ext
()
Grand Pli First
4.6
4.6
0.3
1.4
Rise Second
3.0
5.8
7.4
5.3
Grand Battemant Derriere
8.3
9.1
2.9
6.5
0.9
0.2
4.3
1.1
Temps Lev
5.5
5.4
2.5
4.6
Mean difference from professional dancer group mean
5.3
6.2
7.4
5.9
Teachers that detected error (All)
61%
59%
24%
33%
Teachers that detected error (
78%
88%
40%
60%
Teachers that detected error (1-9 years experience)
59%
48%
18%
27%
Teachers that detected error (0 years experience)
48%
46%
17%
17%
Table 4.15. Average difference between individual dancer joint angle data and professional dancer group
mean, and percentage of teachers that detected error. Variables for which z-value 1.
- 243 -
4.3.3.3 Effect of Professional Performance Experience and Teacher Training on

Accuracy of Identification of Correct or Incorrect Technique
Independent-samples t-test data for the group of professional teachers who were at least
24 years of age (N = 14), (i.e. those of an age at which they could have completed at least
6 years of professional performance experience as a classical ballet dancer), indicate that
there is no statistically significant difference (p = 0.783) in accuracy of identification of
correct and incorrect ballet technique between teachers who have and have not
previously worked as a professional ballet dancer (Table 4.16).
Not Former Professional Ballet Dancer
Mean
Percentage
Accuracy
63.3%
Former Professional Ballet Dancer
64.4%
Group
Std. Dev.
Percentage
Accuracy
6.1%
Std.
Error
Mean
2.5%
7.8%
2.7%
95% CI
Lower Upper
-9.093
t-test
p value
7.010
0.783
Table 4.16. Results of independent-samples t-test for effects of professional performing experience on
percentage accuracy.
An independent-samples t-test also shows that there is no statistically significant

difference (p = 0.855) in accuracy of identification of correct and incorrect ballet
technique between teachers who have and have not completed a teacher training program
(Table 4.17).
No Teacher Training
Mean
Percentage
Accuracy
63.6%
Teacher Training
64.3%
Group
Std. Dev.
Percentage
Accuracy
5.6%
8.4%
95% CI
Std. Error
Lower Upper
Mean
2.1%
-9.139 7.710
3.2%
t-test
p value
0.855
Table 4.17 Results of independent-samples t-test for effects of teacher training program on percentage
accuracy.
4.3.4. Accuracy of Identification of Type of Correction Required

The mean number of variables rated as incorrect by each group, of the possible 11,
were 3.9, 4.7, and 7.7 for participants with no professional ballet teaching experience, 1
9 years professional ballet teaching experience, and 10 or more years of professional
ballet teaching experience, respectively. Results of a oneway ANOVA reveal that
teachers with 10 or more years of experience identified significantly more kinematic
- 244 -
variables as being incorrect compared to less experienced teachers (F = 13.8; p =

0.000).
Results of the binary logistic regression model (Table 4.18) show that there is no
significant main effect of ballet teaching experience (p = 0.087) or step observed (p =
0.777) on the accuracy of identification of the type of correction required, but data
indicate that there is a significant main effect for body region observed (p = 0.006). In
particular, the corrections given for the knee (p = 0.027) and pelvis (p = 0.009) were more
accurate than corrections given for the thorax.
120%
100%
80%
60%
Accurate
Inaccurate
40%
20%
0%
0 years experience
1-9 years
experience
>=10 years
experience

Figure 4.13 Percentage accurate and inaccurate responses by group for identification of type
of correction required. Data include all responses for all steps. Error bars = +1 standard
deviation.
- 245 -
120%
100%
80%
60%
Accurate
Inaccurate
40%
20%
0%
Plie
Rise
Grand
Battement
Temps Leve
Step Observed
Figure 4.14 Percentage accurate and inaccurate responses by step observed for identification of
type of correction required. Data include all responses for all subjects. Error bars = +1
standard deviation.
120%
100%
80%
60%
Accurate
Inaccurate
40%
20%
0%
Thorax
*
Pelvis
*Knee
Body Region Observed

Figure 4.15 Percentage accurate and inaccurate responses by body region observed for
identification of type of correction required. Data include all responses for all subjects. Error
bars = +1 standard deviation.
- 246 -
B
Experience
Step
Body Region*
Constant
S.E.
df
0 years
Sig. (p)
0.087
Exp(B)
Odds
Ratio
95% C.I.for Exp(B)

Lower
Upper
1-9 years
-1.196
0.549
0.029
0.302
0.103
0.887
-0.646
0.54
0.231
0.524
0.182
1.509
0.777
Rise
0.513
0.57
0.368
1.67
0.546
5.108
Battement
0.205
0.558
0.713
1.227
0.411
3.663
Temp Lev
0.475
0.561
0.398
1.607
0.535
4.831
0.006
Pli
Thorax
Pelvis*
1.107
0.423
0.009
3.027
1.321
6.937
Knee*
2.387
1.081
0.027
10.886
1.309
90.552
1.089
0.588
0.064
2.971
accuracy of type of correction. Each category for each independent variable (Experience; Step; Body
Region) is compared to the reference category (i.e. first category listed). (*p < 0.05)
- 247 -
4.4 Discussion
The accuracy with which professional and student classical ballet teachers identify
correct and incorrect classical ballet technique was investigated by comparing their
qualitative responses with 3D quantitative data. Data were analysed with respect to three
groups of teaching experience: 0 years experience; 1-9 years experience; and \HDUV
experience.
4.4.1 Accuracy and Reliability of Identification of the Body Region Most in Need of
Correction
In their first observation of each ballet step, participants were not given specific
instructions as to which part of the body to observe, but were instead given the
opportunity to use their own observation strategy. In watching each step the participants
were asked which body region, if any, would you concentrate on correcting. Very low
percentage accuracy scores were achieved for all groups for this task (i.e. 15%, 20% and
28% for 0, 1-9 and \HDUVH[SHULHQFHUHVSHFWLYHO\ZLWKQRVWDWLVWLFDOO\VLJQLILFDQW
difference between groups. Amongst studies that have investigated the accuracy of
qualitative analysis through comparison with quantitative data (Section 2.3), no study was
identified in which participants were able to choose where and what to observe. As far as
the author is aware therefore, this is the first study of its kind in which participants were
not directed to observe specific body regions, or look for particular movement qualities
when asked to identify correct or normal movements. Consequently, it is not possible
to compare the results for this style of question with existing published data.
The most experienced group of teachers achieved greater interrater reliability (i.e. fair)
than both less experienced groups (i.e. poor), indicating that when using their own
observation strategy there is greater consistency amongst the more highly experienced
teachers. Teachers with at least 10 years experience chose to correct the pelvis 48% of the
time and the thorax 32% of the time, compared to 0-8% for all other regions. It was body
regions or movements relating to the fundamental principles of placement and
alignment therefore, which were most frequently identified as needing correction by the
most experienced teachers. For both less experienced groups, there was a more even
- 248 -
distribution of body regions selected for correction. It therefore appears that despite their
low accuracy in terms of agreement with the quantitative data, as a group the most
experienced teachers are more consistent and perhaps more systematic in their method of
analyses. Results from studies on the visual search strategies used by movement coaches
and judges have found that novices and experts differ in their search patterns. Visual
search strategy refers to the way that the eyes move around the display in an attempt to
direct visual attention towards relevant sources of information (Williams, 2002, p169). It
has been reported that compared to novices, expert movement coaches and judges have
fewer fixations and that these are of longer duration (Bard et al., 1980; Moreno et al.,
2002; Avila and Moreno, 2003). Observer experience has also been found to influence
the body regions which are most commonly fixated upon (Bard et al., 1980; Petrakis,
1986). For example, expert gymnastic judges fixated on the upper body (i.e. head, arms),
while the novice viewed the lower body (i.e. legs) (Bard et al., 1980). It has generally
been concluded therefore that experts have a more compact scan path (Petrakis, 1986,
p258) and are more selective in their search strategy as they know which are the most
informative areas of the display and ignore areas that do not provide important
information. Moreover, it has been suggested that to effectively facilitate movement skill
acquisition, instructors and coaches should have a well-developed internal template or
model of the kinematic characteristics of sound technique (Sherman et al., 2001, p258).
It is possible therefore, that in the current study the increased frequency of selection of
the pelvis and thorax by the more experienced teachers reflects prolonged fixation on
these areas, due to these teachers internal template being based primarily on the
movements of these regions. That is, perhaps for the experienced teachers, pedagogical
experience has shaped the development of their internal template, thereby influencing
the way in which they prioritise observation of and focus on particular body regions.
(Sherman et al., 2001)
4.4.2 Accuracy of Diagnosis of Correct or Incorrect Technique
All groups of ballet teachers achieved better accuracy of qualitative analysis when
directed to look at specific body regions, and answer either yes or no to whether the
dancers needed correction in this area. Overall percentage accuracy scores were 51%,
53%, and 65% for this question for teachers with 0, 1-9 and \HDUV H[SHULHQFH
respectively, compared to 15%, 20% and 28% when allowed to choose their own area of
observation. This improvement in accuracy of observation is consistent with the findings
- 249 -
discussed in the literature review (Section 2.3), where it was concluded that the number
of parameters observed, and the number of scale categories or answer options affects
accuracy of observation. In the first style of question, teachers were essentially required
to observe the dancers whole body and select from one of seven possible categories. In
subsequent questions however, teachers were directed to observe only one specific body
region, were told exactly which movement variable to assess, and were only provided
with two answer categories. The improvement in accuracy obtained for this type of
question is therefore consistent with existing literature in regards to the factors that affect
accuracy of qualitative analysis of human movement. However, despite this
improvement, it should be noted that with only two options to select from, the increased
scores were still only slightly better than chance (i.e. 50%), particularly for the less
experienced groups.
In considering the effect of teacher experience on accuracy of identification of correct

and incorrect classical ballet technique, teachers with at least 10 years experience
performed significantly better than both groups of teachers with less than 10 years
experience. A similar effect of teaching experience on accuracy of observation of skilled
movement was reported by Imwold and Hoffman (1983) and Plessnier and Schallies
(2005) for analysis of gymnastics, however, Knudson (1999) found that teaching
experience did not significantly affect accuracy of observation for general physical
education activities such as the vertical jump. It has been reported (Biscan and Hoffman,
1976) that familiarity with and exposure to particular movements increases the ability to
compare these movements to an ideal movement prototype or template. Instruction of
activities such as gymnastics or classical ballet requires knowledge and familiarity with
very specific and precise movement styles, more so than general physical education
activities. The increased accuracy of observation demonstrated by the more experienced
group, may therefore reflect increased familiarity enabling more accurate comparison of
the observed movements with the ideal prototype.
In the current study, significant improvements in accuracy of qualitative analysis

occurred between 1 and 9 years experience, however this improvement plateaued beyond
10 years experience. It is therefore the earlier rather than later phases of a dance teaching
career when significant improvement in qualitative analysis is made. Chase and Simon
- 250 -
(1973) proposed that the attainment of expertise in any skilled activity was the result of
acquiring vast amounts of knowledge and ability during many years of experience in a
particular domain. Specifically, it has been stated (Chase and Simon, 1973; Bloom, 1985)
that a 10 year rule of necessary preparation can be generalised to several different
domains. According to this rule, not even the most talented of individuals can attain
elite performance without approximately 10 years of preparation (Chase and Simon
1973, p240). Perhaps it is this rule that is responsible for the better accuracy
demonstrated in the current study by teachers with at least 10 years experience, however
it does not necessarily explain the plateau in accuracy for these teachers beyond the 10
year mark.
Ericsson and Charness (1994, p783) state that individuals improve their performance
and attain an expert level, not as an automatic consequence of more experience with an
activity but rather through structured learning and effortful adaptation. Perhaps in the
earlier phase of their professional teaching career (1-9 years) ballet teachers put more
effort and interest into improving their skills, and that after 10 years of experience this
effort declines Thus, simply continuing to instruct ballet without concerted effort being
placed on improving specific components of teaching practice, may not necessarily result
in improved skill with continued experience. According to Ericsson et al. (1993), the
primary mechanism responsible for creating expert performance in a domain is deliberate
practice, during which optimal opportunities for learning and skill acquisition are
provided. Most people spend a very small amount of time on deliberate efforts to improve
their performance once it has reached an acceptable level. Under these conditions only
weak relations between experience and performance would be predicted (Ericsson et al.,
1993; Ericsson and Charness, 1994). So, according to this suggestion, it may be that after
10 years of experience, classical ballet teachers feel satisfied with their level of
competence and therefore less deliberate effort is directed towards skill improvement.
Despite the improvement in accuracy when teachers were required to observe only one
body region, and the significantly greater accuracy achieved by the most experienced
group, in effect the percentage accuracy scores are actually surprisingly low. An average
result of 65% accuracy for the teachers with at least 10 years experience is modest, and
indicates that these experienced teachers were inaccurate in rating ballet technique as
- 251 -
either correct or incorrect for over one third of observations, and only 15% better
than through pure chance. Other studies that have reported percentage accuracy of
qualitative analysis, based on comparison of responses with quantitative kinematic data,
have presented higher accuracy results. For qualitative gait analysis Patla and Clouse
(1988) obtained percentage accuracy data ranging from 40% to 84% and Wren et al.
(2005) reported values of 61% to 88%. In comparison to the highest percentage accuracy
values obtained for these clinical qualitative observations, the data obtained in the current
study are low. Given that some of the teachers in this group had as much as 20 and 30
years teaching experience, this level of inaccuracy does not engender confidence in the
qualitative analysis and diagnostic abilities of highly experienced professional ballet
teachers.
A possible explanation for the relatively low accuracy of qualitative analysis achieved
even for experienced teachers may be that most of the dancers appearing in the videos
only deviated slightly from the ideal or correct position. The dancers used in the videos
were all of a professional level, and while they deviated from the group mean by more
than one standard deviation for nearly half of the variables observed, they were still
highly skilled dancers. Research (Wren et al., 2005; Plessner and Schallies, 2005; Passier
et al., 2010; Dallas et al., 2011) has shown that the degree of deviation from an ideal or
easily identifiable angle (e.g. neutral) influences the level of accuracy of observation. For
example, accuracy of visual observation of joint position increases with increasing
deviation of a joint angle from neutral. This suggests that if less experienced dancers with
greater deviations from the ideal appeared in the videos, the teachers would have
achieved better percentage accuracy scores. The results indicate that at the more highly
skilled and elite end of movement instruction (i.e. instruction of those that have already
achieved an elite level), the naked eye is not well enough equipped to detect subtle
differences and problems in technique. At a high level of movement ability corrections
become more refined and subtle, consequently the ability to accurately diagnose
technique at this level is extremely important.
Although only achieving modest accuracy even after 10 years experience, it was the
identification of incorrect technique that is primarily responsible for the overall
improvement demonstrated during the 1 to 9 year phase of experience. Student teachers
- 252 -
with no professional experience performed poorly in identifying incorrect technique

and therefore obtained the highest proportion of Type II Errors, while teachers with at
least 10 years experience obtained significantly fewer errors of this kind. That is, with
increasing experience from 0 years to over 10 years experience, the teachers
sensitivity to the detection of incorrect technique increased from 39% to 75%.
However, in relation to accurate identification of correct technique (i.e. specificity), a
similar trend of increased accuracy was not observed. In fact, although not to a significant
degree, specificity declined slightly with experience, indicating that more experienced
teachers are more picky, and thus have a higher threshold for identification of
technique as being correct.
Interestingly, the findings obtained in relation to observer experience and error type are
contradictory to some existing literature. Armstrong and Hoffman (1979, p326) stated
that novice teachers create more false alarms, i.e. they reported the presence of an error
when the error was not demonstrated more often than experienced teachers. It was
concluded that experienced teachers are therefore less willing than inexperienced teachers
to take the chance of making a diagnostic error in identifying technique as incorrect.
Similarly, Knudson (2000, p19) stated that many novice instructors are on a mission to
identify errors and tend to over emphasise corrective feedback. This tendency was
defined as a corrective complex. However, the results of the current study do not
concur with the suggestions of either of these authors. In fact, the current study found an
opposite tendency, in that compared to the less experienced teachers, the more
experienced teachers (i.e. \HDUV SURIHVVLRQDO WHDFKLQJ H[SHULHQFH VKRZHG D JUHDWHU
tendency to rate correct elements as incorrect.
(Gordon, 1970; Ste-Marie, 1999)
Gordon (1970, p369) reported that the actual correctness of behaviour observed is a major
factor in determining the accuracy of ratings, and developed the term Differential
Accuracy Phenomena (DAP). This term was defined as the numerical difference in
raters ability to identify correct and incorrect forms of behaviour or movement. Gordon
(1970) and Osborne and Gordon (1972) both reported that ratings were more accurate
when the behaviour or movement was actually correct rather than incorrect, a finding
that would result in a positive DAP. Such a result is consistent with the data obtained for
the less experienced teachers (i.e. 0 years and 1-9 years experience) who demonstrated
- 253 -
higher specificity than sensitivity, but not for teachers with at least 10 years experience.
The most experienced group obtained greater sensitivity (75%) than specificity (52%)
indicating that they were more accurate in identifying incorrect technique. Ste-Marie
(1999) obtained data consistent with this finding for experienced gymnastics judges in
that expert judges were found to be more accurate in identifying form errors in
performance than novice judges.
(Locke, 1972)
Amongst the professional teachers, there was no significant effect of having had a
professional performance career on accuracy of identification of correct and incorrect
ballet technique. However, previous authors (Locke, 1972, p380) have observed that
among physical education teachers, the one with the analytical advantage seems more
often than not to be the one who has performed, practised and studied specific
components of the skill.
Similarly, Sherman (2001, p259) reported that a close
functional relationship exists between level of physical performance and the ability to
observe and interpret movement patterns. Osborne and Gordon (1972, p55) however,
stated that accuracy of ratings was not linked with the physical skill level of the rater. The
current data are in agreement with Osborne and Gordon (1972), in that it was experience
in ballet instruction rather than ballet performance that influenced accuracy of qualitative
analysis of classical ballet. Also of note is the finding that participation in a teacher
training program had no effect on accuracy of qualitative analysis amongst professional
ballet teachers. Perhaps this indicates that curricula including specific training on
methods for qualitative analysis of dance technique have been lacking in teacher training
programs to date.
4.4.3. Accuracy of Identification of Type of Correction Required

If teachers did identify that correction was required for a particular step, they were asked
to specify what type of correction was most appropriate. They were given two possible
correction options to choose from. Overall, percentage accuracy results for responses to
this question were 85%, 70% and 81% for 0, 1-9 and \HDUVH[SHULHQFHUHVSHFWLYHO\
Thus, qualitative analysis of the type of correction required was more accurate than
identification of whether or not a correction was required. It is interesting that neither
teaching experience nor the step observed significantly influenced the accuracy of
identification of the type of correction required, however there was a significant effect of
- 254 -
body region observed. In particular, the corrections given for the knee and pelvis were
more accurate than those given for the thorax. Perhaps there is more agreement among
ballet teachers as to what constitutes correct technique for pelvic placement and knee
extension, than for thoracic alignment. Alternatively, since review of the literature on
accuracy of qualitative analysis of human movement (Section 2.3) indicates that the body
region observed influences accuracy of observation, it could be that when assessing
classical ballet technique, observation of the pelvis and knee is typically more accurate
than for the thorax.
4.4.4 Limitations
The main limitation associated with this study is the number of professional dancers from
which gold standard kinematic data were obtained. Data averaged from 12 professional
dancers were used as a reference for determination of correct classical ballet technique.
Inclusion of data from more professional ballet dancers would certainly add validity to
the method by which correct and incorrect technique were identified. A specific aim
for additional research of this kind would therefore be to increase the sample size on
which the gold standard kinematic data are based.
It is possible that the very low accuracy results obtained for the question on which body
region, if any, would you correct, are due to the fact that the quantitative method by
which the most incorrect body region was identified was not based on anatomical or
pedagogical theory or knowledge. Instead it was based purely on the highest numerical
measure of deviation, with no judgement or prioritisation of the importance of each body
region in achieving correct classical ballet technique. In selecting their answers however,
the participating teachers were likely to have used some level of logic or reason, based on
their understanding of classical ballet technique. Perhaps therefore, this style of question
and analysis was not a valid means of measuring accuracy of qualitative analysis, but
instead exposes the teachers judgments on the prioritisation of corrections.
So as not to influence their qualitative responses, participants were not informed of the
proportions of dance videos that contained correct and incorrect technique. Given
that they were taking part in a scientific experiment, it is possible that the teachers could
have assumed that there were in fact errors in technique in most of the videos they
- 255 -
observed. Therefore despite efforts to eliminate potential bias, it is possible that the
circumstance itself could have introduced bias into the participants responses. However,
the finding that there was a significant difference between groups in the number of
variables identified as incorrect, indicates that it is likely that the experimental
condition was not a significant limitation.
Given that academic literature to date has not provided conclusive evidence on the effect
of the mode of observation (i.e. live vs. video) on the accuracy of qualitative analysis of
human movement (Section 2.3), it is difficult to know whether the use of video analysis
in the current study can be perceived as a limitation. It is acknowledged that dance
teachers typically perform live rather than video observations of their students, and are
able to move around the dance studio to achieve what they consider to be the best vantage
point. Therefore, in an effort to reproduce this situation as closely as possible, and in
accordance with the finding that camera or viewing angle does affect accuracy of
observation (Section 2.3), a synchronised dual-perspective view was provided for each
video in order to minimise any limitations introduced by the methods used.
4.4.5 Conclusions
The conclusion emerging from this study is that experience in classical ballet instruction
has a significant effect on accuracy of qualitative analysis of classical ballet technique.
Participation in a teacher training program or attainment of a professional classical ballet
performance career however, does not significantly influence the accuracy of assessment
of classical ballet technique.
When asked to focus on one body region only, teachers with at least 10 years experience
demonstrated significantly better accuracy of identification of correct and incorrect
ballet technique than teachers with less than 10 years experience. Amongst the teachers in
the earlier phase of their teaching careers (i.e. 1-9 years experience), a significant
relationship was identified between years experience and accuracy of qualitative analysis,
however this was not the case beyond 10 years of experience. Instead, a plateau effect
was observed, where accuracy of observation did not continue to increase with increasing
experience beyond 10 years. The increase in accuracy of qualitative analysis during the 19 year phase of experience occurred as a result of increasing sensitivity to the detection
- 256 -
of incorrect technique. However, a similar trend was not observed for identification of
correct technique (i.e. specificity).
Despite improvements in accuracy during the first 10 years of teaching experience,
overall, the most experienced group of teachers only achieved an average percentage
accuracy score of 65%. This modest result indicates that when observing highly skilled
dancers, the qualitative analysis abilities of classical ballet teachers may be ineffective,
and suggests that increased training in this skill is warranted. Given the relatively low
percentage accuracy scores achieved, it is proposed that development and use of a
quantitative assessment tool, based on 3D motion analysis, could provide a more
objective and systematic process for evaluation of classical ballet technique. The
following chapter will describe the analyses conducted during development of such a
tool, and will present examples of the information the tool could potentially provide.
- 257 -
Chapter 5 Development of a Quantitative Assessment Tool for Classical Ballet Technique
CHAPTER 5
DEVELOPMENT OF A QUANTITATIVE TOOL

FOR ASSESSMENT OF CLASSICAL BALLET
TECHNIQUE
- 258 -
CHAPTER 5
DEVELOPMENT OF A QUANTITATIVE TOOL FOR
ASSESSMENT OF CLASSICAL BALLET TECHNIQUE
5.1 Introduction
The use of instrumented quantitative techniques to analysis dance movements has been
steadily increasing over the last few decades. Of particular note is the growing use of 3D
motion analysis techniques in dance science research during the past five to seven years
(Section 2.2.2). Dance science researchers have used 3D motion analysis to measure and
describe the movements of various dance genres.
The aims of this research have
included: assessing how certain movements may be associated with injury; assessing how
different types of floors or shoes may influence movement patterns; analysing how
different training programmes may affect performance; investigating aspects of motor
control; and to gain further insight into the general kinematics and kinetics of specific
dance movements. Despite this increasing use of 3D analysis, it appears that the
technology is not yet being used as a tool to routinely assess or screen individual dancers
on basic aspects of their technical competency. To date, assessment and evaluation of
competency in the fundamental principles of ballet technique occurs primarily through
qualitative analysis, undertaken by dance instructors in the classroom or dance studio.
This being the case, it is therefore crucial that dance teachers possess the necessary skills
to conduct accurate and effective qualitative analysis.
The preceding chapter investigated the ability of both experienced and inexperienced
ballet teachers to observe the performance of basic classical ballet steps and to detect
correct or incorrect technique. That is, the degree of accuracy of the teachers qualitative
analysis skills were assessed. Comparison of the qualitative responses provided by the
teachers with quantitative kinematic data obtained from 3D motion analysis, revealed that
even ballet teachers with at least 10 years professional experience only achieved a mean
overall percentage accuracy score of 65%. This result indicates that for up to one third of
observations, experienced teachers were inaccurate in rating ballet technique as either
- 259 -
correct or incorrect, and this is only 15% above a result that could be obtained
through chance alone. It could therefore be argued that a quantitative assessment tool in
which student dancers joint movement patterns are objectively compared to a gold
standard of technique, as demonstrated by professional ballet dancers, could be very
useful in providing a systematic method for the diagnosis of technical competency in
classical ballet.
Despite the value of the increased biomechanical detail that can be obtained from fullbody 3D motion analysis, it can be a laborious and time consuming process for all
involved. In order to obtain accurate and meaningful biomechanical data, subject
anthropometric data must be carefully measured, and reflective markers must be
meticulously placed on specific anatomical bony landmarks. Motion analysis cameras
must be accurately calibrated, and appropriate biomechanical models must be identified
and correctly used. As is the case for 3D clinical gait analysis, currently routinely used in
many hospitals throughout the world, 3D dance analysis could potentially take up to 2
hours to be conducted correctly. If a 3D quantitative tool was to be devised for analysis of
classical ballet technique, in order for it to be a feasible routine practice, the process
would need to be as efficient and cost effective as possible. The aim of this section of the
research project then is to develop such a 3D quantitative tool consisting of a minimum
number of independent quantitative kinematic variables and a minimum number of
independent ballet steps for analysis. To do this, the kinematic data collected from a
group of professional ballet dancers will be used as the gold standard for the practical
execution of correct classical ballet technique. That is, a practical biomechanical
benchmark of correct technique will be used, rather than theoretical principles.
Kinematic data from individual non-professional dancers will be used as pilot data to
assist with development of the assessment tool. The quantitative assessment tool will
include the minimum number of kinematic variables that adequately represent the
principles of alignment, placement, turnout, and extension. Careful consideration
will be given to the number and type of steps chosen for inclusion in the tool, with the
aim of selecting a minimum number of steps that can effectively detect deviations in the
relevant kinematic variables.
- 260 -
5.2 Methods
The kinematic data used for analysis in this section were obtained during the data
collection sessions described in Chapter 3. The procedures for ethics approval, participant
recruitment and consent, anthropometric measurement procedures, instrumentation,
experimental protocol, and data processing were therefore the same as described in
Section 3.2.
5.2.1 Data Extraction

Kinematic data collected from the group of professional dancers were used as a gold
standard, against which data from individual non-professional dancers were compared.
It was therefore necessary to average the data across all professional dancers. To do this,
time-normalised (0 to 100% of the movement) data sets were generated for each step for
each dancer according to the step-specific start and end criteria described in Chapter 3
(Section 3.2.5). All data between the step start and end were divided into 20 evenly
spaced time points (i.e. 5% intervals), and these were then averaged across the group of
professional dancers. Time-normalised standard deviation data were also calculated for
the professional group. This process was conducted for each of the nine kinematic
variables that were identified in Chapters 2 and 3 as being the minimum number of
variables that thoroughly represent the four fundamental principles of classical ballet
technique.
Having obtained time-normalised group mean data for the professional dancers,
individual data for each non-professional dancer were compared to the group data by
calculation of z-scores. The z-score is a standard score expressed in terms of standard
deviation units on a normal curve (Hastad and Lacy, 1994). This value indicates how far
above or below a group mean an individual raw score is located. The z-score for an
individual value is calculated using the group mean and standard deviation, according to
the following formula:
z=
where = group mean, = raw score , and s = standard deviation.

- 261 -
5.2.2 Consideration of Steps for Inclusion in the Quantitative Assessment Tool

In designing a quantitative assessment tool for classical ballet technique, it was very
important to carefully consider which steps would be most appropriate for inclusion.
Practically, the number of steps needed to be kept to a minimum, yet the steps needed to
be effective in detecting deviations in the kinematic parameters that relate to the four
fundamental principles of classical ballet technique. Thus, the process of step selection
required identification of steps that were best able to detect differences between the
professional and non-professional dancers in the following kinetic variables.
x
Thoracic anterior/posterior tilt
Thoracic side tilt
Pelvic anterior/posterior tilt
Pelvic obliquity
Pelvic-thoracic transverse rotation
Hip rotation
Ankle rotation
Knee extension
To identify which steps could best detect deviations in these parameters, a number of
calculations based on z-scores were performed. Firstly, the z-scores at each of the 20
time-normalised data points that comprised a complete step were calculated. Then a
single value representing the average z-score across all 20 time points was determined.
This was done for each of the 9 kinematic variables, for each of the 14 steps, for all 14
non-professional dancers (Figure 5.1 Phase 1). From there, the average z-score across
all 14 dancers was calculated for each kinematic variable, and then an average z-score per
step was calculated across all kinematic variables (Figure 5.1 Phase 2). Because the
magnitude, rather than the direction of the z-scores was of most interest (i.e. the amount
of deviation of the non-professional data from the professional data, not the direction of
deviation), the absolute values of the z-scores were calculated. The resultant overall zscore obtained for each step thus provided an indication of how well each step was able to
detect deviations between the professional dancers and non-professional dancers across
all nine kinematic variables throughout all time points within a single step cycle (Figure
5.1 Phase 3).
- 262 -
Phase 1
*Pli in First Position - Thorax Anterior/Posterior Tilt
Time point
Dancer
NP001
Dancer
NP002
Dancer
NP014
Dancer
NP015
0%
z-score
z-score
z-score
z-score
100%
z-score
Mean
z-score for
thorax
ant/post
tilt for pli
in first
z-score
Mean
z-score for
thorax
ant/post
tilt for pli
in first
z-score
Mean
z-score for
thorax
ant/post
tilt for pli
in first
z-score
Mean
z-score for
thorax
ant/post tilt
for pli in
first
Dancer
NP001
Dancer
NP002
Dancer
NP014
Dancer
NP015
A: *Phase 1 conducted for

all 9 kinematic variables
for all 14 steps for all
dancers
Phase 2
**Pli in First Position
Kinematic Variables
Mean
z-score
across all
dancers

Thorax Side Tilt
Pelvic Obliquity
Hip Rotation
Ankle Plant/Dorsi Flexion
C: Overall mean
z-score taken
across all kinematic
variables
Ankle Rotation
B: Mean z-score taken across all dancers for each
kinematic variable for pli in first position
Overall
mean
z-score for
step
Phase 3
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Rise
(Relev )
Jump
(Saut )
Glide
(Glissade)
Dart
(Elanc )
Turn
(Tour)
Step
Overall mean
z-score for
step
D: **Phase 2 conducted for all 14 steps
Pli in First Position

Pli in Second Position
Battement Tendu Derriere
Battement Jet to 45 Front
Rise in First Position
Rise in Second Position
Saut s in First Position
Temps Lev s
Glissades Over
Glissades Under
Jet Ordinaire
Grande Jet Elanc En Avants
Pirouette En Dedans
Pirouette En Dehors
Figure 5.1 Flow chart depicting process for calculation of overall mean z-score per step
- 263 -
From this process (Figure 5.1), a table listing a single overall z-score for each step was
constructed. In addition, the table also names and presents z-scores for any individual
kinematic variables for which an overall z-score of at least 1.0 was obtained across the
step. From this data it was possible to determine which steps were best able to detect
overall deviations in kinematic variables between the professional and non-professional
dancers, and more specifically, the type of kinematic variables for which each step could
detect deviations.
The primary aim in designing the quantitative assessment tool was to establish a
minimum number of ballet steps that could effectively detect deviations of the relevant
kinematic variables from the gold standard. This minimum number of steps was
determined by investigating the effect of various numbers and combinations of steps on
the rank order of the non-professional dancers in terms of the deviation of their kinematic
data from that of the professional dancers. To do this the rank order of the overall zscores (based across all nine kinematic variables) for the 14 non-professional dancers was
compared between various pilot versions of the assessment tool, each of which contained
a different number and combination of steps. A desirable outcome would be for high
correlations of rank order between dancers to be obtained, whether all 14 steps were
included in the assessment tool, or only a few steps, four for example. Such an outcome
would indicate that fewer steps were equally as effective in detecting incorrect or less
than ideal movement patterns, thus eliminating redundancy of data from inclusion of
unnecessary steps, thereby reducing the time and cost for data collection and processing.
The various numbers and combinations of steps selected for each pilot version of the
assessment tool were based on: the overall z-scores obtained for each step (i.e. how well
each step was able to detect overall deviations in kinematic variables between the
professional and non-professional dancers); and the type of individual kinematic variables
for which each step detected deviations. Appropriate numbers of steps selected for
inclusion in each pilot version were 14, 7 and 4. As indicted in Chapter 3, the group of 14
steps initially chosen for analysis contained two steps from each category of the Seven
Movements of Dance (Noverre, 1760). The seven steps selected in the first stage of step
reduction therefore consisted of one step from each movement category. Each of the four
steps selected for the final step reduction stage, also came from a different movement
- 264 -
category. This process ensured that the assessment tool would contain diversity in the
types of steps included. Consequently, to select the steps for the 7-step assessment tool,
the step from each movement category with the highest overall z-score was chosen. For
the 4-step assessment tool, one step from each of the four movement categories with the
highest z-scores was selected. In addition, to investigate any possible effect of the actual
number of steps, as opposed to the combination of steps, on the dancer rankings, another
combination of four steps was selected for analysis. This additional 4-step assessment
tool contained a group of steps selected from the remaining movement categories that
were omitted from the initial 4-step tool.
5.2.3 Analysis of the Effect of the Number and Combination of Steps Included
To investigate the effect of the number and types of steps included in each pilot
assessment tool, it was necessary to calculate an overall z-score for each non-professional
dancer as an indication of the deviation of their kinematic data from the professional data.
To do this, a single value representing the average z-score across all 20 time points was
determined for each kinematic variable, for each step, for each dancer (Figure 5.2 Phase
1). Then an average z-score for each kinematic variable was calculated across all the
relevant steps (i.e. for 14 steps, 7 steps or 4 steps) included in each version of the
assessment tool. Finally, for each pilot tool, an overall z-score was calculated by taking
an average across the z-scores for all nine individual kinematic parameters. (Figure 5.2
Phase 2). It was this overall z-score than was used to rank the dancers, with the highest zscore indicating greatest deviation from the gold standard.
For each different version of the assessment tool (i.e. 14-step tool, 7-step tool, or 4-step
tool), a table was constructed in which the 14 non-professional dancers were listed in
rank order from best (lower z-score) to worst (highest z-score). An additional table
listed individual kinematic variables for which a dancer achieved a z-score of at least 1.0,
thereby indicating the joint or segment movements for which each dancers data deviated
by at least one standard deviation from the professional data. Dancer rank orders for each
of the pilot assessment tools were compared using Spearmans rank correlation
coefficients. This analysis was conducted using SPSSTM version 18.0.
- 265 -
Phase 1
* Dancer NP001 - Thorax Anterior/Posterior Tilt
Time point
0%
100%
Pli in First
Position
Pli in
Second
Position
Pirouette
En Dehors
Pirouette En
Dedans
z-score
z-score
z-score
z-score
z-score
Mean
z-score for
thorax
ant/post
tilt for
Dancer
NP001
z-score
Mean
z-score for
thorax
ant/post
tilt for
Dancer
NP001
z-score
Mean
z-score for
thorax
ant/post
tilt for
Dancer
NP001
z-score
Mean
z-score for
thorax
ant/post
tilt for
Dancer
NP001
Plis in
First
Position
Plis in
Second
Position
Pirouette
En Dehors
Pirouette
En Dedans
A: *Phase 1 conducted for

all 9 kinematic variables
for all dancers
Phase 2
**Dancer NP001
Kinematic Variables
Mean
z-score
across all
steps

Thorax Side Tilt
Pelvic Obliquity
Hip Rotation
Ankle Plant/Dorsi Flexion
C: Overall mean
z-score taken
across all
kinematic
variables
Ankle Rotation
B: Mean z-score taken across all steps for each kinematic

variable for Dancer NP001
Overall
mean
z-score for
Dancer
NP001
Phase 3
Dancer ID
Overall mean
z-score for
dancer
(14 steps)
***Overall
mean z-score
for dancer
(7steps)
***Overall
mean z-score
for dancer
(4 steps)
NP001
NP002
NP003
NP004
NP005
NP006
NP007
NP008
NP010
NP011
NP012
NP013
NP014
D: **Phase 2 conducted for all 14 dancers
E: ***Phase 1 and 2 repeated using only

7 steps, then again using only 4 steps
NP015
Figure 5.2 Flow chart depicting process for calculation of overall mean z-score per dancer
- 266 -
5.2.4 Development of the Assessment Tool Report

While it was important that the assessment report contain specific results for individual
kinematic variables, it was also essential to indicate which fundamental biomechanical
principle each variable related to. Therefore, the individual report constructed for each
dancer consisted of a table listing all nine kinematic variables, the overall z-score
achieved for each kinematic variable, and the fundamental principle with which each
variable was associated, e.g. alignment, placement, turnout, or extension.
In
receiving their results it would be important for individual dancers to get an indication of
how, i.e. in which direction, their joint and segment movements differed from those of the
professionals. Consequently, actual values of the z-scores (i.e. positive or negative),
rather than absolute values were displayed. Z-scores of 1.0 or -1.0 were chosen as the
threshold above or below which a kinematic variable would be considered to be
incorrect, and thus in need of attention. These cut-off values, that is a divergence of at
least one standard deviation from the gold standard or normal population, represents
a considerable proportion (i.e. at least 68%) of the population in question, and can
therefore be considered as an appropriate cut-off value for the current study. Thus, for
each variable for which a z-score of at least 1.0 or -1.0 was obtained, the general results
table also contained comments indicating how this biomechanical parameter differed
from the professional data, suggestions of possible causes, and ways in which the
deviation could potentially be corrected.
To provide a detailed visual representation of the kinematic variables with z-scores of at

least 1.0 or -1.0, data for each of these variables were presented in a graphical format.
For these graphs, percentage complete movement (i.e. 0-100%) was displayed along the
horizontal axis, and degrees of joint or segment rotation appeared along the vertical axis.
To indicate how individual data differed from the averaged professional data, curves for
the mean professional values as well as one standard deviation above and one standard
deviation below the mean were also included in the graphical display. To aid in data
interpretation, generic stick figures depicting the various movement stages of each ballet
step appeared alongside each graph. Basic subject and test date details were also included
in the report. For the purposes of this thesis, all subject information data were included in
a de-identifiable format, e.g. subject ID rather than subject name.
- 267 -
5.3 Results
5.3.1 Consideration of Steps for Inclusion in the Quantitative Assessment Tool
Movement
category
Bend
(Pli)
Stretch
(Battement)
Rise
(Relev )
Jump
(Saut )
Glide
(Glissade)
Dart
(Elanc )
Turn
(Tour)
Step
Overall
z-score
Pli in First
Position
0.910
Pli in Second
Position
0.839
Battement
Tendu
Derriere
Battement
Jet to 45
Front
Kinematic variables with z-score

(z-Score)
Alignment
Placement
Turnout
Extension
Thorax
Ant/Post Tilt
(1.65)
Thorax
Ant/Post Tilt
(1.56)
Hip Rotation
0.794
(1.28)
Hip Rotation
0.770
Rise in First
Position
0.934
Rise in
Second
Position
1.014
Saut s in
First Position
0.635
Temps Lev
0.762
Glissades
Over
0.614
Glissades
Under
0.610
Jet
Ordinaire
0.672
Grande Jet
Elanc En
Avants
0.530
Pirouette En
Dedans
0.444
Pirouette En
Dehors
0.542
(1.01)
Pelvic
Obliquity
(1.25)
Ankle
Rotation
(1.51)
Ankle
Rotation
(1.20)
Pelvic
Obliquity
(1.34)
Pelvic
Obliquity
(1.09)
Ankle
Rotation
(1.01)
Pelvic
Ant/Post Tilt
(1.16)
Hip Rotation
(1.04)
Hip Rotation
(1.07)
Pelvic
Ant/Post Tilt
(1.03)
Ankle
Rotation
(1.07)
Table 5.1 Overall z-scores and deviant kinematic variables for each step and associated movement category.
Table 5.1 presents the overall z-scores for each step, and the kinematic variables for
which a z-score of at least 1.0 was obtained. It can be seen that the steps in the relev
category obtained the highest overall z-scores (0.934 and 1.014), and were able to detect
- 268 -
deviations of at least one standard deviation (i.e. z-score IRU WKH PRVW NLQHPDWLF
variables. These steps detected deviations in kinematic variables that related to the
principles of placement (i.e. pelvic obliquity; pelvic anterior/posterior tilt) and turnout
(i.e. hip rotation; ankle rotation).
The plis ranked next in their ability to detect
deviations in kinematic variables (z-scores: 0.910 and 0.839), followed by the battements
(z-scores: 0.794 and 0.770. The plis were able to detect deviations in alignment (i.e.
thorax anterior/posterior tilt), while the battements picked up deviations in turnout (i.e.
hip rotation). The temps lev from the saut category could detect minor deviations in
placement (i.e. pelvic obliquity) and turnout (i.e. ankle rotation), and although the
pirouette en dehors, from the tour category, only obtained an overall z-score of 0.542 it
too could detect deviations in turnout (i.e. ankle rotation).
Based on the data in Table 5.1, the step from each movement category with the highest
overall z-score was chosen for inclusion in the 7-step pilot assessment tool. Similarly,
since highest z-scores were obtained for the relevs, plis, battements and sauts, one step
from each of these categories was chosen for inclusion in the 4-step tool. Since the
individual z-scores for placement (i.e. pelvic obliquity) and turnout (i.e. ankle
rotation) obtained by the temps lev were only just above 1.0 (1.09 and 1.01,
respectively), and because deviations in placement and turnout were also detected by
the relev and battements, it was decided that an additional pilot assessment tool,
containing only 3 steps would also be analysed.
included:
1. Pli in First Position
2. Battement Tendu Derriere
3. Rise in Second Position
4. Temps Lev
5. Glissades Over
6. Jet Ordinaire
7. Pirouette En Dehors
The 4-step assessment tool included:
- 269 -
Thus, the 7-step assessment tool

4. Temps Lev
And an additional 3-step assessment tool included:

The other 4-step assessment tool, containing steps with the lowest z-scores included:
1. Pirouette En Dedans
2. Grande Jet Elanc En Avants
3. Glissades Under
4. Sauts
Since an additional 3-step assessment tool using the 3 steps with the highest z-scores was
investigated, an additional 3-step assessment tool, containing steps with the lowest zscores, was also assessed. The steps included in this tool were:
1. Pirouette En Dedans
2. Grande Jet Elanc En Avants
3. Glissades Under
5.3.2 Comparisons of Dancer Rankings when Including Different Numbers of Steps in

the Assessment Tool
Table 5.2 shows the rank order of the 14 non-professional dancers according to their
overall z-score across different numbers and combinations of steps. Table 5.3 summaries
the results of the correlations (Spearmans r) and tests for significance (p-values) between
the rank orders obtained for the different number and combination of steps used for each
assessment tool.
- 270 -
Dancer rank order by overall z-score across steps

Dancer ID
14 steps
7 Steps
4 Steps
3 Steps
4 Other Steps
3 Other Steps
NP001
10
NP002
11
11
NP003
NP004
10
10
10
12
11
NP005
11
12
NP006
NP007
10
NP008
NP010
NP011
12
NP012
12
13
13
13
11
10
NP013
13
12
12
11
13
13
NP014
NP015
14
14
14
14
14
14
Table 5.2 Dancer rank order by overall z-score across different numbers and combinations of steps.
Rank 1 = best dancer (i.e. lowest z-score)..... Rank 14 = worst dancer (i.e. highest z-score).
Assessment
Tool A
Assessment
Tool B
Spearman's rho
(r)
p value
14 steps
7 steps
0.908**
0.000
14 steps
4 steps
0.868**
0.000
14 steps
3 steps
0.851**
0.000
14 steps
4 other steps
0.582*
0.029
14 steps
3 other steps
0.385
0.175
7 steps
4 steps
0.903**
0.000
7 steps
3 steps
0.732**
0.003
7 steps
4 other steps
0.503
0.067
7 steps
3 other steps
0.389
0.169
4 steps
3 steps
0.859**
0.000
4 steps
3 other steps
0.525
0.054
4 steps
4 other steps
0.644*
0.013
3 steps
3 other steps
0.415
0.140
3 steps
4 other steps
0.657*
0.011
4 other steps
3 other steps
0.851**
0.000
Table 5.3 Spearmans correlation coefficients (r) and p-values for
correlation of Assessment tool A with Assessment tool B.
**Correlation is significant at the 0.01 lev l.
*Correlation is significant at the 0.05 lev l.
- 271 -
Table 5.3 depicts highly significant correlations (p < 0.01) for comparison of dancer rank
orders from use of 14, 7, 4 and 3 steps in the assessment tool. However, correlation
coefficients were lower when comparing the other combinations of 3 or 4 steps (i.e. the 3
or 4 steps that did not independently detect deviations in any of the relevant kinematic
variables). For example, correlations were not significant at the 0.01 level for comparison
between use of 14, 7, 4, or 3 steps in the assessment tool, with either the 3 other steps, or
4 other steps. The data therefore indicate that it is the type of step included in the
assessment tool, rather than the number of steps, that influence the rank order of dancers.
As indicated in Table 5.2, dancer NP015 was the lowest ranked dancer when considering
all pilot assessment tools, and dancer NP006 ranked first or second when using the
majority of assessment tools. Given that the number or combination of steps did not
greatly affect the rank order for these dancers, it is interesting to examine if different
numbers and combinations of steps included in the tool would enable it to detect or
highlight differences in the z-scores obtained for individual kinematic variables. The next
section therefore presents results for the individual kinematic variables for which z-scores
of at least 1.0 were obtained, and displays these variables in relation to their associated
fundamental biomechanical principle.
5.3.3 Analysis of the Effect of the Number and Combination of Steps on Z-Scores for
Individual Kinematic Variables
Table 5.4 displays the overall z-scores achieved by each dancer when using each different
pilot version of the quantitative assessment tool (i.e. 14-step, 7-step, 4-step, and 3-step).
The table also displays the individual kinematic variables for which a dancer achieved a
z-score of at least 1.0, thereby indicating the joint or segment movements for which each
dancers data deviated by at least one standard deviation from the professional data.
These variables have been colour coded to give a visual representation of the fundamental
principle that they each relate to.
- 272 -
Dancer ID
Assessment
Tool

(z-score)
Alignment
Placement
Turnout
Extension
Overall
z-score
14 steps
0.679
7 steps
0.658
4 steps
0.841
3 steps
0.858
4 other steps
0.551
3 other steps
0.530
14 steps
0.762
7 steps
0.800
4 steps
0.852
3 steps
0.840
4 other steps
0.501
3 other steps
0.410
14 steps
0.697
7 steps
0.713
4 steps
0.811
3 steps
0.854
4 other steps
0.438
3 other steps
0.431
NP001
NP002
NP003
Thorax Ant/
Post Tilt
(1.27)
Thorax
Ant/Post Tilt
(1.22)
Thorax
Ant/Post Tilt
(1.80)
Thorax
Ant/Post Tilt
(1.93)
Knee Flexion/
Extension
(1.07)
Knee Flexion/
Extension
(1.18)
Ankle
Rotation
(1.57)
Ankle
Rotation
(1.71)
Ankle
Rotation
(2.10)
Ankle
Rotation
(2.32)
Hip
Rotation
(1.32)
Hip
Rotation
(1.51)
Hip
Rotation
(1.47)
Hip
Rotation
(1.52)
Thorax Side
Tilt
(1.04)
Pelvic
Obliquity
(1.01)
Pelvic
Obliquity
(1.10)
Pelvic
Ant/Post Tilt
(1.34)
Pelvic
Ant/Post Tilt
(1.47)
Pelvic
Ant/Post Tilt
(2.00)
Pelvic
Ant/Post Tilt
(2.26)
Table 5.4 Dancer overall z-scores per assessment tool, and kinematic variables with z-scores
- 273 -
Dancer ID
Assessment
Tool

(z-score)
Alignment
Placement
Turnout
Extension
Overall
z-score
14 steps
0.758
7 steps
0.765
4 steps
0.892
3 steps
0.909
4 other steps
0.605
3 other steps
0.540
14 steps
0.732
7 steps
0.755
4 steps
0.918
3 steps
0.999
4 other steps
0.542
3 other steps
0.529
14 steps
0.599
7 steps
0.589
4 steps
0.644
3 steps
0.680
4 other steps
0.492
3 other steps
0.497
NP004
NP005
NP006
Ankle
Rotation
(1.41)
Ankle
Rotation
(1.53)
Ankle
Rotation
(1.66)
Pelvic-Thorax
Rotation
(1.64)
Pelvic-Thorax
Rotation
(1.20)
Ankle
Rotation
(1.06)
Pelvic
Obliquity
(1.33)
Pelvic
Obliquity
(1.27)
Pelvic
Obliquity
(1.98)
Pelvic
Obliquity
(2.52)
Pelvic-Thorax
Rotation
(1.27)
Pelvic-Thorax
Rotation
(1.19)
Pelvic-Thorax
Rotation
(1.37)
Ankle
Rotation
(1.56)
Ankle
Rotation
(1.19)
Thorax Ant/
Post Tilt
(1.13)
Thorax
Ant/Post Tilt
(1.54)
Thorax
Ant/Post Tilt
(1.38)
Thorax
Ant/Post Tilt
(1.11)
Thorax Ant/
Post Tilt
(1.11)
Thorax Side
Tilt
(1.09)
Hip
Rotation
(1.03)
Hip
Rotation
(1.09)
Hip
Rotation
(1.28)
Pelvic
Obliquity
(1.04)
Pelvic
Obliquity
(1.10)
Pelvic
Obliquity
(1.54)
Pelvic
Obliquity
(1.96)
Table 5.4 (continued) Dancer overall z-scores per assessment tool, and kinematic variables with z-scores
1.0
- 274 -
Dancer ID
Assessment
Tool

(z-score)
Alignment
Placement
Turnout
Extension
Overall
z-score
14 steps
0.681
7 steps
0.766
4 steps
0.853
3 steps
0.839
4 other steps
0.492
3 other steps
0.486
14 steps
0.703
7 steps
0.715
4 steps
0.809
3 steps
0.866
4 other steps
0.523
3 other steps
0.462
14 steps
0.628
7 steps
0.668
4 steps
0.741
3 steps
0.786
4 other steps
0.505
3 other steps
0.536
NP007
NP008
NP010
Ankle Dorsi/
Plantar Flex
(1.16)
Ankle Dorsi/
Plantar Flex
(1.46)
Thorax
Ant/Post Tilt
(1.59)
Thorax
Ant/Post Tilt
(1.72)
Thorax
Ant/Post Tilt
(1.17)
Ankle Dorsi/
Plantar Flex
(1.40)
Ankle Dorsi/
Plantar Flex
(1.47)
Pelvic Ant/
Post Tilt
(1.03)
Pelvic
Obliquity
(1.28)
Pelvic
Obliquity
(1.27)
Pelvic
Obliquity
(1.86)
Pelvic
Obliquity
(2.15)
Pelvic
Ant/Post Tilt
(1.08)
1.0
- 275 -
Dancer ID
Assessment
Tool

(z-score)
Alignment
Placement
Turnout
Extension
Overall
z-score
14 steps
0.691
7 steps
0.751
4 steps
0.811
3 steps
0.735
4 other steps
0.546
3 other steps
0.559
14 steps
0.785
7 steps
0.881
4 steps
1.032
3 steps
1.153
4 other steps
0.594
3 other steps
0.539
14 steps
0.824
7 steps
0.849
4 steps
0.972
3 steps
0.967
4 other steps
0.702
3 other steps
0.681
NP011
NP012
NP013
Ankle
Rotation
(1.17)
Ankle
Rotation
(1.46)
Ankle
Rotation
(1.65)
Ankle
Rotation
(1.52)
Hip
Rotation
(1.04)
Hip
Rotation
(1.03)
Hip
Rotation
(1.20)
Hip
Rotation
(1.43)
Pelvic
Obliquity
(1.10)
Pelvic
Obliquity
(1.18)
Thorax
Ant/Post Tilt
(1.71)
Thorax
Ant/Post Tilt
(2.11)
Hip
Rotation
(1.01)
Thorax
Ant/Post Tilt
(1.15)
Pelvic
Obliquity
(1.57)
Pelvic
Obliquity
(1.71)
Hip
Rotation
(1.12)
Pelvic Ant/
Post Tilt
(1.33)
Pelvic Ant/
Post Tilt
(1.48)
Ankle
Rotation
(1.42)
Ankle
Rotation
(1.55)
Ankle
Rotation
(1.53)
Ankle
Rotation
(1.42)
Pelvic
Obliquity
(1.00)
Pelvic
Obliquity
(1.07)
Pelvic
Obliquity
(1.52)
Pelvic
Obliquity
(1.32)
Knee Flexion/
Extension
(1.08)
Hip
Rotation
(1.30)
Ankle
Rotation
(1.25)
Hip
Rotation
(1.30)
Hip
Rotation
(1.19)
Ankle
Rotation
(1.17)
Hip
Rotation
(1.03)
Knee
Flexion/
Extension
(1.14)
Ankle
Rotation
(1.40)
Ankle
Rotation
(1.43)
1.0
- 276 -
Dancer ID
Step
Combinations

(z-score)
Alignment
Placement
Turnout
Extension
Overall
z-score
14 steps
0.573
7 steps
0.598
4 steps
0.635
3 steps
0.681
4 other steps
0.535
3 other steps
0.496
14 steps
0.937
7 steps
0.974
4 steps
1.114
3 steps
1.185
4 other steps
0.751
3 other steps
0.709
NP014
NP015
Thorax Side
Tilt
(1.16)
Ankle
Rotation
(1.74)
Ankle
Rotation
(1.79)
Ankle
Rotation
(1.91)
Ankle
Rotation
(2.11)
Ankle
Rotation
(1.49)
Ankle
Rotation
(1.58)
Hip
Rotation
(1.26)
Hip
Rotation
(1.36)
Hip
Rotation
(1.57)
Hip
Rotation
(1.73)
Thorax Side
Tilt
(1.04)
Pelvic-Thorax
Rotation
1.02)
Thorax Side
Tilt
(1.20)
Thorax Side
Tilt (1.54)
Pelvic
Ant/Post Tilt
(1.18)
Pelvic
Ant/Post Tilt
(1.31)
Ankle Dorsi/
Plantar Flex
(1.16)
Ankle Dorsi/
Plantar Flex
(1.12)
1.0
Table 5.4 indicates that when using different versions of the assessment tool (i.e. different
numbers and combinations of steps), there are some notable differences in the individual
kinematic variables for which deviations of greater than one standard deviation were
detected. That is, the number and combination of steps did affect the ability of the
assessment tools to detect deviations in the kinematic variables that relate to the four
fundamental principles of classical ballet technique. For 13 of the 14 non-professional
dancers analysed, the most kinematic variables were detected as being incorrect (i.e. zscore ZKHQRQO\VWHSVZHUHLQFOXGHGLQWKHDVVHVVPHQWWRRO,WZDVWKHVWHSVi.e.
grand pli in first position; battement tendu derriere; rise in second position), that in the
- 277 -
inter-step analysis (Table 5.1) independently achieved the highest overall z-values (i.e.
greatest deviation) rather than the 3 steps (i.e. glissades under; grand jet elanc en
avants; pirouette en dedans) that achieved the lowest overall z-values, for which the most
deviations were detected for each dancer. Table 5.4 also indicates that the least number of
kinematic variables were detected as being incorrect (i.e. z-score , when these
other 3 steps (i.e. glissades under; grand jet elanc en avants; pirouette en dedans) were
used. This result indicates that it is the type of steps, rather than the number of steps,
included in the assessment tool that affect the tools ability to detect deviations in the
kinematic variables that relate to the fundamental principles of classical ballet technique.
5.3.4 Sample Assessment Tool Report

Since the 3-step assessment tool containing the grand pli in first position, battement
tendu derriere, and rise in second position was able to detect deviations in the
fundamental principles of classical ballet technique better than all other numbers and
combinations of steps, this tool was therefore used to create sample assessment reports
for each of the 14 non-professional dancers whose data were used during this analysis.
The following pages contain an example assessment report for dancer NP015, who was
ranked lowest for all assessment tools that were examined. As indicated in the report,
deviations in multiple kinematic variables were detected for dancer NP015. These
variables related to all four fundamental principles of classical ballet technique, and
included ankle rotation and hip rotation (turnout), thoracic side tilt (alignment), pelvic
anterior tilt (placement), and ankle plantarflexion (extension).
- 278 -
Dancer ID: NP015

Date: 6/10/2009
Classical Ballet Technique Quantitative Assessment Report
Dancer ID: NP015
Dancer DOB: 21/04/1991
Assessment Date: 6/10/2009
This report summarises results for three-dimensional (3D) joint movements for the thorax, pelvis,
hips, knees and ankles during performance of the following classical ballet steps:
x
x
x
Grand pli in first position

Battement tendu derriere
Rise in second position
Section 1 - General Results Summary

Principle
Kinematic Measurement
Variable
Thoracic Anterior/Posterior Tilt
Thoracic Side Tilt
Overall
z-score
0.79
-1.54*
Comments
x Satisfactory.
x Increased thoracic side tilt towards left side.
x Possibly due to functional or structural scoliosis,
and/or weak abdominal and spinal musculature.
x Work to strengthen abdominal and spinal
musculature.
Placement Pelvic Anterior/Posterior Tilt
1.31*
x Increased anterior pelvic tilt.
x Possibly due to functional or structural increased
lumber lordosis and/or weak abdominal
musculature.
x Work to strengthen abdominals and lengthen
the coccyx (tailbone) down towards the floor.
Pelvic Obliquity
0.60
x Satisfactory.
Pelvic-Thoracic Rotation
0.71
x Satisfactory.
Turnout
Hip External Rotation
1.73*
x Decreased external hip rotation.
x Possibly due to functional or structural
limitations within the hip joint.
x Work to actively engage the deep rotators of hip
Ankle External Rotation
2.11*
x Decreased external ankle rotation.
limitations within the ankle joint.
x Work to strengthen the musculature across the
ankle to achieve the desired foot and ankle
position and line.
Extension
Knee Extension
0.76
x Satisfactory.
Ankle Plantarflexion
1.12*
x Decreased ankle plantarflexion.
limitations within the ankle joint.
x Work to strengthen the ankle plantarflexors and
the intrinsic muscles of the foot.
Overall z-score calculated as average across all analysed steps.
*z-score RU]-score -1: Joint/segment movement is at least one standard deviation from mean value for
professional ballet dancers, and thus in need of attention or correction.
-1 < z-score < 1: Within satisfactory range.
Page 1
Alignment
- 279 -
Dancer ID: NP015

Date: 6/10/2009
Section 2 Graphical Data for Variables in Need of Attention
1. Ankle Rotation (Overall z-score = 2.11)
x Decreased external ankle rotation
Grand Pli in First Position

Ankle Rotation (Internal / External)
20
Degrees
10
0
0
10
20
30
40
50
60
70
80
90
100
-10
-20
-30
-40
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

20
Degrees
10
0
0
10
20
30
40
50
60
70
80
90
100
-10
-20
-30
-40
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

20
Degrees
10
0
0
10
20
30
40
50
60
70
80
90
100
-10
-20
-30
-40
- 280 -
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD
Page 2
Dancer ID: NP015

Date: 6/10/2009
2. Hip Rotation (Overall z-score = 1.73)
x Decreased external hip rotation

0
0
10
20
30
40
50
60
70
80
90
100
Degrees
-10
-20
-30
-40
-50
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

0
0
10
20
30
40
50
60
70
80
90
100
Degrees
-10
-20
-30
-40
-50
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

0
0
10
20
30
40
50
60
70
80
90
100
Degrees
-10
-20
-30
-40
-50
- 281 -
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD
Page 3
Dancer ID: NP015

Date: 6/10/2009
3. Thoracic Side Tilt (Overall z-score = -1.54)
x Increased thoracic side tilt towards left side
Grand Pli in First
Degrees
Thorax Side Tilt (Right / Left)

10
8
6
4
2
0
-2 0
-4
-6
-8
-10
10
20
30
40
50
60
70
80
90
100
NP015
PR mean
PR + 1SD
PR - 1SD
Degrees
Thorax Side Tilt (Working/Supporting Side)

10
8
6
4
2
0
-2 0
-4
-6
-8
-10
10
20
30
40
50
60
70
80
90
100
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD
Rise in First Position
Degrees
Thorax Side Tilt (Right / Left)

10
8
6
4
2
0
-2 0
-4
-6
-8
-10
10
20
30
40
50
60
- 282 -
70
80
90
100
NP015
PR mean
PR + 1SD
PR - 1SD
Page 4
Dancer ID: NP015

Date: 6/10/2009
4. Pelvic Anterior/Posterior Tilt (Overall z-score = 1.31)
x Increased anterior pelvic tilt

Pelvic Tilt (Anterior / Posterior)
40
35
Degrees
30
25
20
15
10
5
0
0
10
20
30
40
50
60
70
80
90
100
NP015
PR mean
PR + 1SD

40
35
Degrees
30
25
20
15
10
5
0
0
10
20
30
40
50
60
70
80
90
100
NP015-L
NP015-R
PR mean
PR + 1SD

40
35
Degrees
30
25
20
15
10
5
0
0
10
20
30
40
50
60
70
- 283 -
80
90
100
NP015
PR mean
PR + 1SD
PR - 1SD
Page 5
Dancer ID: NP015

Date: 6/10/2009
5. Ankle Plantarflexion (Overall z-score = 1.12)
x Decreased ankle plantarflexion

40
Degrees
20
0
0
10
20
30
40
50
60
70
80
90
100
-20
-40
-60
-80
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

40
Degrees
20
0
0
10
20
30
40
50
60
70
80
90
100
-20
-40
-60
-80
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD

40
Degrees
20
0
0
10
20
30
40
50
60
70
80
90
100
-20
-40
-60
-80
- 284 -
NP015-L
NP015-R
PR mean
PR + 1SD
PR - 1SD
Page 6
5.4 Discussion
In this chapter the minimum number and appropriate combination of ballet steps for
inclusion in a quantitative assessment tool for classical ballet technique were identified.
This process eliminated redundancy of data for inclusion in the tool, thereby making data
collection and processing more cost and time efficient. This analysis procedure initially
required investigation of the ability of particular movement categories or steps to detect
deviations in kinematic variables between professional and non-professional dancers.
5.4.1 Identification of Steps for Inclusion in the Assessment Tool

The plis, battements, and rises were the three movement categories that detected the
most kinematic deviations. Specifically, rises detected differences in the principles of
placement (i.e. pelvic anterior/posterior tilt and pelvic obliquity) and turnout (i.e. hip
rotation and ankle rotation), battements also picked up differences in turnout (i.e. hip
rotation), while the plis were sensitive to differences in alignment (i.e. thorax
anterior/posterior tilt). In Chapter 3, a similar analysis was conducted based on kinematic
data obtained from only the start and peak of each step, rather than from the whole step,
as in the current chapter. Comparison of data from each method of analysis reveals that
the same three movement categories were identified as being most sensitive to kinematic
differences, and in fact for the most part identified differences in the same kinematic
variables. Some notable differences were that when only assessing the starts and peaks,
the rises did not detect differences in placement as well as turnout, as was the case for
analysis of the whole step. However for the battements, analysis of only the start and
peaks revealed differences in the principle of extension (knee extension), in addition to
turnout. Both methods of analysis revealed that the plis were effective in picking up
differences in alignment (i.e. thorax anterior/posterior tilt) but no other kinematic
variables. The finding that the same three steps are most sensitive to kinematic deviations
whether only the start and peak or the whole step is analysed, adds support for their
inclusion in the quantitative assessment tool.
- 285 -
5.4.2 Affect of the Number and Type of Steps Included in the Assessment Tool
Before simply opting to include only three steps in the assessment tool, it was necessary
to investigate how reducing the number of steps would affect the rank order of nonprofessional dancers as well as the kinematic deviations detected for individual dancers.
Spearmans rank order correlation coefficients revealed significant correlations between
rank order of the 14 non-professional dancers when comparing use of 14 steps, 7 steps, 4
steps or just the 3 steps specified above. This result indicates that exclusion of 11 steps
from the assessment tool does not significantly affect the ranking of the non-professional
dancers against the gold standard practical data produced by the professionals.
Interestingly, despite no significant differences in rank order, the reduction in steps to
only three did influence the specific kinematic variables for which deviations were
detected for each dancer. An absolute z-score of greater than 1.0 was used as the
threshold for a kinematic variable being identified as deviating from the gold standard
and thus being incorrect. For all but one of the 14 non-professional dancers, the most
incorrect kinematic variables were detected when only 3 steps were included in the
assessment tool, thereby proving the 3-step tool to be the most sensitive to kinematic
deviations.
To examine the possibility that it was the number, rather than type of steps included in
the assessment tool, that influenced the number of incorrect variables detected,
additional analysis were conducted using combinations of 3 and 4 other steps. The
ranks orders of dancers obtained from an assessment tool based on these other steps did
not correlate well with the 14, 7, 4 or 3 steps used in the initial versions of the tool, and
interestingly, use of only these 3 or 4 other steps resulted in detection of the fewest
number of incorrect variables for all 14 dancers. This result ruled out the possibility
that it was the number of steps that was affecting how many incorrect variables were
identified. Therefore, a reduction in the number of steps included in the tool to only three
does not alter the rank order of dancers, and in fact increases the sensitivity of the
assessment tool at detecting deviations in the relevant variables, as long as the most
appropriate steps are selected.
- 286 -
5.4.3 Comparison With Other Tools Used For Assessment of Classical Ballet
Technique
Other examples of quantitative assessment tools for classical ballet technique were not
identified in the academic literature. Comparison of the newly developed tool with
existing quantitative tools is therefore not possible. However, some assessment tools
based on qualitative analysis were identified for classical ballet technique (Liederbach,
1997; Molnar and Esterson, 1997; Gibbs et al., 2006; Potter and Baas, 2011).
Interestingly, there were similarities between the steps included in these qualitative
assessment tools and those systematically selected for inclusion in the quantitative
assessment tool developed in the current study. Nine different ballet steps were included
across the four qualitative assessment tools identified, these are listed in Table 5.5. Of
these steps, three steps (port de bras; pass; developp) were not included in the analyses
conducted in any of the phases of the current project. Of those that were analysed in the
current project (pli; battement tendu; rise; saut; jet; pirouette), the pli, battement
tendu and rise were the steps most frequently included in the four qualitative assessment
tools for classical ballet technique, and correspond directly with the steps selected for the
quantitative assessment tool.
Step
Molnar and
Esterson
(1997)
Liederbach
(1997)
Gibbs et al.
(2006)
Potter and
Bass (2011)
Total "3
3"
Pli
3
3
3
Battement Tendu
3
3
Rise
3
3
Port de Bras
3
3
Pass
3
Developp
3
3
Saut
3
Jet
3
Pirouette
3
Table 5.5 Classical ballet steps in included qualitative screening tools.

3step included in assessment tool; step not included in assessment tool.
Although similarities were found between the steps included in the qualitative and
quantitative analysis methods, there are some differences in the specific elements each
step was used to assess. For example, the current study found that the pli was effective
- 287 -
in detecting deviations in alignment, specifically in thoracic anterior/posterior tilt,

however Gibbs et al. (2006) described lumbo/pelvic stability, mal-alignment of the knee
with respect to the feet, and the position of the ankle joint to be the crucial elements to
look for when qualitatively assessing the pli. Gibbs et al. (2006) also listed these same
elements as being important factors when observing the rise. With respect to the rise
these critical features are in closer agreement with some of the kinematic variables best
detected by the rise in the quantitative assessment tool (i.e. pelvic anterior/posterior tilt
and ankle rotation). The critical features described by the authors of the other qualitative
assessment tools were more ambiguous and could not be easily translated into specific
kinematic variables.
Molnar and Esterson (1997) reported that elements such as
alignment and stability should be assessed, and Potter and Baas (2011) state that it
is essential to look for attributes of faulty movement patterning. Despite some
uncertainties as to the specific reasons for inclusion of some of the steps in the qualitative
assessment tools, the finding that the three steps chosen for inclusion in the newly
developed quantitative tool are also commonly included in existing qualitative assessment
tools, provides encouragement for the content validity of the new quantitative tool.
5.4.4 Potential Use of the Quantitative Assessment Tool

The qualitative assessment tools described in the previous section typically form part of
larger screening tools used to assess the overall health, physical condition and potential
injury risk of dancers as they enter into a dance training program or commence
employment with a professional dance company. The baseline data obtained through
these screenings can be used as a reference to assess each dancers progression toward
optimal conditioning and development of optimal technique for their dance career
(Solomon, 1997). Data from multiple time points are thus typically compared during a
dancers affiliation with a particular company or educational institution. A similar model
of use could also be adopted for the quantitative tool developed in the current study. That
is, it could be used to systematically measure and record the progression of classical
ballet technique for individual dancers across education semesters or dance company
seasons. Such a tool could help assess the effectiveness of the implementation of specific
training programs, or in fact the influence of different dance teachers on the development
of correct classical ballet technique. Using data obtained from the non-professional
dancers who participated in this research project, Figure 5.3 provides an example of the
- 288 -
way in which data from a single cohort of dancers (e.g. first year entry into a dance
program or company) could be presented to provide an overall indication of the technical
competency of the group. The numbers displayed in each bar represent the z-scores for
each kinematic variable for each dancer. The data are colour coded with respect to the
kinematic variables and the fundamental principle with which each variable is associated.
A sum of the z-scores across all dancers in the cohort for each kinematic variable is listed
in the legend. For each individual dancer, higher z-scores and more variables displayed
on the graph indicate less competent technique, while the absence of any bars next to a
dancers ID indicate excellent technique for that dancer.
Z-Scores per Kinamatic Variable per Dancer in NP Class at Sept/Oct 2009
NP015
1.54
54
NP014
1.31
2.11
1.73
Thorax Ant/Post Tilt (8.25)

25)**
1.12
Thorax Side Tilt (3.79)
1.16
Pelvic Ant/Post Tilt (7.16)
NP013
Pelvic Obliquity (9.44)
NP012
2.11
NP011
1.43
NP010
1.48
1.71
1.3
1.17
(1.64)
Hip Rotation (7.64)
1.52
1.08
NP008
Ankle Rotation (8.30)
2.15
NP007
1.72
NP006
Knee Extension (1.18)
1.03
1.47
1.96
NP005
1.38
NP004
1.11
NP003
2.52
1.09
1.28
1.64
Alignment
1.56
Placement
2.26
NP002
1.1
NP001
Turnout
1.52
1.93
0
2.32
Extension
1.18
2
10
Figure 5.3 Example of assessment results for a cohort of dancers for a specific time point. * Z-score for
each kinematic variable per dancer. ** Cumulative z-score per each kinematic variable across all dancers.
Lower z-scores and fewer kinematic variables indicate better technique.
From the data presented in Figure 5.3, dancer NP013 has the best technique, with no
deviations or incorrect kinematic variables identified. Dancers NP015 and NP012
however, have the least competent technique with incorrect data being detected for 5 of
the 9 kinematic variables assessed. This figure also indicates that across the group the
- 289 -
variables with the greatest deviations from the gold standard were pelvic obliquity,
thoracic anterior/posterior tilt and ankle rotation. This style of presentation is therefore
useful for comparing classical ballet technique between dancers, but it can also
potentially be used to compare technique between groups. Given that a particular group
or class of dancers often have the same teacher for a whole semester or season, intergroup comparisons could enable investigation of the effects of individual teachers on
particular kinematic variables across the whole group, across multiple time points.
In addition to inter-dancer and inter-group comparisons, the data obtained from the
quantitative tool could also be used to assess the progression of technique of individual
dancers across multiple time points. Figure 5.4 provides an example of how quantitative
data could be presented to allow for efficient analysis of changes in an individual
dancers technical competency over time. Data from Dancer NP015 were used for the
Term 1 2009 time point, but it should be pointed out that all data moving forward from
this time point is hypothetical and has been added to the graph for the purposes of
demonstrating potential uses of the tool.
Z-Scores for Dancer NP015 Across Multiple Semesters

Thorax Side Tilt
Term 2
2011
1.2
Term 1
2011
1.3
Term 2
2010
1.3
Term 1
2010
1.5
1.2
1.5
1.8
Term 2
2009
1.5
1.3
1.5
Term 1
2009
1.54
1.2
1.5
Pelvic Obliquity
1.1
1.5
1.2
Hip Rotation
1.1
Ankle Rotation
1.6
1.2
Knee Extension
1.01
Alignment
1.1
Placement
1.31
2
1.1
2.11
1.73
4
1.12
8
Turnout
10
Extension
Figure 5.4 Example of assessment results by time point for individual dancer NP015. * Z-score
Z score for each
Zkinematic variable time point. Lower z-scores and fewer kinematic variables indicate better technique.
(Data for Term 2 2009 Term 2 2011 are fictional).
In this hypothetical example, the reduction in the z-scores and number of kinematic
variables for each successive time point indicates that this dancer has improved her
- 290 -
classical ballet technique across the 6 semesters displayed. Figure 5.4 indicates that on
her initial assessment (Term 1, 2009), Dancer NP015 had problems, or deviations from
the gold standard, in thorax side tilt, pelvic anterior/posterior tilt, pelvic obliquity, hip
rotation, ankle rotation, and ankle plantarflexion. The highest z-scores for hip rotation
and ankle rotation at this time point indicate that turnout is the fundamental principle for
which this dancer exhibits the most deviation. Hypothetically, by her final semester
(Term 2, 2011) this dancer made considerable improvements in ankle plantarflexion,
pelvic obliquity, and pelvic anterior/posterior tilt, as indicted by omission of these
variables from the graph at this final time point. Some improvements in the remaining
kinematic variables are also indicated by the decrease in z-scores compared to the
baseline data. This style of data presentation therefore has the potential to provide
encouraging information to individual dancers on their technical development, and to
provide a clear and concise summary of specific areas in which a dancer and his/her
teacher may need to focus their attention in order to improve classical ballet technique.
5.4.5
Limitations
As was the case for Chapter 4, the main limitation associated with this study and the
development of the quantitative assessment tool is the number of professional dancers
from which gold standard kinematic data were obtained. Data averaged from 12
professional dancers were used as the practical benchmark for indication of correct
classical ballet technique. Inclusion of data from more professional ballet dancers would
add to the validity of the tool, in that the variability amongst the reference data would
potentially be reduced, promoting increased confidence in the identification of correct
and incorrect technique. A specific goal therefore in developing this tool further would
be to increase the sample size on which the gold standard kinematic data are based.
Given that there were limitations in laboratory, equipment and participant availability
throughout the duration of this project, the quantitative tool that has been produced
provides a reasonable first attempt at development of such a tool, and provides some
interesting conclusions that can be used for continued development.
5.4.6 Conclusions
A minimum of three steps was determined to be appropriate for inclusion in the
quantitative assessment tool. Specifically, these steps were the grand pli in first position,
- 291 -
the battement tendu derriere, and the rise in second position. These steps were consistent
with steps that have been included in previously developed qualitative assessment tools
for classical ballet technique. The combination of these three steps enabled the most
effective detection of deviations in the kinematic variables that relate to the fundamental
principles of classical ballet technique. The rise was most effective in detecting
differences in the principles of placement (i.e. pelvic anterior/posterior tilt and pelvic
obliquity) and turnout (i.e. hip rotation and ankle rotation), battements also picked up
differences in turnout (i.e. hip rotation), while the plis were sensitive to differences in
alignment (i.e. thorax anterior/posterior tilt).
A sample quantitative assessment report was presented. Such a report provides individual
dancers with specific data (i.e. overall z-scores) for all nine kinematic variables assessed
in the tool. Summaries of the magnitudes and directions of deviations for incorrect
variables are provided in graphical format, with stick figures included to assist with data
interpretation. Written summaries of elements of incorrect technique, and suggestions
for their cause and management are also provided. This style of data presentation,
together with graphical summaries of changes in technical competency over multiple time
points (Figure 5.4) provides individual dancers and their teachers with concise
information on their abilities and development in classical ballet technique. Inter-dancer
and inter-group comparisons are also facilitated by the quantitative assessment tool
(Figure 5.3), thereby allowing individual dancer assessment and ranking, as well as
comparison of technical competency between groups or classes.
The development of this quantitative tool has come as the culmination of multiple phases
of analysis presented in this research thesis. Four fundamental theoretical principles of
classical ballet technique were identified through review of relevant dance literature. The
relevance of these theoretical principles with respect to current highly skilled practical
execution of classical ballet technique was then investigated. The accuracy with which
classical ballet teachers are able to qualitatively identify correct and incorrect
technique was then studied. The results of this analysis indicated that development of an
objective quantitative assessment tool for classical ballet technique was in fact necessary.
The implications on classical ballet teaching practice of the findings from each phase of
this research project will be discussed in the next and final chapter.
- 292 -
Chapter 6 - Conclusions and Implications for Teaching Practice
CHAPTER 6
CONCLUSIONS AND IMPLICATIONS

FOR TEACHING PRACTICE
- 293 -
CHAPTER 6
CONCLUSIONS AND IMPLICATIONS
FOR TEACHING PRACTICE
6.1 Specific Conclusions and Implications

Following a comprehensive review of the literature, results of linked projects have been
presented in this thesis. Each project is themed around classical ballet technique, with the
aim of contributing new knowledge to the biomechanics of classical ballet technique and
to dance teaching practice. This thesis has presented the first biomechanical analysis of
classical ballet technique with respect to is fundamental principles assessed across
multiple steps. The thesis has also presented original work investigating the accuracy of
qualitative analysis skills in assessing classical ballet technique.
Development of a
unique protocol for quantitative screening of classical ballet technique was also
discussed. This section will summarise the conclusions for each project with respect to
their implications for ballet teaching practice.
6.1.1 Comparison of Theoretical and Practical Execution of Classical Ballet Technique

Classical ballet teachers typically instruct their students by providing verbal cues to
describe or correct movements. Ideally, the information given should be accurate with
respect to anatomical and biomechanical knowledge and principles, as well as aligning
with classical ballet teaching objectives. These cues should have a biomechanical basis
that is logical, and include instructions or descriptions that are anatomically and
physically possible. Given that the fundamental principles of classical ballet technique are
deeply rooted in historical tradition, it is not surprising that much about classical ballet
teaching practice has also been passed down through many generations in a non-scientific
form. Although the longevity and frequency of use of certain instructional theories and
cues may imply they have validity as useful teaching aids, the question posed was
whether or not they would in fact stand up to scientific scrutiny. The aim of the first
project was to examine quantitatively the long-held beliefs about correct classical ballet
- 294 -
technique through comparison of theoretical information with kinematic data measured

on professional and non-professional dancers.
In relation to the four fundamental principles of classical ballet technique identified in the
dance literature, it is evident that for the principles of extension and turnout, the
practical data are in close agreement with the theoretical concepts. In contrast however,
some discrepancies were observed for the principles of alignment and placement. The
findings for each fundamental principle and associated implications for ballet teaching
practice will be outlined in the following sections.
6.1.1.1 Alignment
It was found that ballet dancers typically hold their thorax in a position of posterior
thoracic tilt, rather than the vertical position typically described in the literature. They
achieve the desired grand or regal appearance by taking their thorax beyond the
posture typically exhibited by most people during everyday life. In considering the
ramifications of this finding on current teaching practice, care should be taken in
translating these biomechanical results into useful verbal cues or instructions. For
example, in helping students achieve the aesthetic of grand manners it may not in fact
be advisable to instruct them to tilt their thorax posteriorly (backward), as such a cue
could invite increased lumbar extension and posterior shift of the centre of gravity (i.e.
body weight). A more useful cue that still aligns with the practical data could be to
slightly lift the sternum (breast bone) up, as this may have the effect of subtly altering
the orientation of the thorax, thereby achieving the desired alignment and aesthetic
without introducing other functional problems.
Also in relation to alignment was the finding that when functionally necessary
professional ballet dancers bend the rules of the theoretical principle and utilise an
increased range of anterior/posterior thoracic movement. Specifically, this discrepancy
between theory and practice was observed for the jumps (i.e. sauts and temps levs).
From a biomechanical perspective, it is a natural and instinctive movement response to
move the trunk forward in preparing for a jump, as this facilitates increased hip extensor
power during push-off. The long-held belief that in performing jumps correctly the
dancer must mask the effort by keeping the body erect (Royal Academy of Dance,
- 295 -
1997, p79), has informed much of the teaching practice to date. However the instruction
to aim for maximum elevation off the floor while simultaneously using limited trunk
flexion is functionally contradictory and therefore extremely challenging. If dance
teachers had increased awareness of how some biomechanical principles may introduce
certain challenges to achieving aesthetic aims then they could potentially provide better
assistance to their students in achieving a satisfactory compromise between functional
and aesthetic requirements.
6.1.1.2 Placement
In terms of standard clinical and biomechanical conventions, professional and nonprofessional dancers hold their pelvis in a position of at least 10 anterior tilt during
execution of most ballet steps. This pelvic position is not dissimilar to that used during
normal human gait. Thus, the centred or neutral pelvic placement often referred to in
the theoretical dance literature is not demonstrated practically by ballet dancers, and is
therefore not an accurate description in the biomechanical sense. Since, in executing
classical ballet technique correct placement is not achieved by altering the pelvis from
its usual position for daily activities, cues that place considerable emphasis on pelvic
anterior/posterior tilt may in fact be not always necessary. Instructions should be limited
to cues that focus on optimal usage of the musculature around the pelvis, rather than
pelvic orientation. In instances where individual students do have excessive anterior
pelvic tilt or lumbar lordosis attention should be drawn to pelvic sagittal placement.
However, if the dancers pelvic tilt is within a normal range then over emphasising
sagittal plane placement may result in excess tucking of the pelvis (i.e. movement
towards posterior tilt) beyond a functionally efficient position.
The finding that, on average, from start to peak of the steps professional dancers
demonstrate a smaller range of pelvic sagittal plane motion than non-professionals,
supports the theoretical idea of minimal displacement of the pelvis, and indicates that
more experienced dancers are better able to control or isolate their sagittal pelvic motion
in relation to the motion of their lower limb. Instruction to stabilise the pelvis and
minimise its anterior/posterior range of motion does therefore appear to be appropriate.
Specific training of the correct use of the abdominals, gluteals, and hamstrings would
help achieve this aim of minimal pelvic displacement.
- 296 -
In contrast to the finding for sagittal plane pelvic motion, the coronal and transverse plane
motion demonstrated by professional dancers is not consistent with the theoretical
literature. Professional and non-professional dancers exhibit raised pelvic obliquity on the
working side when performing a 45 leg elevation to the front, and interestingly, despite
the theoretical ideal that the hips and shoulders must always face the same plane
(Lawson, 1979, p34), the professionals display greater pelvic-thoracic rotation during
execution of this step than the non-professionals. More pelvic-thoracic rotation than
would be theoretically expected was also observed for the battement tendu derriere. For
this movement, close to 10 of transverse pelvic rotation was exhibited by both groups of
dancers. This finding indicates therefore that even for single leg extensions where the
working leg remains in contact with the floor (i.e. battement tendu) there is concurrent
motion of the pelvis and thigh during even the most basic battement movement. Ballet
dancers are commonly taught to "keep the pelvis square" (i.e. the pelvis should not move
with lower extremity movements) and this cue is often strongly emphasised when
performing leg movements behind the body. The current finding of 10 transverse pelvicthoracic rotation for highly skilled professional ballet dancers even for an a terre (i.e.
floor bound) posterior leg movement, adds scientific rigour to the suggestion that the
pelvis does in fact open during posteriorly directed leg movements in classical ballet.
Given that the arabesque (i.e. extension of one leg behind the body) is a signature pose or
position of the classical ballet repertoire, these findings have significant implications for
classical ballet instruction, in that the "square pelvis" cue has been shown to be inaccurate
with respect to transverse plane motion and is therefore not a useful instruction.
6.1.1.3 Turnout
Professional ballet dancers perform classical ballet steps with close to 5 more external
transverse hip rotation than non-professional dancers. This finding is in agreement with
the theoretical ideal of maximal turnout, in that highly skilled dancers are able to
demonstrate this principle to a greater extent than less skilled dancers. However, the
actual values of external hip rotation demonstrated by professional dancers in this study
(i.e. highest mean value of 40) are below the ideal values of 60-90 presented in the
literature. Thus, it appears that the theoretical ideal may be beyond the limitations of
many highly skilled professional ballet dancers. In relation to external ankle rotation,
again, the professional dancers demonstrate increased external rotation compared to non- 297 -
professionals, indicating that transverse rotation at the ankle joint does in fact contribute
to the overall turnout of the lower limb.
The conclusions for the principle of turnout have implications with respect to dancer
safety in classical ballet instruction. It appears that the theoretical data for hip external
rotation in classical ballet are based primarily on passive static measurement methods
(Champion and Chatfield, 2008). While these data may in fact be accurate for this
measurement condition, their relevance and use as a benchmark for the level of active
external hip rotation that can be achieved during dynamic dance movements is
questionable. The theoretical values typically presented in the dance literature may in fact
give ballet teachers misguided expectations as to what is an appropriate degree of active
external hip rotation for dancers to achieve. In striving to create the desired aesthetic,
dancers often force their hip rotation beyond the limits of their personal safety, sometimes
resulting in injury (Negus et al., 2005). While turnout is undeniably a crucial
requirement for execution of correct classical ballet technique, the 20 difference between
active practical and passive theoretical hip rotation revealed in the current study, indicate
that dance teachers should be discouraged from habitually repeating cues relating to hip
turnout, without careful consideration of the physical limitations of the individuals they
are addressing.
The heel forward or slightly winged position of the foot and ankle, particularly for a
fully extended working leg, is encouraged in classical ballet instruction as it creates an
appealing aesthetic. The finding that the degree of ankle external rotation that contributes
to this position is influenced by concurrent plantarflexion or dorsiflexion has implications
for classical ballet instruction. Providing dance teachers with more detailed information
about the mechanics of the ankle joint complex and the interplay of simultaneous
movement across multiple planes, may assist teachers in understanding why it is that for
some dancers the heel forward position with simultaneous plantarflexion is particularly
difficult to achieve. Increased information and understanding in this area could therefore
help teachers deliver appropriate exercises to encourage safe and aesthetically pleasing
ankle rotation.
- 298 -
6.1.1.4 Extension
Professional and non-professional dancers demonstrate knee hyperextension when
required to perform maximum knee extension. However, professional dancers exhibit a
significantly greater degree of knee hyperextension than non-professionals. The position
achieved by the professionals (up to 13 hyperextension) is extreme in relation to the
knee positions typically used during everyday life, and is in agreement with the
theoretical principle of maximum extension. While this extreme knee hyperextension
may reflect the desired theoretical aesthetic described in the literature, care should be
taken with respect to how this position is dealt with from a practical teaching perspective.
In the most severe cases, constant knee hyperextension can lead to damage to the anterior
cruciate ligament (ACL), due to strain of the postero-lateral bundle of the ligament
(Fornalski et al., 2008; Vogel, 2007). While some knee hyperextension is considered
desirable in achieving the aesthetics of classical ballet, there are differing opinions, with
respect to knee joint safety, about the degree of knee hyperextension that is considered
appropriate. Some instructors and clinical therapists believe that any degree of weightbearing knee hyperextension places undue stress across the knee joint and is therefore
potentially injurious, while others feel some controlled knee hyperextension is acceptable
(Grieg, 1994). The knee extension data obtained for the steps analysed in the current
study indicate that the highest measures for knee hyperextension generally were obtained
for both groups when the legs were in a weight-bearing second position (i.e. feet
separated), and specifically, at the start of the rise in second position and the grand pli in
second position. This finding indicates that the body position adopted can determine the
resulting degree of knee hyperextension.
From a biomechanical perspective, the increased knee hyperextension occurring in

weight-bearing second position occurs as result of the separation of the feet. That is, an
increased knee extensor torque/moment is placed across the knee joint due to the
increased horizontal displacement of the bodys line of gravity from the knee
flexion/extension axis. If not appropriately counteracted with the necessary knee flexor
muscle activation (i.e. hamstrings and gracilis), then increased uncontrolled knee
extension results, with increased stress being placed on knee joint ligaments. To prevent
knee joint injury, it is crucial therefore that ballet teachers have a thorough understanding
of the implications of the positions of the legs on the stresses invoked at the knee. Given
- 299 -
that knee hyperextension in classical ballet can be visually appealing, dancers with
natural knee hyperextension are typically selected into vocational training programs. In
training such dancers it is important therefore that teachers develop the skills and
knowledge to effectively instruct their students in the correct use of the specific
musculature (i.e. hamstrings and gracilis) that can protect the knee from potentially
unsafe weight-bearing hyperextension.
Professional ballet dancers also demonstrate increased ankle plantarflexion compared to

non-professionals, a finding that is also consistent with the theoretical concept that
extreme elongation (extension) of the lower limb is a characteristic of elite classical
ballet technique. As is likely the case for knee hyperextension, the degree of ankle
plantarflexion and the ankle shape achieved during maximal ankle extension is largely
determined by genetics. Therefore, although some increased plantarflexion can be
achieved to an extent during the earlier years of ballet training, there are varying degrees
of structural limitation at the ankle joint for different individuals. Such limitations must
be duly acknowledged and respected by dance teachers in the provision of safe dance
training, and less pointed feet should not be forced beyond the limits of safety. At the
other extreme, the highly arched or flexible ankle joint often seen in professional
dancers, and suggested by the current data, introduces significant safety issues with
regard to dancing en pointe. Although aesthetically pleasing, this type of foot can initially
be very unstable en pointe. It is very important therefore that teachers are fully aware of
the safety issues associated with this type of ankle joint, and equip their students with the
necessary knowledge and skills to gain adequate strength and control across their feet and
ankles.
In summarising the conclusions and associated implications of the first project, the key
feature that comes forth is the importance for classical ballet teachers to develop some
understanding of the basic biomechanical concepts that relate to the safety and efficacy of
dance-related movement. Knowledge of concepts such as joint forces, torques/moments
and powers, and how these are influenced by relative positions of body segments, joint
axes, and the bodys centre of gravity would certainly enhance a teachers ability to assist
their students acquire safe and efficient technique. Although there has been an increased
awareness of safe dance practice over recent decades (e.g. Ausdance Safe Dance Report 1
- 300 -
(Geeves, 1990); Ausdance Safe Dance Report 2 (Geeves, 1997); Ausdance Safe Dance
Report
(Crookshank,
1999);
The
Dancer
Wellness
Project
2002-2012
(http://www.dancerwellnessproject.com), similar focus has not been given to instruction

of biomechanically efficient dance practice. To date it appears that basic biomechanical
knowledge is not deemed by the dance community to be an integral component of skilful
dance instruction. However, it is suggested by the author that such knowledge,
communicated to dancers in an appropriate and accessible style, could not only contribute
to injury prevention, but also to highly efficient motor skill acquisition.
In addition, the dissemination of the findings of the literature review on classical ballet
technique to classical ballet teachers is also suggested by the author to have important
implications for classical ballet instruction. It would be beneficial to highlight that the
crucial elements of correct classical ballet can in fact be simplified into a few
fundamental principles. Thinking about the technique and describing it in terms of these
few key elements may reduce complexity and confusion in the teaching cues typically
employed, thereby promoting increased student understanding of the instructions
provided and of the physical goals they are aiming to achieve.
6.1.2 Qualitative Analysis of Classical Ballet Technique

Movement instructors provide their students with feedback on their performance by
informing them how their movement response compares to the ideal template or
prototype for their particular discipline. The efficacy of this feedback depends largely on
the instructors ability to identify the aspects of the response which are preventing the
learner from attaining the skill objective (Armstrong and Hoffman, 1979, p318). It
follows then that instructors who lack proficiency in identifying correct or incorrect
technique are unable to access critical information on which to base constructive
feedback. The value of their feedback, and consequent capacity for skill acquisition by
their students, would therefore be greatly diminished. This project aimed to assess the
accuracy with which ballet teachers are able to identify correct and incorrect classical
ballet technique, and the effect of teaching experience on this capacity.
(Armstrong and Hoffman, 1979)
Comparison of the ballet teachers qualitative responses with quantitative kinematic data
obtained from 3D motion analysis, revealed that accuracy of qualitative analysis of
- 301 -
classical ballet technique improved significantly between 1 and 9 years experience, but
plateaued beyond 10 years experience. Improvement in accuracy from 1 to 9 years
experience was due to increasing sensitivity to detection of incorrect ballet technique. It
has been argued (Osborne and Gordon, 1972) that it is far more important that incorrect
movements, as opposed to correct movements, be identified accurately so that faulty
movements can be remedied, thereby facilitating efficient progression towards successful
acquisition of a skill. The current study therefore provides encouraging data to indicate
that with increasing experience ballet teachers become more discriminating in their
judgement of what constitutes incorrect technique, and therefore do in fact develop
increasing skill in an element of qualitative analysis that is highly important for effective
teaching practice.
A noticeable trend was for the more experienced teachers to rate correct technique as
incorrect more so than less experienced teachers. That is, with increasing experience
teachers demonstrated decreased specificity for identification of correct technique.
Although identification of incorrect technique appears to be given more value in
teaching practice (Osborne and Gordon, 1972), it is also important to recognise the
importance of accurate identification of correct technique. Failure by a movement
instructor to recognise that a particular technical element is actually correct and to instead
rate it as incorrect will place unnecessary attention on a body region or movement quality
that does not need to be corrected. The adverse outcome of such misdiagnosis is that
perfectly adequate movement patterns may be unnecessarily altered and/or the students
focus and efforts distracted from other elements that may actually be incorrect and
therefore in need of attention. Moreover, from a psychological perspective, the
recognition and positive reinforcement of correct technique may provide encouragement,
increased motivation and confidence for students. Rist (2001) states that a useful
feedback tool is the specific praise method, in which the dancer is praised for effort
and told specifically what was good about it. Praise given in this way actually reinforces
the change in technique, thereby anchoring the achievement (Rist, 2001, p431). Results of
the current study however, suggest that perhaps more experienced teachers are too
picky or are reluctant to acknowledge correct technique. Failure by experienced
teachers to recognise correct technique has negative implications for teaching practice in
that valuable positive feedback may be constantly withheld from students.
- 302 -
(Rist, 2001)
Despite increasing accuracy of qualitative analysis with increasing teaching experience, it
should be highlighted that even teachers with at least 10 years experience only achieved a
mean overall percentage accuracy score of 65%, a result which is only 15% better than
that produced by chance alone. It therefore took nearly 10 years of experience for ballet
teachers to acquire only a relatively low level of observational accuracy. Even more
concerning is the finding that beyond 10 years of experience the accuracy of qualitative
analysis did not significantly improve. Ericsson and colleagues (Ericsson and Charness,
1994; Ericsson and Lehmann, 1996) argued that expertise is not attained as an automatic
consequence of experience, but rather, through structured and effortful adaptation
affected by training. They therefore suggested that specific training methods should be
developed that facilitate the development of the critical domain-specific skills that lead to
expertise. In addition, it was stated (Ericsson and Charness, 1994) that in order for
improvement to occur, regular training activities should offer accurate, preferably
immediate, feedback or opportunities for corrected repetitions. In line with this argument,
the results of the current study support the suggestion for implementation of increased
training in qualitative analysis of highly skilled dance movement. The fact that there was
no significant effect of participation in a teacher training program on the accuracy of
qualitative analysis, suggests that to date training in qualitative analysis is not a major
component of the curricula in existing teacher training programs. If implemented, such
training should consist of goal directed activities with specific feedback, as this will lead
to improved acquisition of qualitative analysis skill compared to non-planned non goal
directed activities.
The fact that employment as a professional classical ballet dancer did not significantly
affect the accuracy of qualitative analysis of classical ballet technique also has substantial
implications for teaching practice.
It appears to be an accepted belief that a highly
accomplished dancer will automatically have the skills required to be a highly effective
teacher. However, the results of the current study suggest that, with respect to
identification of correct and incorrect technique, this is not in fact the case. Attaining
an expert level of practical skill in classical ballet is not necessarily accompanied by a
corresponding level of skill in qualitative analysis relative to teachers who did not
perform professionally. Therefore dancers transitioning from the position of elite dancer
- 303 -
to dance teacher, may not immediately possess all of the skills required for expert
movement instruction. It should be recognised that sufficient time, training and
experience is required to develop these important skills, and retiring professional dancers
should therefore be encouraged to complete teacher training programs incorporating goaldirected tuition in qualitative analysis.
The implications of the results of this study on classical ballet teaching practice are that
all aspiring ballet teachers, whether coming from a professional performance background
or not, would benefit from participation in training programs designed specifically to
improve qualitative analysis of ballet technique. Teacher training programs appear to be
currently lacking in such curricula, and the dance teaching profession will benefit greatly
from attention being turned towards this area of teaching practice. The implementation of
such programs will increase the rate of improvement in qualitative analysis, and on-going
education in this area will ensure that accuracy of observation does not plateau beyond a
certain level of experience. Although accurate identification of incorrect technique is
critical, the importance for the student of identification and reinforcement of correct
technique should also be duly acknowledged as an essential component of effective
feedback.
6.1.3 Development of a Quantitative Assessment Tool for Classical Ballet Technique

In response to the relatively low accuracy of qualitative analysis demonstrated by
professional ballet teachers, a quantitative assessment tool based on 3D motion analysis
data was developed for the evaluation of classical ballet technique. A specific aim in
developing this tool was to determine the minimum number of ballet steps required to
enable effective detection of deviations in the kinematic variables relating to the four
fundamental principles of classical ballet technique. The need for determining the
minimum number of steps was driven by recognition of the potential financial and time
costs of 3D motion analysis. It was determined that a minimum of three steps were
required, and in fact the use of only three appropriate steps produced an assessment tool
that was more sensitive to deviations in the relevant kinematic variables than a tool
containing 14 steps. The three steps that were identified as being most effective in
detecting deviations between professional and non-professional dancers, were the grand
pli in first position, the battement tendu derriere, and the rise in second position.
- 304 -
Given the inaccuracies in qualitative analysis of classical ballet technique identified in the
previous project, development of a quantitative assessment tool designed specifically for
classical ballet technique is a potentially valuable contribution to the dance teaching
profession. The tool provides a mechanism to obtain systematic, objective and concise
information on the technical competencies of ballet dancers, as well as the teaching
abilities of dance instructors. Inter-dancer comparisons could provide a means for ranking
and assessing dancers within a given group, while inter-group comparisons would enable
evaluation of the effects of different teachers or teaching styles on different groups, or the
same groups over multiple time points. For example, specific training programs or
techniques could be implemented for defined periods of time and the quantitative
assessment tool could provide an objective and accurate means to assess their outcome or
efficacy. In addition, the tool could also enable systematic evaluation of the progress, or
lack thereof, of individual dancers over time.
Other potential benefits of the quantitative assessment tool to classical ballet teaching
practice include its use in the evaluation of the accuracy of qualitative analysis. Similar to
the process conducted in the second project, qualitative responses provided by teachers
could be compared to quantitative data obtained from the assessment tool. Teachers
participating in this process would receive feedback on the accuracy of their observations,
and specific areas for which their accuracy is poor, and thus in need of attention. Given
the importance of structured learning in skill acquisition (Ericsson et al., 1993), this tool
could therefore be a valuable source of feedback and as such a very useful resource for
teacher training. The results of subsequent re-evaluation of accuracy of qualitative
analysis after implementation of training programs designed to improve observation
skills, could also have important implications on classical ballet teaching. Such a process
could provide valuable information for the development and evaluation of dance teacher
training programs.
In practice, if each dancer performed three steps, a cohort of 20 dancers could be assessed
in a motion analysis laboratory in one to two days. The subsequent data analysis and
reporting would be relatively rapid, an additional two days, if a modern motion analysis
system and standard automated protocol, as suggested by the author, were used.
- 305 -
It is acknowledged that the equipment and technical expertise required to perform 3D

quantitative motion analysis is currently not readily available to most dance institutions
and instructors. However, the information that has been obtained through development of
this quantitative assessment tool is also applicable to regular qualitative analysis. This
study has identified the kinematic variables for which deviations typically occur in the
pli, battement, and rise categories of classical ballet. If these same movements were to
be included in the qualitative assessment tools regularly used in dance teaching practice,
there would be systematically obtained quantitative data to indicate the specific kinematic
parameters that teachers should be looking for when observing these steps.
In summary, the quantitative assessment tool developed as part of this research project
has many implications for classical ballet instruction. These implications relate to
enhancing and evaluating dancer progress and technical development, evaluating and
improving teaching practice, and developing curricula for teacher training programs.
- 306 -
6.2 General Conclusions

Four fundamental theoretical principles of classical ballet technique were identified
through review of historical and current dance literature, these are:

4) Extension -maximum elongation of the lower limbs
Through quantitative biomechanical analysis it was concluded that professional dancers

demonstrate kinematic variables relating to turnout and extension in agreement with
these theoretical principles, however some deviations between practice and theory were
observed in the areas of alignment and placement. The deviations observed were
considered to be due primarily to functional necessity. That is, although classical ballet
technique is based on distinct aesthetic ideals, it appears that in some instances it is
necessary to compromise on these aesthetic ideals in order to achieve the functional
biomechanical requirements of particular movements or steps. It is suggested therefore
that increased understanding by ballet teachers of basic biomechanical and anatomical
principles and how they influence human movement function will greatly enhance
teaching practice. Increased awareness of the ways in which an appropriate balance
between the aesthetic ideals and functional requirements of the technique can be achieved
will facilitate efficient skill acquisition and safe practice in classical ballet instruction.
Additionally, attainment of more detailed biomechanical understanding by dance
educators, would lead to a shift in the manner in which the principles of classical ballet
technique are typically presented and discussed.
The accuracy with which classical ballet teachers are able to identify correct and
incorrect ballet technique in relation to gold standard practical kinematic data was
also investigated.
A positive relationship between years of professional teaching
experience and accuracy of qualitative analysis was demonstrated from 1 to 9 years of

experience, however this plateaued after 10 years experience. The improvement in
- 307 -
accuracy with increasing experience was due to increased sensitivity to the detection of
incorrect technique. Despite this improvement, even the most experienced teachers,
some with as much as 20 and 30 years experience, only achieved moderate accuracy.
Perhaps it is due to a deficiency in understanding of basic biomechanical concepts that
ballet teachers only demonstrate moderate accuracy of qualitative analysis.
It has been stated (Wilson, 2009, p11) that dance teachers train students to develop both
artistry and skill, relying on their innate understanding of the body moving through
space. However, in considering the conclusions reached from the projects, it appears that
teachers innate understanding of movement may in fact not be enough to facilitate
expert teaching practice. Moreover, it has been argued that expertise is not necessarily
attained as an automatic consequence of experience, but, rather, through structured and
effortful adaptation affected by training (Ericsson and Charness, 1994). It is proposed
therefore, that dance teaching practice would benefit greatly from the development of
training programs containing carefully constructed curricula that includes basic
biomechanics and principles of qualitative analysis. It is suggested that well planned,
goal-directed training in these areas would equip teachers with more information with
which to conduct accurate and efficient qualitative analysis, which would in turn promote
effective feedback and successful skill acquisition. (Wilson, 2009)
A quantitative tool was developed to provide an objective and systematic means for
analysis of classical ballet technique. Using kinematic data from professional dancers as a
reference for correct technique, a minimum of three steps (i.e. pli in first position;
battements tendu derriere; rise in second position) were identified as being adequate for
the detection of deviations in kinematic variables relating to alignment, placement
turnout, and extension. This tool could potentially be used to evaluate the progress and
technical development of individual dancers, to evaluating teaching practice, and to assist
with developing curricula for teacher training programs.
- 308 -
6.3 Future Directions

In completing each project, a number of areas for future research were identified. Firstly,
as already mentioned in previous chapters, it would be beneficial to increase the number
of professional ballet dancers on whom the benchmark or reference data for correct
technique are based. This process would add further evidence to support the validity of
the techniques used to assess accuracy of qualitative analysis, and also to the validity of
the data obtained from the quantitative assessment tool. In addition, it would be
interesting to expand the geographical location from which professional dancers are
recruited. Ballet dancers from different countries and regions of the world are known
within the international dance community to have distinct movement characteristics or
attributes developed within specific training systems. Compilation and comparison of
kinematic data from professional ballet dancers trained within different regional systems
or syllabi, would enable identification of the universal kinematic features as well as
highlighting regional differences.
Having determined that the implementation of training programs in basic biomechanics

and qualitative analysis would be beneficial for ballet teachers, it would be interesting to
assess the efficacy of such programs once established and delivered. This would involve
using techniques similar to those described in Chapter 4 to obtain baseline data for the
accuracy of qualitative analysis for a sample of teachers, and then performing repeat
assessments at one or more subsequent time points. Comparison of data from a control
group who did not take part in the training program could highlight any specific effects of
the program.
Examination of the visual search strategies used by ballet teachers when observing ballet
technique could also provide some useful data. Results from Chapter 4 of this thesis
showed that when asked to select a body region in need of correction, greater consistency
in responses was obtained amongst more experienced ballet teachers. Specifically, it was
the pelvis that was most frequently selected by the more experienced teachers. Visual
search behaviour is typically examined using an eye movement registration system. These
systems, which can be floor- or head-mounted, record eye movements as well as the
- 309 -
interspersed visual fixations as they perform the task. The duration of each fixation is
presumed to represent the amount of cognitive processing, whereas the point of gaze is
assumed to indicate areas of interest (Williams, 2002, p169). Investigation of visual
search strategies would therefore reveal which body regions experienced and nonexperienced teachers observe the most, and what are the durations and frequencies of
these fixations with respect to body regions chosen for correction. Such analyses could
provide valuable information to enhance teacher training. (Williams, 2002)
Finally, if logistically possible, pilot testing of the regular use of the quantitative
assessment tool in a dance education institution would enable evaluation of its efficacy
and feasibility. To assess the feasibility and practicality of the tool it would be necessary
to monitor the time required for data collection, data processing and report production.
Methods for assessing and maintaining data quality would also need to be established. In
addition, a survey of the user-friendliness or ease of understanding of the data presented
in the assessment tool report would also assist in enhancing the tool further. Once the
procedures and protocols associated with the tool are fine-tuned, its efficacy in
systematically measuring the progression of individual dancers or efficacy of individual
teachers or training programs could be assessed.
In conclusion, there is much scope for future research of many important factors that can
enhance teaching practice in classical ballet. The current project has provided an
important and significant contribution towards these future scientific endeavours.
- 310 -
References
REFERENCES
- 311 -
References
REFERENCES
Abendroth-Smith, J., Kras, J. & Strand, B. (1996). Get aboard the B-boat
(biomechanically based observation and analysis for teachers). Journal of Physical
Education, Recreation and Dance, 67(8), 20-23.
Albers, D., Hu, R., McPoil, T. & Cornwall, M. W. (1992). Comparison of foot plantar
pressures during walking and en pointe. Kinesiology and Medicine for Dance, 15(1), 2532.
Alderson, J., Hopper, L., Elliott, B. & Ackland, T. (2009). Risk factors for lower back
injury in male dancers performing ballet lifts. Journal of Dance Medicine and Science,
13(3), 83-89.
Anderson, J. (1992). Ballet and Modern Dance: A Concise History, New Jersey:
Princeton Book Company.
Andrews, D. M., Arnold, T. A., Weir, P. L., van Wyk, P. M. & Callaghan, J. P. (2008).
Errors associated with bin boundaries in observation-based posture assessment methods.
Occupational Ergonomics, 8(1), 11-25.
Arend, S. & Higgins, J. R. (1976). Strategy for the classification, subjective analysis, and
observation of human movement. Journal of Human Movement Studies, 2(1), 36-52.
Armstrong, C. W. & Hoffman, S. J. (1979). Effects of teaching experience, knowledge of
performer competence, and knowledge of performance outcome on performance error
identification. Research Quarterly, 50(3), 318-327.
Armstrong, T. J., Foulke, J. A., Joseph, B. S. & Goldstein, S. A. (1982). Investigation of
cumulative trauma disorder in a poultry processing plant. American Industrial Hygiene
Association Journal, 43(2), 103-116.
Avila, F. & Moreno, F. J. (2003). Visual search strategies elaborated by tennis coaches
during execution error detection processes. Journal of Human Movement Studies, 44(3),
209-224.
Baluyut, R., Genaidy, A. M., Davis, L. S., Shell, R. L. & Simmons, R. L. (1995). Use of
visual perception in estimating static postural stresses: Magnitudes and sources of errors.
Ergonomics, 38(9), 1841-1850.
Bao, S., Howard, N., Spielholz, P. & Silverstein, B. (2007). Two posture analysis
approaches and their application in a modified Rapid Upper Limb Assessment evaluation.
Ergonomics, 50(12), 2118-2136.
- 312 -
References
Bao, S., Howard, N., Spielholz, P., Silverstein, B. & Polissar, N. (2009). Interrater
reliability of posture observations. Human Factors, 51(3), 292-309.
Bard, C., Fleury, M., Carriere, L. & Halle, M. (1980). Analysis of gymnastics judges'
visual search. Research Quarterly for Exercise and Sport, 51(2), 267-273.
Barnes, M. A., Krasnow, D., Tupling, S. J. & Thomas, M. (2000). Knee rotation in
classical dancers during the grand pli. Medical Problems of Performing Artists, 15(4),
140-147.
Beaumont, C. W. & Idzikowski, S. (1932). The Cecchetti Method of Classical Ballet,
New York: Dover.
Beijjani, F. J., Halpern, N. & Pavlidis, L. (1990). Spinal motion and strength
measurements in flamenco dancers. Medical Problems of Performing Artists, 5(3), 121124.
Berardi, G. (2005). Finding Balance: Fitness, Training and Health for a Lifetime in
Dance, London: Routledge.
Bernhardt, J., Bate, P. J. & Matyas, T. A. (1998). Accuracy of observational kinematic
assessment of upper-limb movements. Physical Therapy, 78(3), 259-270.
Bernhardt, J., Matyas, T. & Bate, P. (2002). Does experience predict observational
kinematic assessment accuracy? Physiotherapy: Theory and Practice, 18(3), 114-149.
Bertucco, M. & Cesari, P. (2010). Does movement planning follow Fitts' law? Scaling
anticipatory postural adjustments with movement speed and accuracy. Neuroscience,
171(1), 205-213.
Biscan, D. V. & Hoffman, S. J. (1976). Movement analysis as a generic ability of
physical education teachers and students. Research Quarterly, 47(2), 161-163.
Bjrklund, K., Skld, C., Andersson, L. & Daln, N. (2006). Reliability of a criterionbased test of athletes with knee injuries; where the physiotherapist and the patient
independently and simultaneously assess the patient's performance. Knee Surgery, Sports
Traumatology, Arthroscopy, 14(2), 165-175.
Bland, A. (1976). A History of Ballet and Dance in the Western World, London: Barrie &
Jenkins.
Blasis, C. (1820). An Elementary Treatise upon the Theory and Practice of the Art of
Dancing, translated by Mary Stewart Evans, London, Reproduced, New York: Dover,
1968.
- 313 -
References
Blasis, C. (1828). The Code of Terpsichore : A Practical and Historical Treatise, on the
Ballet, Dancing, and Pantomime, translated by R. Barton, Reproduced, New York:
Dance Horizons.
Bloom, B. S. (1985). Generalizations about talent development. In: Bloom, B. S. (ed.)
Talent Development in Young People. New York: Ballantine.
Bober, T., Morecki, A., Fidelus, A., Kedzior, K. & Wit, A. (1981). Biomechanical
aspects of sports technique. Biomechanics VII-A. Baltimore: University Park Press.
Boyd, R. N. & Graham, H. K. (1999). Objective measurement of clinical findings in the
use of botulinum toxin type A for the management of children with cerebral palsy.
European Journal of Neurology, 6(S4), S23-S35.
Bronner, S., Kaminski, T. R. & Gordon, A. M. (2002). Effect of anterior cruciate
ligament reconstruction on the passe movement in elite dancers. Journal of Dance
Medicine and Science, 6(4), 110-118.
Bronner, S. & Ojofeitimi, S. (2006). Gender and limb differences in healthy elite dancers:
passe kinematics. Journal of Motor Behavior, 38(1), 71-79.
Bronner, S. & Ojofeitimi, S. (2011). Pelvis and hip three-dimensional kinematics in grand
battement movements. Journal of Dance Medicine and Science, 15(1), 23-30.
Brown, C. R., Hillman, S. J., Richardson, A. M., Herman, J. L. & Robb, J. E. (2008).
Reliability and validity of the Visual Gait Assessment Scale for children with hemiplegic
cerebral palsy when used by experienced and inexperienced observers. Gait and Posture,
27(4), 648-652.
Brunnekreef, J. J., van Uden, C. J., van Moorsel, S. & Kooloos, J. G. (2005). Reliability
of videotaped observational gait analysis in patients with orthopedic impairments. BMC
Musculoskeletal Disorders, 6(17), 1-9.
Bruyneel, A. V., Mesure, S., Pare, J. C. & Bertrand, M. (2010). Organization of postural
equilibrium in several planes in ballet dancers. Neuroscience Letters, 485(3), 228-232.
Buchholz, B., Paquet, V., Punnett, L., Lee, D. & Moir, S. (1996). PATH: a work
sampling-based approach to ergonomic job analysis for construction and other nonrepetitive work. Applied Ergonomics, 27(3), 177-187.
Burdorf, A., Derksen, J., Naaktgeboren, B. & van Riel, M. (1992). Measurement of trunk
bending during work by direct observation and continuous measurement. Applied
Ergonomics, 23(4), 263-267.
- 314 -
References
Burt, S. & Punnett, L. (1999). Evaluation of interrater reliability for posture observations
in a field study. Applied Ergonomics, 30(2), 121-135.
Camomilla, V., Donati, M., Stagni R. & Cappozzo, A. (2009) Non-invasive assessment
of superficial soft tissue local displacements during movement: A feasibility study.
Journal of Biomechanics, 42(7), 931-937.
Cappozzo, A., Della Croce, U., Leardini, A. & Chiari, L. (2005). Human movement
analysis using stereophotogrammetry: Part 1: theoretical background. Gait and Posture,
21(2), 186-196.
Carr, G. (1997). Mechanics of Sport, Champaign: Human Kinetics.
Champion, L. M. & Chatfield, S. J. (2008). Measurement of turnout in dance research: A
critical review. Journal of Dance Medicine and Science, 12(4), 121-135.
Chase, W. G. & Simon, H. A. (1973). Visual Information Processing, New York:
Academic Press.
Chatfield, S. J. (1993). Electromyographic response of dancers to isokinetic work and
select dance movements. Kinesiology and Medicine for Dance, 16(1), 60-82.
Chatfield, S. J. (1996). Electromyographic and kinematic analysis of movement
repatterning in dance. Impulse: International Journal of Dance Science, Medicine and
Education, 4(3), 220-234.
Chatfield, S. J., Krasnow, D. H., Herman, A. & Blessing, G. (2007). A descriptive
analysis of kinematic and electromyographic relationships of the core during forward
stepping in beginning and expert dancers. Journal of Dance Medicine and Science, 11(3),
76-84.
Chockley, C. (2008). Ground reaction force comparison between jumps landing on the
full foot and jumps landing en pointe in ballet dancers. Journal of Dance Medicine and
Science, 12(1), 5-8.
Chow, J. W. & Knudson, D. V. (2011). Use of deterministic models in sports and
exercise biomechanics research. Sports Biomechanics, 10(3), 219-233.
Chung, C. Y., Park, M. S., Lee, S. H., Kong, S. J. & Lee, K. M. (2010). Kinematic
aspects of trunk motion and gender effect in normal adults. Journal of Neuroengineering
and Rehabilitation, 7(9), 9-15.
Clippinger-Robertson, K. S., Hutton, R. S., Miller, D. I. & Nichols, T. R. (1985).
Mechanical and anatomical factors relating to the incidence and etiology of
patellofemoral pain in dance. Kinesiology for Dance, 7(3), 7-9.
- 315 -
References
Conyn, C. (1948). Three Centuries of Ballet, Sydney: Australasian Publishing Co.

Cooper, J. M., Adrian, M. & Glassow, R. B. (1976). Kinesiology, St Louis: Mosby.
Corry, S. I., Cosgrove, A. P., Duffy, C. M., McNeill, S., Taylor, T. C. & Graham, H. K.
(1998). Botulinum toxin A compared with stretching casts in the treatment of spastic
equinus: A randomised prospective trial Journal of Pediatric Orthopaedics, 18(3), 304311.
Couillandre, A., Lewton-Brain, P. & Portero, P. (2008). Exploring the effects of
kinesiological awareness and mental imagery on movement intention in the performance
of demi-plie. Journal of Dance Medicine and Science, 12(3), 91-98.
Crookshank, D. (1999). Safe Dance Report III - A Report on the Occurence of Injury in
the Australian Professional Dance Population, Sydney: Australian Dance Council
(Ausdance).
Dallas, G., Mavidis, A. & Chairopoulou, C. (2011). Influence of angle of view on judges'
evaluations of inverted cross in men's rings. Perceptual and Motor Skills, 112(1), 109121.
Davis, R. B., Ounpuu, S., Tyburski, D. & Gage, J. R. (1991). A gait analysis data
collection and reduction technique. Human Movement Science, 10(5), 575-587.
de Bruijn, I., Engels, J. A. & van der Gulden, J. W. J. (1998). A simple method to
evaluate the reliability of OWAS observations. Applied Ergonomics, 29(4), 281-283.
de Looze, M. P., Toussaint, H. M., Ensink, J., Mangnus, C. & van der Beek, A. J. (1994).
The validity of visual observation to assess posture in a laboratory-simulated, manual
material handling task. Ergonomics, 37(8), 1335-1343.
Deckert, J. L., Barry, S. M. & Welsh, T. M. (2007). Analysis of pelvic alignment in
university ballet majors. . Journal of Dance Medicine and Science, 11(4), 110-117
Della Croce, U., Cappozzo, A. & Kerrigan, D. C. (1999). Pelvis and lower limb
anatomical landmark calibration precision and its propagation to bone geometry and joint
angles. Medical and Biological Engineering and Computing, 37(2), 155-161.
Denis, D., Lortie, M. & Bruxelles, M. (2002). Impact of observers' experience and
training on reliability of observations for a manual handling task. Ergonomics, 45(6),
441-454.
Dickens, W. E. & Smith, M. F. (2006). Validation of a visual gait assessment scale for
children with hemiplegic cerebral palsy. Gait and Posture, 23(1), 78-82.
- 316 -
References
Dozzi, P. A. & Winter, D. A. (1993). Biomechanical analysis of the foot during rises to
full pointe: implications for injuries to the metatarsal-phalangeal joints and shoe redesign.
Kinesiology and Medicine for Dance, 16(1), 1-11.
Dworak, L. B. (2005). A study characterising dynamic overloads of professional dancers.
Biomechanical Approach. Acta of Bioengineering and Biomechanics, 7(1), 77-83.
Eastlack, M. E., Arvidson, J., Snyder-Mackler, L., Danoff, J. V. & McGarvey, C. L.
(1991). Interrater reliability of videotaped observational gait-analysis assessments.
Physical Therapy, 71(6), 465-472.
Ebersole, M. L. & Armstrong, T. J. (2006). Analysis of an observational rating scale for
repetition, posture, and force in selected manufacturing settings. Human Factors, 48(3),
487-498.
Ericsson, K. A. & Charness, N. (1994). Expert performance: Its structure and acquisition.
The American Psychologist, 49(8), 725-747.
Ericsson, K. A., Krampe, R. T. & Tesch-Roemer, C. (1993). The role of deliberate
practice in the acquisition of expert performance. Psychological Review, 100(3), 363-406.
Ericsson, K. A. & Lehmann, A. C. (1996). Expert and exceptional performance: Evidence
of maximal adaptation to task constraints. Annual Review of Psychology, 47(1), 273-305.
Feipel, V., Dalenne, S., Dugailly, P.-M., Salvia, P. & Rooze, M. (2004). Kinematics of
the lumbar spine during classic ballet postures. Medical Problems of Performing Artists,
19(4), 174-180.
Fiolkowski, P. & Bauer, J. (1997). The effects of different dance surfaces on plantar
pressures. Journal of Dance Medicine and Science, 1(2), 62-66.
Fleiss, J. L. (1986). Design and Analysis of Clinical Experiments, New York: John Wiley
& Sons.
Flett, P. J., Stern, L. M., Waddy, H., Connell, T. M., Seeger, J. D. & K., G. S. (1999).
Botulinum toxin A versus fixed cast stretching for dynamic calf tightness in cerebral
palsy Journal of Paediatrics and Child Health, 35(1), 71-77.
Fornalski, S., McGarry, M. H., Csintalan, R. P., Fithian, D. C. & Lee, T. Q. (2008).
Biomechanical and anatomical assessment after knee hyperextension injury. American
Journal of Sports Medicine, 36(1), 80-44.
Gamboian, N., Chatfield, S. J. & Woollacott, M. H. (2000). Further effects of somatic
training on pelvic tilt and lumbar lordosis alignment during quiet stance and dynamic
dance movement. Journal of Dance Medicine and Science, 4(3), 90-98.
- 317 -
References
Gamboian, N., Chatfield, S. J., Woollacott, M. H., Barr, S. & Klug, G. A. (1999). Effect
of dance technique training and somatic training on pelvic tilt and lumbar lordosis
alignment during quiet stance and dynamic dance movement. Journal of Dance Medicine
and Science, 3(1), 5-14.
Gantz, J. (1990). Evaluation of faulty dance technique patterns: a working model.
Geeves, T. (1990). Safe Dance Report I - A Report of Dance Injury Prevention and
Management in Australia, Sydney: Australian Dance Council (Ausdance).
Geeves, T. (1997). Safe Dance Report II - National Injury and Lifestyle Survey of
Australian Adolescents in Per-professional Dance Training, Sydney: Australian Dance
Council (Ausdance).
Genaidy, A. M., Al-Shedi, A. A. & Karwowski, W. (1994). Postural stress analysis in
industry. Applied Ergonomics, 25(2), 77-87.
Genaidy, A. M., Simmons, R. J., Guo, L. & Hidalgo, J. A. (1993). Can visual perception
be used to estimate body part angles? Ergonomics, 36(4), 323-329.
Gibbs, R., Cassella, M. & Southwick, H. (2006). An annual post-hire health screen for
professional dancers: phase one of a preventative healthcare program for professional
dance companies (from the Dance/USA Taskforce on Dancer Health). In: Solomon, R. &
Solomon, J. (eds.) 16th Annual Meeting of the International Association for Dance
Medicine and Science. Florida, USA: University of Oregon.
Golomer, E., Bouillette, A., Mertz, C. & Keller, J. (2008). Effects of mental imagery
styles on shoulder and hip rotations during preparation of pirouettes. Journal of Motor
Behavior, 40(4), 281-290.
Golomer, E., Dupui, P., Sereni, P. & Monod, H. (1999). The contribution of vision in
dynamic spontaneous sways of male classical dancers according to student or
professional level. Journal of Physiology - Paris, 93(3), 233-237.
Golomer, E. & Fery, Y. (2001). Unilateral jump behaviour in young professional female
ballet dancers. International Journal of Neuroscience, 110(1-2), 1-7.
Golomer, E., Gravenhorst, R. M. & Toussaint, Y. (2009a). Influence of vision and motor
imagery styles on equilibrium control during whole-body rotations. Somatosensory and
Motor Research, 26(4), 105-110.
Golomer, E., Toussaint, Y., Bouillette, A. & Keller, J. (2009b). Spontaneous whole body
rotations and classical dance expertise: How shoulder-hip coordination influences
supporting leg displacements. Journal of Electromyography and Kinesiology, 19(2), 314321.
- 318 -
References
Goodkin, R. & Diller, L. (1973). Reliability among physical therapists in diagnosis and
treatment of gait deviations in hemiplegics. Perceptual and Motor Skills, 37(3), 727-734.
Gordon, M. E. (1970). The effect of the correctness of behaviour observed on the
accuracy of rating. Organizational Behavior and Human Performance, 5(4), 366-377.
Grant, G. (2008). Technical Manual and Dictionary of Classical Ballet, Lexington: BN
Publishing.
Grieg, V. (1994). Inside Ballet Technique: Separating Anatomical Fact from Fiction in
the Ballet Class, Hightstown: Princeton Book Company.
Guillou, E., Dupui, P. & Golomer, E. (2007). Dynamic balance sensory motor control and
symmetrical or asymmetrical equilibrium training. Clinical Neurophysiology, 118(2),
317-324.
Hagins, M., Pappas, E., Kremenic, I., Orishimo, K., F. & Rundle, A. (2007). The effect of
an inclined landing surface on biomechanical variables during a jumping task. Clinical
Biomechanics, 22(9), 1030-1036.
Hamilton, W. G. (2006). The New York City Ballet orthopedic screening exam. In:
Solomon, R. & Solomon, J. (eds.) Proceedings of the 16th Annual Meeting of the
International Association for Dance Medicine and Science. Eugene, Oregon: University
of Oregon.
Hammond, P. E. & Hammond, S. N. (1979). Internal logic of dance - A Weberian
perspective on the history of ballet. Journal of Social History, 12(4), 591-608.
Hammond, S. N. (2007). The Rise of Ballet Technique and Training. In: Kant, M. (ed.)
The Cambridge Companion to Ballet. New York: Cambridge University Press.
Hammond, S. N. & Hammond, P. E. (1989). Technique and autonomy in the
development of art - A case study in ballet. Dance Research Journal, 21(2), 15-24.
Hardaker, W., Erickson, L. & Myers, M. (1984). The pathogenesis of dance injury. In:
Shell, C. (ed.) The Dancer as Athlete. Champaign IL: Human Kinetics.
Harley, Y. X. H., St.Clair Gibson, A., Harley, E. H., Lambert, M. I., Vaughan, C. L. &
Noakes, T. D. (2002). Quadriceps strength and jumping efficiency in dancers. Journal of
Dance Medicine and Science, 6(3), 87-94.
Harrison, E. L., Duenkel, N., Dunlop, R. & Russell, G. (1994). Evaluation of single-leg
standing following anterior cruciate ligament surgery and rehabilitation. Physical
Therapy, 74(3), 245-252.
- 319 -
References
Hastad, D. N. & Lacy, A. C. (1994). Measurement and Evaluation in Physical Education

and Exercise, Scottsdale: Gorsuch Scarisbrick.
Haupstein, T. & Goldie, P. (2000). Visual judgements of steadiness in one-legged stance:
reliability and validity. Physiotherapy Research International, 5(3), 141-156.
Hay, J. G. & Reid, J. G. (1982). The Anatomical and Mechanical Bases of Human
Motion, New Jersey: Prentice-Hall.
Heinsalmi, P., Corlett, E. N., Wilson, J. & J., M. (1986). Methods to measure working
posture loads at working sites. The Ergonomics of Working Postures. London: Taylor &
Francis.
Hillman, S. J., Hazlewood, M. E., Loudon, I. R. & Robb, J. E. (1998). Can transverse
plane rotations be estimated from video tape gait analysis? Gait and Posture, 8(2), 87-90.
Hillman, S. J., Hazlewood, M. E., Schwartz, M. H., van der Linden, M. L. & Robb, J. E.
(2007). Correlation of the Edinburgh Gait Score with the Gillette Gait Index, the Gillette
Functional Assessment Questionnaire, and dimensionless speed. Journal of Pediatric
Orthopaedics, 27(1), 7-11.
Hinson, M., Buckman, S., Tate, J. & Sherrill, C. (1978). Grand-jete-en-tournant-entrelace
(tour jete) - analysis through motion photography. Dance Research Journal, 10(1), 9-13.
Holt, K. M., Welsh, T. M. & Speights, J. (2011). A within-subject analysis of the effects
of remote cueing on pelvic alignment in dancers. Journal of Dance Medicine and
Science, 15(1), 15-22.
Homans, H. (2010). Apollo's Angels: A History of Ballet, New York: Random House
Hughes, K. A. & Bell, F. (1994). Visual assessment of hemiplegic gait following stroke:
pilot study. Archives of Physical Medicine and Rehabilitation, 75(10), 1100-1107.
Imperial Society of Teachers of Dancing (2011). Cecchetti Society Classical Ballet
Society - Syllabus Outline of Cecchetti Examinations, London Imperial Society of
Teachers of Dancing.
Imura, A., Iino, Y. & Kojima, T. (2008). Biomechanics of the continuity and speed
change during one revolution of the fouette turn. Human Movement Science, 27(6), 903913.
Imura, A., Iino, Y. & Kojima, T. (2010). Kinematic and kinetic analysis of the fouette
turn in classical ballet. Journal of Applied Biomechanics, 26(4), 484-492.
- 320 -
References
Imwold, C. H. & Hoffman, S. J. (1983). Visual recognition of a gymnastics skill by

experienced and inexperienced instructors. Research Quarterly for Exercise and Sport,
54(2), 149-155.
Innes, E. & Straker, L. (1999). Validity of work-related assessments. Work, 13(2), 125152.
Kadaba, M. P., Ramakrishnan, H K., & Wootten, M. E., Gainey, J., Gorton, G. &
Cochran, G. V. (1989). Repeatability of kinematic, kinetic, and electromyographic data in
normal adult gait. Journal of Orthopaedic Research, 7(6), 849-860.
Kadaba, M. P., Ramakrishnan, H K., & Wootten, M. E. (1990). Measurement of lower
extremity kinematics during level walking. Journal of Orthopaedic Research, 8(3), 383392.
Kant, M. (2007). The soul of the shoe In: Kant, M. (ed.) The Cambridge Companion to
Ballet. New York: Cambridge University Press.
Karhu, O., Kansi, P. & Kuorinka, I. (1977). Correcting working postures in industry: A
practical method for analysis. Applied Ergonomics, 8(4), 199-201.
Karsavina, T. (1962). Classical Ballet: The Flow of Movement, London: Adam & Charles
Black.
Kawamura, C. M., de Morais Filho, M. C., Barreto, M. M., de Paula Asa, S. K., Juliano,
Y. & Novo, N. F. (2007). Comparison between visual and three-dimensional gait analysis
in patients with spastic diplegic cerebral palsy. Gait and Posture, 25(1), 18-24.
Keenan, A. M. & Bach, T. M. (1996). Video assessment of rearfoot movements during
walking: a reliability study. Archives of Physical Medicine and Rehabilitation, 77(7),
651-655.
Kent, M. (1984). The Oxford Dictionary of Sports Sciences and Medicine, Oxford:
Oxford University Press.
Ketola, R., Tiovonen, R. & Viikari-Juntura, E. (2001). Interobserver repeatability and
validity of an observation method to assess physical loads imposed on the upper
extremities. Ergonomics, 44(2), 119-131.
Keyserling, W. M. (1986). Postural analysis of the trunk and shoulders in simulated real
time. Ergonomics, 29(4), 569-583.
Kilbom, A., Persson, J. & Jonsson, B. G. (1986). Disorders of the cervicobrachial region
among female workers in the electronics industry. International Journal of Industrial
Ergonomics, 1(1), 37-48.
- 321 -
References
Kirstein, L. (1987). A Short History of Classical Theatrical Dancing, Princeton, New

Jersey: Dance Horizons.
Knudson, D. (1999). Validity and reliability of visual ratings of the vertical jump.
Perceptual and Motor Skills, 89(2), 642-648.
Knudson, D. V. (2000). What can professionals qualitatively analyze? Journal of
Physical Education, Recreation and Dance, 71(2), 19-23.
Knudson, D. V. & Morrison, C. S. (1997). Qualitative Analysis of Human Movement,
Champaign IL: Human Kinetics.
Kociolek, A. M. & Keir, P. J. (2010). Reliability of distal upper extremity posture
matching using slow-motion and frame-by-frame video methods. Human Factors, 52(3),
441-455.
Koman, L. A., Mooney, J. F., 3rd, Smith, B. P., Goodman, A. & Mulvaney, T. (1994).
Management of spasticity in cerebral palsy with botulinum-A toxin: report of a
preliminary, randomized, double-blind trial. Journal of Pediatric Orthopaedics, 14(3),
299-303.
Kostrovitskaya, V. & Pisarev, A. (1978). School of Classical Dance: Textbook of the
Vaganova Choreographic School, London: Dance Books.
Krasnow, D. & Chatfield, S. J. (2009). Development of the "Performance Competence
Evaluation Measure": assessing qualitative aspects of dance performance. Journal of
Dance Medicine and Science, 13(4), 101-107.
Krasnow, D. H. & Chatfield, S. J. (1997). Imagery and conditioning practices for dancers.
Dance Research Journal, 29(1), 43-64.
Kraus, R., Chapman - Hilsendager, S. & Dixon, B. (1997). History of the Dance in Art
and Education, Boston: Allyn & Bacon.
Krebs, D. E., Edelstein, J. E. & Fishman, S. (1985). Reliability of observational kinematic
gait analysis. Physical Therapy, 65(7), 1027-1033.
Krosshaug, T., Nakamae, A., Boden, B., Engebretsen, L., Smith, G., Slauterbeck, J.,
Hewett, T. E. & Bahr, R. (2007). Estimating 3D joint kinematics from video sequences of
running and cutting maneuvers--assessing the accuracy of simple visual inspection. Gait
and Posture, 26(3), 378-385.
Kulig, K., Fietzer, A. L. & Popovich, J. M. (2011). Ground reaction forces and knee
mechanics in the weight acceptance phase of a dance leap take-off and landing. Journal
of Sports Sciences, 29(2), 125-131.
- 322 -
References
Kwon, Y.-H., Wilson, M. & Joong-Hyun, R. (2007). Analysis of the hip joint moments in
grand rond de jambe en l'air. Journal of Dance Medicine and Science, 11(3), 93-99.
Landis, J. R. & Koch, G. G. (1977). The measurement of observer agreement for
categorical data. Biometrics, 33(1), 159-174.
Lau, M. H. & Armstrong, T. J. (2011). The effect of viewing angle on wrist posture
estimation from photographic images using novice raters. Applied Ergonomics, 42(5),
634-643.
Laws, K. (1986). The mechanics of the fouette turn. Kinesiology for Dance, 28(4), 22-24.
Laws, K. (1998). Momentum transfer in dance movement. Medical Problems of
Performing Artists, 13(4), 136-145.
Laws, K. & Fulkerson, L. (1992). The slowing of pirouettes. Kinesiology and Medicine
for Dance, 15(1), 72-80.
Laws, K. & Lee, K. (1989). The grand jete - a physical analysis. Kinesiology for Dance,
11(4), 12-13.
Laws, K. & Petrie, C. (1999). Momentum transfer in dance movement, vertical jumps: A
research update. Medical Problems of Performing Artists, 14(3), 138-149.
Laws, K. L. (1979). An analysis of turns in dance. Dance Research Journal, 11(1-2), 1219.
Lawson, J. (1979). The Principles of Classical Dance, London: Adam & Charles Black.
Lawson, J. (1983). The Teaching of Classical Ballet: Common Faults in Young Dancers
and their Training, London: A & C Black.
Leardini, A. (2005). Human movement analysis using stereophotogrammetry: Part 3. Soft
tissue artifact assessment and compensation. Gait and Posture, 21(2), 212-225.
Lee, C. (1983). An Introduction to Classical Ballet, New Jersey: Lawrence Erlbaum
Associates.
Lee, C. (2002). Ballet in Western Culture: A History of its Origins and Evolution, New
York: Routledge.
Lees, A., Barton, G. & Robinson, M. (2010). The influence of Cardan rotation sequence
on angular orientation data for the lower limb in the soccer kick. Journal of Sports
Sciences, 28(4), 445-450.
- 323 -
References
Lees, A., Vanrenterghem, J. & De Clercq, D. (2004). The maximal and submaximal
vertical jump: implications for strength and conditioning. Journal of Strength and
Conditioning Research, 18(4), 787-791.
Lepelley, M. C., Thullier, F., Koral, J. & Lestienne, F. G. (2006). Muscle coordination in
complex movements during jete in skilled ballet dancers. Experimental Brain Research,
175(2), 321-331.
Leskinen, T., Hall, C., Rauas, S., Ulin, S., Tonnes, M., Viikari-Juntura, E. & Takala, E.
(1997). Validation of Portable Ergonomic Observation (PEO) method using
optoelectronic and video recordings. Applied Ergonomics, 28(2), 75-83.
Li, G. & Buckle, P. (1999). Current techniques for assessing physical exposure to workrelated musculoskeletal risks, with emphasis on posture-based methods. Ergonomics,
42(5), 674-695.
Liederbach, M. (1997). Screening for functional capacity in dancers: designing
standardized, dance-specific injury prevention screening tools. Journal of Dance
Lin, C. F., Su, F. C. & Wu, H. W. (2005). Ankle biomechanics of ballet dancers in releve
en pointe dance. Research in Sports Medicine, 13(1), 23-35.
Liu, A., Meiring, W. & Wand, Y. (2004). Testing generalized linear models using
smoothing spline methods. Statistica Sinica, 15, 235-256.
Locke, L. F. (1972). Skill learning and performance. Implications for physical education.
Research Quarterly, 43(3), 374-386.
Loong, T.-W. (2003). Understanding sensitivity and specificity with the right side of the
brain. British Medical Journal, 327(7417), 716-719.
Lord, S. E., Halligan, P. W. & Wade, D. T. (1998). Visual gait analysis: the development
of a clinical assessment and scale. Clinical Rehabilitation, 12(2), 107-119.
Lowe, B. D. (2004a). Accuracy and validity of observational estimates of shoulder and
elbow posture. Applied Ergonomics, 35(2), 159-171.
Lowe, B. D. (2004b). Accuracy and validity of observational estimates of wrist and
forearm posture. Ergonomics, 47(5), 527-554.
Mackey, A. H., Lobb, G. L., Walt, S. E. & Stott, N. S. (2003). Reliability and validity of
the Observational Gait Scale in children with spastic diplegia. Developmental Medicine
and Child Neurology, 45(1), 4-11.
- 324 -
References
Marques-Bruna, P., Lees, A. & Grimshaw, P. (2008). Structural Principal Components

Analysis of the Kinematics of the Soccer Kick Using Different Types of Rating Scales.
International Journal of Sports Science and Coaching, 3(1), 73-85.
Masso, N., Germain, A., Rey, F., Costa, L., Romero, D. & Guitart, S. (2004). Study of
muscle activity during releve in first position and sixth positions. Journal of Dance
Mayers, L., Bronner, S., Agraharasamakulam, S. & Ojofeitimi, S. (2010). Lower
extremity kinetics in tap dance. Journal of Dance Medicine and Science, 14(1), 3-10.
McAtamney, L. & Corlett, E. N. (1993). RULA: A survey method for the investigation of
work-related upper limb disorders. Applied Ergonomics, 24(2), 91-99.
McGinley, J. L., Goldie, P. A., Greenwood, K. M. & Olney, S. J. (2003). Accuracy and
reliability of observational gait analysis data: judgments of push-off in gait after stroke.
Physical Therapy, 83(2), 146-160.
McGinley, J. L., Morris, M. E., Greenwood, K. M., Goldie, P. A. & Olney, S. J. (2006).
Accuracy of clinical observations of push-off during gait after stroke. Archives of
Physical Medicine and Rehabilitation, 87(6), 779-785.
McNitt-Gray, J. L., Koff, S. R. & Hall, B. L. (1992). The influence of dance training and
foot position on landing biomechanics. Medical Problems of Performing Artists, 7(3), 8791.
McPherson, M. (1990). A systematic approach to skill analysis. Sports: Science
Periodical on Research & Technology in Sport, 11(1), 1-9.
McPherson, M. N. (1996). Qualitative and quantitative analysis in sports. American
Journal of Sports Medicine, 24(S6), S85-S88.
Midgett, V. D. (1993). Effects of arm position on hang time during grande jete.
Miller, C. D., Paulos, L. E., Parker, R. D. & Fishell, M. (1990). The ballet technique
shoe: a preliminary study of eleven differently modified ballet technique shoes using
force and pressure plates. Foot and Ankle International, 11(2), 97-100.
Miyazaki, S. & Kubota, T. (1984). Quantification of gait abnormalities on the basis of
continuous foot-force measurement: correlation between quantitative indices and visual
rating. Medical and Biological Engineering and Computing, 22(1), 70-76.
Molnar, M. & Esterson, J. (1997). Screening students in a pre-professional ballet school.
Journal of Dance Medicine and Science, 1(3), 118-121.
- 325 -
References
Moreno, F. J., Reina, R., Luis, V. & Sabido, R. (2002). Visual search strategies in
experienced and inexperienced gymnastic coaches. Perceptual and Motor Skills, 95(3),
901-902.
Morrison, C. S. (2000). Thoughts regarding the qualitative analysis of human movement.
Perceptual and Motor Skills, 91(1), 259-260.
Morrison, C. S., Knudson, D., Clayburn, C. & Haywood, P. (2005). Accuracy of visual
estimates of joint angle and angular velocity using criterion movements. Perceptual and
Motor Skills, 100(3), 599-606.
Morrow, J. R., Jackson, A. W., Disch, J. G. & Mood, D. P. (2005). Measurement and
Evaluation in Human Performance. , Champaign IL: Human Kinetics.
Moseley, A. M., Descatoire, A. & Adams, R. D. (2008). Observation of high and low
passive ankle flexibility in stair descent. Perceptual and Motor Skills, 106(1), 328-340.
Mouchnino, L., Aurenty, R., Massion, J. & Pedotti, A. (1992). Coordination between
equilibrium and head-trunk orientation during leg movement: a new strategy build up by
training. Journal of Neurophysiology, 67(6), 1587-1598.
Mouchnino, L., Mesure, S., Lizee, E., Landjerit, B. & Massion, J. (1998). Is the spinal
column a rigid or articulated ais during leg movement? Human Movement Science, 17(3),
289-306.
Mullineaux, D. R., Bartlett, R. M. & Bennett, S. (2001). Research design and statistics in
biomechanics and motor control. Journal of Sports Sciences, 19(10), 739-760.
Mumford, L. (1952). Art and Technics, New York: Columbia University Press.
Negus, V., Hopper, D. & Briffa, N. K. (2005). Associations between turnout and lower
extremity injuries in classical ballet dancers. Journal of Orthopaedic and Sports Physical
Therapy, 35(5), 307-318.
Neumann, W. P., Wells, R. P., Norman, R. W., Kerr, M. S. & Shannon, J. F. (2001).
Trunk posture: reliability, accuracy, and risk estimates for low back pain from a video
based assessment method. International Journal of Industrial Ergonomics, 28(6), 355365.
Nicholls, R., Fleisig, G., Elliott, B., Lyman, S. & Osinski, E. (2003). Accuracy of
qualitative analysis for assessment of skilled baseball pitching technique. Sports
Biomechanics, 2(2), 213-226.
Nordera, M. (2007). Ballet de cour. In: Kant, M. (ed.) The Cambridge Companion to
Ballet. New York: Cambridge University Press.
- 326 -
References
Novacheck, T. F., Stout, J. L. & Tervo, R. (2000). Reliability and validity of the Gillette
Functional Assessment Questionnaire as an outcome measure in children with walking
disabilities. Journal of Pediatric Orthopaedics, 20(1), 75-81.
Noverre, J. G. (1760). Letters on Dancing and Ballet, translated by Cyril W. Beaumont.
London 1830, Reproduced, New York Dance Horizons 1968.
Olsson, E. C. (1990). Methods of studying gait. In: Schmidt, G. L. (ed.) Gait in
Rehabilitation. New York: Churchill Livingstone.
Orishimo, K. F., Kremenic, I. J., Pappas, E., Hagins, M. & Liederbach, M. (2009).
Comparison of landing biomechanics between male and female professional dancers.
American Journal of Sports Medicine, 37(11), 2187-2193.
Osborne, M. M. & Gordon, M. E. (1972). An investigation of the accuracy of ratings of a
gross motor skill. Research Quarterly, 43(1), 55-61.
Oxford Online Dictionary (2011). http://oxforddictionaries.com.
Pappas, E. & Hagins, M. (2008). The effects of "raked" stages on standing posture in
dancers. Journal of Dance Medicine and Science, 12(2), 54-58.
Paquet, V. L., Punnett, L. & Buchholz, B. (2001). Validity of fixed-interval observations
for postural assessment in construction work. Applied Ergonomics, 32(3), 215-224.
Paskeva, A. (2002). Ballet: From the First Plie to Mastery, an Eight-Year Course, New
York: Routledge.
Paskeva, A. (2005). Beyond Ballet Tradition, New York: Routledge.
Passier, L. N., Nasciemento, M. P., Gesch, J. M. & Haines, T. P. (2010). Physiotherapist
observation of head and neck alignment. Physiotherapy: Theory and Practice, 26(6), 416423.
Patla, A. E. & Clouse, S. D. (1988). Visual assessment of human gait: Reliability and
validity. Rehabilitation Research, 8796.
Patla, A. E., Proctor, J. & Morson, B. (1987). Observation of aspects of visual gait
assessment: a questionnaire study. Physiotherapy Canada, 39(5), 311-316.
Perrin, P., Deviterne, D., Hugel, F. & Perrot, C. (2002). Judo, better than dance, develops
sensorimotor adaptabilities involved in balance control. Gait and Posture, 15(2), 187194.
- 327 -
References
Perry, J. (1992). Gait Analysis: Normal and Pathological Function, New Jersey: SLACK
Inc.
Petrakis, E. (1986). Visual observation patterns of tennis teachers. Research Quarterly for
Exercise and Sport, 57(3), 254-259.
Piripis, M., Ugoni, A., Starr, R., Rodda, J., Boyd, R. N., Nattras, G. R. & Graham, H. K.
(2001). The 'Physician Rating Scale' - Validity and reliability. Gait and Posture, 13(3),
293.
Plastino, J. G. (2005). Physical Screening Procedures. In: Solomon, R. (ed.) Preventing
Dance Injuries. 2nd ed. Champaign, IL: Human Kinetics.
Plessner, H. & Schallies, E. (2005). Judging the cross on rings: A matter of achieving
shape constancy. Applied Cognitive Psychology, 19(9), 1145-1156.
Poggini, L., Losasso, S., Cerreto, M. & Cesari, L. (1997). Jump ability in novice ballet
dancers before and after training. Journal of Dance Medicine and Science, 1(2), 46-50.
Portney, L. G. & Watkins, M. P. (1993). Foundations of Clinical Research: Applications
to Practice, Norwalk: Appleton & Lange
Potter, K. & Baas, J. (2011). Screening for improved dance function. The International
Association of Dance Medicine and Science Bulletin for Teachers, 3(1), 14-17.
Randolph, J. J. (2005). Free-marginal multirater kappa: An alternative to Fleiss' fixedmarginal multirater kappa. Proceedings of the Joensuu University Learning and
Instruction Symposium, October 14-15th, 2005, p3-20.
Randolph, J. J. (2008). Online Kappa Calculator, http://justus.randolph.name/kappa.
Rasmussen, C. C. & Hay, J. G. (1993). The illusion of suspension in selected dance leaps.
Ravn, S., Voigt, M., Simonsen, E. B., Alkjaer, T., Bojsen-Mller, F. & Klausen, K.
(1999). Choice of jumping strategy in two standard jumps, squat and countermovement
jump--effect of training background or inherited preference? Scandinavian Journal of
Medicine and Science in Sports, 9(4), 201-208.
Read, H. S., Hazlewood, M. E., Hillman, S. J., Prescott, R. J. & Robb, J. E. (2003).
Edinburgh visual gait score for use in cerebral palsy. Journal of Pediatric Orthopaedics,
23(3), 296-301.
Reinsch, C. (1967). Smoothing by spline functions. Numerische Mathematik, 10(2), 177183.
- 328 -
References
Ridd, J. E., Nicholson, A. S. & Montan, A. J. (1989). A portable microcomputer based

system for on site activity and posture recording. In: Megaw, E. D. (ed.) Contemporary
Ergonomics. London: Taylor & Francis.
Rist, R. (2001). Giving feedback when teaching. Dancing Times, 91(1085), 431.
Romei, M., Galli, M., Motta, F., Schwartz, M. & Crivellini, M. (2004). Use of the
normalcy index for the evaluation of gait pathology. Gait and Posture, 19(1), 85-90.
Royal Academy of Dance (2011). Royal Academy of Dance Examinations Board
Specification, London: Royal Academy of Dance.
Runeson (1984). Perceiving people through their movements. In: Kirkcaldy, B. (ed.)
Individual Differences in Movement. Lanchester: MIT Press.
Ryman, R. S. (1978). Kinematic analysis of selected grand allegro jumps. In: Woodruff,
D. L. (ed.) Essays in Dance Research from the 5th CORD Conference, New York,
Congress on Research in Dance, 1978, p. 231-242.
Ryman, R. S. & Ranney, D. A. (1979). A preliminary investigation of two variations of
the grand-battement-devant. Dance Research Journal, 11(1-2), 2-11.
Saleh, M. & Murdoch, G. (1985). In defence of gait analysis. Observation and
measurement in gait assessment. Journal of Bone and Joint Surgery (British Volume),
67(2), 237-241.
Sartor, C., Alderink, G., Greenwald, H. & Elders, L. (1999). Critical kinematic events
occurring in the trunk during walking. Human Movement Science, 18(5), 669-679.
Schlaich, J. & DuPont, B. (1993). The Art of Teaching Dance Technique, Reston
Virginia: American Alliance for Health, Physical Education, Recreation and Dance.
Schorer, S. (1999). Suki Schorer on Balanchine Technique, New York: Random House.
Schutte, L. M., Narayanan, U., Stout, J. L., Selber, P., Gage, J. R. & Schwartz, M. H.
(2000). An index for quantifying deviations from normal gait. Gait and Posture, 11(1),
25-31.
Schwartz, M. (2004). Kinematics of normal gait. In: Gage, J. R. (ed.) The Treatment of
Gait Problems in Cerebral Palsy. Suffolk: Mac Keith Press.
Shan, G. (2005). Comparison of repetitive movements between ballet dancers and martial
artists: risk assessment of muscle overuse injuries and prevention strategies. Research in
Sports Medicine, 13(1), 63-76.
- 329 -
References
Sherman, C. A., Sparrow, W. A., Jolley, D. & Eldering, J. (2001). Coaches' perceptions
of golf swing kinematics. International Journal of Sport Psychology, 32(3), 257-270.
Shippen, J. M. & May, B. (2010). Calculation of muscle loading and joint contact forces
during the rock step in Irish dance. Journal of Dance Medicine and Science, 14(1), 11-18.
Siev-Ner, I., Barak, A., Heim, M., Warshavsky, M. & Azaria, M. (1997). The value of
screening. Journal of Dance Medicine and Science, 1(3), 87-92.
Simmons, R. W. (2005a). Neuromuscular responses of trained ballet dancers to postural
perturbations. International Journal of Neuroscience, 115(8), 1193-1203.
Simmons, R. W. (2005b). Sensory organization determinants of postural stability in
trained ballet dancers. International Journal of Neuroscience, 115(1), 87-97.
Simpson, K. J., Jameson, E. G. & Odum, S. (1996). Estimated patellofemoral
compressive forces and contact pressures during dance landings. Journal of Applied
Biomechanics, 12(1), 1-14.
Simpson, K. J. & Kanter, L. (1997). Jump distance of dance landings influencing internal
joint forces: I. Axial forces. Medicine and Science in Sports and Exercise, 29(7), 916927.
Simpson, K. J. & Pettit, M. (1997). Jump distance of dance landings influencing internal
joint forces: II. Shear forces. Medicine and Science in Sports and Exercise, 29(7), 928936.
Solomon, R. (1997). A pro-active screening program for addressing injury prevention in a
professional ballet company. Journal of Dance Medicine and Science, 1(3), 113-117.
Spielholz, P., Silverstein, B., Morgan, M., Checkoway, H. & Kaufman, J. (2001).
Comparison of self-report, video observation and direct measurement methods for upper
extremity musculoskeletal disorder physical risk factors. Ergonomics, 44(6), 588-613.
Stagni, R., Fantozzi, S. Cappello, A. & Leardini, A. (2005) Quantification of soft tissue
artefact in motion analysis by combining 3D fluoroscopy and stereophotogrammetry: a
study on two subjects. Clinical Biomechanics, 20(3), 320-329.
Staheli, L. T. & Engel, G. M. (1972). Tibial torsion: a method of assessment and a survey
of normal children. Clinical Orthopaedics and Related Research, 86(3), 183-186.
Ste-Marie, D. M. (1999). Expert - novice differences in gymnastic judging: An
information-processing perspective. Applied Cognitive Psychology, 13(2), 269-281.
- 330 -
References
Stensrud, S., Myklebust, G., Kristianslund, E., Bahr, R. & Krosshaug, T. (2011).
Correlation between two-dimensional video analysis and subjective assessment in
evaluating knee control among elite female team handball players. British Journal of
Sports Medicine, 45(7), 589-595.
Stott, N. S., Atherton, W. G., Mackey, A. H., Galley, I. J., Nicol, R. O. & Walsh, S. J.
(2005). The reliability and validity of assessment of sagittal plane deviations in children
who have spastic diplegia. Archives of Physical Medicine and Rehabilitation, 86(12),
2337-2341.
Teitz, C. C., Harrington, R. M. & Wiley, H. (1985). Pressures on the foot in pointe shoes.
Foot and Ankle, 5(5), 216-221.
Teschke, K., Trask, C., Johnson, P., Chow, Y., Village, J. & Koehoorn, M. (2009).
Measuring posture for epidemiology: comparing inclinometry, observations and selfreports. Ergonomics, 52(9), 1067-1078.
Thullier, F. & Moufti, H. (2004). Multi-joint coordination in ballet dancers. Neuroscience
Letters, 369(1), 80-84.
Toro, B., Nestor, C. & Farren, P. (2003). A review of observational gait assessment in
clinical practice. Physiotherapy: Theory and Practice, 19(3), 137-149.
Trepman, E., Gellman, R. E., Micheli, L. J. & De Luca, C. J. (1998). Electromyographic
analysis of grand-pli in ballet and modern dancers. Medicine and Science in Sports and
Exercise, 30(12), 1708-1720.
Trepman, E., Gellman, R. E., Solomon, R., Ramesh Murthy, K., Micheli, L. J. & De
Luca, C. J. (1994). Electromyographic analysis of standing posture and demi-plie in ballet
and modern dancers. Medicine and Science in Sports and Exercise, 26(6), 771-782.
Vaganova, A. (1953). Basic Principles of Classical Ballet: Russian Ballet Technique, 2nd
English Edition., London: A & C Black.
van der Beek, A. J., van Gaalen, L. C. & Frings-Dresen, M. H. (1992). Working postures
and activities of lorry drivers: a reliability study of on-site observation and recording on a
pocket computer. Applied Ergonomics, 23(5), 331-336.
van Wyk, P. M., Weir, P. L., Andrews, D. M., Fiedler, K. M. & Callaghan, J. P. (2009).
Determining the optimal size for posture categories used in video-based posture
assessment methods. Ergonomics, 52(8), 921-930.
Vanrenterghem, J., Lees, A. & De Clercq, D. (2008). Effect of forward trunk inclination
on joint power output in vertical jumping. Journal of Strength and Conditioning
Research, 22(3), 708-714.
- 331 -
References
Village, J., Trask, C., Luong, N., Chow, Y., Johnson, P., Koehoorn, M. & Teschke, K.
(2009). Development and evaluation of an observational Back-Exposure Sampling Tool
(Back-EST) for work-related back injury risk factors. Applied Ergonomics, 40(3), 538544.
Vincent, W. J. (1995). Statistics in Kinesiology, Champaign, IL: Human Kinetics.
Vogel, D. (2007). Understanding knee trauma. Dance Spirit, 11(1), 51-52.
Voloshin, A. S., Beijjani, F. J., Halpern, M. & Frankel, V. H. (1989). Dynamic loading on
flamenco dancers: A biomechanical study. Human Movement Science, 8(5), 503-513.
von Porat, A., Holmstrom, E. & Roos, E. (2008). Reliability and validity of videotaped
functional performance tests in ACL-injured subjects. Physiotherapy Research
International, 13(2), 119-130.
Walczak, B. & Kai, U. (2008). Balanchine the Teacher: Fundamentals that Shaped the
First Generation of New York City Ballet Dancers, Gainesville, FL: University Press of
Florida.
Warrens, M. (2010). Inequalities between mulit-rater kappas. Advances in Data Analysis
and Classification, 4(4), 271-286.
Watelain, E., Froger, J., Rousseaux, M., Lensel, G., Barbier, F., Lepoutre, F. X. &
Thevenon, A. (2005). Variability of video-based clinical gait analysis in hemiplegia as
performed by practitioners in diverse specialties. Journal of Rehabilitation Medicine,
37(5), 317-324.
Weir, J. P. (2005). Quantifying test-retest reliability using the intraclass correlation
coefficient and the SEM. Journal of Strength and Conditioning Research, 19(1), 231-240.
White, J. (1996). Teaching Classical Ballet, Gainesville, FL: The University Press of
Florida.
Whiting, H. T. (1972). Overview of the skill learning process. Research Quarterly, 43(3),
266-294.
Wikipedia (2011). http://en.wikipedia.org.
Williams, A. M. (2002). Visual search behaviour in sport. Journal of Sports Sciences,
20(3), 169-170.
Wilmerding, V., Gurney, B. & Torres, V. (2003). The effect of positive heel inclination
on posture in young children training in flamenco dance. Journal of Dance Medicine and
Science, 7(3), 85-90.
- 332 -
References
Wilmerding, V., Heyward, V. H., King, M., Fielder, K. J., Stidely, C. A., Pett, S. B. &
Evans, B. (2001). Electromyographic comparison of the developpe devant at the barre
and centre. Journal of Dance Medicine and Science, 5(3), 69-74.
Wilson, M. (2009). Applying biomechanics research in the dance studio. The
International Association of Dance Medicine and Science Bulletin for Teachers, 1(2), 1113.
Wilson, M., Bee-Oh, L. & Young-Hoo, K. (2004). A three-dimensional kinematic
analysis of grand rond de jambe en l'air. Journal of Dance Medicine and Science, 8(4),
108-115.
Wilson, M., Ryu, J.-H. & Kwon, Y.-H. (2007). Contribution of the pelvis to gesture leg
range of motion in a complex ballet movement grand rond de jambe en l'air en dehors.
Journal of Dance Medicine and Science, 11(4), 118-123.
Winter, D. A. (1985). Concerning the scientific basis for the diagnosis of pathological
gait and for rehabilitation protocols. Physiotherapy Canada, 37(4), 245-252.
Woltring, H. J. (1985). On optimal smoothing and derivative estimation from noisy
displacement data in biomechanics. Human Movement Science, 4(2), 229-245.
Woltring, H. J. (1986). A Fortran package for generalized cross-validatory spline
smoothing and differentiation. Advances in Engineering Software, 8(2), 104-113
Wood, G. A. & Jennings, L. S. (1979). On the use of spline functions for data smoothing.
Journal of Biomechanics, 12(1), 477-479.
Wood G. A. (1982). Data smoothing and differentiation procedures in biomechanics.
Exercise and Sport Sciences Reviews, 10(3), 308-362.
Woodhull-McNeal, A. P., Clarkson, P. M., James, R., Watkins, A. & Barrett, S. (1990).
How linear is dancers' posture. Medical Problems of Performing Artists, 5(4), 151-154.
Wren, T. A., Rethlefsen, S. A., Healy, B. S., Do, K. P., Dennis, S. W. & Kay, R. M.
(2005). Reliability and validity of visual assessments of gait using a modified physician
rating scale for crouch and foot contact. Journal of Pediatric Orthopaedics, 25(5), 646650.
Wu, G. & Cavanagh, P. R. (1995). ISB recommendations for standardization in the
reporting of kinematic data. Journal of Biomechanics, 28(10), 1257-1261.
Xu, X., Chang, C. C., Faber, G. S., Kingma, I. & Dennerlein, J. T. (2011). The validity
and interrater reliability of video-based posture observation during asymmetric lifting
tasks. Human Factors, 53(4), 371-382.
- 333 -
References
Yen, T. Y. & Radwin, R. G. (2000). Comparison between using spectral analysis of

electrogoniometer data and observational. Ergonomics, 43(1), 106-132.
Youdas, J. W., Carey, J. R. & Garrett, T. R. (1991). Reliability of measurements of
cervical spine range of motion-comparison of three methods. Physical Therapy, 71(2),
98-104.
- 334 -
Appendices
APPENDICES
- 335 -
Appendices
APPENDICES
Appendix A: Ethics Approval for Chapter 3 Comparison of Theoretical and Practical
Execution of Classical Ballet Technique
- 336 -
Appendices
Appendix B: Information and Consent Form for Non-professional Ballet Dancers
- 337 -
Appendices
Appendix B continued:
- 338 -
Appendices
Appendix B continued:
- 339 -
Appendices
Appendix C: Information and Consent Form for Professional Ballet Dancers
- 340 -
Appendices
Appendix C continued:
- 341 -
Appendices
Appendix C continued:
- 342 -
Appendices
Appendix D: Dancer Questionnaire
- 343 -
Appendices
Appendix D continued:
- 344 -
Appendices
- 345 -
Appendices
- 346 -
Appendices
- 347 -
Appendices
Appendix E: Chapter 3 - Additional Information on Cubic Spline Curve Interpolation
Interpolation (gap filling) of the kinematic data was achieved using cubic spline curve
interpolation, The interpolation procedure was run using the Vicon Workstation software.
The term spline has its origins in a pliable strip of wood or rubber used by draftsmen in
patterning curves, but the mathematical form was popularised as an approximation
procedure during the 1960s when it was shown that a spline function was the smoothest of
all functions for fitting N data within specified limits (Reinsch, 1967).
In general, one need only specify the degree of the spline, the required accuracy of the fit,
and the number and positions of the knots. Spline functions are an extremely useful
analytical tool for biomechanists, as they are the ideal interpolative function to use when a
set of time histories must be synchronised in order to process data (Wood, 1982). Specific
detail on the cubic spine function is provided below.
The cubic spine function is a form of interpolation where the interpolant is a special type of
polynomial called a spline (Wood & Jennings, 1979). A spline consists of a number of
polynomials, all of some low degree m, that are pieced together at points in time called
knots (xj: j = 1, 2, ........n) and joined in such a way as to provide a continuous function
g(t) with m-1 continuous derivatives. When m equals 3, as is most common, the resulting
cubic spline function consists of n-1 cubic polynomials, each of the form
g(t) = pj(t) = a j + b j(t) + cj(t)2 + dj(t)3
spanning an interval xj-1 W[j and satisfying the continuity condition
pkj (x j) = pkj+1; (k = 0,1,2; j = 1,2,......n)
where pkj denotes the kth derivative of the jth polynomial piece. The condition by which the
function has m-1 continuous derivatives ensures that it is smooth itself but, unlike a global
polynomial, its piecewise nature enables it to adapt quickly to changes in curvature
(Wood, 1982).
- 348 -
Appendices
Appendix F: Chapter 3 - Additional Information on Woltring General Cross-Validatory

(GCV) Quintic Spline Routine
Woltring's general cross-validatory (GCV) quintic spline routine (Woltring, 1985;
Woltring, 1986) was used for the filtering process. As previously indicate in Appendix E,
spline functions are a series of polynomial curves through one or more points joined (or
pieced together) at points called knots. Many spline techniques are derived from the
work of Reinsch (1967) and have a knot at each data point, therefore requiring the use to
choose optimal knot positions. That is, the user has to specify a weighting factor for each
data point, and select the value of the smoothing parameter which controls the extent of the
smoothing.
The GCV quintic spline routine does not require the user to specify the amount of error in
the data to be smoothed, but instead automatically selects an optimum smoothing
parameter (Wolting, 1985), thereby allowing selective filtering of the data signal more
where it is required, and less where it does not require as much. The GCV quintic spline
routine can thus accommodate data points sampled at unequal time intervals.
,I LV D VPRRWKLQJ SDUDPHWHU FRQWUROOLQJ WKH WUDGH-off between the smoothness of the
estimate and the goodness of fit, WKHQ 9 LV WKH *&9 HVWLPDWH RI WKH VPRRWKLQJ
parameter.
)RU
, the GCV function is:
(Liu et al., 2004)
The Woltring GCV filter was run using the Vicon Workstation software.
- 349 -
Appendices
Appendix G: Chapter 3 - Graphical Outputs for Gait Data Displayed in Vicon Polygon
Authoring Tool
Shaded area represents mean normal adult gait data 1 standard deviation
- 350 -
Appendices
Appendix H: Chapter 3 - Summary of Descriptive Statistics for Kinematic Variables
Descriptive Statistics for Thorax Anterior/Posterior Tilt

Thorax Anterior-Posterior Tilt
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Time
point
Step
Mean
Difference ()
(NP - PR)
t-test
CV
Mean (SD) ()
CV
Start
-7.1 (6.7)
0.94
-12.1(3.4)
0.28
5.0
0.040*
Peak
4.5 (8.2)
1.82
-2.6 (6.7)
2.58
7.1
0.039*
Start
-7.2 (6.8)
0.94
-12.4 (3.4)
0.27
5.2
0.036*
Peak
4.5 (9.4)
2.09
-0.5 (4.7)
9.40
5.0
0.136
Start
-6.9 (2.1)
0.30
-6.8 (5.3)
0.78
-0.1
0.958
Peak
-7.6 (2.4)
0.32
-7.6 (5.0)
0.66
0.0
1.000
Start
-8.7 (3.3)
0.38
-6.5 (6.0)
0.92
-2.2
0.286
Peak
-7.0 (3.8)
0.54
-6.2 (6.0)
0.97
-0.8
0.707
Start
-7.1 (2.7)
2.63
-5.7 (5.7)
1.00
-1.4
0.480
Peak
-7.3 (2.6)
0.36
-6.8 (6.3)
0.93
-0.5
0.827
Start
-6.8 (2.5)
0.37
-8.3 (2.0)
0.24
1.5
0.142
Peak
-7.5 (2.5)
0.33
-8.9 (2.5)
0.28
1.4
0.185
Start
-0.5 (4.3)
8.60
1.3 (4.8)
3.69
-1.8
0.383
Peak
-6.9 (1.7)
0.25
-9.1 (3.3)
0.36
2.2
0.067
Start
0.0 (3.3)
0.9 (4.8)
5.33
-0.9
0.607
Peak
-5.1 (3.5)
0.69
-9.2 (3.0)
0.33
4.1
0.008*
Start
-3.4 (4.8)
1.41
-4.5 (4.3)
0.96
1.1
0.582
Peak
-1.2 (3.8)
3.17
-4.3 (4.4)
1.02
3.1
0.091
Start
-4.3 (3.0)
0.70
-6.6 (3.4)
0.52
2.3
0.111
Peak
-0.2 (3.5)
17.50
-2.6 (4.0)
1.54
2.4
0.156
Start
3.4 (4.3)
1.26
2.7 (5.9)
2.19
0.7
0.777
Peak
-3.6 (3.2)
0.89
-6.9 (5.2)
0.75
3.3
0.084
Grand Jet Elanc

En Avants
Start
1.6 (5.0)
3.13
-1.0 (9.1)
9.10
2.6
0.411
Peak
-2.7 (7.0)
2.59
-3.8 (11.4)
3.00
1.1
0.789
Double Pirouette
En Dehors
Start
-2.9 (5.6)
1.93
-0.9 (4.3)
4.78
-2.0
0.362
Peak
1.2 (6.7)
5.58
0.3 (8.1)
27.00
0.9
0.777
Start
3.4 (6.5)
1.91
6.7 (5.7)
0.85
-3.3
0.210
Peak
-2.9 (6.5)
2.24
1.1 (7.2)
6.55
-4.0
0.185
Grand Pli 1
Position
st
Grand Pli 2
Position
nd
Battement Tendu
Back
Battement Jet 45
Front
st
Rise
(Relev)
nd
Rise 2
Position
st
Saut 1 Position
Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
Professionals (PR)
(N=11)
Mean (SD) ()
Rise 1 Position
Dart
(Elanc)
Non-professionals (NP)
(N=11)
Double Pirouette
En Dedans
SD = Standard Deviation
CV = Coefficient of Variation
* Significant difference between non-professional and professional groups (t-test: p<0.05).
- 351 -
p-value
Appendices
Appendix H continued:
Descriptive Statistics for Thorax Side Tilt
Thorax Side Tilt
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Professionals (PR)
(N=12)
Mean
Difference ()
(NP - PR)
Mean (SD) ()
CV
Mean (SD) ()
CV
t-test
p-value
Start
-0.5 (1.9)
3.80
0.5 (1.7)
3.40
-1.0
0.252
Peak
0.8 (2.9)
3.63
0.0 (1.9)
0.8
0.438
Grand Pli 2nd

Position
Start
0.1 (1.9)
19.00
0.0 (2.0)
0.1
0.917
Peak
0.3 (2.5)
8.33
0.2 (1.9)
9.50
0.1
0.925
Start
0.8 (1.8)
2.25
1.1 (2.1)
1.91
-0.3
0.747
Peak
0.5 (1.7)
3.40
0.7 (2.1)
3.00
-0.2
0.738
Start
-0.2 (1.2)
6.00
-0.7 (2.0)
2.86
0.5
0.436
Peak
-0.5 (1.4)
2.80
-0.7 (2.1)
3.00
0.2
0.809
Start
0.2 (1.4)
7.00
0.5 (1.2)
2.40
-0.3
0.631
Peak
0.3 (1.6)
5.33
0.6 (1.8)
3.00
-0.3
0.624
Start
0.2 (1.4)
7.00
0.3 (1.4)
4.67
-0.1
0.881
Peak
0.5 (1.6)
3.20
0.5 (1.6)
3.20
0.0
1.000
Start
0.5 (1.4)
2.80
0.7 (1.2)
1.71
-0.2
0.634
Peak
-0.5 (1.3)
2.60
-0.4 (2.4)
6.00
-0.1
0.914
Start
-8.4 (3.7)
0.44
-6.5 (2.5)
2.00
-1.9
0.194
Peak
-4.8 (3.0)
0.63
-3.1 (2.6)
0.38
-1.7
0.166
Start
2.0 (7.0)
3.50
3.5 (9.5)
2.71
-1.5
0.668
Peak
-6.3 (3.1)
0.49
-6.5 (5.2)
0.80
0.2
0.922
Start
-1.2 (4.8)
4.00
1.9 (8.6)
4.53
-3.1
0.310
Peak
-5.7 (3.1)
0.54
-7.5 (4.9)
0.65
1.8
0.331
Start
0.5 (3.6)
7.20
0.1 (2.0)
20.00
0.4
0.771
Peak
-2.4 (3.9)
1.63
-0.4 (2.1)
5.25
-2.0
0.147
Grand Jet Elanc

En Avants
Start
5.7 (4.2)
0.74
5.0 (4.2)
0.84
0.7
0.689
Peak
2.5 (3.1)
1.24
-5.7 (7.2)
1.26
8.2
0.002*
Double Pirouette
En Dehors
Start
3.5 (7.8)
2.23
4.2 (6.8)
1.62
-0.7
0.797
Peak
-3.9 (6.3)
1.62
-2.7 (5.1)
1.89
-1.2
0.635
Start
-3.5 (6.8)
1.94
-1.6 (4.5)
2.81
-1.9
0.449
Peak
0.9 (6.4)
7.11
-0.5 (8.5)
17.00
0.4
0.655
Battement Tendu
Back
Battement Jet
45 Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
(N=13)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
Double Pirouette
En Dedans
- 352 -
Appendices
Descriptive Statistics for Pelvic Anterior-Posterior Tilt

Pelvic Anterior-Posterior Tilt
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Professionals (PR)
(N=12)
Mean (SD) ()
CV
Mean (SD) ()
CV
Mean
Difference ()
(NP - PR)
t-test
p-value
Start
16.6 (4.6)
0.28
15.2 (4.4)
0.29
1.4
0.434
Peak
10.4 (6.7)
0.64
6.1 (4.8)
0.79
4.3
0.080
Grand Pli 2nd

Position
Start
15.3 (3.9)
0.25
13.8 (5.4)
0.39
1.5
0.410
Peak
11.4 (6.8)
0.60
6.1 (7.1)
1.16
5.3
0.066
Start
21.1 (3.7)
0.18
18.7 (4.4)
0.24
2.4
0.164
Peak
30.5 (3.9)
0.13
28.8 (4.4)
0.15
1.7
0.313
Start
15.4 (3.3)
0.21
15.1 (4.2)
0.28
0.3
0.841
Peak
7.8 (4.2)
0.54
5.6 (4.6)
0.82
2.2
0.212
Start
20.2 (3.6)
0.18
17.8 (3.4)
0.19
2.4
0.100
Peak
19.6 (3.6)
0.18
15.9 (3.1)
0.19
3.7
0.012*
Start
19.2 (4.3)
0.22
16.5 (3.5)
0.21
2.7
0.104
Peak
18.5 (4.4)
0.24
15.1 (4.2)
0.28
3.4
0.051
Start
22.1 (3.3)
0.15
19.4 (5.4)
0.28
2.7
0.145
Peak
17.3 (4.3)
0.25
17.7 (4.9)
0.28
0.2
0.846
Start
20.2 (4.2)
0.21
21.4 (3.8)
0.18
-1.2
0.468
Peak
17.6 (4.4)
0.25
17.8 (4.5)
0.25
-0.2
0.940
Start
-4.5 (2.5)
0.56
-1.8 (3.8)
2.11
-2.7
0.044*
Peak
-0.6 (3.8)
6.33
-0.2 (2.9)
14.50
-0.4
0.748
Start
-3.8 (2.9)
0.76
-0.4 (5.3)
13.25
-3.4
0.055
Peak
1.8 (3.3)
1.83
-0.5 (4.1)
8.20
2.3
0.131
Start
20.5 (4.6)
0.22
21.1 (5.2)
0.25
-0.6
0.816
Peak
20.1 (3.6)
0.18
19.1 (5.4)
0.28
1.0
0.587
Grand Jet Elanc

En Avants
Start
15.5 (5.7)
0.37
14.8 (5.7)
0.39
0.7
0.732
Peak
24.8 (4.5)
0.18
28.1 (4.6)
0.16
-3.3
0.095
Double Pirouette
En Dehors
Start
-2.5 (11.4)
4.56
-7.6 (9.5)
1.25
5.1
0.238
Peak
0.8 (10.2)
12.75
-2.9 (16.3)
5.62
3.7
0.501
Start
-15.5 (11.2)
0.72
-12.3 (9.7)
0.79
-3.2
0.454
Peak
5.4 (14.6)
2.70
-2.4 (12.7)
5.29
-3.0
0.168
Battement Tendu
Back
Battement Jet
45 Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
(N=13)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
Double Pirouette
En Dedans
- 353 -
Appendices
Descriptive Statistics for Pelvic Obliquity

Pelvic Obliquity
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Professionals (PR)
(N=12)
Mean (SD) ()
CV
Mean (SD) ()
CV
Mean
Difference ()
(NP - PR)
t-test
p-value
Start
0.3 (2.3)
7.67
0.8 (1.2)
1.50
-0.5
0.552
Peak
0.1 (1.7)
17.00
0.3 (1.4)
4.67
-0.2
0.786
Grand Pli 2nd

Position
Start
0.2 (1.8)
9.00
0.0 (1.7)
0.2
0.745
Peak
0.1 (2.3)
23.00
-0.1 (1.8)
18.00
0.2
0.846
Start
2.8 (2.2)
0.79
3.3 (1.5)
0.45
-0.5
0.46
Peak
1.2 (2.9)
2.42
0.5 (1.2)
2.40
0.7
0.427
Start
1.8 (2.4)
1.33
1.9 (1.0)
0.53
-0.1
0.926
Peak
9.3 (3.1)
0.33
9.3 (2.6)
0.28
0.0
0.982
Start
0.2 (2.0)
10.00
1.0 (1.3)
1.30
-0.8
0.237
Peak
0.0 (1.9)
0.5 (1.0)
2.00
-0.5
0.419
Start
0.0 (2.0)
1.0 (1.2)
1.20
-1.0
0.142
Peak
0.2 (2.0)
10.00
0.9 (1.2)
1.33
-1.1
0.274
Start
-0.2 (2.1)
10.50
0.6 (2.3)
3.83
-0.8
0.405
Peak
0.0 (2.3)
0.2 (1.6)
8.00
-0.2
0.987
Start
-1.7 (4.7)
2.76
1.1 (3.6)
3.27
-2.8
0.112
Peak
-12.4 (3.8)
3.26
-10.8 (3.1)
0.29
-1.6
0.257
Start
18.5 (4.7)
0.25
18.5 (4.9)
0.26
0.0
0.984
Peak
16.3 (4.7)
0.24
15.3 (6.3)
0.41
1.0
0.663
Start
18.7 (4.5)
0.24
17.6 (4.3)
0.24
1.1
0.534
Peak
18.2 (3.7)
0.20
16.8 (4.8)
0.29
1.4
0.397
Start
6.9 (3.3)
0.48
3.9 (3.8)
0.97
3.0
0.046*
Peak
4.4 (4.0)
0.91
4.5 (3.0)
0.67
-0.1
0.936
Grand Jet Elanc

En Avants
Start
2.8 (4.6)
1.64
2.1 (5.2)
2.48
0.7
0.702
Peak
4.6 (4.9)
1.07
5.1 (6.4)
1.25
-0.5
0.839
Double Pirouette
En Dehors
Start
-10.2 (7.4)
0.73
-8.4 (9.8)
1.17
-1.8
0.605
Peak
12.2 (10.5)
0.86
8.6 (12.8)
1.49
3.6
0.441
Start
8.1 (9.3)
1.15
4.8 (9.6)
2.00
3.3
0.399
Peak
-3.2 (15.0)
0.21
-2.1 (14.6)
6.95
-1.1
0.859
Battement Tendu
Back
Battement Jet
45 Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
(N=13)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
Double Pirouette
En Dedans
- 354 -
Appendices
Descriptive Statistics for Pelvic - Thoracic Transverse Rotation

Pelvic - Thoracic Transverse Rotation
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Professionals (PR)
(N=12)
Mean (SD) ()
CV
Mean (SD) ()
CV
Mean
Difference ()
(NP - PR)
t-test
p-value
Start
-0.3 (2.0)
6.67
0.8 (3.2)
0.40
-1.1
0.348
Peak
-0.1 (2.7)
27.00
-0.6 (2.7)
4.50
0.5
0.703
Grand Pli 2nd

Position
Start
0.1 (2.1)
21.00
-0.1 (2.4)
24.00
0.2
0.853
Peak
-0.3 (3.1)
10.33
-1.1 (2.6)
2.36
0.8
0.512
Start
-1.5 (3.5)
2.33
0.6(3.0)
5.00
-2.1
0.152
Peak
-6.6 (3.2)
0.48
-9.2 (3.4)
0.37
2.6
0.077
Start
-1.8 (2.7)
1.50
-1.9 (3.0)
1.58
0.1
0.941
Peak
1.5 (4.3)
2.87
5.7 (3.1)
0.54
-4.2
0.041*
Start
0.4 (3.2)
8.00
0.3 (2.4)
8.00
0.1
0.941
Peak
0.9 (2.6)
2.89
0.0 (2.6)
0.9
0.421
Start
0.1 (2.7)
27.00
0.5 (2.2)
4.40
-0.4
0.731
Peak
0.0 (2.2)
0.4 (2.8)
7.00
-0.4
0.74
Start
-0.1 (2.6)
26.00
-0.6 (2.7)
4.50
0.5
0.638
Peak
-0.6 (3.3)
5.50
-0.6 (3.4)
5.67
0.0
1.000
Start
-2.3 (4.9)
2.13
-3.6 (3.2)
0.89
1.3
0.477
Peak
-2.9 (4.6)
1.59
-2.3 (4.7)
2.04
-0.6
0.751
Start
-3.1 (5.8)
1.87
-1.0 (3.8)
3.80
-2.1
0.329
Peak
-0.7 (4.3)
6.14
2.9 (5.0)
1.72
-3.6
0.287
Start
-6.7 (4.2)
0.63
-4.7 (5.6)
1.19
-2.0
0.354
Peak
1.6 (3.9)
2.44
-1.0 (4.2)
4.20
0.8
0.156
Start
-3.6 (4.3)
1.19
-4.4 (3.6)
0.82
0.8
0.634
Peak
-3.6 (4.7)
1.31
-2.4 (2.8)
1.17
-1.2
0.446
Grand Jet Elanc

En Avants
Start
2.7 (7.1)
2.63
-7.7 (8.5)
0.56
10.4
0.005*
Peak
32.5 (5.8)
0.18
21.6 (9.8)
0.45
10.9
0.005*
Double Pirouette
En Dehors
Start
-2.9 (5.6)
1.93
-0.9 (4.3)
4.78
-2.0
0.362
Peak
1.2 (6.7)
5.58
0.3 (8.1)
27.00
0.9
0.777
Start
3.4 (6.5)
1.91
6.7 (5.7)
0.85
-3.3
0.21
Peak
-2.9 (6.5)
2.24
1.1 (7.2)
6.55
-4.0
0.185
Battement Tendu
Back
Battement Jet
45 Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
(N=13)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
Double Pirouette
En Dedans
- 355 -
Appendices
Descriptive Statistics for Hip Transverse Rotation

Hip Transverse Rotation
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Grand Pli 1st

Position
Grand Pli 2nd
Position
Battement Tendu
Back
Battement Jet 45
Front
Rise 1st Position
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Dart
(Elanc)
Turn
(Tour)
Grand Jet Elanc

En Avants
Double Pirouette En
Dehors
Double Pirouette En
Dedans
Time
point
(N=13)
Professionals (PR)
(N=12)
Mean
Difference ()
(NP - PR)
t-test
Mean (SD) ()
CV
Mean (SD) ()
CV
Start
-11.9 (2.8)
0.24
-17.8 (6.6)
0.37
5.9
0.008*
Peak
-21.7 (7.9)
0.36
-23.4 (11.3)
0.48
1.7
0.660
Start
-20.8 (6.2)
0.30
-28.8 (6.4)
0.22
8.0
0.004*
Peak
-27.5 (8.2)
0.30
-25.3 (13.8)
0.55
-2.2
0.615
Start
-14.9 (5.0)
0.34
-22.2 (6.2)
0.28
7.3
0.004*
Peak
-17.2 (2.4)
0.14
-24.8 (6.7)
0.27
7.6
0.001*
Start
-9.5 (3.8)
0.40
-17.3 (6.8)
0.39
7.8
0.002*
Peak
-23.6 (4.3)
0.18
-30.3 (7.6)
0.25
6.7
0.012*
Start
-12.3 (4.0)
0.33
-17.5 (5.7)
0.33
5.2
0.014*
Peak
-17.7 (2.3)
0.13
-23.6 (7.0)
0.30
5.9
0.009*
Start
-19.0 (4.2)
0.22
-22.6 (6.1)
0.27
3.6
0.099
Peak
-25.0 (4.4)
0.18
-30.8 (5.7)
0.19
5.8
0.010*
Start
-20.6 (4.8)
0.23
-20.8 (6.0)
0.29
0.2
0.951
Peak
-25.5 (4.2)
0.16
-32.0 (8.4)
0.26
6.5
0.021*
Start
-9.5 (4.5)
0.47
-10.3 (5.8)
0.56
0.8
0.734
Peak
-26.9 (5.7)
0.21
-32.2 (7.1)
0.22
5.3
0.052
Start
-16.2 (5.3)
0.33
-23.6 (6.8)
0.29
7.4
0.006*
Peak
-37.8 (6.7)
0.18
-39.0 (11.1)
0.28
1.2
0.738
Start
-12.4 (7.2)
0.58
-18.0 (6.7)
0.37
5.6
0.055
Peak
-36.5 (5.5)
0.15
-41.2 (10.5)
0.25
4.7
0.171
Start
-29.4 (5.9)
0.20
-25.8 (9.3)
0.36
-3.6
0.262
Peak
-24.8 (5.4)
0.22
-30.1 (8.0)
0.27
5.3
0.062
Start
1.8 (5.4)
3.00
-3.8 (4.8)
1.26
5.6
0.013*
Peak
-2.2 (10.6)
4.82
3.3 (12.7)
3.85
-5.5
0.258
Start
-14.1 (7.6)
0.54
-19.4 (7.0)
0.36
5.3
0.083
Peak
-12.6 (6.9)
0.55
-21.3 (6.8)
0.32
8.7
0.005*
Start
-12.4 (4.7)
0.38
-19.5 (8.0)
0.41
7.1
0.012*
Peak
-15.2 (3.4)
0.22
-21.5 (5.5)
0.26
6.3
0.002*
- 356 -
p-value
Appendices
Descriptive Statistics for Ankle Transverse Rotation

Ankle Transverse Rotation
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
Mean (SD) ()
CV
Mean (SD) ()
CV
Mean
Difference ()
(NP - PR)
t-test
p-value
Start
-16.8 (8.3)
0.49
-16.3 (5.0)
0.31
-0.5
0.899
Peak
5.5 (10.6)
1.93
-1.9 (11.3)
5.95
7.4
0.217
Grand Pli 2nd

Position
Start
-7.1 (6.5)
0.92
-10.0 (5.3)
0.53
2.9
0.367
Peak
-15.9 (12.0)
0.75
-13.7 (5.8)
0.42
-2.2
0.673
Start
-13.1 (6.3)
0.48
-17.6 (7.5)
0.43
4.5
0.234
Peak
2.0 (4.2)
2.10
1.4 (4.2)
3.00
3.3
0.797
Start
-16.8 (7.8)
0.46
-20.1 (5.3)
0.26
3.3
0.348
Peak
4.3 (5.3)
1.23
4.0 (4.3)
1.08
0.3
0.922
Start
-11.3 (8.5)
0.75
-15.4 (3.6)
0.23
4.1
0.250
Peak
7.9 (4.5)
0.71
1.6 (2.7)
1.69
6.3
0.007*
Start
-6.3 (7.8)
1.24
-13.9 (3.3)
0.24
7.6
0.032*
Peak
7.8 (5.1)
0.65
1.1 (2.6)
2.36
6.7
0.009*
Start
-30.9 (7.2)
0.23
-36.9 (10.0)
0.27
6.0
0.201
Peak
5.9 (5.1)
0.86
6.6 (4.2)
0.64
-0.7
0.78
Start
-29.5 (12.0)
0.71
-40.7 (9.8)
0.24
11.2
0.072
Peak
5.1 (4.0)
0.78
2.6 (5.4)
2.08
2.5
0.312
Start
-27.4 (7.2)
0.26
-35.7 (6.2)
0.17
8.3
0.033*
Peak
3.3 (4.5)
1.36
-2.4 (7.3)
3.04
5.7
0.090
Start
-26.6 (11.7)
0.44
-35.3 (6.6)
0.19
8.7
0.107
Peak
2.0 (3.2)
1.60
-1.1 (6.2)
5.64
3.1
0.230
Start
4.0 (4.8)
1.20
0.9 (6.6)
7.33
3.1
0.309
Peak
4.6 (2.4)
0.52
5.1 (5.8)
1.08
-0.5
0.820
Grand Jet Elanc

En Avants
Start
0.4 (8.8)
22.00
1.9 (6.4)
3.37
-1.5
0.718
Peak
7.6 (6.7)
0.88
7.7 (8.4)
1.09
-0.1
0.982
Double Pirouette En
Dehors
Start
-0.8 (5.0)
62.50
-9.0 (5.1)
0.57
8.2
0.007*
Peak
-3.1 (9.5)
3.06
-3.9 (6.3)
1.62
0.8
0.866
Start
11.4 (7.8)
0.68
3.4 (6.8)
2.00
8.0
0.057
Peak
15.0 (6.8)
0.45
10.4 (7.2)
0.69
4.6
0.229
Battement Tendu
Back
Battement Jet 45
Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
Professionals (PR)
(N=12)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
(N=13)
Double Pirouette En
Dedans
- 357 -
Appendices
Descriptive Statistics for Knee Extension

Knee Extension
Movement
Category
Bend
(Pli)
Stretch
(Battement)
Mean (SD) ()
CV
Mean (SD) ()
CV
Start
-6.4 (3.7)
0.58
-10.4 (3.8)
0.37
4.0
0.014*
Peak
n/a
n/a
n/a
n/a
n/a
n/a
Start
-9.3 (3.2)
0.34
-12.3 (3.5)
0.28
3.0
0.039*
Peak
n/a
n/a
n/a
n/a
n/a
n/a
Battement Tendu
Back
Start
-4.5 (3.6)
0.80
-8.9 (5.3)
0.60
4.4
0.021*
Peak
-3.9 (2.0)
0.51
-6.5 (3.8)
0.58
2.6
0.044*
Battement Jet 45
Front
Start
-2.8 (4.5)
1.61
-8.8 (5.2)
0.59
6.0
0.005*
Peak
-5.8 (3.3)
0.57
-8.7 (3.1)
0.36
2.9
0.040*
Start
-5.6 (3.5)
0.63
-8.8 (4.9)
0.56
3.2
0.078
Peak
-7.4 (3.3)
0.45
-8.6 (4.4)
0.51
1.2
0.447
Start
-10.2 (3.8)
0.37
-12.8 (4.2)
0.33
2.6
0.130
Peak
-8.2 (5.2)
0.63
-10.1 (4.5)
0.45
1.9
0.348
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-6.5 (2.5)
0.38
-8.3 (3.1)
0.37
1.8
0.144
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-5.9 (2.5)
0.42
-9.8 (3.0)
0.31
3.9
0.002*
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
0.5 (6.5)
13.00
-4.1 (5.2)
1.27
4.6
0.065
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-3.2 (3.3)
1.03
-6.4 (4.4)
0.69
3.2
0.046*
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-4.5 (3.4)
0.76
-6.6 (3.0)
0.45
2.1
0.113
Grand Jet Elanc

En Avants
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-1.5 (4.3)
2.87
-7.7 (5.3)
0.69
6.2
0.004*
Double Pirouette
En Dehors
Start
-1.8 (4.9)
2.72
-2.8 (5.7)
2.04
1.0
0.676
Peak
2.8 (10.6)
3.79
0.3 (12.1)
40.33
2.5
0.597
Start
-2.8 (3.3)
1.18
-6.3 (4.8)
0.76
3.5
0.049*
Peak
-4.2 (6.4)
1.52
-6.1 (5.3)
0.87
1.9
0.422
Grand Pli 1st

Position
Grand Pli 2nd
Position
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
Professionals (PR)
(N=12)
Mean
Difference ()
(NP - PR)
Step
Rise 1st Position
Dart
(Elanc)
(N=13)
Double Pirouette
En Dedans
Time
point
- 358 -
t-test
p-value
Appendices
Descriptive Statistics for Ankle Plantarflexion

Movement
Category
Bend
(Pli)
Stretch
(Battement)
Step
CV
Mean (SD) ()
CV
t-test
p-value
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
n/a
n/a
n/a
n/a
n/a
n/a
Grand Pli 2nd

Position
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
n/a
n/a
n/a
n/a
n/a
n/a
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-52.5 (5.0)
0.10
-56.7 (4.8)
0.08
4.2
0.045*
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-56.5 (3.9)
0.07
-59.4 (4.8)
0.08
2.9
0.113
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-48.8 (2.9)
0.06
-51.7 (5.0)
0.10
2.9
0.094
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-50.1 (3.3)
0.07
-52.2 (3.4)
15.35
2.1
0.132
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-56.4 (4.8)
0.09
-59.7 (4.7)
0.08
3.3
0.099
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-52.2 (5.4)
0.10
-57.3 (5.3)
0.09
5.1
0.026*
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-52.4 (4.1)
0.08
-55.9 (4.9)
0.09
3.5
0.063
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-52.5 (3.5)
0.07
-56.4 (4.3)
0.08
3.9
0.021*
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-53.2 (3.5)
0.07
-56.7 (5.2)
0.09
3.5
0.058
Grand Jet Elanc

En Avants
Start
n/a
n/a
n/a
n/a
n/a
n/a
Peak
-50.5 (5.1)
0.10
-56.8 (6.4)
0.11
6.3
0.012*
Double Pirouette En
Dehors
Start
-38.0 (7.7)
0.20
-41.2 (3.6)
0.09
3.2
0.206
Peak
-25.8 (19.5)
0.76
-30.8 (17.0)
0.55
5.0
0.497
Start
-39.9 (6.6)
0.17
-40.2 (6.8)
0.17
0.3
0.928
Peak
-38.7 (10.4)
0.27
-41.7 (13.2)
0.32
3.0
0.535
Battement Tendu
Back
Battement Jet 45
Front
Rise
(Relev)
Rise 2nd Position
Saut 1st Position

Jump
(Saut)
Temps Lev
Glissade Under
Glide
(Glisse)
Glissade Over
Jet Ordinaire
Turn
(Tour)
Mean (SD) ()
Mean
Difference ()
(NP - PR)
Professionals (N=12)
Grand Pli 1st

Position
Rise 1st Position
Dart
(Elanc)
Time
point
(N=13)
Double Pirouette En
Dedans
- 359 -
Appendices
Appendix I: Chapter 3 - Summary of Mean Difference in Joint Angle for Steps and
Kinematic Variables for which a Statistical Significance was Achieved
Mean Joint Angle at Start or Peak of Step ()
Nonprofessionals
Professionals
Difference between nonprofessional and

professional dancers
-7.1
-12.1
5.0
Grand Pli 1 Position - Peak
4.5
-2.6
7.1
Grand Pli 2nd Position - Start
-7.2
-12.4
5.2
Temps Lev - Peak
-5.1
-9.2
4.1
Thorax Anterior/Posterior Tilt

Grand Pli 1st Position - Start
st
Mean
5.4
Thorax Side Tilt

Grand Jet Elanc En Avants - Peak
2.5
-5.7
8.2
Mean
8.2
Pelvic Anterior-Posterior Tilt

Rise 1st Position - Peak
19.6
15.9
3.7
Glissade Under - Start
-4.5
-1.8
2.7
Mean
3.2
Pelvic Obliquity
Jet Ordinaire - Start
6.9
3.9
3.0
Mean
3.0
Pelvic - Thoracic Transverse Rotation

Battement Jet 45 Front - Peak
1.5
5.7
4.2
Grand Jet Elanc En Avants - Start
2.7
-7.7
10.4
32.5
21.6
10.9
Mean
8.5
Ankle Transverse Rotation

7.9
1.6
6.3
Rise 2nd Position - Start
-6.3
-13.9
7.6
Rise 2nd Position - Peak
7.8
1.1
6.7
-27.4
-35.7
8.3
Double Pirouette En Dehors - Start
-0.8
-9
8.2
Mean
- 360 -
7.4
Appendices
Appendix I continued:
Mean Joint Angle at Start or Peak of Step ()

Difference between nonNonProfessionals
professional and
professionals
professional dancers
Hip Transverse Rotation
-11.9
-17.8
5.9
-20.8
-28.8
8.0
Battement Tendu Back - Start
-14.9
-22.2
7.3
Battement Tendu Back - Peak
-17.2
-24.2
7.0
Battement Jet 45 Front - Start
-9.5
-17.3
7.8
-23.6
-30.3
6.7
Rise 1st Position - Start
-12.3
-17.5
5.2
-17.7
-23.6
5.9
Rise 2nd Position - Peak
-20.5
-30.8
10.3
Saut 1st Position - Peak
-25.5
-32
6.5
-16.2
-23.6
7.4
Grand Jet Elanc En Avants - Start
1.8
-3.8
5.6
Double Pirouette En Dehors - Peak
-12.6
-21.3
8.7
Double Pirouette En Dedans - Start
-12.4
-19.5
7.1
Double Pirouette En Dedans - Peak
-15.2
-21.5
6.3
Mean
7.0
Knee Extension
-6.4
-10.4
4.0
-9.3
-12.3
3.0
Battement Tendu Back - Start
-4.5
-8.9
4.4
-3.9
-6.5
2.6
Battement Jet 45 Front - Start
-2.8
-8.8
6.0
-5.8
-8.7
2.9
Temps Lev - Peak
-5.9
-9.8
3.9
Glissade Over - Peak
-3.2
-6.4
3.2
-1.5
-7.7
6.2
Double Pirouette En Dedans - Start
2.8
0.3
2.5
Mean
3.9
-52.5
-56.7
4.2
Temps Lev - Peak
-52.2
-57.3
5.1
Glissade Over - Peak
-52.5
-56.4
3.9
-50.5
-56.8
6.3
Mean
- 361 -
4.9
Appendices
Appendix J: Ethics Approval for Chapter 4 Accuracy of Qualitative Analysis of Classical

Ballet Technique
- 362 -
Appendices
Appendix K: Information and Consent Form for Professional Ballet Teachers
- 363 -
Appendices
Appendix K continued:
- 364 -
Appendices
Appendix K continued:
- 365 -
Appendices
Appendix L: Information and Consent Form for Student Ballet Teachers
- 366 -
Appendices
Appendix L continued:
- 367 -
Appendices
Appendix L continued:
- 368 -
Appendices
Appendix M: Ballet Teacher Questionnaire
- 369 -
Appendices
Appendix M continued:
- 370 -
Appendices
- 371 -
Appendices
- 372 -
Appendices
Appendix N: Qualitative Analysis Questionnaire Introductory information and

questions presented for Step 1 only. Same questions and format were
repeated for Steps 25
- 373 -
Appendices
Appendix N continued:
- 374 -
Appendices
- 375 -
Appendices
- 376 -
Appendices
- 377 -
Appendices
- 378 -
Appendices
Appendix O: Chapter 4 - Percentages of Accurate Responses, Type I and Type II Errors

Teaching Experience: 0 years (N=12)
Correct Technique
Incorrect
Technique
Total
Rated Incorrect
Rated Correct
Total
15%
(Type I Error)
30%
(Accurate)
45%
21%
(Accurate)
34%
(Type II Error)
55%
36%
64%
100%
Rated Incorrect
Rated Correct
Total
17%
(Type I Error)
28%
(Accurate)
45%
25%
(Accurate)
30%
(Type II Error)
55%
42%
58%
100%
Rated Incorrect
Rated Correct
Total
21%
(Type I Error)
24%
(Accurate)
45%
41%
(Accurate)
14%
(Type II Error)
55%
62%
38%
100%
Teaching Experience: 1-9 years (N=11)
Correct Technique
Incorrect
Technique
Total
Teaching Experience: E
Correct Technique
Incorrect
Technique
Total
- 379 -
Appendices
Appendix P: Chapter 5 - Correlation Matrix for Different Versions of Quantitative

Assessment Tool
14 Steps
14 Steps
Correlation
Coefficient
1.000
Sig. (2-tailed)
7 Steps
4 Steps
3 Steps
4 Other Steps
3 Other Steps
7 Steps
4 Steps
3 Steps
4 Other
Steps
3 Other
Steps
0.908**
0.868**
0.851**
0.582*
0.385
0.000
0.000
0.000
0.029
0.175
1.000
0.903**
0.732**
0.503
0.389
0.000
0.003
0.067
0.169
1.000
0.859**
0.644*
0.525
0.000
0.013
0.054
1.000
0.657*
0.415
0.011
0.140
1.000
0.851**
Correlation
Coefficient
0.908**
Sig. (2-tailed)
0.000
Correlation
Coefficient
0.868**
0.903**
Sig. (2-tailed)
0.000
0.000
Correlation
Coefficient
0.851**
0.732*
0.859**
Sig. (2-tailed)
0.000
0.003
0.000
Correlation
Coefficient
0.582*
0.503
0.644*
0.657*
Sig. (2-tailed)
0.029
0.067
0.013
0.011
Correlation
Coefficient
0.385
0.389
0.525
0.415
0.851**
Sig. (2-tailed)
0.175
0.169
0.054
0.140
0.000
**Correlation is significant at the 0.01 level.

*Correlation is significant at the 0.05 level.
- 380 -
0.000
1.000

Biomechanic Ballet

Uploaded by

Copyright:

Available Formats

Biomechanic Ballet

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Biomechanic Ballet

Uploaded by

Copyright:

Available Formats

What are the four fundamental biomechanical principles of classical ballet technique identified in the literature?

What are the four fundamental biomechanical principles of classical ballet technique identified in the literature?

What were the results when comparing the kinematic data of professional ballet dancers to non-professional ballet dancers performing core ballet steps?

What were the results when comparing the kinematic data of professional ballet dancers to non-professional ballet dancers performing core ballet steps?

BIOMECHANICAL PERSPECTIVES ON

CLASSICAL BALLET TECHNIQUE AND

Rachel Evelyn Ward

A thesis in fulfilment of the requirements for the degree of

School of Risk and Safety Sciences

Abbrevootion for de~ree a _given In the UniVersity calendar PhD

Sohoor. Stlloot ol Rls~ &nd Safety Sciences

Faculty. FaO<Jity of Science

TIUB! Biometllanfcal poll!pediVas On classical ballet techniqUe and

Abstract 350 words maximum: (PLEASE TYPE)

pretequ!"ite for a SiJCGe$$ful career-as a professional clnSsical ballal aa"~' Given

wa~ IUW!Illgah~d with

respt:ct to 'experience In classical ballet lnstrudJon_ Teachers qualitative re.spon.lut$ ware.

data oblaincd rrorr, 30 mc>l!on oua\ySts. A fJ0$1Uve

bt:twettn yeuf3' ot teaching eJ(penence and aCcuracy of

programs- ror dance teachers.

a quantllallve assessment 1001 was developed In

Date or completion of requirements for Award:

THIS S~EET IS iO 8E GLIJEO TO TfiE' INSIDE FRONT COVER OF THE THESIS

COPYRIGHT STAT EMENT

Signed ...... ... ...

.r/..~.:R.( .... /1/.~. . . .. . . .. . . .

Date ~ " .. ~ ...

Signed ... .. ...

&.~4./. . . .o/..C?.e:::.. ..................... . .

Pate ......... ......

Thank you to my supervisors Associate Professor Andrew McIntosh and Dr John

1.2 Aims of the thesis.....................................................................................................

1.3 Structure of the thesis..............................................................................................

CHAPTER 2 - LITERATURE REVIEW..................................................................

2.1 Foundations of Classical Ballet Technique...........................................................

2.1.2 The History and Development of Classical Ballet Technique......................

2.1.3 The Fundamental Principles of Classical Ballet Technique.........................

2.2 Quantitative Biomechanical Analysis in Dance....................................................

2.2.2 Instrumentation Used in Quantitative Dance Biomechanics Research.........

2.2.3 Purpose and Aims of Quantitative Dance Biomechanics Research..............

2.2.4 Dance Genres and Quantitative Dance Biomechanics Research..................

2.2.5 Quantitative Biomechanics Research in Classical Ballet Technique............

2.3 Qualitative Analysis of Human Movement.............................................................

2.3.2 Application of Qualitative Analysis of Human Movement............................

2.3.3 Reliability and Accuracy of Qualitative Analysis of Human Movement.......

2.3.4 Investigation of the Accuracy of Qualitative Analysis of Human Movement 96

2.4 Summary of Literature Review and Project Rationale.......................................... 115

CHAPTER 3 - COMPARISON OF THEORETICAL AND PRACTICAL

3.2.1 Participant Recruitment................................................................................

3.2.3 Kinematic Measurement Variables...............................................................

3.2.4 Experimental Protocol..................................................................................

3.2.5 Data Extraction.............................................................................................

3.2.6 Statistical Analysis........................................................................................

3.3.1 Participant Demographics.............................................................................

3.3.3 Between Group Analysis of Fundamental Principles with Respect to Joint

3.3.4 Analysis of Fundamental Principles with Respect to Movement Categories

3.4.5 Analysis of Fundamental Principles with Respect to Movement Categories

4.2.1 Participant Recruitment.................................................................................

4.2.2 Development of the Qualitative Assessment Questionnaire..........................

4.2.3 Video Data Selection and Preparation...........................................................