Original Research

Kinematic Differences in Shoulder Roll and Hip Roll

at Different Front Crawl Speeds in National
Level Swimmers
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw

Jordan T. Andersen,1 Peter J. Sinclair,1 Carla B. McCabe,2 and Ross H. Sanders1

CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

1
Physical Activity, Lifestyle, Aging and Wellness, Faculty of Health Sciences, The University of Sydney, Sydney, Australia; and 2Sport
and Exercise Sciences Research Institute, School of Sport, Faculty of Life & Health Sciences, Ulster University, Newtownabbey,
Northern Ireland

Abstract
Andersen, JT, Sinclair, PJ, McCabe, CB, and Sanders, RH. Kinematic differences in shoulder roll and hip roll at different front crawl
speeds in National Level Swimmers. J Strength Cond Res 34(1): 20–25, 2020—Dry-land strength training is a common component of
swimming programs; however, its efficacy is contentious. A common criticism of dry-land strength training for swimming is a lack of
specificity. An understanding of movement patterns in swimming can enable dry-land strength training programs to be developed to
elicit adaptations that transfer to improvements in swimming performance. This study aimed to quantify the range and velocity of hip
roll, shoulder roll, and torso twist (produced by differences in the relative angle between shoulder roll and hip roll) in front crawl at
different swimming speeds. Longitudinal torso kinematics was compared between sprint and 400-m pace front crawl using 3D
kinematics of 13 elite Scottish front crawl specialists. The range (sprint: 78.1˚; 400 m: 61.3˚) and velocity of torso twist (sprint: 166.3˚·s2
1
; 400 m: 96.9˚·s21) were greater at sprint than 400-m pace. These differences were attributed to reductions in hip roll (sprint: 36.8˚;
400 m: 49.9˚) without corresponding reductions in shoulder roll (sprint: 97.7˚; 400 m: 101.6˚) when subjects swam faster. Shoulder roll
velocity (sprint: 190.9˚·s21; 400 m: 139.2˚·s21) and hip roll velocity (sprint: 75.5˚·s21; 400 m: 69.1˚·s21) were greater at sprint than 400-
m pace due to a higher stroke frequency at sprint pace (sprint: 0.95 strokes·s21; 400 m: 0.70 strokes·s21). These findings imply
that torques acting to rotate the upper torso and the lower torso are greater at sprint than 400-m pace. Dry-land strength training
specificity can be improved by designing exercises that challenge the torso muscles to reproduce the torques required to generate the
longitudinal kinematics in front crawl.
Key Words: torso twist, biomechanics, sprint, middle-distance, performance

Introduction with swimming speed. Furthermore, the influence of swimming

speed on the rate of change (or velocity) of hip roll, shoulder roll,
To maximize the probability that strength training adaptations will
and torso twist has never been reported. Considering the associa-
transfer to improvements in performance, training must be based on
tion between torso muscle activity and the magnitude and speed of
the demands of a sport (12,13). The lack of effectiveness of many dry-
twisting motions of the spine (16,19), differences in the range and
land strength training programs in improving swimming performance
velocity of torso twist in front crawl may influence the demands on
is often attributed to a lack of specificity in training (11,32,33).
the torso muscles. Our understanding of the torso muscle
Transference of strength training gains to performance can be en-
requirements in front crawl may therefore be limited by the lack of
hanced by designing exercises that match the demands associated with
the movement patterns used within a sport (37). Dry-land strength evidence of torso twist characteristics in front crawl swimming.
training specificity for swimming can therefore be improved with Total hip roll, a measurement of the range of hip roll from one
a better understanding of the movement patterns used in swimming. side to the other, tends to decrease as swimming speed increases,
Longitudinal body rotation is essential for maximizing perfor- whereas total shoulder roll, which is the range of shoulder roll
mance in front crawl swimming (6,17). Rotation of the shoulders from one side to the other, does not seem to change with in-
and hips about the body’s longitudinal axis, known, respectively, creasing speed as much as total hip roll. McCabe and Sanders (21)
as shoulder roll and hip roll, depends on swimming speed (27). reported a total hip roll of 57° at 1.50 m·s21 during a 400-m
Some characteristics of shoulder roll and hip roll remain consistent maximal effort, whereas Psycharakis and Sanders (26) reported
across different front crawl speeds; for example, the shoulders roll a total hip roll of 44° at 1.68 m·s21 in the first 50 m of a 200-m
through a greater range of motion than the hips regardless of maximal front crawl test. Psycharakis and McCabe (25) reported
swimming speed (3,35). The effect of swimming speed on several an even lower total hip roll of 39° at 1.81 m·s21 during a maximal
features of longitudinal rotation in front crawl, however, remains 25-m sprint. Despite the 18° difference in total hip roll between
unclear. For example, it is unknown how torso twist produced by 400-m pace and sprint front crawl swimming, total shoulder roll
differences in the relative angles of hip roll and shoulder roll varies remained between 105 and 111° across all 3 studies. Differences
in the range or timing of hip roll and shoulder roll require twist
Address correspondence to Jordan T. Andersen, [email protected]. within the torso. Data from the studies by McCabe and Sanders
Journal of Strength and Conditioning Research 34(1)/20–25 (21), Psycharakis and Sanders (26), and Psycharakis and McCabe
ª 2019 National Strength and Conditioning Association (25) indicate that the range of torso twist is likely to increase with

20

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.
 Shoulder and Hip Roll at Different Front Crawl Speeds (2020) 34:1 | www.nsca.com

swimming speed; however, differences in torso twist from the Subjects

same group of swimmers swimming at different front crawl
Three-dimensional coordinate data of a 15-segment whole-body
speeds have never been examined.
model of 13 national- and international-level male Scottish front
The time for the hips and shoulders to roll from one side to the
crawl specialists (mean 6 SD: age: 17.54 6 1.98 years, range
other and back again is determined by the duration of the arm
15–22 years; height: 181.18 6 4.98 cm; body mass: 71.58 6 6.26
stroke cycle (SC) (28,34). The velocities of hip roll and shoulder
kg) were analyzed from a data set that was previously used in the
roll are therefore influenced by the range of hip roll and shoulder
studies of McCabe et al. (20) and McCabe and Sanders (21).
roll, respectively, and the number of SCs per unit of time, or
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw

Subjects had specialized in front crawl for a minimum of 2 years,

stroke frequency. It is well documented that stroke frequency
were not currently injured or recovering from injury, and held
increases as swimming speed increases (4,7,29,30); however, the
a short-course personal best time of either less than 24.60 seconds
influence of swimming speed on hip roll velocity and shoulder roll
CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

for 50 m or less than 4 minutes 10 seconds for 400 m. The pro-

velocity is unknown. Changes in the relative angle between hip
tocols and procedures were approved by the University of Edin-
roll and shoulder roll and differences in stroke frequency across
burgh ethics committee. All subjects were informed of the risks
front crawl speeds suggest that torso twist velocity may also
and benefits of the study and provided written consent before
change with swimming speed, but torso twist velocity has yet to
data collection. For subjects aged 18 years younger, subjects and
be quantified in the scientific literature.
a parent or guardian provided written consent.
Although twist of the shoulders and hips relative to each other
is influenced by the torques produced by the actions of the upper
Procedures
and lower limbs, it may be hypothesized that the differences be-
tween shoulder and hip rotation, manifest in changing torso twist The data collection by McCabe was conducted in an indoor 25-m
angles, are also influenced by the actions of the torso muscles pool. Subjects were marked to enable identification of the fol-
connecting the shoulders and hips. Therefore, it is likely that lowing anatomical landmarks: the vertex of the head (on top of
differences in torso twist rates of change, that is, torso twist ve- the swim cap), the left and right: tip of the third distal phalanx of
locities, may reflect differences in demands on the torso muscles to the finger, wrist axis, elbow axis, shoulder axis, hip axis, knee
control posture and maintain stability of the swimmer’s torso. axis, ankle axis, lateral aspect of the fifth metatarsophalangeal
Furthermore, if there are differences in the relative magnitudes joint, and tip of first phalanx of the foot (big toe). After an in-
and velocities of shoulder and hip roll between paces, demands on dividualized warm-up, subjects swam 4 3 25 m at sprint pace and
the torso muscles are likely to differ between swimming speeds. one 400-m effort at a pace that would result in the fastest time
Therefore, insights into these demands may be gained by quan- possible. After each sprint trial, subjects swam back to the start
tifying the differences in the range and velocity of torso twist at position at recovery pace and rested in-water for 2 minutes before
different swimming paces. beginning the next trial. The order of swimming pace was ran-
Although the ranges of hip roll and shoulder roll at different domized, and subjects swam for at least 5 minutes to recover after
swimming speeds have been examined in separate studies, the completing the first pace and then exited the pool for an addi-
differences in the velocities of hip roll and shoulder roll between tional 10-minute rest before warming up again and completing
swimming speeds have never been reported. Moreover, the range the second pace.
and velocity of torso twist produced by differences in hip and As subjects swam through a calibration volume (4.5 m long,
shoulder roll at different front crawl speeds have never been ex- 1.0 m wide, and 1.5 in height) located 15.25 m from the starting
amined to the best of our knowledge. These gaps in swimming wall, and their motion was captured by 6 synchronized JVC
research present a barrier to understanding the movement pat- KY32 CCD cameras (4 below and 2 above the water surface) at
terns in front crawl swimming that can be used to improve the a frame rate of 50 Hz. Each trial began from a push start, and
specificity of dry-land strength training for swimmers. Therefore, subjects were required to not breathe as they swam through the
the purpose of this study was to quantify the range and velocity of calibration volume to avoid any effect of the breathing actions on
hip roll, shoulder roll, and torso twist in front crawl at different their swimming technique (25,30). Swimmers familiarized
swimming speeds. The differences in the longitudinal kinematics themselves with the breath-holding requirement during warm-up.
between speeds will further our understanding of the movement All subjects used a 6-beat flutter kick at both swimming paces.
patterns in front crawl swimming, which can be used to develop
insight into the demands on the torso muscles in front crawl Data Processing. One SC was defined as the moment the tip of the
swimming. third digit of one hand entered the water to the subsequent entry
of that digit on the same hand performed completely within the
calibrated space. At sprint pace, 1 SC was analyzed for each of the
Methods four 25-m trials. During the 400-m effort, 1 SC was recorded
from the first 25-m length of each 50-m lap. Stroke cycles from
Experimental Approach to the Problem
laps 2, 3, 4, and 5 during the 400-m effort were analyzed, totaling
This cross-sectional study of 3D kinematics enabled analysis of the 4 observations per swimmer at 400-m pace. These laps were se-
movement patterns of high level front crawl swimming for 2 different lected to align with previous findings that laps 1, 7, and 8 were
event distances (i.e., 50 and 400 m freestyle). National- and consistently different from laps 2–6 (21). Lap 6 was excluded to
international-level swimmers were recruited because of their ability further minimize the effect of fatigue on swimming technique.
to produce movement patterns that can provide insights into the Owing to marker occlusion during data collection that prevented
requirements for high-level swimming performance. Although ex- digitization of landmarks over several consecutive frames, 1 trial
perienced swimmers are known to reliably produce consistent from 1 subject at 400-m pace (P4) was discarded. Data were
swimming technique, multiple trials at both swimming paces were retained for all 4 trials at both paces from every other subject.
collected to account for individual variability inherent of human Three-dimensional reconstruction from manual digitization of
movement. the anatomical landmarks was conducted using the Ariel

21

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.
 Shoulder and Hip Roll at Different Front Crawl Speeds (2020) 34:1

Performance Analysis System (direct linear transformation algo- ability to compare and interpret differences between swimming
rithms from Abdel-Aziz and Karara (1)). Errors due to digitiza- paces by providing a range about the mean of each kinematic
tion for the variables used in the current study were considered variable in which the true mean was likely to fall for either pace.
small from digitization reliability tested in a previous study (20). To The Shapiro-Wilk test indicated that all variables were normally
prevent data loss during filtering, an additional 30 frames were distributed. A separate paired t-test was conducted to evaluate the
extrapolated by reflection. Fourier truncation was used to filter the differences in stroke frequency, range of hip roll, range of
position data of the body landmarks. This filtering strategy was shoulder roll, range of torso twist, average hip roll velocity, av-
deemed appropriate because the cyclic nature of movements in erage shoulder roll velocity, and average torso twist velocity be-
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw

front crawl swimming results in periodic data (2). Residual analysis tween sprint pace and 400-m pace. Effect sizes were determined
indicated that a 6-Hz cutoff was suitable to smooth the data. Stroke using Cohen’s d and interpreted with the following recom-
cycle length was then standardized to 201 points using a Fourier mendations: small 0.2, moderate 0.5, and large 0.8 (5). Post hoc
CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

transform and inverse transform so that each datum represented power analysis was conducted using open-source software
a half percentage of the SC (i.e., 0–100%). (G*Power 3.1) (9).
The filtered anatomical landmark data were entered into a be-
spoke MATLAB (Mathworks, Inc.) analysis program written by
the last author. The orthogonal external reference system was Results
defined by the horizontal X-axis pointing in the swimming di-
Intraclass correlations were high for stroke frequency (sprint
rection, the Y-axis pointing vertically up, and the horizontal Z-
pace: 0.93; 400-m pace: 0.98) range of hip roll (sprint pace: 0.90;
axis pointing to the swimmer’s right. Shoulder roll and hip roll
400-m pace: 0.93), range of shoulder roll (sprint pace: 0.85; 400-
were calculated independently for each percentile of the SC as the
m pace: 0.94), range of torso twist (sprint pace: 0.82; 400-m pace:
angle, expressed in degrees, between the Z-axis and vectors
0.91), average hip roll velocity (sprint pace: 0.90; 400-m pace:
connecting the shoulders and hips, respectively, projected onto
0.91), average shoulder roll velocity (sprint pace: 0.83; 400-m
the YZ plane.
pace: 0.96), and average torso twist velocity (sprint pace: 0.84;
400-m pace: 0.89) at both paces.
Data Analyses. The average swimming velocity, calculated by
Time series for ensemble averages of hip roll, shoulder roll, and
dividing the horizontal component of the center of mass dis-
torso twist are shown in Figure 1, and time series for ensemble
placement by SC time, was 1.81 6 0.06 m·s21 at sprint pace and
averages of hip roll velocity, shoulder roll velocity, torso twist
1.47 6 0.06 m·s21 at 400-m pace.
velocity are shown in Figure 2 for 1 SC at sprint and 400-m pace.
Stroke frequency was determined using the inverse of the time
Table 1 shows means, 95% confidence intervals using the t-
to complete 1 SC (stroke·s21). Torso twist was the difference in
distribution of the sample mean, effect sizes, and statistical power
the relative angles of shoulder roll and hip roll and was calculated
for comparisons between paces of stroke frequency, range of hip
for each percentile of the SC in degrees. Hip roll velocity, shoulder
roll, range of shoulder roll, range of torso twist, average hip roll
roll velocity, and torso twist velocity were the rate of change of
velocity, average shoulder roll velocity, and average torso twist
hip roll, shoulder roll, and torso twist, respectively, and were
velocity. Stroke frequency was greater at sprint pace than at 400-
expressed as angular velocities (in degrees per second) using the
m pace (t (12) 5 12.27, p , 0.01) with a large effect size. Range of
time derivatives of hip roll, shoulder roll, and torso twist with the
hip roll was greater at 400-m pace than at sprint pace (t (12) 5
central difference method.
6.77, p , 0.01) with a large effect size, whereas range of shoulder
Range of hip roll, range of shoulder roll, and range of torso
roll was similar between paces (p 5 0.14). Range of torso twist (t
twist was determined separately for each trial by summing the
(12) 5 6.88, p , 0.01), average shoulder roll velocity (t (12) 5
maximum magnitude of hip roll, shoulder roll, and torso twist,
9.17, p , 0.01), and average torso twist velocity (t (12) 5 12.30, p
respectively, to the left side and to the right side. Averages for hip
, 0.01) was greater at sprint pace than at 400-m pace with large
roll velocity, shoulder roll velocity, and torso twist velocity were
effect sizes. Average hip roll velocity was also greater at sprint
calculated using the mean of the absolute values of hip roll ve-
pace than at 400-m pace (t (12) 5 2.98, p , 0.05) but with
locity, shoulder roll velocity, and torso twist velocity, re-
a moderate effect size.
spectively, over each entire SC.

Discussion
Statistical Analyses
The purpose of this study was to quantify the range and velocity
Statistical tests were performed using IBM SPSS Statistics 24 (a 5 of hip roll, shoulder roll, and torso twist in front crawl at different
0.05), with the exception of effect sizes which were calculated swimming speeds. The differences in hip roll and shoulder roll
manually (10). Intraclass correlations between swimming trials that contributed to the changes in the range and velocity of torso
were determined using a single-rating, absolute agreement, 2-way twist between paces will further understanding of the demands on
mixed random-effects model analysis (14) for stroke frequency, the torso muscles in front crawl swimming. The findings from this
range of hip roll, range of shoulder roll, range of torso twist, study contribute to the knowledge of movement patterns in front
average hip roll velocity, average shoulder roll velocity, and av- crawl that can be used to improve the specificity of dry-land
erage torso twist velocity at sprint pace and 400-m pace. strength training for swimmers.
Means and 95% confidence intervals (i.e., the t-value for the The larger range of torso twist at sprint pace than at 400-m
sample size (n 5 13) multiplied by the standard error of the pace seemed to be the result of a reduction in hip roll without
sample mean) were calculated at both swimming paces for stroke a corresponding reduction in shoulder roll when subjects were
frequency, range of hip roll, range of shoulder roll, range of torso swimming faster. The range of hip roll and range of shoulder roll
twist, average hip roll velocity, average shoulder roll velocity, and observed in the current study are consistent with trends of total
average torso twist velocity. Confidence intervals improved our hip roll and total shoulder roll across different swimming speeds

22

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.
 Shoulder and Hip Roll at Different Front Crawl Speeds (2020) 34:1 | www.nsca.com
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw
CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

Figure 1. Time series with ensemble averages for hip roll, shoulder roll, and torso twist at sprint pace and 400-m pace. Dashed
lines represent 95% confidence intervals calculated using the t-value and standard error of the sample mean. Positive values
indicate rotation to the swimmer’s left (i.e., in the anticlockwise direction when viewing the swimmer from behind), and
negative values indicate rotation to the swimmer’s right (i.e., in the clockwise direction when viewing the swimmer from
behind). Swimmers began these SCs with the right hand. Time series for SCs beginning with the left hand were similar to this
figure. SC 5 stroke cycle.

from previous findings (26). The similar range of shoulder roll Furthermore, the difference in torso twist velocity between
between paces and the higher stroke frequency at sprint pace than swimming paces implies the swimmers in this study rotated their
at 400-m pace meant the swimmers rolled their shoulders faster as upper torso with respect to their lower torso more rapidly at
swimming speed increased. This was reflected in an average sprint pace than at 400-m pace. Increases in the magnitude and
shoulder roll velocity that was 37% greater at sprint pace than at speed of rotation between the upper and lower torso are associ-
400-m pace (Table 1). Despite the smaller range of hip roll at ated with higher torso muscle activity (18,19). These findings
sprint pace than at 400-m pace, the higher stroke frequency suggest that the demands on the torso muscles are likely to be
resulted in an increase in hip roll velocity as swimming speed higher at faster swimming speeds, but this cannot be stated with
increased; however, average hip roll velocity was only 9% greater confidence without further research measuring the muscle activity
at sprint pace than at 400-m pace. Moreover, the effect size of the at different paces.
difference in average hip roll velocity was moderate, whereas all Torques that produce rotation of the upper torso must have
other statistically significant differences between paces had large been higher at sprint pace than at 400-m pace for the swimmers to
effect sizes (Table 1). The difference in torso twist velocity be- achieve a similar range of shoulder roll at both paces, considering
tween swimming paces therefore seemed to be the result of the the increase in stroke frequency as swimming speed increased.
swimmers’ ability to maintain their range of shoulder roll, despite Hydrodynamic and buoyancy torques associated with the arm
an increase in stroke frequency, and to reduce their range of hip stroke produce longitudinal body rotation (23,34,36) and could
roll as they increased swimming speed. have contributed to the differences in shoulder roll velocity ob-
The patterns of hip roll, shoulder roll, and torso twist in served in the current study. Although the shoulders and hips roll
Figure 1 suggest the magnitude of rotation between the upper and somewhat independently in front crawl (28), longitudinal rota-
lower torso was greater at sprint pace than at 400-m pace. tion is likely transferred from the shoulders to the hips. For

23

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.
 Shoulder and Hip Roll at Different Front Crawl Speeds (2020) 34:1
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw
CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

Figure 2. Time series with ensemble averages for hip roll velocity, shoulder roll velocity, and torso twist velocity at sprint pace
and 400-m pace. Dashed lines represent 95% confidence intervals calculated using the t-value and standard error of the
sample mean. Positive values indicate rotation to the swimmer’s left (i.e., in the anticlockwise direction when viewing the
swimmer from behind), and negative values indicate rotation to the swimmer’s right (i.e., in the clockwise direction when
viewing the swimmer from behind). Swimmers began these SCs with the right hand. Time series for SCs beginning with the left
hand were similar to this figure. SC 5 stroke cycle.

example, motion can be transferred along the torso during to rotate the upper torso, may have therefore been required to
twisting motions of the spine through passive mechanisms (e.g., reduce the range of hip roll as swimming speed increased. Sanders
by using connective tissue and intervertebral discs) (15,16) or and Psycharakis (28), for instance, hypothesized that hip roll is
with the assistance of muscle torques (19,24,31). Greater torque “dampened” compared with shoulder roll from torques associ-
acting to rotate the lower torso, separate from the torques acting ated with the flutter kick. Considering swimmers tend to increase

Table 1
Stroke frequency, range of hip roll, range of shoulder roll, range of torso twist, average absolute hip roll velocity, average absolute
shoulder roll velocity, and average absolute torso twist velocity at sprint pace and 400-m pace.
Sprint pace 400-m pace
Mean 95% confidence interval Mean 95% confidence interval Effect size (Cohen’s d) Power (n 5 13)
Stroke frequency (stroke·s21) 0.95** 0.04 0.70 0.04 3.73 1.0
Range of hip roll ( ˚ ) 36.8** 3.1 49.9 5.6 21.58 1.0
Range of shoulder roll ( ˚ ) 97.7 3.1 101.6 5.9 20.46 0.40
Range of torso twist ( ˚ ) 78.1** 3.4 61.3 4.7 2.23 1.0
Average absolute hip roll velocity ( ˚·s21) 75.5* 7.1 69.1 7.9 0.52 0.66
Average absolute shoulder roll velocity ( ˚·s21) 190.7** 9.9 139.2 11.5 2.92 1.0
Average absolute torso twist velocity ( ˚·s21) 166.3** 10.0 96.9 8.2 4.13 1.0
Significantly different from 400-m pace (*p , 0.05, **p , 0.01).

24

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.
 Shoulder and Hip Roll at Different Front Crawl Speeds (2020) 34:1 | www.nsca.com

kicking frequency as swimming speed increases (4,8,22), torques 8. de Jesus K, Sanders RH, de Jesus K, et al. The effect of intensity on 3-
from the flutter kick acting on the lower torso may have been dimensional kinematics and coordination in front-crawl swimming. Int J
greater at sprint pace than at 400-m pace, which could have Sports Physiol Perform 11: 768–775, 2016.
9. Faul F, Erdfelder E, Lang AG, Buchner AG. *Power 3: A flexible statistical
contributed to the reduction in hip roll as swimming speed in- power analysis program for the social, behavioral, and biomedical scien-
creased. The differences in the longitudinal kinematics presented ces. Behav Res Methods 39: 175–191, 2007.
here indicate that the torques acting to rotate the upper torso and 10. Fritz CO, Morris PE, Richler JJ. Effect size estimates: Current use, cal-
the torques acting to rotate the lower torso may be greater at culations, and interpretation. J Exp Psychol Gen 141: 2–18, 2012.
sprint pace than at 400-m pace. This may also indicate that the 11. Girold S, Maurin D, Dugue B, Chatard JC, Millet G. Effects of dry-land vs.
Downloaded from http://journals.lww.com/nsca-jscr by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCyw

demands on the torso muscles increase as swimming speed resisted-and assisted-sprint exercises on swimming sprint performances.
J Strength Cond Res 21: 599, 2007.
increases. Quantification of the torques acting on the upper torso 12. González-Badillo JJ, Sánchez-Medina L. Movement velocity as a measure of
and lower torso in front crawl is required to test this hypothesis.
CX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 07/05/2024

loading intensity in resistance training. Int J Sports Med 31: 347–352, 2010.
13. Izquierdo M, Häkkinen K, Gonzalez-Badillo JJ, Ibanez J, Gorostiaga EM.
Effects of long-term training specificity on maximal strength and power of
Practical Applications the upper and lower extremities in athletes from different sports. Eur J
Appl Physiol 87: 264–271, 2002.
This is the first study to investigate the velocity of hip roll, 14. Koo T, Li M. A guideline of selecting and reporting intraclass correlation
shoulder roll, and torso twist in front crawl swimming. Coaches coefficients for reliability research. J Chiropr Med 15: 155–163; 2016.
15. Kumar S, Narayan Y, Garand D. An electromyographic study of iso-
can use these findings to guide recommendations for changes to
kinetic axial rotation in young adults. Spine J 3: 46–54, 2003.
swimming technique between sprint and middle-distance 16. Kumar S, Narayan Y, Zedka M. An electromyographic study of unresisted
swimming. For example, swimmers can be encouraged to trunk rotation with normal velocity among healthy subjects. Spine 21:
maintain their range of shoulder roll as stroke frequency 1500–1512, 1996.
increases with swimming speed. From the differences in the 17. Maglischo EW. Guidelines for increasing propulsion and reducing re-
range and velocity of torso twist between swimming paces, sistance. In: Swimming Fastest. Champaign, IL: Human Kinetics, 2003. pp.
80–83.
torques acting to produce rotation of the upper torso and the
18. Marras W, Davis K, Granata K. Trunk muscle activities during asym-
lower torso are likely to increase as swimming speed increases. metric twisting motions. J Electromyogr Kinesiol 8: 247–256, 1998.
Dry-land strength training specificity may be improved by de- 19. Marras WS, Granata KP. A biomechanical assessment and model of axial
signing exercises that challenge the torso muscles to generate twisting in the thoracolumbar spine. Spine 20: 1440–1451, 1995.
torques that produce or resist longitudinal rotation of the upper 20. McCabe CB, Psycharakis S, Sanders RH. Kinematic differences between
torso and the lower torso. Coaches are encouraged to consider front crawl sprint and distance swimmers at sprint pace. J Sports Sci 29:
115–123, 2011.
the differences in the demands placed on swimmers competing
21. McCabe CB, Sanders RH. Kinematic differences between front crawl sprint
over different distances when designing dry-land strength and distance swimmers at a distance pace. J Sports Sci 30: 601–608, 2012.
training. For instance, torques required from the torso muscles 22. Millet G, Chollet D, Chalies S, Chatard J. Coordination in front crawl in
may be greater at faster swimming speeds than at slower elite triathletes and elite swimmers. Int J Sports Med 23: 99–104, 2002.
swimming speeds. Acknowledgement of the differences in 23. Payton CJ, Bartlett RM, Baltzopoulos V, Coombs R. Upper extremity
demands between swimming speeds could increase the likeli- kinematics and body roll during preferred-side breathing and breath-
holding front crawl swimming. J Sports Sciences 17: 689–696, 1999.
hood that benefits from dry-land strength training will transfer
24. Pink M, Perry J, Jobe FW. Electromyographic analysis of the trunk in
to improvements in swimming performance. golfers. Am J Sports Med 21: 385–388, 1993.
25. Psycharakis S, McCabe C. Shoulder and hip roll differences between
breathing and non-breathing conditions in front crawl swimming.
Acknowledgments J Biomech 44: 1752–1756, 2011.
26. Psycharakis S, Sanders RH. Shoulder and hip roll changes during 200-m
The authors thank the swimmers and coaches for their front crawl swimming. Med Sci Sports Exerc 40: 2129–2136, 2008.
participation in data collection. There are no conflicts of interest 27. Psycharakis S, Sanders RH. Body roll in swimming: A review. J Sports Sci
to declare. 28: 229–236, 2010.
28. Sanders RH, Psycharakis SG. Rolling rhythms in front crawl swimming
References with six-beat kick. J Biomech 42: 273–279, 2009.
29. Schnitzler C, Seifert L, Ernwein V, Chollet D. Arm coordination adapta-
1. Abdel-Aziz Y, Karara H. Direct linear transformation from comparator tions assessment in swimming. Int J Sports Med 29: 480–486, 2008.
coordinates into object space coordinates in close-range photogrammetry. 30. Seifert L, Chollet D, Bardy B. Effect of swimming velocity on arm co-
Presented at Proceedings of the symposium on close-range photogram- ordination in the front crawl: A dynamic analysis. J Sports Sci 22:
metry, Falls Church, VA, 1971. 651–660, 2004.
2. Bartlett R. Quantitative analysis of movement. In: Introduction to Sports 31. Shaffer B, Jobe FW, Pink M, Perry J. Baseball batting: An electromyo-
Biomechanics: Analysing Human Movement Patterns. New York, NY:
graphic study. Clin Orthop Relat Res 292: 285–293, 1993.
Routledge, 2007. p. 136.
32. Tanaka H, Costill DL, Thomas R, Fink WJ, Widrick JJ. Dry-land re-
3. Cappaert JM, Pease DL, Troup JP. Three-dimensional analysis of the
sistance training for competitive swimming. Med Sci Sports Exerc 25:
men’s 100-m freestyle during the 1992 Olympic games. J Appl Biomech
11: 103–112, 1995. 952–959, 1993.
4. Chollet D, Chalies S, Chatard JC. A new index of coordination for the 33. Tanaka H, Swensen T. Impact of resistance training on endurance per-
crawl: Description and usefulness. Int J Sports Med 21: 54–59, 2000. formance. Sports Med 25: 191–200, 1998.
5. Cohen J. The concepts of power analysis. In: 2nd, ed. Statistical Power 34. Yanai T. What causes the body to roll in front-crawl swimming? J Appl
Analysis for the Behavioral Sciences. Hillsdale, MI: Erlbaum Associates, Biomech 17: 28–42, 2001.
1988. pp. 20–27. 35. Yanai T. Stroke frequency in front crawl: Its mechanical link to the fluid
6. Counsilman JE, Counsilman BE. The crawl stroke. In: The New Science of forces required in non-propulsive directions. J Biomech 36: 53–62, 2003.
Swimming. Englewood Cliffs, NJ: Prentice-Hall Inc., 1994. p. 58. 36. Yanai T. Buoyancy is the primary source of generating bodyroll in front-
7. Craig AB, Skehan PL, Pawelczyk JA, Boomer WL. Velocity, stroke rate, crawl swimming. J Biomech 37: 605–612, 2004.
and distance per stroke during elite swimming competition. Med Sci Sports 37. Young WB. Transfer of strength and power training to sports perfor-
Exerc 17: 625–634, 1985. mance. Int J Sports Physiol Perform 1: 74–83, 2006.

25

Copyright © 2019 National Strength and Conditioning Association. Unauthorized reproduction of this article is prohibited.