See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258967110

Phonetic symmetry in sound systems

Article in Symmetry Culture and Science · January 1993

CITATIONS READS

0 6,157

1 author:

Jan Tent
Australian National University
110 PUBLICATIONS 467 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jan Tent on 03 June 2014.

The user has requested enhancement of the downloaded file.

Symmetry:
- ~n~ o

The Quarterly of the Editors:

International Society for the Gy6rgy Darvas and Ddnes Nagy.
Interdisciplinary Study of Symmetry
(ISIS-Symmetry) Volume 4, Number 4, 1993

DLA fractal cluster

of 10~ particles
 Symmetry: Culture and Science
Vol 4, No. 4, 1993, 345-368

SYMMETRY." CUL TURE AND SCIENCE

PHONETIC SYMMETRY IN SOUND SYSTEMS

Jan Tent

Linguist, (b. Amsterdam~ The Netherlands, 1952).

Address: Department of Literature and Language, School of Humanities, The University of the South
Pacific, P.O. Box 1168, Suva, Fiji; E-mail: [email protected].
Fields of interest: Linguistics -- phonetics/phonology, English syntax, socio-linguisties, symmet~, in
language.

INTRODUCTION
The sounds of a language are traditionally - and perhaps most easily - described
and categorised in terms of how and where in the oral cavity they are articulated.
Consonants are described and categorised in terms of place (point) of articulation
and manner (method) of articulation. The former refers to the location in the oral
cavity where the obstruction or modification of the airstream occurs which
produces consonantal sounds, and the latter to the manner in which that
obstruction or modification of the airstream is made.
Based on this method of description, the consonant phonemesI of a language are
generally set out in the form of a chart or matrix according to general phonetic
taxonomic categories of place and manner of articulation. The various places of
articulation are represented as individual columns, whilst the various manners of
articulation are exemplified in rows2:

1 Aphoneme is defined as a minimal and contrastive unit of sound in the sound system of a language
which cannot be analysed into smaller linear units.
2 The phonetic symbols used in the charts below provide a unique written representation of each sound
(phoneme) independent of the orthographies of particular languages. Each phonetic symbol
corresponds exclusively to a particular vertical and horizontal position in a diagram.
In the vowel patterns on the following pages, the symbols used do not refer to any fixed phonetic
quality. The vowels of each language represented have their own distinct quality.
See also Figures 1, 2 and 3, Appendix.
346

bilabial alveolar velar

plosives (voiceless) 13 t k
(voiced) 13 cl g

nasals m n rj

fricatives (voiceless) qb s x
(voiced) 13 z ~’

Vowels, on the other hand, do not lend themselves to such a relatively easy method
of description and classification - at least not in articulatory terms. The
articulation of most consonants offers enough sensory feedback to determine the
place in theoral cavity where the obstruction or modification of the airstream
occurs. The articulation of vowels does not offer such sensory feedback. They are
articulated in a small and restricted area in the oral cavity, known as the ’vowel
space’. Very small vertical and horizontal movements of the tongue within this
space can differentiate one vowel quality from another. The vowel space is
generally represented as a quadrilateral (see Figures 2 and 3, Appendix), and the
relative position the tongue assumes for the articulation of individual vowels is
plotted within this quadrilateral.
The vowels of most languages tend to be evenly and widely distributed within the
vowel space. This configuration helps to provide for maximum phonetic contrast,
and as Disner (1980, p. 91) reports, about 86% of the 317 languages in the UPSID
(the UCLA Phonological Segment Inventory Database) have "vowel systems that are
built on a basic framework of evenly dispersed peripheral vowels" and that "10%
[of the languages] approach this specification".

PHONETIC SYMMETRY
Consonant charts and vowel quadrilaterals often reveal phonetic symmetries in a "
language’s inventory of phonemes3. Indeed, there is a strong tendency for symmetry
in the phonological inventory of most languages.
Phonetic symmetry refers to the occurrence of sounds in parallel series, so that the
sounds of one type, which occur at certain points of articulation, are parallelled by
sounds of another type at those same points of articulation. For example, if a
language has three voiceless plosives, three voiced plosives and three voiced nasals,
they would most likely be:

3 These inventories do not usually include allophonic variants of phonemes. Allophonic variants are
noticeable variations inthe [orm o[ a phoneme which do not affect the phoneme’s functional identity.
The variations are due to the influence of neighbouring sounds.
 PHONETIC SYMMETRY IN SOUND SYSTEMS 347

b d 0

The phonetic symmetry exhibited in such a language’s inventory of bilabial,

alveolar and velarplosives is parallelled in its inventory of nasals.
French and Spanish also have three voiceless plosives, three voiced plosives and
three voiced nasals, however, their inventory of these sound sets is asymmetrical:
p t k
b d g

The places of articulation of the plosives are not mirrored by the places of
articulation of the nasals. Instead of having a velar nasal, French and Spanish have
a palatal nasal.

SYMMETRICAL (NATURAL) SOUND SYSTEMS

1. Symmetrical Vowel Systems

The number of vowel phonemes that.any of the world’s known languages may have
ranges from a minimum of three to a maximum of approximately 20 - 24.
Languages with a three vowel system will almost always have some form of [i], [a]
and [u]. This makes sense phonetically because these three vowels provide for
maximum phonetic contrast since they have maximum dispersal and are articulated
towards the extremes (i.e., outer edges) of the vowel space. These extremes are:
[] the front of the tongue high in the oral cavity, close to the hard palate
(together with spread lips) which produces a close (or high) front vowel
like [i], as heard English heat, German wie, and French si.
[] the centre of the tongue low in the oral cavity (together with open lips)
which produces an open (or low) central vowel like [a], as heard in
Southern British English hard, German fahren, and French pas (short),
pale (long).
[] the back of the tongue high in the oral cavity, close to the soft palate
(together with rounded lips) which produces a close (or high) back vowel
like [u], as heard in English book, German Hund, and French tout.

Three vowel systems are quite uncommon since they limit the total number of
possible words a language may have. ~
348 . J. TENT

At the other end of the spectrum, !Xfi (a Khoisan language spoken in Botswana
with a total of 141 phonemes) has 24 vowel phonemes4. The larger the number of
vowels in a system, the smaller is the degree of dispersal. This stands to reason, as
the vowel space is such a restricted area, and the more divisions made within this
area the less widely distributed vowels will be. The end result is less acoustically
distinct vowels, and in theory at least a less efficient system.
The question arises as to What the optimum number of vowels in a system may be.
This ~s impossible to answer as the speakers of languages with a large number of
vowels have no trouble distinguishing them. The majority of the world’s languages,
however, have .between five and seven vowels, with the five vowel system being the
most common.
The following examples of symmetrical vowel systems show that the height of the
front vowels mirrors that of the back vowels independently of any central vowels
that may occur in the system. This type of symmetry is evident in most of the vowel
systems of the world’s languages.

3 Vowel @stemss
Quechua (Peru and Equador), Greenlandic Eskimo, Classical Arabic, Moroccan
Arable, Iatmul (Sepik, PNG), and a large number of Australian Aboriginal
languages.

4 Vowel @stem (relatively rare)

Arapaho* (U.S.A. Wyoming): i IJ
0

Squamish* (U.S.A. Washington):

4 This includes both monophthongs and diphthongs.

5 Phoneme inventories marked with * are adapted from Ruhlen (1976)~ Those marked with **. are
adapted from UPSID: UCLA Phonological Segment Inventory Database (1981).
 PHONETIC SYMMETRY IN SOUND SYSTEMS 349

5 Vowel @stems
(the most common vowel system approximately ¼ of the world’s languages)
Spanish, Modem Greek, Arabic, Latin, Czech, Mandarin, Japanese, Russian,
Polish, Basque, Malayalam, Telugu, Tlingit (SE Alaska), kiSwahili and some of the
other Bantu languages, and most of the languages of Oceania etc.

i U

6 Vowel @stems

Persian*; Chamorro** (Guam): i U

Malay*: i u

e e o
a

Lapp** (Lappland): i i u

~ 0

a
~50

7 Vowel Systems
Italian*: i u

Rumanian*: i i U

Ewe (Ghana)**: i u

8 Vowel Systems
Bahasa Indonesia*: i u
o
 PHONETIC SYMMETRY IN SOUND SYSTEMS 351

Burmese*:

9 Vowel System

Maasai** (SW Kenya): i u

t o

e o

12 Vowel System

English (British/Australian): i u

~ 0 o

~ D

a
¯ 352 J. TENT

Symmetrical Consonant Systems

Symmetry never occurs throughout a language’s entire consonant inventory, but

rather:
(a) within certain classes of consonants, e.g., symmetry within the plosive,
nasal or fricative inventories, and
(b)across certain classes of consonants, e.g.,_ where the places of
articulation in the plosive inventory are mirrored by those in the inventory
of nasals. This type of symmetry is a very common occurrence.

Every language has ’holes’ in i~s phoneme inventonj. These are due to:
(a) The general human inability to articulate certain speech sounds. For
instance, no language has bilabial, labio-dental or velar laterals since they
are impossible to articulate. All laterals are produced by allowing the
airstream to escape between the sides of the tongue’s blade (or dorsum)
and the alveolum (or hard palate). This cannot be achieved at the lips or
the soft palate, and hence, you will never see the space on a consonant
chart intersecting at "bilabial’ and ’lateral’ being occupied.
(b) A language simply not ’choosing’ to incorporate a particular sound or
class of sounds in its inventory. For instance, English (see below) has
’chosen’ not to include the palatal plosives/c, ~t/, or the palatal nasal/t~ / in
its inventory. Malay, on the other hand (see below), has ’chosen’ to d’-o so
together with all the other plosives and nasals English possesses.
(c) Genuine asymmetries in the system (see below).

A phonetic feature that often displays extensive sometimes almost perfect

symmetry is that of voice. This can be seen in a number of the plosive and fricative
inventories of the examples below.

English: p~ t~ k~
b d g

v~z3

I
~jw
354

Lapp*:

p: p:i t t t~ t: t:j i k ki k: k:i

b: b:i d d (~ d: d: g~ g: g:i

rnn~ m:m:in _n ni n: n:i ~1 rJi r.l: ~j:i

t~ t~i t:,
d~ d~i ]i

f fi f: f:i s si s: s:i J j~, j-: [:i. x xi x: x:i h I~ h: h:i

v vi v: v:i z zi z: z:i 3 5~ 5:5:~ "

I Ii l: l:i
r rj r: r:j

Arosi* (Solomon Islands): P pW t k k~’

b bw d g g"

m mw-

ASYMMETRICAL SYSTEMS
Asymmetrical phoneme inventories have ’holes’ or ’gaps’ in unexpected places. For
instance, if a language only has three plosives, they would most likely be/p t k/
(which are voiceless) or ib d g/(which are voiced). The three plosives in each set
articulated at the lips (i.e., bilabial plosives), the alveolum (i.e., alveolar plaosives)
and the velum (i.e., velar plosives). These three places of articulation provide for
maximum acoustic contrast which insures a maximum degree of differentiation
between them.
 PHONETIC SYMMETRY IN SOUND SYSTEMS

1. Asymmetrical Consonant Systems

Arapaho*: t k 7
b

Kiribati* (i.e. Gilbertese):

bbW

m m: mw rn:w n n:

Palauan* (Palau, Micronesia): t k

b d

Hawai’ian*:

m
,3~6

Rotokas** (Bougainville, Melanesia):

2. Asymmetrical Vowel Systems

Blackfoot & Cree* (U.S.A.):

Hopi** (U.S.A.): i

Malagasy** (Madagascar):

Khoi-Khoi* (i.e. Hottenlo0: i IJ

0
 PHONETIC SYMMETRY IN SOUND SYSTEMS 357

The vowel systems of virtually all languages have front vowels that are articulated
with unfounded lips and back vowels articulated with varying degrees of lip
rounding. In addition to this, some languages (e.g., Dutch, German, Swedish and
French) also possess a set of rounded front vowels and/or a set of unfounded back
vowels (e.g., Vietnamese). The vowel inventory of such languages is always
asymmetrical because the back vowels never have a set of corresponding
unfounded vowels (or in the case of languages with unrounded back vowels, the
front vowels don’t have a set of corresponding rounded front vowels)6:

Dutch*: iy u
t¥

eo o o

a a

German*: iy u
I. Y ~

eo o 0

French (Parisian)’*: iy u

e~ O 0

~ oe o

6 Note: The French nasalised ~owels/,~, ~’, ~, ~/are not included in this inventory.
358 J. TENT

Vietnamese*: U
LU

e 0 Y

O A

REASONS FOR SYMMETRY

The tendency for symmetrical phonological systems can be explained as a striving
for ’economy’, because symmetrical systems make the most economic use of
phonetic features. If phonemes (and the symbols used to represent them) are
regarded not as indivisible entities, but rather as bundles of distinctive phonetic
features, the repetition of the corresponding parameters in symmetrical phoneme
inventories can be explained.
Symmetrical systems have a simple set of rules that maximise the use of a limited
number of phonetic features by combining and reusing them to define an optimum
number of phonemes.
Some simple calculations on the consonant inventories of Lapp (see above) -
which is a symmetrical system - and Rotokas (see above) - which is a very
asymmetrical system - will instantly reveal the advantages a symmetrical system
has over an asymmetrical one.
The Lappish system is very economical as it employs only 18 phonetic features:
6 primary places of articulation: Bilabial [p, b, m]
Labio-denlal [t, v]
Alveolar [I, d, n, s, z, I, r]
Velar [k, g, .rl, x]
Palato-alveolar [~, i, .[, 3 ]
Glottal [hi
2 secondary places of articulation: Palatalised [i]
Half-palalalised [_]
 PHONETIC SYMMETRY IN SOUND SYSTEMS 3.59

6 manners of articulation: Plosive [p, b, t, d, k,

Nasal.. Ira,
Affricate ItS, dz~ ~, ]]
Fricative [l, v, $, z, j’, 3, x, hi
Lateral [I]
Trill [r]

2 values for voicing: Voice [b, d, g, m, n, ~j, d2, | v, z, 3, I, r]

voiceless [p, t, k, t’, ~, f, s, x, h]

2 values for length: Long [:]

¯: Nonrlong
to define a staggering 87 consonant phonemes.
Rotokas,. on the other hand, has a very Uneconomical phoneme inventory as!t
employs a total of 8 phonetic features:

3 places of articulation: Bilabial [p, [31

Alveolar It]
Velar JR, g]

3 manners of articulation: Plosive [p, t, k, g]

Fricative []3]
Tap [r]

2 values for voicing: Voiced [g, [3, r]

" Voiceless. [p, t, k]

to define a mere 6 consonant phonemes. This phoneme inventory is full of ’holes’,

as it imposes complicated rules to guarantee that only phonemes with admissible
combinations of phonetic features are defined, e.g., ’only the velar plosive, bilabial
fricative and alveolar tap can be voiced’; ’only the plosives can be voiceless’. No
such ad hoc restrictive statements are needed for a symmetrical system like Lapp.
There are some ’hoIes’ in the Lappish system too as no language has an entirely
symmetrical consonant system. However, a number of these ’holes’ can be
accounted for as they are entirelydue to the general human inability to articulate
such sounds. ¯
Because of their economic use of phonetic features, symmetrical phoneme systems
als0 reduce the burden of memorisation during language acquisition even when
such systems: have many more phonemes than asymmetrical systems (as in our
examples above). A limited number of phonetic features need only be learned since
they are constantly reused in defining phonemes grouped in natural classes (i.e.,
phonemes which are phonetically related). It is preferable to have a system with .a
limited number of phonetic features, and phonemes that form natural classes than
a system in which the phonemes have little in common with each other. The latter
36O 1. TENT

requires the mastering: of a large number of phonetic features relative to the

number of phonemes they are permitted to define in return.
The preference for symmetrical systems, can also be attributed to the tendency to
have the largest possible margin of safety between contrasting sounds so that they
can be acoustically/perceptually distinguished with maximum efficiency.
Asymmetry makes differentiation of contrasting sounds more difficult. Symmetry
insures maximum degrees of differentiation between sounds.

REASONS FOR ASYMMETRY

Sound change is perhaps the principal cause of asymmetry. Languages are living,
evolving, and dynamic systems, in each of which there is a constant interplay
between a set structure that displays symmetry, and asymmetry. Though symmetry
and maintaining a margin of safety between contrasting sounds may be a universal
tendency, sound change often leads to asymmetrical or undifferentiated systems.
That is, the tendencies do not provide restrictions to sound change, but may,
however, act in response to sound change, motivating further sound changes which
eliminate lack of symmetry. The following example of a changing five vowel system
illustrates this:

tJ

The loss of/e/from this system, as a consequence from its merging with/i/(/e/~/if)
would result in an imbalanced 4 vowel system. Thus,

In time, it would not be surprising to see/o/shift to/u/to match the change --

exhibited in the front vowels - which originally produced the imbalance (Crowley
1992 p. 201).
Sound changes occur over centuries. The diachronic study of language is only a
relatively new science - dating back to the mid 19th century. Linguists have
therefore not been able to study and trace sound changes for any appropriate
length of time. One other problem is that linguists do not precisely know what
causes a language to undergo a sound change in the first place.
 PHONETIC SYMMETRY IN SOUND SYSTEMS 361

CONCLUSION
There is always a danger that a discussion of symmetry in phonological systems will
be more concerned with patterns on paper than with genuine insights into the
phonetic or phonological nature of the system itself. Ruhlen (1976, p. 27) points
out that we need to recognise both attention towards symmetry and orthographical
convenience have significantly influenced the drafting of phoneme inventories. He
cites the example of French for which the nasalised vowels are seldom accurately
transcribed phoneticallyT: ~

are usually represented as:

or as:

Even though it may very be convenient to represent vowel systems as triangular or

squared arrays and consonant systems in matrices; it must be appreciated that while
these representations have some merit in displaying certain contrasts, they also
have some serious drawbacks. These include the following:
m They do not necessarily reflect, in any realistic way, the manner 0r place
of articulation. For instance, as a consequence of the ’asymmetry’ of the
oral cavity, the area in which back vowels are articulated is considerably
narrower than the corresponding front vowel area; the tongue has less
room for vertical movement in the back of the mouth than it has in the
front. Hence, the back vowels are more ’compressed’. Stylised vowel
diagrams do not reveal this in any realistic manner.
m The nature of the contrasts usually shown in the charts also do not
accurately reflect significant qualitative differences. For instance, the

7 It is interesting to note that Ruhlen’s example contains only three nasalised vowels when French
actually has four such vowels.
362

apparent phonologically equivalent 5 vowel systems of Spanish, Russian

and Japanese are quite different in their actual phonetic character. The
Japanese/u/is articulated with unrounded lips, whereas the Spanish and
Russian/u/are produced with the lips rounded.
"Finally, it is"important to: understand that the patterns and symmetries
often seen in these phoneme charts must, in part, be attributed to the
following three artefacts: .: ~
- The phonetic interpretations and transcription methods of the field
linguist will ultimately determine whether or not the phonemes of a
particular language are represented in sufficient phonetic detail, as well
as the final nature and configuration of the phoneme inventory (see
Footnote 7).
-The taxonomic principles used by a linguist in analysing and
categorising the sounds of a language will also have a direct bearing on
the composition and final shape Of the phoneme inventory.

-The manner in which the phoneme inventory is arranged will, of

course, affect the final configuration of the phonemes in the chart. One
arrangement may reveal symmetries that another may not.
For instance, a comparison of the phoneme inventories of Maori and
Hawai’ian (two closely related Polynesian languages) as supplied by
different linguists reveals the following:
Maori*: i u

e o

Maori**: u
 PHONETIC SYMMETRY IN SOUND SYSTEMS 363

Hawai’ian*: i u

e o

Hawai’ian** :

Although the two Maori inventories are asymmetrical, the phonetic detail
of each system is quite different. On the other hand, the two Hawai’ian
inventories differ in their phonetic judgements to such an extent that one
interpretation shows a symmetrical system whilst the other an
asymmetrical one.

A more striking example resulting from these artefacts can be seen in the
following inventories of Burmese, where the first interpretation reveals a
five vowel system whilst the second an eight vowel systerhS:
Burmese*:

e o

BHrrllese**~ i u

e o

a
8 These inventories do not include the three nasal vowels if, ~’,
 J. TENT
364

There is, therefore, a very real danger in over-emphasising the significance of the
search for symmetry and regularity in the phonological systems of languages.
Language, after all, is human behaviour, which is more often than not disorderly or
irregular in one way or another. A systematic analysis of language will always reveal
loose ends.
It must also be remembered that living languages are evolving, dynamic systems in
which there is a constant interplay between set structures which display symmetry
and asymmetry. This is clearly illustrated by phonological change, especially with
vowels.

APPENDIX

GLOSSARY

Place of Articulation

Bilabial Articulated with both lips.

Labio-dental Articulated with the bottom lip against the top teeth:
Interdental Articulated between the teeth.
Dental Articulated with the tip of the tongue on or near the top front
teeth.
Alveolar Articulated with the tip of the tongue on or near the alveolum
(i.e., the ridge just behind the top front teeth).
Aieveo-palatal Articulated with the blade of the tongue on or near the
alveolum and the hard palate.
Retroflex Articulated with the tip of the tongue on or near the hard
palate.
Palatal Articulated with the dorsum of the tongue on or near the hard
palate.
Velar Articulated with the back of the tongue on or near the velum
(i.e., the soft palate).
Labio-velar A simultaneous articulation involving the lips and the back of
the tongue on or near the velum.
Uvular Articulated with the back of the tongue on or near the uvula.
Pharyngeal Articulated with the root of the tongue drawn back towards the
back wall of the pharynx.
Glottal Articulated at the vocal folds.
 PHONETIC SYMMETRY IN SOUND SYSTEMS 365

Manner of Articulation

Plosive An articulation involving three stages: (a) complete closure of

the oral cavity blocking off the airstream, (b) compression of" ¯
the airstream behind the blocakage, (c) sudden release of the
blockage.
Nasal An articulation involving a complete blockage in the oral cavity
allowing the airstream to escape through the nasal cavity.
Fricative An articulation involving a narrowing in the oral cavity which
results in audible friction of the airstream.
Affricate A plosive with a fricative release.
Approximant An articulation involving a narrowing in the oral cavity without
resultant audible friction of the airstream.
Lateral An articulation in which the airstream is allowed to escape
along the sides of the tongue.
Flap/Tap An articulation in which the tongue makes one rapid ’tap’
against the alveolum or hard palate.
Trill An articulation involving the rapid ’tapping! or ’tilling’ of the
tongue against the alveolum, or of the uvula against the back of
the tongue.
Voiced/
Voiceless Voiced sounds are articulated with vibration of the vocal folds;
voiceless sounds are articulated without vibration of the vocal
folds, for example the z in zoo is voiced whilst the s in sue is
voiceless.
366 J. TENT

F~u~ 1
PHONETIC SYMMETRY IN SOUND SYSTEMS 367

Figure 2: Vowels -- Unfounded

Figure 3: Vowels -- Rounded

 J. TENT
368

REFERENCES

Crowley, T. (1992) An Introduction to Historical Linguistics, 2rid ed., Auckland: Oxford University Press.
Disner, S. F. (1980) Insights on vowel spacing, UCLA Working Papers in Phonetics, 50, August 1980,
70-92.
Hock, H. H. (1986) Principles of HistoricalLinguistics, Berlin: Mouton de Gruyter.
Ruhlen, M. (1976) A Guide to the Languages of the World, Language Universals Project, Stanford:
- Stanford University.
UPSID: UCLA Phonological Segment Inventory Database, Data and Index, UCLA Working Papers in
Phonetics, 53, November 1981.

View publication stats