View
1
Download
0
Category
Preview:
Citation preview
Toronto Working Papers in Linguistics (TWPL), Volume 38
©2017 Doug Hitch
Vowel spaces and systems* Doug Hitch Whitehorse, Yukon
Human language uses two kinds of vowel space: acoustic
defined by the F1 vs. F2 quadrilateral, and psychological
defined by spatial features (high, low, back, front) which
themselves are derived in relation to an often extralinguistic
neutral vowel. Lip rounding is not a primary feature of vowel
systems. Phonetically front rounded and back unrounded
vowels may serve in the psychological space as central
vowels. Spatial features define vocalic planes with between
two and nine vowels. The interplay between the asymmetrical
acoustic space and the symmetrical psychological space
produces the known variety in size and shape of vowel
systems among the world’s languages.
1 Introduction
A vowel may be described in several ways, through the physiology of articulation (articulatory
phonetics), through acoustic properties (acoustic phonetics), and through psychologically distinctive traits
(phonemics). In my view, for the description of vowel systems, only acoustics and psychology are relevant.
Beyond the facts that vowels have to be humanly produceable and hearable, physiology is not relevant to
the description of phonemic vowel systems. In spite of this, certain descriptive terms that are historically
based on articulation, high (close), low (open), back, front, and round are retained for consistency with
tradition and to simplify the presentation.
As is well-known, the number of possible shapes of vowel systems in the world’s languages is
relatively limited. Twenty-one shapes are presented here and, depending on the analysis, the number will
be a little lower or higher. The number and shape of distinct configurations is limited by the interplay of
two sets of parameters, acoustic and psychological.
1.1 Purpose
The analytical system presented here has both theoretical and descriptive applications. It helps us to
better understand how vowel systems work. It makes reference only to acoustic phonetics and not at all to
articulatory phonetics. Because of the strict reliance on acoustics, it bears directly on the debate concerning
the definition of distinctive features. If one component of the phonology, the vowels, can be successfully
defined using only acoustic parameters, this supports the view that all distinctive features are acoustic.
Several theoretical claims should be testable. For instance, it is claimed that no vocalic plane will feature
more than three degrees of phonemic height or depth. Language descriptions where four or more phonemic
contrasts on one axis have been claimed should be subject to reanalysis. Several such cases have been
addressed below. An unorthodox claim is that languages can have a vocalic plane containing front rounded
vowels like /y ø œ/. Some support is given here, but ideally one would like to find a range of languages
where vowels on such a plane behave as a natural class.
* This is a study I have been intending to write for a long time. The idea was originally inspired about 40 years ago
by “Typology and universals in vowel systems” by John Crothers (1978). In the intervening years I have not been a
student of phonological theory and the first draft was written in ignorance of some relevant scholarship. An
anonymous TWPL reviewer has greatly helped make this paper look less like it was written on a desert island. The
TWPL editor, Ruth Maddeaux, made significant contributions to the organization and presentation.
DOUG HITCH
2
This theory simplifies the analysis and description of particular languages. It enables a principle-
based typology of possible vowel systems. Every vowel system will conform to one of these patterns.
Currently, a typical description of a language contains an IPA-style vocalic trapezoid with the vowels
arranged according to acoustic phonetic properties. Alongside this, there should now be one or more square
vocalic planes representing psychological properties. Currently descriptions are both phonetic and
phonemic, but the use of planes may be a useful improvement.
2 Vowel system acoustics
The acoustic analysis of vowels and vowel systems depends on the intersection of the formants F1
and F2. Low vowels have a higher frequency F1 while high vowels have a lower F1. Back vowels have a
lower F2 frequency, while front vowels have higher F2 frequency. Figure 1 plots IPA front unrounded
vowels, back rounded vowels, and most low unrounded vowels, as pronounced by Bruce Hayes of UCLA.
Figure 1: F1 and F2 of some IPA vowels.1
The acoustic analysis of vowels seen in Figure 1 is now usually schematized as a quadrilateral (also
called trapezoid or trapezium). In Figure 2 a quadrilateral is superimposed on the chart of Figure 1.
1 http://www.linguistics.ucla.edu/people/hayes/103/Charts/VChart/, accessed 10 Jan 2017.
VOWEL SPACES AND SYSTEMS
3
Figure 2: Acoustic quadrilateral
The two-dimensional space defined by the quadrilateral is not abstract but physically real. It defines
the physical limits of vowel production for a particular speaker. Bruce Hayes will not normally pronounce
a vowel with F1 lower than [i] or [u], that is, above the top of the quadrilateral. Similarly, he will not in
normal speech produce a vowel with an F1 higher than those along the bottom of the quadrilateral. The
sides of the shape require reference to both formants. For instance, there will not be a vowel with the F1 of
[e] and a higher F2, or with the F1 of [o] and a lower F2. All vowels produced by this speaker will fall
within the acoustic space defined by the quadrilateral.
It is standard practice to use a quadrilateral with a right angle at the bottom right corner in the
presentation of vowel qualities as in the well known IPA trapezium of Figure 3.
Figure 3: IPA vowel trapezium2
While the shape of the quadrilateral may vary by presenter, every presentation shares similar aspects.
The left or front line is always longer than the right or back line, and the top or high line is always longer
than the bottom or low line which is the shortest of the four. The acoustic quadrilateral is asymmetrical.
2 https://www.internationalphoneticassociation.org/content/ipa-vowels, accessed 10 Jan 2017.
DOUG HITCH
4
2.1 Vowel system psychology
In addition to describing the vowel space in terms of acoustic properties, F1 vs. F2, within individual
languages it is useful to describe the space in terms of psychologically distinctive properties. These
properties do ultimately rest on acoustics but they are distinguished by the mind and organized by the mind
into binary oppositions according to certain principles.
The first principle is that of the phonetically neutral vowel. It is assumed that all speakers can produce
and recognize a neutral vowel. This is a vowel made when the articulatory apparatus is at rest except for
the vibration of the glottis. It is the vowel heard when a speaker, uncertain of what to say, says uh [ə],
sometimes in a long, drawn-out fashion, or um [əm]. It is common in extra-linguistic expressions, often
with a non-spatial phonetic feature added, like nasalization in English unh-unh ‘no’ or creaky voice in ugh
‘that’s disgusting’. Speakers of languages with no phonemic /ə/ may still produce phonetic [ə] and
recognize it in extra-linguistic situations.
A second principle is that all phonemic vowels are distinguished in relation to the phonetically neutral
vowel. A prototypically low vowel will have a higher F1 than the neutral vowel; a high vowel will have a
lower F1. Similarly, a front vowel will have a higher F2 and a back vowel a lower F2.
A useful and appropriate way to visualize the psychological or psycho-acoustic space created by these
principles is by comparing the geometry of geographic directions as illustrated in the map in Figure 4.
Figure 4: Directions from the starting point
NW N NE
W here E
SW S SE
By convention, the direction north is at the top of the map. The entire top edge of the map represents
its northern region. Similarly, there are eastern, western and southern regions. All regions are defined in
relation to the neutral starting point which is given as ‘here’ in Figure 4. Two directions overlap at each of
the four corners, giving northwest (NW), northeast (NE), southwest (SW), and southeast (SE) regions or
corners. The regions along the edges of the map and between the corners are defined by a single direction,
north (N), south (S), east (E) and west (W). This arrangement creates nine regions; one neutral region and
eight regions defined in relation to the neutral region.
The psychological vowel geography is essentially identical, as illustrated in Figure 5.
Figure 5: The nine vowel regions
high
front high
high
back
front ə back
low
front low
low
back
This principle-derived vowel space geometry has useful implications. One is that no language will
exhibit more than three degrees of height or depth as defined purely in spatial terms. It implies that
descriptions of languages which assume more than three degrees of phonemic height or backness are
somehow defective. Ladefoged and Maddieson (1996: 289) mention Danish as a candidate for exhibiting
four degrees of height. But other analyses of this language are possible. For instance, Grønnum (1998: 101)
VOWEL SPACES AND SYSTEMS
5
lists ten vowel phonemes, “/i e ɛ a y ø œ u o ɔ/”. If the front rounded vowels /y ø œ/ occupy a secondary
plane (see §2.4 below), then there are seven vowels defined purely with spatial features on the primary
plane as shown in Figure 6.
Figure 6: Danish phonemic vowel space
i u
e o
ɛ a ɔ
Ladefoged and Maddieson (1996: 289–290) also mention that the dialect of Bavarian spoken in
Amstetten, Austria may have five vowel heights. It is perhaps surprising that this typologically unusual
vowel system has not attracted more attention. The phonemic complexity may be reduced with a close
reading of Traunmüller’s (1983) description. For instance, the four front rounded vowels [y ø œ ɶ] arise
through l-vocalization and do not occur in underlying forms (Traunmüller, 1983: 16). He also notes that the
dialect area features a kind of “dynamic diglossia” in which there are several socially stratified forms of
language in use and speakers are not rigidly attached to any particular one. For instance, “Loans from higher
or lower ranked sociolects fulfill the function of expressing respect or disdain towards somebody or
something” (Traunmüller, 1983: 15–16). It is plausible that the extensive code-switching would make it
difficult to establish a phonemic inventory for a particular lect.
Descriptions of languages with more than three heights are not uncommon, but the position here is
that alternative analyses are possible with a maximum of three phonemic heights on a single plane.
Similarly, there will be no languages with four or more phonemic degrees of depth in a single plane.
Liljencrants and Lindblom (1972: 846–852) refer to a number of languages which had been reported to
feature more than three degrees of phonemic depth. Many of these are from the Turkic, Mongolian and
Uralic families which, probably areally, feature front-back or palatal-velar vowel harmony which here
involves parallel harmonic planes. A few feature a plane for front rounded vowels like Danish (see above
and §2.4 below), and the rest do not have four or more depths in more recent analyses. Specifically, by the
Liljencrants and Lindblom vowel system number:
1. (5d) Chacobo (Chakobo~Chácobo-Pakawara; Panoan, Bolivia) has /i ɨ o a/ (Prost, 1967: 285).
2. (6f) Mandarin (Modern Standard Chinese) has a vowel system which is debated but no scholar
currently argues for four depths (cf. §3.1 below).
3. (7h) “central Chinese Siang-tang dialect” refers to a presumably outdated 1938 study
(Trubetzkoy, 1939: 11 fn. 4).
4. (8a) Turkish, Balkar, Chuvash, Tatar, Kirghiz (all Turkic).
5. (8c) Cheremis (now Mari, Uralic, Russia), Khalka (Mongolian).
6. (9h) Akha (Loloish, Laos-Burma-Thailand-China) has a 9-vowel primary plane /i ɯ u e ɤ o ɛ a
ɔ/ and a secondary plane with /y œ/ (Lewis, 1968: 10; Hayashi, 2016: 75 with /ø/ for /œ/).
7. (9i) Swedish long vowels constitute a seven-vowel primary plane /iː ʉː uː eː oː ɛː ɒː/ and a
secondary /yː øː/ plane (Bruce & Engstrand, 2006: 20; see §2.4 below).
8. (10a) Koibal (Khakas Turkic), Karagin (if Koryak then /i u e̞ o̞ a̟/ U2709); Korean Seoul /i ɨ u
e~ɛ o a ʌ/, Kyungsang /i ɨ u e o a/ (Lee & Jongman, 2016: 158–160).
9. (11c) Selkup (Samoyedic, Siberia) has contrasts of length and tenseness; the short tense vowels
feature primary /i ɨ u e ə o æ a/ and secondary /y ø/ (Helimski, 1998: 552–553).
10. (12a) Tibetan, dialect unspecified but Central Tibetan has /i y u e ø o ɑ/ (R2125).
DOUG HITCH
6
11. (12b) Akha (see 9h).
I notice only one proposal for a four-way depth contrast that does not have an obvious alternative
analysis. Kato analyzes two Pwo Karen dialects from Burma as having four high vowels /i ɨ ɯ u/:
Kyonbyaw (Kato 1995: 69) and Tavoy (Kato 1995: 84). Kyonbyaw has a three-way contrast among the
mid and low vowels /e ɣ o ɛ a ɔ/ while Tavoy has a three-way mid contrast and one low vowel /e ɣ o a/.
Phonetically back /ɤ/ is the only vowel which occurs in atonic syllables (73, 88) which suggests it functions
like schwa, that is, it is phonemically non-back. These three-way contrasts make a four-way high contrast
look suspicious. Kato’s descriptions are thorough and I see no way to reduce the contrast to the three-way
theoretical maximum for one plane.
The idea that human language maximally distinguishes nine vowels has been presaged by previous
scholarship on differing grounds. Crothers (1978: 113) notes that “Languages with more than nine basic
vowel qualities are quite uncommon”. The study by Schwartz et al. (1997: 251) “focuses on ‘structural’
trends of the occupation of the acoustic space by vowels” and they observe that “primary systems mainly
consist of three to nine vowels”.3
2.2 Psycho-acoustic planes
Latin is often described as having five short /ɪ ʊ ɛ ɔ a/ and five long /iː uː eː oː aː/ vowels. The two
sets are precisely parallel from a phonemic point of view and are sometimes described as /i u e o a/ and /iː
uː eː oː aː/, using the same five symbols with one set marked for length. Both units of the pair /i iː/ occupy
the same position in the phonemic square relative to the neutral vowel. They differ in a non-spatial feature,
in this case length. Latin may be described as having two vocalic planes, one for short vowels and one for
long. Within a plane vowels are defined only by spatial features based on F1 and F2.
Visualizing vowel systems as a series of planes is not a new idea. For instance, Crothers (1978) refers
to a “basic” set which is essentially a primary plane. Schwartz et al. (1997: 237–238) refer to a “basis” or
“primary” system and “parallel” or “secondary” subsystems distinguished by quantity or nasality. A simple
but effective three-dimensional chart is presented by Mukherjee et al. (2008: 373 Fig. 1). This features a
primary plane P with plain /i u a/, a secondary plane P′ with nasalized /ĩ ũ ã/, and a tertiary plane P″ with
long nasalized /ĩː ũː ãː/. A fourth plane for implied long /iː uː aː/ is not included, presumably because it
would complicate the diagram.
A wide array of features can distinguish the secondary spatial vocalic planes. Ladefoged and
Maddieson (1996: 98 and Table 9.8) list a dozen non-spatial vocalic features: nasalization, advanced tongue
root (ATR), pharyngealization, stridency, rhotacization, fricativity, voicelessness, breathy voice, slack
vocal cords, stiff vocal cords, creaky voice, and length. Other scholars might add distinctions to the list,
perhaps most notably, tense/lax. Ahi (Tibeto-Burman, Yunnan) is described as having constricted and
unconstricted /i ɯ u e o ɛ ɑ/ (TB4:8). The secondary vocalic plane may exactly parallel the primary one, or
feature a different number of vowels in a different array. Tonkawa (isolate, Oklahoma-Texas-New Mexico)
has short /i u e o a/ and long /iː uː eː oː aː/ (U6752). In the Khoisan languages only the back vowels /u o a/
have pharyngealized counterparts (Ladefoged & Maddieson, 1996: 308). These will constitute a secondary
plane /uˁ oˁ aˁ/ forming a vertical three-vowel system congruent to Danish /y ø œ/. Jalapa Mazatec may have
six parallel planes. Alongside primary /i u æ o a/ there are spatially identical planes with breathy voice,
creaky voice, nasalization, breathy voice with nasalization, and creaky voice with nasalization (Silverman
et al., 1995: 72–73).
3 In this connection Schwartz et al. (1997: 242) mention an earlier work which I have not seen: “We have ourselves
suggested (Vallée, 1994) that the highest limit for a viable primary system is nine”. [Vallée, N. (1994) Systèmes
vocaliques: de la typologie aux prédictions. Thèse de Doctorat en Sciences du language, Université Stendhal,
Grenoble.] 4 Bibliographic abbreviation (TB=Namkung 1996, R=Ruhlen 2004, U=UPSID, W=Wikipedia) are defined in §3
below.
VOWEL SPACES AND SYSTEMS
7
2.3 Harmonic planes
The processes subsumed under the heading of vowel harmony seem to be of two types. The simplest
is a kind of assimilation among vowels on one plane. One assimilatory example is /a/ > /i/ in Jingulu (NC
Australia), a three-vowel /i u a/ language. A suffix with /i/ or /u/ will raise preceding /a/ vowels in the root
to /i/ unless a high vowel intervenes, for example, bardarda ‘younger brother’ + -rni > birdirdirni ‘younger
sister’, ngamurla ‘big’ + -rni > ngamurlirni ‘big (fem)’ (Pensalfini, 2002: 562–563). Another assimilatory
example showing a contrasting process, a root vowel influencing a suffix vowel, may be seen in Wikchamni
where a suffixal /i/ will assimilate to stem /u/ or /ü/: tan- + -ši > tanši ‘made’, hud- + -ši > hudšu ‘knew’,
düʔs- + -ši > düʔüšü ‘made’ (Archangeli, 1985: 354). A last example, Chamorro (Malayo-Polynesian,
Guam) with /i u e o æ a/ features fronting harmony in which stressed /u>i/, /o>e/ and /a>æ/ in a root when
preceded by a prefix or bound morpheme containing a front vowel: sæn- ‘direction of’ + hulo’ ‘up’ > sæn-
hilo’ ‘in the direction up’; mi- ‘lots of’ + otdot ‘ant’ > mi-etdot ‘lots of ants’; si definite, focus-marking
personal article + nana ‘mother’ > si næna ‘Mother’ (Topping, 1968: 68–69).
The second type of vowel harmony involves two harmonic planes. In these languages, the vowels in
a particular word will all be from one plane or the other. It seems that the choice of plane is determined at
the lexical level. In the lexicon, the words contain archiphonemes spanning both planes, and each word is
marked with a feature indicating plane membership. There appear to be two distinct sub-types with regard
to the phonetic vowel space; one vertical and one horizontal. The vertical type is associated with the
articulatory feature ATR while the horizontal type is often described as palatal-velar or front-back. The
vertical type has separate +ATR and -ATR planes while the horizontal has separate front and back vocalic
planes. Ideally, the articulatory terms here would be replaced by acoustic definitions.
ATR harmony is most common in Africa. Ogbia (Atlantic-Congo, Nigeria) has ten vowels, five
+ATR /i e u o ɐ/ and five -ATR /ɪ ʊ ɛ ɔ a/ (U4124). Lugbara (Central Sudanic, Uganda-DR Congo) similarly
has +ATR /i u e o ɜ/ and -ATR /ɪ ʊ ɛ ɔ a/ sets (U4215). Both languages could be more abstractly described
as having a five-vowel /I U E O A/ archiphonemic set with phonemic realization dependent on the ATR
marking of the word. Nez Perce (Sahaptian, NW USA), has the surface vowels /i u o æ ɑ/. A word may
contain vowels from either /i u æ/ or /i o ɑ/. Hall and Hall (1980: 212–217) show that this is ATR vowel
harmony and assume an underlying -ATR vowel which merges with /i/ on the surface. Following
Mackenzie and Dresher (2003: 285), the underlying vowel is here characterized as /ɛ/, and there are two
three-vowel harmonic planes, +ATR /i u æ/ and -ATR /ɛ o ɑ/. Again, archiphonemic /I U A/ is realized
according to the ATR marking of the root.
Horizontal vocalic harmony seems to mostly occur in languages which also feature rounding
harmony which involves a more complicated planar structure. But several Finnic languages feature just
front and back planes. Because of historical change and the rise of ‘neutral’ vowels, the planar structure has
been somewhat obscured. Finnish has three harmonic classes: front /y ø æ/, back /u o ɑ/ and neutral /i e/. A
word (uncompounded) contains vowels which are either front and neutral or back and neutral. Back and
front cannot co-occur (Suomi et al., 2008: 51). Karelian (Finnic, Russia-Finland) has exactly the same
inventory and harmonic pattern with neutral /i e/ (Harrison, 2001). Võro (S Estonia) harmony is similar but
because it has unrounded back mid /ɤ/, front /e/ is not neutral. The classes are front /y e ø æ/, back /u ɤ o
ɑ/, and neutral /i/ (W). Votic (Russia) has the identical harmonic pattern and neutral vowel, orthographically
<ü e ö ä>, <u õ o a> and <i> (Chernjavskij, 2005: 10). The neutral vowels are reminiscent of Nez Perce /i/
(above) and here we may also assume underlying back vowels that merge into the front correspondents on
the surface. Also, like Nez Perce, the underlying vowel(s) may have existed historically. Figure 7 compares
the Proto-Finnic5 harmonic planes with those in Võro, Votic, Finnish and Karelian. The underlying vowels
are in curly brackets {}.
5 The Proto-Finnic inventory appears to be traditionally constructed as identical to modern Finnish with ‘neutral’ /i
e/ (cf. Laakso, 2001: 182–183). However, Proto-Ugric may have had two high unrounded vowels /i/ and /ɨ/
(symbolized as i̮) and back unrounded mid /ɤ/ (e̮) is thought of as appearing in the southern Finnic languages after
DOUG HITCH
8
Figure 7: Some Finnic harmonic planes
Proto-Finnic Võro and Votic Finnish and Karelian front back front back front back
i y ɨ u i y {ɨ} u i y {ɨ} u
e ø ɤ o e ø ɤ o e ø {ɤ} o
æ ɑ æ ɑ æ ɑ
Like the African ATR languages Ogbia and Lugbara discussed above, these four Finnic languages
could have their vowels described archiphonemically as /I U E O A/, but in these cases the words would be
marked for frontness or backness rather than for ATR.
Turkish and many other Turkic languages also feature front-back harmony. Turkish harmonically
contrasts the set /ɪ ʏ ɛ œ/ with the set /ɯ u ɑ o/ (R1935, W), orthographically <i ü e ö> and <ı u a o>. Native
Turkish words have vowels only from one set or the other. Within a stem, the vowels may be described
archiphonemically as /I U A O/, with frontness or backness indicated at the lexical level. In addition,
Turkish has two types of harmony in suffixes. The two-fold type involves suffixes with the archiphoneme
/A/ which manifests on the surface as ‘e’ /ɛ/ or ‘a’ /ɑ/ according to the harmonic plane of the stem. For
example, the plural suffix -lAr appears as either -ler or -lar: kediler ‘cats’, köpekler ‘dogs’, kapılar ‘doors’,
odalar ‘rooms’, koyunlar ‘sheep (pl.)’, köyler ‘villages’. To this point in the description the harmony is
congruent with that of the Finnic languages discussed above. The second type of harmony involving
suffixes, the four-fold, appears to be of the assimilatory type. An archiphoneme /I/ appears on the surface
as /ɪ~ɯ~ʏ~u/ <i~ı~ü~u>. It follows the harmonic plane for backness, but it also assimilates in rounding to
the preceding vowel. For instance, the deadjectival noun suffix -lIk appears as -lik, -lük, -lık or -luk: özellik
‘speciality’, üzgürlük ‘freedom’, kıtlık ‘famine’, uzunluk ‘length’.
A possible third type of planar vowel harmony may be exhibited by Somali. There seem to be two
horizontally related planes, one containing /i ʉ e ø æ/ and the other /ɪ u ɛ o a/. Within a word the vowels are
from one plane or the other. The chief difference between the harmonic pairs /i~ɪ/, /ʉ~u/, /e~ɛ/, /ø~o/, and
/æ~a/ is F2 or backness. While the harmony is frequently described as ATR, this is different from the ATR
harmony described above in which the chief difference between pairs is F1 or height.6 It is an open question
how to properly characterize the harmony, but archiphonemically the planes have a classic five-vowel
system /I U E O A/.
2.4 Rounding and psycho-acoustic planes
The articulatory feature of lip rounding is often listed as a primary feature of vowel systems, along
with height and depth (cf. Ladefoged & Maddieson, 1996: 282). In the analysis presented here, rounding is
not relevant to the description of the primary vocalic plane of a language. For example, in the extremely
common five-vowel system of the type /i u e o a/, the segments /u o/ are phonetically but not phonemically
round. The vowels are psychologically defined by the spatial criteria illustrated above. As is well known,
the rounding in /u o/ serves to improve the perceptual distinctiveness of these vowels (cf. Hall, 2001: 16–
21). In a perceptual sense, it increases their distance from the neutral vowel. It phonetically enhances a
phonemic contrast. It is not necessary to include rounding as a distinctive feature in the phonemic
description of an /i u e o a/ system. Alekano (Goroka, Papua New Guinea) has a five-vowel system /i ɯ e
ɤ ɑ/ with no rounded vowels (Deibler, 1992: 2) but which may be defined by the same distinctive features
as the common /i u e o a/. Similarly, Alawa (Arnhem (?), N Australia) has a four-vowel system /ɪ ɯ̞ e a/
Proto-Finnic (ibid. 183). The inventory used here offers a simpler, more systematic explanation for the development
of the later systems and the harmonically neutral surface vowels. 6 Also previously pointed out by the unnamed scholar responsible for
http://www.ling.upenn.edu/courses/Spring_1998/ling202/atr.html.
VOWEL SPACES AND SYSTEMS
9
(U8354) with no rounding but may have the same distinctive features as /i u̞ e a/. The Tillamook (Salish,
NW Oregon) system /i ə æ ɑ/ also has no rounded vowels (Thompson & Thompson, 1966: 318; and see
§3.5.1 below). The presence of phonetic rounding among vowels does not imply the presence of phonemic
rounding.
Rounding can have a second phonetic enhancement function. This appears to be related to the first.
Some languages feature phonetically front rounded vowels as phonemically central vowels. Most
commonly, /y ø/ correspond to what in most languages would be /ɨ ə/. Three examples of this phenomenon,
Mbiywom from Australia, Albanian from central Europe, and Chechen from the Caucasus, are given in
Figure 8.
Figure 8: Rounding as perceptual centering of front vowels
Mbiywom Albanian Chechen
i y u i y u i y u
e o ɛ ə ɔ e ø o
a a æ a
In congruent fashion, phonetically back unrounded vowels can function as central vowels. The
examples in Figure 9 are Mbabaram from Australia, Fe’fe’ from Cameroon, and Lao from SE Asia.
Figure 9: Unrounding as perceptual centering of back vowels
Mbabaram Fe’fe’ Lao
i ɯ u i ɯ u i ɯ u
e ɤ o e ɤ o
a a̘ ɑ ɛ a ɔ
Whether one finds rounded /y ø/ or unrounded /ɯ ɤ/ functioning as central vowels in the
psychological space of those languages just discussed, it is not necessary to include rounding as a distinctive
feature in their descriptions. This claim may be surprising in the case of Chechen, which has more
phonetically rounded vowels than unrounded ones.
Rounding does appear to be phonemically necessary in a couple of circumstances. It may be
necessary in languages with front-back harmonic planes such as the Finnic languages and Turkish discussed
above. There it would appear that the backness of a vowel is defined by the plane at the lexical level. That
is, the distinctive features related to F2 are attached to the word. As backness is a feature of plane
membership, it likely cannot be used to distinguish vowels within a plane. Within a plane the distinctive
features are those related to F1, height, and to the acoustic correlate of rounding. The four-fold harmony of
Turkish suffixes <i~ü~ı~u> may confirm this. Rounding is available for the suffixes as a distinctive feature
since it is necessary to distinguish vowels within a stem.
The second circumstance in which rounding is phonemically necessary is with the unorthodox
concept of front round vocalic planes. As mentioned above, Danish /y ø œ/ are here regarded as being on a
separate plane from the other seven vowels which are defined purely on spatial grounds. On the secondary,
rounded plane, /y ø œ/ constitute a vertical three-vowel system defined by height. Three of the languages
listed by Liljencrantz and Lindblom (1972) as having more than three horizontal vowels, Akha, Selkup,
and Swedish, are here regarded as featuring front round vocalic planes. The concept of a front round vocalic
plane may be intuitively less appealing than that of a plane of nasalized or long vowels. There are some
indications that front roundedness may constitute a feature which is just as orthogonal to the primary plane
as those other features. There is a special association between F3 and front rounded vowels. As Crothers
DOUG HITCH
10
(1978: 96–97) puts it, “Lip rounding lowers all formants, primarily F3 for palatal vowels, F2 for velar
vowels”. Languages with front round vowels often have an acoustic vowel space which is much more
crowded in the front area. Grønnum (1998: 100), in her ‘Illustration of the IPA’ for Danish, has ten
allophones in the front area of the acoustic chart and five in the back. All varieties of Swedish crowd at
least four distinctive vowels into the acoustic high front region /iː ɪ yː ʏ/ and two at high back /uː ʊ/ (Bruce
& Engstrand, 2006: 20). The view here is that the front acoustic crowding in these languages reflects the
intersection of four planes: short, long, front-round-short and front-round-long. Central Standard Swedish
and some other varieties actually crowd five vowels in this space, as shown in Figure 10.
Figure 10: The vowel phonemes of Central Standard Swedish7
The fifth vowel, /ʉ̟ː/, has puzzled describers since, like /yː/ it is phonetically front, round, and high.
It is well known that these vowels feature different kinds of rounding and here may lie key evidence. /yː/
has outrounding or protrusion while /ʉ̟ː/ has inrounding or compression (Bruce & Engstrom, 2006: 22).
Acoustically then, there must be some quality which distinguishes these vowels, and phonemically there
must be something at play other than front, high and round. There is further phonological evidence. /ʉ̟ː/
patterns with other non-front vowels in some processes. It does not turn a preceding velar stop into a
fricative, for example, kyla [ˈɕyːla] ‘cold’ and kula [ˈkʉ̟ːla] ‘ball; bullet’, and it takes part in umlaut, for
example, hus [hʉ̟ːs] ‘house’ vs. hysa [ˈhyːsa] ‘accommodate’ (ibid.). Bruce & Engstrom (2006: 20) treat
/ʉ̟ː/ phonologically as a central vowel in the same way that the languages in Figure 8 above have
phonetically front rounded vowels as phonemically central vowels. Swedish may be described as having
four planes as shown in Figure 11.
Figure 11: The four vowel planes of Swedish
short long
ɪ ʊ ʏ iː ʉ̟ː uː yː
ɛ ɵ ɔ œ eː oː øː
a ɛː ɑː
7 Chart redrawn from Engstrand (1999: 140).
VOWEL SPACES AND SYSTEMS
11
The concept of front rounded vocalic planes may offer a principled way of explaining the typological
observations made by Rice:
Within the front vowels, contrasts may exist between an unrounded and a rounded vowel
in the language.
Within the central and back vowels, no single language makes a contrast between
unrounded and rounded vowels. (1995: 89)
More research may be needed to conclusively prove that front-roundedness has the potential to be
orthogonal to a primary plane. Articulatory and acoustic phonetic research can be useful, but decisive
evidence would be offered by languages in which front-rounded vowels behave phonologically as a natural
class.
2.5 Symmetry and asymmetry
As mentioned above, the number of distinct configurations of vowel systems among the world’s
languages is limited by the interplay of two sets of parameters; acoustic and psychological. These
parameters give rise to two distinct vowel spaces; an acoustic quadrilateral and a psychological square. The
acoustic space is asymmetrical with the least area in the low region. When the acoustic space is dominant,
the vowels tend to distribute evenly in the acoustic quadrilateral, and a downward-pointing triangular
configuration results. Figure 12 provides examples of triangular shapes containing from three to seven
vowels. Two-vowel systems are always rectangular (see below).
Figure 12: Triangular systems from three to seven vowels
3 4 5 6 7
Aleut Kerek Mongsen Ao Lemoro Romanian
i u i u i ʉ u i u i ɨ u
ə ə e ə o e ə o
a a a a a
In contrast, the psychological space is symmetrical and when this space is dominant, the vowels
distribute in a symmetrical, rectangular fashion. Figure 13 exemplifies symmetrical systems involving four
to nine vowels. Three-vowel systems are always triangular (see below).
Figure 13: Rectangular systems from four to nine vowels
4 5 6 7 8 9
Tuscarora Wadjiginy Anuak Wantoat Cayuvava Cham
ɪ ʊ i u i u i u i ɨ u i ɨ u
ø e o e ə o ɛ o e ə o
ɛ ɑ e a æ ɔ æ ɑ æ a ɔ ɛ a ɔ
All of the examples in Figure 13 are symmetrical in two dimensions. There may be a third category.
There are seven- and eight-vowel systems with triangular acoustic but square phonemic shapes. This occurs
when there are three low vowels and the central vowel is lower than the front and back ones. In Figure 14
are the seven-vowel system of Yoruba adapted from Bamgboṣe (1969: 166) and the eight-vowel system of
DOUG HITCH
12
Standard Eastern Catalan adapted from Carbonell & Llisterri (1999: 62) together with their phonemic
squares.
Figure 14: Triangular acoustic and square phonemic systems
Yoruba Standard Eastern Catalan
i u i u
e o e ə o
ɛ a ɔ ɛ a ɔ
Any seven-, eight-, or nine-vowel system with three low vowels could be regarded as being
acoustically triangular. This acoustic shape often leads to the assumption of four vowel heights, which is
correct acoustically but not phonemically.
The triangular shapes as in Figure 12 reflect the dominance of the acoustic space while the rectangular
shapes in Figure 13 reflect the dominance of the psychological space. This provides an explanation for
Trubetskoy’s observation that vowel systems have linear (see below), quadrilateral, and triangular
structures (Trubetskoy, 1939: 87, 1971: 97).
The contrast in shape between the asymmetrical acoustic space and the symmetrical psychological
space can be used to explain other phenomena related to vowel systems such as typological patterns and
the relationship between systems in historical or geographic dialects (see §4.0 below). For instance, the
universal that no language has more back vowels than front is due to the fact that the area in the front of the
acoustic space is larger than the area in the back. Congruently, languages often have equal numbers of front
and back vowels because of the symmetry of the psychological space.
2.6 Transcribing the psychological space
Transcription of the vowels according to their position in the acoustic space is reasonably
straightforward. Most scholarship today employs the vowel symbols of the IPA as shown in Figure 3. This
involves 28 symbols, but still greater phonetic precision may be achieved through the use of diacritics.
Transcription of the segments as they occur in the psychological space involves different principles. The
focus is on a psychologically real segment rather than on phonetic reality. Generally speaking, to represent
a phoneme, scientists try to select the phonetic symbol representing the most frequent phonetic allophone.
But it often happens that two reports on a vowel system will agree on the number of vowel phonemes and
their general phonetic behaviour, but disagree on the symbols used to represent them. Additionally, a
particular phonetic segment may receive differing phonological treatment in different languages. An
example is /y/, as discussed above, which either appears on a secondary plane, or fills the role of a high
central vowel on a primary plane.
In theory, to represent the occupants of the psychological space one could use a symbology unrelated
to phonetics. For instance, one could use the directions as in Figure 4. Instead of /i/ there could be NW,
instead of /ɨ/ N, or instead of /u/ NE, etc. Another method would be to use numerals, and number the nine
possible slots in the space as shown in Figure 15.
Figure 15: Numbering the psychological space
1 2 3
4 5 6
7 8 9
VOWEL SPACES AND SYSTEMS
13
Each of the numbered slots can feature a range of phonemic symbols, and some comment is useful.
Slot 1 probably has the least variety of symbology. It is most often /i/, less often /ɪ/.
Slot 2 can have true phonetic high central symbols /ɨ ʉ/, but also may feature phonetically front round
/y/ or phonetically back unround /ɯ/ symbols if the phonological system treats them as non-front or non-
back, respectively (see §2.4 above).
Slot 3 is prototypically /u/, less often /ʊ/ and occasionally /ɯ/. Systems that do not distinguish /u/ and
/o/, and have one segment that can vary between [u] and [o] in pronunciation, are often described as if /o/
is the highest back vowel. For instance, three-vowel systems /i u a/ are often given as /i o a/.
Slot 4 is prototypically /e/. Often /ɛ/ is given as the mid vowel. In cases where the describer gives
both /e/ and /ɛ/, the former fills slot 4 and the latter fills slot 7.
Slot 5 is prototypically /ə/. IPA recognizes four heights, distinguishing close-mid and open-mid. In
this slot among the central vowels I have also noticed close-mid round /ɵ/, open-mid unround /ɜ/ and open-
mid round /ɞ/, but not close-mid unround /ɘ/. In addition, this slot may feature phonetically front round
close-mid /ø/ or open-mid /œ/ symbols, if the language treats such a segment as central, and, conversely,
we may find the back unrounded close-mid /ɤ/ or open-mid /ʌ/ here if the language treats the segment as
central (see §2.4 above).
Slot 6 is prototypically /o/ although as described above under slot 3, that symbol may be used where
slots 3 and 6 form an /o~u/ unit. An /ɔ/ may be found here in some descriptions, but elsewhere /ɔ/ clearly
denotes a low vowel and belongs in slot 9. Occasionally unround back close-mid /ɤ/ can be found here, but
I have not noticed unround back open-mid /ʌ/ in this slot.
Slot 7 is perhaps prototypically /æ/ but often /ɛ/ is found here and sometimes /a/. The symbol /a/ may
appear for any unround low vowel. Sometimes this symbol may be used out of convenience without any
phonological implication other than low.
Slot 8 is perhaps most often filled by /a/, even though the IPA confusingly considers the symbol to
mark a front, not a central vowel. In these pages, /a/ is essentially prototypically central. Frequently /ɐ/ is
found here, but I suspect it is by scholars wishing to use an IPA symbol for a clearly central vowel, even
though IPA places the vowel above the low (open) line in its quadrilateral (Figure 3). Very rarely one finds
the open-mid central unrounded /ɜ/ or open-mid unrounded back /ʌ/ here.
Slot 9 should be prototypically unround /ɑ/ or round /ɒ/ but we often find /ɔ/ here, and even /a/.
The numbering of the slots also provides a convenient standard for a linear listing of a vowel system
where a chart is inconvenient. For example, for a prototypical five-vowel system the listing is /i u e o a/ or
/1 3 4 6 8/. In other words, the phonemes are given as they occur in the psychological space from left to
right and top to bottom.
3 Systems by number of vowels
Below are described and listed the known spatial systems for vowels, organized in sections according
to the number of vowels in the system. Each section begins with a discussion of the proposed types. Starting
with the triangular three-vowel systems, there is also a list of examples for each type in Appendix A. Each
example begins with one or more language names. Where there are two or more commonly used names
they are separated with a tilde. Next, in parentheses, there is an indication of language family, and of
geographic location(s) of speakers. These indications are intended to aid in understanding the genetic and
geographic relationships of a system and may not be consistent. Multiple locations are separated by hyphen.
The next item is a linear list of phonemes in slashes following the numbering of slots given above.
Sometimes there are indications of secondary planes where this would be useful. The final item is a second
set of parentheses containing the reference(s) and any comment. Three major lists of phonological systems
consulted here are given abbreviations. An item from Merritt Ruhlen’s (2004) “Global Typological
Database” is referred to as R+the database number; the first item is R1 Hadza and the last is R5736
Interlingua. An item from the UCLA Phonological Segment Inventory Database (UPSID; Maddieson &
Precoda ongoing; Brasington 2013; Reetz undated) is referred to as U+the UPSID number; for instance,
Klamath is U6707. UPSID /ɷ/ is replaced by current IPA /ʊ/. Language information from the “Phonological
DOUG HITCH
14
Inventories of Tibeto-Burman Languages” (Namkung, 1996) uses TB:+the page number from that list, so
that Amdo Tibetan is TB:390. On occasion it has been useful to refer to Wikipedia (W) either where the
language is well known, or where the information is credible but it has not been possible to find published
sources. I have not consulted Ruhlen (2004) past central Mexico and did not encounter the aid at phoible.org
until late in the process. Undoubtedly many useful examples of inventories could be added. Orthographic
representations are in angle brackets < >.
3.1 Zero-vowel systems
Aert Kuipers first proposed in 1960 that Kabardian, a Northwest Caucasian language, could be
regarded as vowelless. Morris Halle refuted the possibility in 1970 but Kuipers maintained it again in 1976.
Edwin G. Pulleyblank in 1984 accepted the logical possibility of a vowelless Kabardian and further
proposed that Modern Standard Chinese as well as Middle Chinese could be analyzed as vowelless
(Pulleyblank, 1984a, 1984b: 21–22). He treated Pekingese schwa as epenthetic, and the other vowels as
syllabic forms of glides with which those vowels alternate morphophonemically (Pulleyblank, 1984b: 57).
John Colarusso (2014: 293–454) in a specialized work on NW Caucasian phonology shows with extensive
evidence that those languages all feature two-vowel systems /a ə/ with some featuring also long /aː/ with
phonological charts of 30 dialects all with /ə a/.
Daniel Barreteau (1983: 387–388) proposes an analysis of Mofu-Gudur (Chadic, Cameroon) as a
vowelless language. His more conservative treatment offers three short /e ə a/ and two long /eː aː/ vowels
(Barreteau 1983: 250 ff.). Such a three-vowel system would be typologically unique. Mofu-Gudur
apparently exhibits front-back vowel harmony with words containing either /ə e eː/ or /ə a aː/. The
allophones of /ə/ are [i ɨ ʉ u ə], of /e/ are [ε ø œ], and of /a/ are [o ɔ a] (Barreteau 1983: 364). Barreteau
calls the harmony a prosodic trait and suggests using a superscript y (ʸ) to show words containing /e/ rather
than /a/, for example, ɗVɗVwV = ɗaɗawa ‘heart’ vs. ʸɗVɗVwV = ɗeɗewe ‘Leptadenia hastata’. /ə/ appears
to be harmonically neutral, for example, wápə́rá ‘fall badly’ (Barreteau 1983: 384), mévərkétéleŋ
‘dragonfly’ (Barreteau 1983: 385). Conceivably, the language has a system of two, two-vowel harmonic
planes: a back /ə a/ and a front (‘palatalized’) /ə e/.
A zero-vowel language would insert vowels according to rules of epenthesis, then colour the vowels
according to phonetic context. It sounds theoretically possible, but no completely convincing cases have
yet been identified.
3.2 One-vowel systems
There seem to be no indisputable examples of one-vowel systems on a primary plane. A system with
one vowel would not require distinctive features reflecting F1 and F2. The single vowel would only need
to be marked [+syllabic]. Comrie (1991: 396) suggests that Harui~Waibuk (Piawi, Papua New Guinea)
“can be analysed, with some degree of plausibility, as having only a single vowel, the mid central vowel
ö”. This involves several abstractions from the seven-vowel system he begins with (using IPA ɵ for his ö)
/i ɨ u e ɵ o a/. One abstraction is striking: “Haruai consonants have syllabic allophones, e.g., /p/ has, in
addition to its basic nonsyllabic allophone [p], a syllabic allophone [pi]” (Comrie 1991: 394). This is a
device I have not seen before and would likely be subject to dispute.
There are languages that have a secondary plane with one vowel.
It appears that if a language has only one long vowel, it will be /aː/. The UPSID database search for
languages with only one long vowel produces four languages, and all four have only long /aː/: Angaatiha
(Angan, Papua New Guinea) U8627; Lai~Li (Li-Kam-Tai, Guangxi) U2432; Mien (Hmong-Mien, China-
Viet Nam-Laos-Thailand) U2517; Yay (Li-Kam-Tai, NW Vietnam-S Yunnan) U2402. Several of the NW
Caucasian languages have a two-vowel system /ə a/ with long /aː/ Ubykh R1601; Adyghe R1605; Kabardian
R1606. I did not systematically search for this feature in R or TB (they do not have electronic database
versions) but also noticed Chepang TB:78 (Sino-Tibetan, Nepal) with just long /aː/.
VOWEL SPACES AND SYSTEMS
15
I suspect that a language with one pharyngealized or laryngealized vowel will base it on /a/. Dargwa
(NE Caucasian, SW Russia) has one pharyngealized vowel /aˤ/ (R1626). Mongsen Ao (Tibeto-Burman, NE
India) has one creaky voiced vowel /a̰/ (Coupe 2008:2).
There appear to be no restrictions concerning a single nasalized vowel. Cherokee (Iroquoian, N
Carolina-Oklahoma) has just /ə̃/ (U6781; although this does not contrast with a non-nasal */ə/ and the
nasalization could be regarded as phonetic enhancement in /i u e ə̃ o a/); Gwari (Volta-Niger, Nigeria) has
just ĩ (U4140).
3.3 Two-vowel systems
Systems with two phonemic vowels defined by space are attested from the Northwest Caucasus,
Nigeria, Australia, and possibly Kansas. There are South Arabian languages with two vertical vowels in a
possibly secondary short vowel plane. Front-rounded planes with /y ø/ have been mentioned above and
appear in the lists below.
As mentioned above in §3.1, all of the Northwest Caucasian languages, have /ə a/. They feature
extensive allophony. For instance, according to the environment, Kabardian /ə/ may appear as [i ɨ u ə ʉ ü]
and /a/ as [e ɑ o a æ ø ɒ] (Kuipers, 1960: 23).
Margi (Chadic, NE Nigeria) has been shown to have a two-vowel system /ɨ a/ (Maddieson, 1987).
Recent loanwords also distinguish [e o] (ibid.: 328).
At least two languages from Australia are still described as featuring a two-vowel system, while there
is doubt about others. Enindhilyagwa~Anindilyakwa (Arnhem (?), N Australia) has high and low central /ɨ
a/ with rich allophony (Leeding, 1989: 38–62). Kaytetye~Kaititj (Arandic, C Australia) has /ə a/ (R4588
and still implied in Harvey et al., 2015). Upper Arrernte (Arandic, C Australia) was once described as
having /ə a/ (and still is on Wikipedia: “Upper Arrernte language”) but current scholarship assigns more
vowels. For Eastern/Central Arrernte Henderson (2002: 102) claims that all dialects have /i ə a/ and less
conservative lects also have /u/. Breen & Dobson (2005: 251) maintain /i u ə a/ for Central Arrernte.
As discussed below in §3.4.1, Wichita may uniquely have an /i a/ system.
Some dialects of Mehri (South Arabian, Oman-Yemen) have short /ə a/ and long /iː uː eː oː ɛː aː/
(Simeone-Senelle, 1997: 84). Harsusi (South Arabian, Yemen) has /ə a/ and /iː uː eː oː aː/ (R1580). In these
languages the long vowels may constitute the primary plane which may mean that the language has full and
reduced vowels, rather than long and short.
A noteworthy two-vowel system is offered by Marshallese. It was originally reported by Bender
(1968: 20) as having a vertical four-vowel system, that is, four phonemes distinguished by four degrees of
height. More recent work recognizes that these four fall into +ATR and -ATR classes (cf. Wilson, 2003: 2–
3). +ATR /ɨ/ has the allophones [i ɯ u], -ATR /ə/ has the allophones [ɪ ɤ ʊ], +ATR /ɜ/ has [e ʌ o], and -ATR
/a/ has [ɛ a ɔ]. Here the system may be analyzed as having two two-vowel vocalic planes: +ATR /ɨ ɜ/ and -
ATR /ə a/.
Crothers (1978) was only aware of the NW Caucasian languages, particularly Kabardian, as having
two-vowel systems. He was suspicious of the analyses and thought the type, “though interesting as an
extreme, has little bearing on the general picture of vowel system typology” (Crothers 1978: 109). But this
type reveals something fundamental about vowels: that [low] is the most basic of the four spatial features.
3.4 Three-vowel systems
3.4.1 Vertical three-vowel systems
The idea of systems with three vertical vowels was perhaps given most currency by Nikolai
Trubetzkoy. In his posthumously published monument, Grundzüge der phonologie, he repeated the idea of
DOUG HITCH
16
Jakovlev from the 1920s8 that Adyghe, Abkhaz and Ubykh had such a system (Trubetzkoy 1939: 87). As
mentioned above in §3.1, all NW Caucasian languages have a two-vowel system.
In his most recent description of Wichita (Caddoan, Oklahoma-Kansas-Texas), David Rood
(1996:1.1) assumes a system of three vertical vowels /i e a/. But he also notes that the short version of his
/e/ “very rarely occurs” and the overlong version “is by far the most frequent overlong vowel” (Rood 1996:
1.1.2). He suggests that overlong [e∙] could be treated as the sequence /ayi/ or /iya/ parallel to overlong [o∙]
= /VwV/, but because the choice between /iya/ and /ayi/ is often not clear he prefers to retain phonemic /e/
(Rood 1996: 1.1.2). However, Rood also notes that only /i a/ occur in final position, never /e/ (Rood 1996:
1.1.6), which is another parallel with [o∙]. It seems to me that Wichita is either two vowel /i a/ or four vowel
/i o e a/.
The neighbouring and plausibly related Papuan language families Sepik and Ramu are sometimes
suggested as exhibiting three vertical vowel phonemes /ɨ ə a/9, but more recent research seems to negate
this. For instance, Staalsen (1966: 1) at first proposed three vertical vowels for Iatmul (Sepik) /ɨ ə a/ but
later revised this to a seven-vowel system /i ɨ u e o ɐ ɑ/ (Staalsen 1992: 3). Most recently, Jendraschek
(2012: 36) also argues for seven “vowel positions” for Iatmul which may correspond to /i ɨ u e o a ɑ/.
As mentioned in §3.3 above, for Eastern/Central Arrernte (Arandic, C Australia) Henderson (2002:
102) claims that all dialects have /i ə a/ and less conservative lects also have /u/, while Breen & Dobson
(2005: 251) maintain /i u ə a/ for Central Arrernte.
I have not been able to find completely convincing evidence that any language operates with a vertical
three-vowel system on a primary plane. Ideally there should be multiple languages from different families
and locations clearly described with a vertical three-vowel system for this type to be recognized. Parisian
French has a vertical three-vowel configuration /y ø œ/ on a front-rounded plane (primary /i u e ə o ɛ a ɔ/
W). While vertical three-vowel systems may not exist, primary plane triangular three-vowel systems are
exceedingly common.
3.4.2 Triangular three-vowel systems
The prototypical triangular three-vowel system is /i u a/. The phonetics may vary with each language,
and may vary with the researcher. Corresponding to /i/, an /e/ may be reported, and corresponding to /u/,
an /o/ may be reported. Some descriptions may use mid vowel symbols as in /e o a/, /i o a/, or even /e u a/.
This variety has caused challenges to other approaches (cf. Disner, 1984: 140–144) but poses no difficulty
to the analytical system used here. In Figure 16, the borders between slots 1 and 4 and between 3 and 6 are
not drawn, to suggest the /i~e/ and /u~o/ variations. In the list of examples in Appendix A, just a small
selection of the many languages with /i u a/ is given. An effort was made to demonstrate the geographic
and genetic range.
Figure 16: The three-vowel system
i u
a
8 Colarusso (2014: 294), when discussing the analysis of a vertical vowel system by attributing phonetic vocalic
difference to the surrounding consonants, writes, “This denuding of the vowels, so to speak, in favor of the
consonants was first proposed by Jakovlev (1923) for Kabardian in one of the earliest phonemic analyses ever
performed on a language. All subsequent workers on these languages have followed the basic principles of
Jakovlev’s analysis.” 9 Cf. https://en.wikipedia.org/wiki/Sepik_languages, https://en.wikipedia.org/wiki/Ndu_languages.
VOWEL SPACES AND SYSTEMS
17
3.5 Four-vowel systems
There appear to be between two and four types of four-vowel systems. Type 4a is a prototypical
triangular four-vowel system with the addition of a phonemic neutral vowel. Type 4b is similarly a three-
vowel system but with what may be a high central vowel like /ɨ ʉ/ added. Some descriptions of 4a /ə/ and
4b /ɨ/ show the allophonic range overlapping, which is suggested in the figure by the border missing between
slots 2 and 5. Maranao, arbitrarily included below as 4b, is given with /ə/ in one description and as /ɨ/ in
another. Type 4c is a rectangular structure with the phonemes clearly in the corners. Type 4d may be
considered to have the same rectangular structure as 4c, but as many systems are reported with mid vowel
symbols like /e o/, it may be useful to give a separate list for this (sub)variety at this stage. Many descriptions
of 4d systems describe the allophonic range of /e/ as going as low as [æ] and of /o/ as going as high as [u],
as suggested by undrawn borders. I suspect /a/ tends to be more back than central. While 4a and 4b require
three features and 4c two, it is not clear if 4d requires two or three features, or if the analysis is language
dependent. Crothers (1978: 109) notes two four-vowel systems, 4:0 /i ɛ a u/ which here would be 4b, and
4:1 /i ɨ a u/ which is here 4c. He also thought they were subtypes of a single proto-type /i ə a u/ which is
attested and is here listed as 4a. Examples of each type are given in Appendix A.
Figure 17: Four-vowel systems
4a 4b 4c 4d
i u i ɨ u i u i
ə e o
a a æ ɒ a
3.5.1 North Straits Salish
Three dialects of North Straits Salish have four-vowel systems that are typologically of interest. The
Saanich~SENĆOŦEN (Vancouver Island) dialect is described by Montler as having “an unusual
symmetrical four vowel system” Montler (1986: §1.2). He places /i ə/ above /e a/ in a chart (ibid.), but the
system cannot straightforwardly be assigned to 4c. It also has no rounded vowels but there is [u] in loans
(ibid.). The Tillamook (Salish, NW Oregon) system /i ə æ ɑ/ also has no rounded vowels (Thompson &
Thompson, 1966: 318). The Lummi (NW Washington) dialect of North Straits Salish /i e ə o/ (R4741)
appears to have no low vowels. It is possible that what is given as /ə/ may be closer to [ɤ] than [ə], and then
these systems may classed as 4c.
3.6 Five-vowel systems
The extremely common classic five-vowel system is given here as type 5a. The next most common
type appears to be 5b, a symmetrical rectangle like 4c with a neutral vowel added. The types 5c–e are
relatively rare. 5c is a symmetrical triangle which in a way combines 4a and 4b, having both /ə/ and /ɨ/.
Type 5d is like 4c with a front mid vowel /e/ added. It is asymmetrical because the acoustic space is
dominant. Type 5e is also like 4c but with a high central vowel added. As for features, 5a, 5c and 5d must
feature both high and low, to distinguish three heights. Types 5c and 5e must feature both back and front
to distinguish three depths. Three features are sufficient to distinguish 5a, 5b, 5d and 5e, while 5c requires
all four features. Crothers (1978) identified two five-vowel systems. Besides the obvious 5a which he
labelled 5:0, he identified a shape *5:1 which he found in five languages. Island-Carib /i u e ɤ a/ and Ocaina
/i ɯ ɛ o a/ are here both 5a. Tolowa /i u ə e a/ and Maranungku /i ʊ ɵ æ a/ are here both 5b. The fifth
language, Nez Perce, was shown by Hall and Hall (1980) to feature ATR vowel harmony, and here is
DOUG HITCH
18
regarded as having two three-vowel planes: +ATR /i u æ/ and -ATR /i o ɑ/ (see §2.3 above). Crothers,
(1978: 140) also noted one **5:1 language, Evenki, which is apparently no longer in UPSID but the
qualities he listed, “/i æ ə u O/”, would be 5b. His one ***5:1 language, Papago, now Tohono O'odham, is
discussed below. Examples of each type are given in Appendix A.
Figure 18: Five-vowel systems
5a 5b 5c 5d 5e
i u i u i ɨ u i u i ɨ
e o ə ə e o
a æ ɔ a æ ɑ ɛ ɔ
3.6.1 Tohono O’odham
Tohono O'odham (formerly Papago, Uto-Aztecan, Mexico-Arizona) appears to not fit the pattern of
any other language and to violate a universal by having more back than front vowels with /i ɨ u o a/
(Miyashita, 2011: 323).
3.7 Six-vowel systems
The triangular systems 6a and 6b essentially reflect the classic five-vowel system 5a /i u e o a/ with
the addition of a central vowel. The salient difference is likely in the number of features required by each.
Type 6a requires all four spatial features while type 6b needs both front and back but just one of high or
low. Rectangular type 6c, in a fashion converse to 6b, requires both high and low but just one of front or
back. There are three six-vowel systems in Crothers (1978). His 6:0 is 6c, his 6:1 is here divided into 6a
and 6b. He also noted a *6:2 with two languages. Hopi is included below as 6a. Chuvash, a Turkic language,
does have six full vowels but these are distributed in front /i y ɛ/ and back /ɯ u ɑ/ three-vowel harmonic
planes (Róna-Tas, 1997: 2, with symbols /i ü e/ and /ï u ɑ/). Examples of each type are given in Appendix
A.
Figure 19: Six-vowel systems
6a 6b 6c
i u i ɨ u i u
e ə o e o e o
a a æ ɑ
3.8 Seven-vowel systems
Any seven-vowel system will require all four spatial features. Of the four seven-vowel systems
identified, just 7a could be called triangular in the psychological space, as it has just one low vowel. Type
7d is phonetically triangular. Phonemically rectangular 7b, 7c and 7d all feature one central vowel, which
is neutral, high or low, respectively. Crothers’s (1978) 7:0 is here 7d and his 7:2 is 7a. His three members
of *7:1 are Sentani~Buyaka 7b, Kwoma~Wahkuk 7c, and Wolof, which, according to Unseth (2010: 1),
has +ATR /i u e ə o/ and -ATR /ɛ ɔ a/ harmonic planes. Examples of each type are given in Appendix A.
VOWEL SPACES AND SYSTEMS
19
Figure 20: Seven-vowel systems
7a 7b 7c 7d
i ɨ u i u i ɨ u i u
e ə o e ə o e o e o
a æ ɑ æ ɑ æ a ɑ
3.9 Eight-vowel systems
There may be only three configurations for eight-vowel systems. Although mathematically there
could be a gap in any position, the gaps seem to be restricted to the central column. Type 8a seems to be
the most common type. It has a gap clearly at the high central position. Type 8b is next most common. In
Figure 21 it is represented with a gap at low central. However, some descriptions would seem to place the
gap at low front, and others at low back. It may be that three types are subsumed here, according to where
the low gap is, but that is not clear. Often, when two descriptions of a dialect differ, it is in the
characterization of the two low vowels. The symbol /a/ often seems to function for any low vowel. Type 8c
is lacking the neutral vowel. It appears to be rather uncommon. Only two examples have been collected. If
the /ɨ/ in these languages allophonically varies [ɨ~ə] then they might reflect 8a. Crothers (1978: 112)
reported that, “None of the eight vowel systems is common enough to be important typologically.” His 8:1
languages are mostly 8a (Ewondo, Javanese, Mianka~Minyanka). His 8:2 Icelandic looks like 6c /i u ɪ ɔ ɛ
a/ with front vocalic /ʏ ø/ (cf. Gussmann, 2011), and Central Tibetan is here 7a. His 8:3 examples Turkish,
Kirghiz (Turkic) and Cheremis (Uralic) all feature vowel harmony with smaller planes. Examples of each
type are given in Appendix A.
Figure 21: Eight-vowel systems
8a 8b 8c
i u i ɨ u i ɨ u
e ə o e ə o e o
æ a ɑ æ ɑ æ a ɑ
3.10 Nine-vowel systems
The maximum number of segments on a vowel plane as defined by spatial features is nine. In other
words, there is only one type of nine-vowel system; that which has all slots filled, as in Figure 22. Not
surprisingly, Crothers (1978: 112) also identifies essentially one type: “9:2 is the only common type with
more than seven vowels.” His 9:2" examples, Akan, Grebo, and Luo, all feature ATR harmony, and of his
four 9:3' examples, Azerbaijani (Turkic) features front-back vowel harmony, Norwegian has a front
rounded plane, Korean inventories are now apparently differently analysed with Seoul /i ɨ u e~ɛ o a ʌ/ and
Kyungsang /i ɨ u e o a/ (Lee & Jongman, 2016: 158–160), and Ostyak (now Khanty, Uralic) also now has
differing inventories (cf. Abondolo, 1998: 360). Examples are given in Appendix A.
DOUG HITCH
20
Figure 22: Nine-vowel system
i ɨ u
e ə o
æ a ɑ
3.11 Crothers’s 10- to 12-vowel systems
Crothers (1978: 143) listed seven languages with more than nine “basic” vowels. All have different
analyses here. Hanoi Vietnamese and Pacoh have nine vowels. Three languages have front-round planes:
Iaai is 8a /i u e ɤ o æ a ɔ/ with /y ø/; Akha is 9 /i ɯ u e ɤ o ɛ a ɔ/ with /y ø/; Parisian French is 8a /i u e ə o
ɛ a ɔ/ with /y ø œ/. Panjabi has seven long vowels in shape 7d /i u e o ɛ a ɔ/ and three short /ɪ ʊ ə/ (Karamat,
2002: 182). Lugbara has +ATR /i u e o ɜ/ and -ATR /ɪ ʊ ɛ ɔ a/ planes (U4215).
4 Final observations
With regard to the primary plane, two-vowel systems are vertical, probably prototypically /ə a/. There
is plausibly just one type of three-vowel structure, triangular /i u a/, which reflects acoustic space
dominance, and there is just one nine-vowel structure which has all of the possible psychological slots
filled. The two- to eight-vowel structures show a range of three to five types with the variation reflecting
the interplay between the acoustic and psychological spaces.
The concept of vocalic planes as implemented here can simplify phonological descriptions, can
provide analyses for vowel systems which have challenged analysts, and can help identify descriptions that
are defective and suggest where to look for improvements.
This study may have implications in other areas of theoretical phonology. For instance, all principles
in operation here are based strictly on acoustic phonetics (even though articulatory terminology is often
used). If the descriptions of vowel systems here are judged to be effective, this opens the possibility that
consonants too may be effectively described strictly in acoustic terms.
With regard to typology and universals, the system here can explain at least some observations in a
principled way. For instance, consider the vowel hierarchy set up by Crothers (1978: 114) based on
implicational statements of the form: “if a language has a vowel phoneme of type z, it also has one of type
w”. The first part of his Figure 10 (ibid.) is adapted here as Figure 23.
Figure 23: Crothers’ hierarchy
a i u
↙ ↘
ɨ ɛ
↓ ↓
ɛ ɔ
↓ ↓
ɔ e
↓ ↓
ə o
The explanation for this implicational hierarchy may be most easily understood if it is read in a
different way. The chart may also be read as a hierarchy defining certain historical relationships between
systems. Implicationally, /ɔ/ implies /ɛ/ while historically a language will not add /ɔ/ to its inventory if it
does not have /ɛ/. A three-vowel system may evolve to a four-vowel system by adding either an ɨ-like or an
VOWEL SPACES AND SYSTEMS
21
ɛ-like vowel. The ɨ-like addition in the left column does not change the basic phonetic shape of the system
or substantially change the phonetic values of the segments, but it does increase the psychological load by
increasing the distinctive features needed from two to three. In contrast, the ɛ-like addition in the right
column might change the phonetics by causing the /a/ to shift toward the back of the acoustic space perhaps
becoming /ɑ/, but it does not affect the number of distinctive features or psychological load. With the next
tier on the chart, the situation reverses. The addition of an ɛ-like vowel on the left side taxes the acoustics
but does not change the features. And on the right side the addition of an ɔ-like vowel creates the
acoustically desirable classic five-vowel system but now increases the number of distinctive features to
three. With each tier the addition processes alternate, in a sense, reflecting the interplay between the acoustic
and psychological spaces. Languages may also evolve to have fewer vowels along the same hierarchy. The
implicational hierarchy parallels the evolutionary. It too reflects the interplay between acoustics and
phonemics.
References
Abenaki-Penobscot pronunciation guide, alphabet and phonology. (n.d.). Retrieved March 8, 2017, from
http://www.native-languages.org/abna_guide.htm.
Abondolo, D. (1998). Khanty. In D. Abondolo (Ed.), The Uralic languages (358–386). London and New
York: Taylor & Francis.
Alves, M. J. (2000). A Pacoh analytic grammar (Doctoral Dissertation). University of Hawai’i, Honolulu,
HI.
Archangeli, D. (1985). Yokuts harmony: Evidence for coplanar representation in nonlinear phonology.
Linguistic Inquiry 16(3), 335–372.
Axelrod, M. (1990). Incorporation in Koyukon Athapaskan. International Journal of American Linguistics
56(2), 179–195.
Bamgboṣe, A. (1969). Yoruba. In E. Dunstan (Ed.), Twelve Nigerian languages (166). New York: Africana
Publishing Corp.
Barreteau, D. (1983). Description du Mofu-Gudur (langue de la famille tchadique parlée au Cameroun)
(Doctoral Dissertation). Université de la Sorbonne Nouvelle Paris III, Paris, France.
Bender, B. W. (1968). Marshallese phonology. Oceanic Linguistics 7(1), 16–35.
Bielenberg, B., & Zhalie, N. (2001). Chokri (Phek dialect) phonetics and phonology. Linguistics of the
Tibeto-Burman Area 24(2), 85–122.
Boraas, A. (2010). An Introduction to Dena’ina grammar: The Kenai (outer inlet) dialect. ms, Soldotna,
AK. http://web.kpc.alaska.edu/denaina/documents/denaina_grammar.pdf.
Brasington, R. (2012). A NetLogo interface to updated UPSID. Retrieved from
http://www.personal.rdg.ac.uk/~llsling1/upsid.netlogo/UPSID.updated.interface.html.
Breen, G., & Dobson, V. (2005). Central Arrernte. Journal of the International Phonetic Association 35(2),
249–254.
Bruce, G., & Engstrand, O. (2006). The phonetic profile of Swedish. Sprachtypologie Und
Universalienforschung 59(1), 12–35.
Carbonell, J. F., & Llisteri, J. (1999). Catalan. In Handbook of the International Phonetic Association: A
guide to the usage of the International Phonetic Alphabet (61–65). Cambridge: Cambridge
University Press.
Chafe, W. (2014). A Grammar of the Seneca Language. Onöndowa’ga:’ Gawë:nö׳ (Language of the People
of the Great Hills) website: Seneca Nation of Indians. Retrieved from http://senecalanguage.com/wp-
content/uploads/Chafe-Seneca-grammar-PDF-final.pdf.
Chelliah, S. L. (1997). A Grammar of Meithei. Berlin; New York: Mouton de Gruyter.
Chen, C.-M. (2006). A comparative study on Formosan phonology: Paiwan and Budai Rukai (Doctoral
Dissertation). University of Texas at Austin, Austin, TX.
DOUG HITCH
22
Chernjavskij, V. [Чернявский, В.]. (2005). Vaďďa tšeeli (Izeõpõttaja). Водский язык (Самоучитель)
[Votic Language (Self-study)]. Памяти академика П. Аристэ [To the Memory of Academician P.
Ariste]. Retrieved from http://www.lingvisto.org/files/votic.pdf.
Chitoran, I. (2002). A perception-production study of Romanian diphthongs and glide-vowel sequences.
Journal of the International Phonetic Association 32(2), 203–222.
Colarusso, J. (2014). The Northwest Caucasian languages. A phonological survey. (online edition of the
1988 original edition). Oxon and New York: Routledge.
Comrie, B. (1991). On Harui vowels. In A. Pawley (Ed.), Man and a half: Essays in Pacific anthropology
in honour of Ralph Bulmer (393–397). Auckland: The Polynesian Society.
Coupe, A. R. (2008). A grammar of Mongsen Ao. Berlin & New York: De Gruyter Mouton.
Cowan, H. K. J. (1965). Grammar of the Sentani language. The Hague: Martinus Nijhoff.
Crawford, J. M. (1973). Yuchi phonology. International Journal of American Linguistics 39(3), 173–179.
Crothers, J. (1978). Typology and universals in vowel systems. In J. H. Greenberg, C. A. Ferguson, & E.
A. Moravcsik (Eds.), Universals of human language. Volume 2: Phonology (93–152). Stanford:
Stanford University Press.
Davis, D. (2004, printed from data checked 1994). Organised phonology data. Wantoat (Wandoat)
language [WNC]. SIL document. Retrieved from http://www-
01.sil.org/pacific/png/pubs/928474542364/Wantoat.pdf.
Deibler, E. (1992, data checked). Organised phonology data: Alekano (Gahuku) language. SIL document.
Retrieved from http://www-01.sil.org/pacific/png/pubs/928474542307/Alekano.pdf.
Disner, S. F. (1984). Insights on vowel spacing. In I. Maddieson (Auth.), Patterns of sounds (136–155).
Cambridge, UK: Cambridge University Press.
Donohue, M. (2007). A Grammar of the Skou language of New Guinea. ms. Monash University, Clayton,
Australia. http://papuan.linguistics.anu.edu.au/Skoufiles/Skounopiccies1.pdf
Ellis, J. S. (2012). Saipan Carolinian, one Chuukic language blended from many (Doctoral Dissertation).
University of Hawai’i at Mānoa, Honolulu, HI.
Elson, B. F. (1992). Reconstructing Mixe-Zoque. In Language in context: Essays for Robert E. Longacre
(577–592). Arlington, Texas: Summer Institute of Linguistics and the University of Texas at
Arlington.
Engstrand, O. (1999). Swedish. In Handbook of the International Phonetic Association: A guide to the
usage of the International Phonetic Alphabet (140–142). Cambridge: Cambridge University Press.
Ford, L. J. (1990). The phonology and morphology of Bachamal (Wogait). ms. The Australian National
University, Canberra, Australia. https://openresearch-
repository.anu.edu.au/bitstream/1885/10815/8/Ford_L_Master_1990.pdf
Godddard, I. (1974). An outline of the historical phonology of Arapaho and Atsina. International Journal
of American Linguistics 40(2), 102–116.
Grønnum, N. (1998). Illustrations of the IPA: Danish. Journal of the International Phonetic Association
28(1–2), 99–105.
Grune, D. (1997). Hopi. Survey of an Uto-Aztecan Language. online pdf. Retrieved from
http://www.cs.vu.nl/~dick/Summaries/Languages/Hopi.pdf.
Gussmann, E. (2011). Getting your head around the vowel system of modern Icelandic. Folia Scandinavica
12, 71–90.
Hall, B. L., & Hall, R. M. R. (1980). Nez Perce vowel harmony: An Africanist explanation and some
theoretical consequences. In R. M. Vago (Ed.), Issues in vowel harmony (201–236). Amsterdam:
John Benjamins.
Hall, D. C. (2001). Phonological contrast and its phonetic enhancement: dispersedness without dispersion.
Phonology 28(1), 1–54.
Halle, M. (1970). Is Kabardian a vowel-less language? Foundations of Language 6(1), 95–103.
Harrison, K. D. (2001). Vowel Harmony. In P. Strazny (Ed.), Encyclopedia of Linguistics. Chicago: Fitzroy
Dearborn.
VOWEL SPACES AND SYSTEMS
23
Harvey, M., Lin, S., Turpin, M., Davies, B., & Demuth, K. (2015). Contrastive and non-contrastive pre-
stopping in Kaytetye. Australian Journal of Linguistics 35(2), 232–250.
Hayashi, N. (2016). A phonological sketch of Akha Buli — A Lolo-Burmese language of Muang Sing,
Laos. Research Institute of Foreign Studies, Kobe City University of Foreign Studies, Researches in
Asian Languages Vol. 10, Annals of Foreign Studies Vol. 92 67–84.
Helimski, E. (1998). Selkup. In D. Abondolo (Ed.), Uralic languages (548–579). London and New York:
Routledge.
Henderson, J. (2002). The word in Eastern/Central Arrernte. In R. M. W. Dixon & A. Y. Aikhenvald (Eds.),
Word, A cross-linguistic typology (100–124). Cambridge: Cambridge University Press.
Jakobson, R., & Waugh, L. R. (2002). The Sound Shape of Language (3rd ed.). Berlin & New York: Mouton
de Gruyter.
Jany, C. (2007). Phonemic versus phonetic correlates of vowel length in Chuxnabán Mixe. Berkeley
Linguistics Society 33(2), 66–76.
Jendraschek, G. (2012). A grammar of Iatmul (Habilitationsschrift). Universität Regensburg, Regensburg,
Germany. Retrieved from https://www.academia.edu/1247243/A_grammar_of_Iatmul.
Karamat, N. (2002). Phonemic inventory of Punjabi. In CRULP Annual Student Report, 2001-2002 (179–
188). Lahore: Center for Research in Urdu Language Processing. Retrieved from
http://www.cle.org.pk/Publication/Crulp_report/CR02_21E.pdf.
Kato, A. (1995). The phonological systems of three Pwo Karen dialects. Linguistics of the Tibeto-Burman
Area 18(1), 63–103.
Kim, D. (2015). The Sarikoli reflexive pronoun. Open Linguistics 2015(1), 788–799.
Kinkade, M. D. (1963). Phonology and morphology of Upper Chehalis: I. International Journal of
American Linguistics 28(3), 181–195.
Kinkade, M. D. (1966). Vowel alternation in Upper Chehalis. International Journal of American Linguistics
32(4), 343–349.
Kirby, J. (2011). Vietnamese (Hanoi Vietnamese). Journal of the International Phonetic Association 41(3),
381–392.
Kossmann, M. G., & Stroomer, H. J. (1997). Berber phonology. In A. S. Kaye (Ed.), Phonologies of Africa
and Asia (461–475). Winona Lake, Indiana: Eisenbrauns.
Kuipers, A. H. (1960). Phoneme and morpheme in Kabardian (Eastern Adyghe). The Hague: Mouton and
Co.
Kuipers, A. H. (1976). Typologically salient features of some North-West Caucasian languages. Lisse:
Peter De Ridder Press.
Kumar, P. (2012). Descriptive and typological study of Jarawa (Doctoral Dissertation). Jawaharlal Nehru
University, New Delhi, India.
Laakso, J. (2001). The Finnic languages. In Ö. Dahl & M. Koptjevskaja-Tamm (Eds.), Circum-Baltic
Languages Vol. 1: Past and Present (179–212). Amsterdam & Philadelphia: John Benjamins.
Ladefoged, P., & Maddieson, I. (1996). The Sounds of the world’s languages. Oxford: Blackwell.
Lee, H., & Jongman, A. (2016). A diachronic investigation of the vowels and fricatives in Korean: An
acoustic comparison of the Seoul and South Kyungsang dialects. Journal of the International
Phonetic Association 46(2), 157–184.
Leeding, V. J. (1989). Anindilyakwa Phonology and morphology (Doctoral Dissertation). University of
Sydney, Sydney, Australia.
Lewis, P. (1968). Akha phonology. Anthropological Linguistics 10(2), 8–18.
Liljencrants, J., & Lindblom, B. (1972). Numerical simulation of vowel quality systems: The role of
perceptual contrast. Language 48(4), 839–862.
Lindskoog, J. N., & Brend, R. M. (1962). Cayapa Phonemics. In B. F. Elson (Ed.), Studies in Ecuadorian
Indian languages: I (31–44). Norman, Oklahoma: Summer Institute of Linguistics of the University
of Oklahoma.
Lobel, J. W., & Riwarung, L. H. S. (2011). Maranao: A preliminary phonological sketch with supporting
audio. Language Documentation & Conservation 5, 31–59.
DOUG HITCH
24
MacKenzie, S., & Dresher, E. (2003). Contrast and phonological activity in the Nez Perce vowel system.
Paper presented at the Annual Meeting of the Berkeley Linguistics Society 29(1), 283–294.
Maddieson, I. (1987). The Margi vowel system and labiocoronals. Studies in African Linguistics 18(3),
327–355.
Maddieson, I. (2009). Patterns of sounds. Cambridge University Press.
Maddieson, I., & Anderson, V. (1994). Phonetic structures of Iaai. UCLA Working Papers in Phonetics 87,
163–182. (Fieldwork Studies of Targeted Languages II.)
Maddieson, I., & Precoda, K. (ongoing). The UCLA Phonological Segment Inventory Database. Retrieved
from http://www.linguistics.ucla.edu/faciliti/sales/software.htm#upsid.
Matisoff, J. A. (2003). Aslian: Mon-Khmer of the Malay Peninsula. Retrieved from
http://stedt.berkeley.edu/pdf/JAM/ASLIAN.pdf.
Miyashita, M. (2011). Diphthongs in Tohono O’odham. Anthropological Linguistics 53(4), 323–342.
Montler, T. (1986). An outline of the morphology and phonology of Saanich, North Straits Salish.
(Hypertext edition 1999). ms. Missoula, MT: University of Montana Linguistics Laboratory.
http://www.cas.unt.edu/~montler/Saanich/Outline/index.htm.
Mortenson, D., & Keogh, J. (2011). Sorbung, an undocumented language of Manipur: Its phonology and
place in Tibeto-Burman. Journal of the Southeast Asian Linguistics Society 4(1), 64–114.
Mukherjee, A., Choudhury, M., Basu, A., Ganguly, N., & Chowdhury, S. R. (2008). Rediscovering the co-
occurence principles of vowel inventories: A complex network approach. Advances in Complex
Systems 11(3), 371–392.
Namkung, J. (1996). Phonological inventories of Tibeto-Burman Languages. Berkeley: University of
California.
Osumi, M. (1995). Tinrin Grammar. Honolulu: University of Hawaii Press.
Pensalfini, R. (2002). Vowel harmony in Jingulu. Lingua 112, 561–586.
Prost, A. (1964). Contributrion à l’étude des langues Voltaiques. Dakar: Institut Francophone de l’Afrique
Noire. Retrieved from Senoufo, Mamara Phonology:
https://archive.org/details/rosettaproject_myk_phon-1
Prost, G. R. (1967). Chacobo. In E. Matteson (Ed.), Bolivian Indian Grammar 1 (285–359). Norman,
Oklahoma: Summer Institute of Linguistics of the University of Oklahoma.
Pulleyblank, E. G. (1984a). Middle Chinese: A study in historical phonology. Vancouver: University of
British Columbia Press.
Pulleyblank, E. G. (1984b). Vowelless Chinese? An application of the three-tiered theory of syllable
structure. In M. K. Chan (Ed.), Proceedings of the Sixteenth International Conference on Sino-
Tibetan Languages and Linguistics: September 16-18, 1983, University of Washington (568–619).
Seattle: University of Washington.
Reetz, H. (undated). Simple interface to the UPSID database. http://web.phonetik.uni-
frankfurt.de/upsid.html: Last accessed 16 March 2016.
Reid, L., A. (1973). Diachronic typology of Philippine vowel systems. In T.A. Sebeok (Ed.), Diachronic,
areal, and typological linguistics (485–506). The Hague and Paris: Mouton and Co. Retrieved from
https://scholarspace.manoa.hawaii.edu/bitstream/10125/32977/1/A13.1973.pdf
Rice, K. (1995). On vowel place features. Toronto Working Papers in Linguistics 14(1), 73–116.
Roddy, K. (2007). A sketch grammar of Satawalese, the language of Satawal Island, Yap State, Micronesia
(MA). University of Hawai’i. Retrieved from
https://scholarspace.manoa.hawaii.edu/bitstream/handle/10125/20678/M.A.CB5.H3_3421_r.pdf?se
quence=2
Róna-Tas, A. (1997). Nutshell Chuvash. Erasmus Mundus Intensive Program, Turkic languages and
cultures in Europe (TLCE), Szeged, Hungary. Retrieved from
https://web.archive.org/web/20110807192758/http://www2.lingfil.uu.se/afro/turkiskasprak/IP2007/
NUTSHELLCHUVASH.pdf
Rood, D. S. (1975). The Implications of Wichita Phonology. Language 51(2), 315–337.
VOWEL SPACES AND SYSTEMS
25
Rood, D. S. (1996). Sketch of Wichita, a Caddoan Language. In I. Godddard (Ed.), Handbook of North
American Indians Vol. 17 (580–608). Washington: Smithsonian Institution.
Rosario, Jr., F. C. (2012). The Vowel Space of Pangasinan. Frontiers of Language and Teaching 3, 264–
270.
Ruhlen, M. (2004). Global Typological Database: A Description. ms. Retrieved from
http://starling.rinet.ru/Texts/typology.pdf
Saint, R., & Pike, K. (1962). Auca Phonemics. In B. F. Elson (Ed.), Studies in Ecuadorian Indian
Languages: I. Norman, Oklahoma: Summer Institute of Linguistics of the University of Oklahoma.
Salminen, T. (2007). Notes on Forest Nenets phonology. Mémoires de La Société Finno-Ougrienne 253,
349–372. Retrieved from http://www.sgr.fi/sust/sust253/sust253_salminen.pdf
Schuh, R., G., & Yalwa, L. D. (1993). Illustrations of the IPA: Hausa. Journal of the International Phonetic
Association 23(2), 77–82.
Schwartz, J.-L., Boë, L.-J., Valée, N., & Abry, C. (1997). Major trends in vowel system inventories. Journal
of Phonetics 25, 233–253.
Silverman, D., Blankenship, B., Kirk, P., & Ladefoged, P. (1995). Phonetic Structures in Jalapa Mazatec.
Anthropological Linguistics 27(1), 70–88.
Simeone-Senelle, M.-C. (1997). The Modern South Arabian Languages. In R. Hetzron (Ed.), The Semitic
Languages (pp. 378–423). London: Routledge.
Staalsen, P. (1966). The Phonemes of Iatmul. In C. I. Frantz, M. E. Frantz, D. Oatridge, R. E. Loving, J.
Swick, A. Pence, P. Staalsen, H. Boxwell, & M. Boxwell (Eds.), Papers in New Guinea Linguistics
No. 5 (69–76). Canberra: Pacific Linguistics, The Australian National University.
Staalsen, P. (1992). Iatmul Organised Phonology Data. Retrieved from http://www-
01.sil.org/pacific/png/pubs/928474542478/Iatmul.pdf.
Street, C. S., & Mollinjin, G. P. (1981). The phonology of Murinbata. Working Papers of SIL-AAB 5, 183–
244.
Suomi, K., Toivanen, J., & Ylitalo, R. (2008). Finnish sound structure: Phonetics, phonology, phonotactics
and prosody. Oulu, Finland: University of Oulu.
Takashi, N. (2009). The structure of the Ponapean language. Nagoya University of Foreign Studies Foreign
Language Department Bulletin (名古屋外国語大学外国語学部紀要) 37, 101–118.
Thompson, L. C., & Thompson, M. T. (1966). A Fresh Look at Tillamook Phonology. International Journal
of American Linguistics 32(4), 313–319.
Thompson, L. C., Thompson, M. T., & Efrat, B. S. (1974). Phonological Developments in Straits Salish.
International Journal of American Linguistics 40(3), 182–196.
Tłı̨chǫ Community Services Agency. (2007). Reading and Writing in Tłı̨chǫ Yatıì. Behchokǫ̀, Northwest
Territories, Canada: Tłı̨chǫ Community Services Agency.
Topping, D. M. (1968). Chamorro Vowel Harmony. Oceanic Linguistics 7(1), 67–79.
Topping, D. M. (1973). Chamorro Reference Grammar. Honolulu: University of Hawaii Press.
Traunmüller, H. (1983). On vowels. Perception of spectral features, related aspects of production and
sociophonetic dimensions (Doctoral dissertation). University of Stockholm, Stockholm, Sweden.
Trubetskoy, N. S. (1939). Grundzüge der Phonologie. Prague: [Jednota Československých Matematiků a
Fysiků].
Unseth, C. (2010). Vowel Harmony in Wolof. Dallas: Graduate Institute of Applied Linguistics. Retrieved
from http://www.gial.edu/documents/opal/No-7-Unseth-Wolof-Vowel-Harmony.pdf.
Ward, M., Sánchez, E. Z., & Marlett, S. A. (2008). Zapoteco de Santa Catarina Quioquitani. In S. A. Marlett
(Ed.), Ilustraciones fonéticas de lenguas amerindias (preliminary edition). Lima: SIL International
y Universidad Ricardo Palma. Retrieved from
http://lengamer.org/publicaciones/trabajos/zapoteco_de_Quioquitani_afi.pdf.
Wilson, H. (2003). A brief introduction to Marshallese phonology. ms. UCLA, Los Angeles, CA. Retrieved
from http://linguistics.byu.edu/faculty/hwills1/ABriefIntroductiontoMarshallesePhonology.pdf.
DOUG HITCH
26
Appendix A: Examples of languages in each vowel system Triangular three-vowel systems
Language name(s) Language family / geographic
location Phonemes Reference
Jukun Niger-Congo, NE Nigeria /i u a/ R491
Ngizim Chadic, Nigeria short /i u a/ &
long /eː oː aː/
U4369
Central Atlas
Tamazight
Berber, Morocco /i u a/ Kossmann & Stroomer,
1997: 463
Eastern Arabic Arabic, Saudi Arabia /i u a/ R1564
Tunisian Arabic Arabic, Tunisia short /ɪ ʊ ə/ &
long /iː uː æː/
R1573
Moroccan Arabic Arabic, Morocco short /e o ɐ/ &
long /iː uː aː/
R1575
Lak NE Caucasian, Russia /i u a/ R1625
Aleut Eskimo-Aleut, Russia-USA /i u a/ R2008
Brahui Dravidian, Pakistan-Afghanistan /i u a/ R2021
Thao Formosan, Taiwan /i u a/ R2608
Pirahā Mura, Brazil /i o a/ U6802
Gadsup Kainantu, Papua New Guinea short /i u ɜ/ &
long /eː oː aː/
U8608
Chukchi Chukotko-Kamchatkan, E Russian /i u e₁/10
Jakobson & Waugh,
2002: 149–150
Nunggubuyu Arnhem, NE Australia /i u ɐ/ R4372
Gadang ~ Worimi Pama-Nyungan, New S Wales short /ɪ ʊ ə/ &
long /iː uː aː/
R4538
Coast Tsimshian Tsimshianic, British Columbia /i u a/ U6774
Blackfoot Algonquian, Alberta-Montana /i o a/ R4689
Cheyenne Algonquian, Montana /e o a/ R4690
Miccosukee ~
Mikasuki
Muskogean, Florida /i o a/ R4848
Sierra Totonac Totonacan, Puebla~Veracruz /i u a/ R4853
Alabama Muskogean, Texas /e o a/; U6759;
/i o a/ W
Amuesha ~ Yanesha’ Arawakan, Peru /e o a/ U6824
Jaqaru Aymaran, Peru /i ɯ a/ U6820
Yanyuwa Pama-Nyungan, N Australia /ɪ ʊ a/ U8347
Yolngu Pama-Nyungan, N Australia /ɪ ʊ ɐ/ U8375
10 “Recessive” vowels harmonically alternate with /e₂ o a/ “dominant” vowels; (/ə/ is also dominant).
VOWEL SPACES AND SYSTEMS
27
Four-vowel systems
Type 4a
Language name(s) Language family / geographic
location Phonemes Reference
Hassaniya Arabic Arabic, Mauritania short /ɪ ʊ ə ɐ/ &
long /iː uː aː/
R1576
Kerek Chukotko-Kamchatkan, E Russia /i u ə a/ R2004
Central Yupik Eskimo-Aleut, Alaska /i u ə a/ R2014
Budai Rukai Formosan, Taiwan /i u ə a/ Chen, 2006: 231
Paiwan Formosan, Taiwan /i u ə a/ Chen, 2006: 69
Pangasinan Malayo-Polynesian, N
Philippines
/i ʊ ə a/ Rosario, 2012: 269
Dena’ina ~ Tanaina Athabaskan, Alaska /i u ə a/ Boraas, 2010: 5, with <e> for /ə/
Penobscot E
Abenaki
Algonquian, Maine /i o ə a/; R4705;
/i̝ u~o ə~ε ɑ/ Abenaki-Penobscot
Pronunciation Guide 2017
Comox Salishan, SW British Columbia /i u ə a/; R4733;
/i~ε u~ɔ ə a/ W
Upper Chinook Chinookan, Washington /i u ə a/ R4800
Siuslaw Oregon Coast Penutian (?),
Oregon
/i u ə a/ R4807
Ivatan Malayo-Polynesian, N
Philippines
/i u ə a/ U2428
Jebero Cahuapanan, Peru /i u ə a/ U6844
Lushootseed Salishan, Washington State /ɪ ʊ ə a/ U6734
Upper Chehalis Salishan, Washington State /e o ə a/ Kinkade, 1963: 193, 1966: 343
Yupik Eskimo-Aleut, Alaska-Siberia /i u ə a/ U6902
Type 4b
Language name(s) Language family /
geographic location Phonemes Reference
Kitja ~ Kija Jarrakan, NW Australia /i ɨ u a/ R4421
Arritinngithigh Paman, N Australia /i ʉ u æ a/ R4460
Maranao Malayo-Polynesian, S
Philippines
/i ɨ o a/; U2445;
/i o ə a/ Lobel & Riwarung, 2011: 35–36
Chacobo ~ Chakobo ~
Chácobo-Pakawara
Panoan, Bolivia /i ɨ o a/ Prost, 1967: 285
DOUG HITCH
28
Type 4c
Language name(s) Language family / geographic
location Phonemes Reference
Ancient Egyptian Egyptian, Egypt /i u e a/ R1317
Dargwa Russia, NE Caucasian /i u e a/ R1626
Turkish Turkic, Turkey front /i y e ø/;
back /ɯ u a o/11
R1935; W
Mzieme Kamarupan Tibeto-Burman, NE India /i u e a/ TB:272
Alawa Arnhem (?), N Australia /i ɯ ɛ a/ R4397
Lardil Pama-Nungan, N Australia /i u e a/ R4448
Koyukon Athabaskan, Alaska /i u æ ɔ/ Axelrod, 1990: 180
Tsuut’ina ~ Sarsi Athabaskan, Alberta /ɪ ʊ a ɒ/ R4678
Arapaho Algonquian, Wyoming-Oklahoma /i u e o/ Goddard, 1979: 115
Pawnee Caddoan, Oklahoma /ɪ ʊ ɛ ɐ/ R4785
Tuscarora Iroquoian, Ontario-New York-
North Carolina
/ɪ ʊ ɛ ɑ/ R4795
Cayapa ~ Cha’palaa Barbacoan, Ecuador /i ʊ ɛ a/ Lindskoog & Brand,
1962: 32, 38
Alawa Arnhem (?), N Australia /ɪ ɯ̞ e a/ U8354; or 4d
Bandjalang Pama-Nyungan, New S Wales /i u e a/ U8368
Moxo Arawakan, Bolivia /i u ɛ a/ U6827
Murrinh-patha S Daly (?), N Australia /i ʊ ɛ a/12 U8349
Quileute Chimakuan, Washington State /i u æ a/; U6732;
short /e o a/ &
long /eː oː æː aː/
W
Shasta Hokan, California-Oregon /i u e a/ U6746
Tiwi Isolate, N Australia /i u a ɔ/; U8351;
/ɪ ʊ a ɑ/; R4373;
/i u a o/ W
11 Turkish harmonically contrasts front /i y e ø/ with back /ɯ u a o/. 12 Street & Mollinjin (1981: 199-200) state /ɛ/ also has allophones [æ, e].
VOWEL SPACES AND SYSTEMS
29
Type 4d
Language
name(s)
Language family / geographic
location Phonemes Reference
Bontok Malayo-Polynesian, N
Phillipines
/i e o a/ R2653
Malagasy Malayo-Polynesian, Madagascar /i e o a/ U2410
Klamath Plateau Penutian, Oregon-
California
/ɪ ɛ o ɐ/ R4808
Seneca Iroquoian, NY State-Ontario-
Oklahoma
/i e o a/ with [u] in vowel
symbolism, & nasal /e o a/
Chafe, 2014: 8
Dogrib Athabaskan, N Canada /i ɛ o a/ Tłı̨chǫ Community Services
Agency, 2007: 10–14
Navajo Athabaskan, SW USA /ɪ ɛ ɔ a/ R4683
Fox Algonquian, C USA-N Mexico /i ε o a/ R4700
Mohawk Iroquoian, Ontario-Quebec-NY /i ɛ o a/ R4790
Nahuatl Uto-Aztecan, Mexico /i ɛ o a/ U6762
Machiguenga
Campa
Arawakan, Peru /i e o a/ U6825
Tacana Tacanan, Bolivia /i e o a/ U6812
DOUG HITCH
30
Five-vowel systems
Type 5a
Language name(s) Language family /
geographic location Phonemes Reference
Hadza Khoisan, N Tanzania /i u e o a/ R1
Soninke Mande, Mali /i u ɛ o a/ R77
Mandinka Mande, Gambia /i u ɛ ɔ a/ R86
Swahili ~ Kiswahili Bantu, E Africa /i u ɛ ɔ ɑ/ W
Hausa Chadic, Niger-Nigeria-Chad /i u e o a/ Schuh & Yalwa, 1993: 78
Musey Chadic, N Cameroon /i u e o a/ R1358
Kafa Omotic, SW Ethiopia /i u e o a/ R1490
Hadiyya Cushitic, C Ethiopia /i u e o a/ R1524
Basque Isolate, N Spain-S France /i u e o a/ R1600
Ingush NE Caucasian, SW Russia /i u e o a/ R1609
Burushaski Isolate, N Pakistan /i u e o a/ R1640
Dari Iranian, N Afghanistan long /iː uː eː oː aː/ &
short /i u a/
R1780
Castillian Spanish Romance, Spain /i u e o a/ R1823
Lithuanian Baltic, Lithuania short /ɪ ʊ ɛ o a/ &
long /iː uː eː oː æː aː/
R1874
Tamil Dravidian, S India /ɪ ʊ e ɔ ɐ/ R2048
Baining East Papuan, Papua New
Guinea
/i u ɛ o a/ U8631
Batak Palawan, Phillipines /i u ɛ o a/ U2413
Even Tungusic, E Russia /ɪ u ɛ o ɑ/ U2167
Hamer Omotic, Ethiopia /i u e o ɑ/ & pharyngealized
/ɪˤ ʊˤ eˤ ɔˤ ɐˤ/
U4365
Hawaiian Polynesian, Hawaii /i u ɛ o a/ U2424
Kharia Munda, Bihar India /i u e o a/ U2301
Yucuna Arawakan, Colombia /i u ɛ o a/ U6843
Gunwinygu ~
Kunwinjku
Arnhem, N Australia /i u ɛ ɔ a/ R4389
Santiam Kalapuya Kalapuyan, Oregon /i u e o a/ R4802
Patwin Wintuan, California /i u e o a/ R4815
Zuni Isolate, New Mexico /i u ɛ ɔ a/ R4838
Tonkawa Isolate, Oklahoma-Texas-
New Mexico
short /i u e o a/ &
long /iː uː eː oː aː/
U6752
Resigaro Arawakan, Peru /i ɯ ɛ ɔ a/ U6838
Island-Carib Arawakan, Lesser Antilles /i u e ɤ a/ U6823
Ocaina Bora-Huitoto, Peru-
Columbia
i ɯ ɛ o a/ U6805
VOWEL SPACES AND SYSTEMS
31
Type 5b
Language name(s) Language family /
geographic location Phonemes Reference
Forest Nenets Samoyedic, N Russia unstressed syllables /i u ə æ ɑ/
(stressed is 6c /i u e o æ ɑ/
Salminen, 2007: 366
Wadjiginy ~ Patjtjamalh
~ Bachamal
Wagaydyic, N
Australia
/i u ɵ e ɑ/; R4411;
/i u ø e a/ Ford, 1990: 35
Maranungku ~ Marranj Western Daly, N
Australia
/i ʊ ɵ æ a/ R4412
Matngele ~ Madngele Eastern Daly, N
Australia
/i u ɵ e ɑ/ R4413
Tolowa Athabaskan, Oregon /i u ə e a/; R4658;
allophony [u~o], [a~ɔ], [ə~ɨ~ʉ] W
Potawotami Algonquian, Michigan-
Kansas
/i o ə ɛ a/; R4697;
with [u~o] allophony W
Diegueño Yuman, California short /ɪ ʊ ə a ɔ/ &
long (4c) /eː oː aː ɔː/
U6743
Type 5c
Language name(s) Language family / geographic location Phonemes Reference
Mongsen Ao NE India, Tibeto-Burman /i ʉ u ə a/13 Coupe, 2008: 2
Sarangani Manobo Malayo-Polynesian, S Philippines /i ɨ o ʌ a/ Reid, 1973: 13
Type 5d
Language name(s) Language family / geographic
location Phonemes Reference
Latvian Baltic, Latvia /ɪ ʊ e æ a/; R1872;
[ɔ] in loans W
Waorani ~ Auka Isolate, E Ecuador /i e o æ a/ Saint & Pike, 1962: 4
Type 5e
Language name(s) Language family / geographic
location Phonemes Reference
Cofan Isolate, Ecuador-Columbia /i ɨ ɛ o a/ U6836
Hixkaryana Carib, Brazil /e ɯ u æ ɔ/ U6842
MalakMalak Isolate, N Australia /i ɨ u ɛ ɐ/ Dorothea Hoffmann, p.c.
27Feb2017
13 Other Ao varieties are type 4a; all creaky voice /a̰/
DOUG HITCH
32
Six-vowel systems
Type 6a
Language name(s) Language family / geographic location Phonemes Reference
Moro Heiban, Sudan /i u e ə o a/ Guest, 1997: 1
Rashad Niger-Congo, Sudan /i u e ə o a/ R71
Badyara Senegambian, Guinea /i u e ə o a/ R132
Lemoro Kainji, Nigeria /i u e ə o a/ R557
Meidob Nubian, NW Sudan /ɪ u ɛ ə ɔ a/ R1188
Gaam E Sudanic, Sudan /i u e ə o a/ R1205
Coptic Egyptian, Egypt /i u e ə o a/ R1318
Ga’anda Chadic, NE Nigeria /i u e ə o a/ R1389
East Gurage Ethiopic, EC Ethiopia /i u e ə o a/ R1587
Svan Kartvelian, NW Georgia /i u e ə o a/ R1597
E Armenian Armenian, Armenia /i u e ə o a/ R1654
Yidgha Iranian, NE Afghanistan /i u e ə o a/ R1743
Bulgarian Slavic, Bulgaria /i u ɛ ə ɔ a/ U2009
Tundra Yukaghir Uralic, NE Russia /i u e ø o a/ R1890
Magar Tibetic, Nepal /i u e ə o a/ R2132
Meithei ~ Manipuri Sino-Tibetan, NE India /i u e ə o a/ Chelliah, 1997: 21
Gaddang Cagayan Valley, N Phillipines /i u e ə o a/ R2626
Asmat Trans-New Guinea, W Papua /i u e ə ɔ a/ U8601
Bodo Sino-Tibetan, NE India /i u e ə o a/ U2515
Kera Chadic, Chad /i u ɛ ə ɔ a/ U4371
Kwak’wala Wakashan, British Columbia /i u ɛ ə ɔ a/ U6731
Tobo-Kube Finisterre-Huon, E Papua New Guinea /i u e ə o a/ R3632
Yir Yoront Paman, NE Australia /i u e ə o a/ R4471
Mi’kmaq ~ Micmac Algonquian, E Canada-E USA /i u e ə o a/ R4703
Lebanese Arabic Arabic, Lebanon /i u e ə o a/ R1568
Hopi Uto-Aztecan, Arizona /i ɯ e ø o a/ Grune, 1997: 3
VOWEL SPACES AND SYSTEMS
33
Type 6b
Language name(s) Language family / geographic
location Phonemes Reference
Tera Chadic, NE Nigeria /i ɨ u e o a/ R1387
Gude Chadic, NE Nigeria /ɪ ɨ ʊ ɛ ɔ a/ R1420
Khwarshi / Xvarshi NE Caucasian, Russia /i ɨ u e o a/ R1619
Hinukh / Hinux NE Caucasian, Russia /i y u e o a/ 1621
Sorbung Tibeto-Burman, NE India /i ʉ u e o a/ Mortenson & Keogh, 2011: 87
Chokri Tibeto-Burman, NE India /i ɨ u e o a/ Bielenberg & Nienu, 2001: 98
Muinane Bora-Witoto, Colombia /i ɨ u ɛ o a/ U6806
Ngalkbon ~ Dalabon Arnhem, N Australia /i ɨ u ɛ ɔ a/ R4390
Mbiywom Paman, NE Australia /i y u e o a/ R4462
Mbabaram Paman, NE Australia /i ɨ u ɛ ɔ a/; W;
/i ɯ u a/ (4b); R4493;
/i u ɛ ɤ ɔ a/ (6a) U8373
Maidu Maiduan, California / i ɨ u ɛ ɔ a/ R4818
Southern Sierra
Miwok
Utian, California /ɪ ɨ ʊ ɛ ɔ a/ R4835
Huave Isolate, Oaxaca /i ɨ u e o a/ R4852
Sierra Popoluca Mixe-Zoquean, S Veracruz /i ɨ u e o a/ Elson, 1992: 586
Sarikoli Shughni Iranian, W Xinjiang /i y u e o a/ R1747
Kyungsang Korean Korean, Korea /i ɨ u e o a/ Lee & Jongman 2016: 158–160
DOUG HITCH
34
Type 6c
Language name(s) Language family /
geographic location Phonemes Reference
Kxoe Khoisan, NE Namibia /i u e o ɛ a/ R26
Higi, Nkafa dialect Chadic, NE Nigeria /i u e o ɛ a/ R1397
Murle Didinga East Sudanic, Sudan /i u e o ɛ a/ R1193
Anuak E Sudanic, SW Ethiopia /i u ɛ o a ɔ/ R1225
Dameli Dardic, Pakistan /ɪ u ɛ ɔ æ ɑ/ R1674
Tehrani Persian Persian, Iran /i u e o æ ɒ/ W
Slovak Slavic, Slovakia /i u e o æ a/ R1884
Uzbek Turkic, C Asia /i u e o a ɑ/ R1942
Telugu Dravidian, SE India long /iː uː eː oː æː aː/ &
short (5a)
R2035
Hakka Chinese, S China-Taiwan /i u ɛ ɔ æ a/ U2502
Chamorro Malayo-Polynesian, Guam /i u e o æ ɑ/ Topping, 1973: 16–24
Kate Finisterre-Huon, Papua New
Guinea
/i u e o a ɔ/ R3616
Chipewyan ~
Dënesųłiné
Athabaskan, N Canada /i u e o ɛ a/ W
Menominee Algonquian, Wisconsin /i u e o ɛ a/ R4698
Yuchi Isolate, Tennessee-Oklahoma /i u e o æ a/14 Crawford, 1973: 174–
175
Ashuslay ~ Nivaclé Matacoan, Argentina /i u e o a ɑ/ U6814
Zapoteco de Santa
Catarina Quioquitani
Zapotec, Oaxaca /i u e o ɛ a/ Ward et al., 2008: 4
14 Crawford’s /ω/ is a product of contraction and is non-phonemic (Crawford 1973: 175).
VOWEL SPACES AND SYSTEMS
35
Seven-vowel systems
Type 7a
Language name(s) Language family / geographic
location Phonemes Reference
Kanuri Nilo-Saharan, NE Nigeria /i ɨ u e ə o a/ R1175
Tigrinya Ethiopic, N Ethiopia /i ɨ u e ə o a/ R1583
Ket Yeniseian, C Russia /i ɨ u ɛ ə ɔ a/ W
Wakhi Iranian, Afghanistan /i ɨ u e ə o a/ R1741
Sarikoli Iranian, W Xinjiang /i ɯ u ɛ ə o a/ system implied in Kim 2015
Albanian Albanian, Albania /i y u ɛ ə ɔ a/ R1789
Romanian Romance, Romania /i ɨ u e ə o a/ Chitoran, 2002: 204
Welsh Celtic, Wales short /ɪ ɨ ʊ ɛ ə ɔ a/
& long (6b)
R1847
Udmurt Finno-Ugric, NW Russia /i ɨ u ɛ ə o a/ R1904
Komi-Zyrian Finno-Ugric, NW Russia /i ɨ u e ə o a/ R1905
Nishi Tibeto-Burman, N India-
Assam
/i ɨ u e ə o a/ R2111
Central Tibetan Tibetic, C Tibet /i y u e ø o ɑ/ R2125
Apatani Tibeto-Burman, NE India /i ɯ u e ə o a/ TB:24
Sa’ban North Sarawakan, Borneo /i ɨ u ɛ ə ɔ a/ U2415
Skou Skou, W Papua /i ʉ u ɛ ø ɔ a/ Donohue, 2004: 46
Jarawa Ongan, Andaman Islands /i ɨ u e ə o a/ Kumar, 2012: 40
Seoul Korean Korean, Korea /i ɨ u e~ɛ ʌ o a/ Lee & Jongman, 2016: 158–160
Type 7b
Language name(s) Language family / geographic location Phonemes Reference
Tamahaq Tuareg, S Algeria /i u e ə o æ ɐ/ R1326
Tigre Ethiopic, Ethiopia /i u e ə o æ a/ R1582
Eastern Pashto Iranian, Afghanistan-Pakistan /i u e ə o æ ɑ/ R1740
Moksha Mordvin Finnic, W Russia /i u e ə o æ a/ R1910
Amdo Zeku
Tibetan
Tibetic, Qinghai /i u e ə o a ɔ/ TB:390
Konyagi ~ Wamey Senegambian, Senegal-Gambia /i u ɛ ə ɔ æ a/ U4145
Wantoat Finisterre-Huon, E Papua New Guinea /i u e ə o æ ɑ/ Davis, 2004: 2
Drehu ~ Lifu Oceanic, New Caledonia /i u e ø o æ ɑ/ R3424
Sentani~Buyaka Sentani, W Papua /i u e ə o ɛ a/ Cowan, 1965: 4
DOUG HITCH
36
Type 7c
Language
name(s)
Language family /
geographic location Phonemes Reference
Swedish North Germanic,
Sweden
long /iː ʉ: u: e: o: ɛ: ɑ:/ with /yː øː/ on a
separate plane, & short (6a) /ɪ ʊ ɛ ɵ ɔ a/
with /ʏ œ/ on a separate plane
See above §2.4;
Bruce & Engstrand,
2006: 20
Eastern Saami Finnic, NW Russia /i ɨ u ɛ o a ɑ/ R1912
Tulu Dravidian, SW India short /i ɯ u e o ɛ a/ & long (6c) R2037
Kwoma Sepik, Papua New
Guinea
/i ɨ u e o ε a/ U8602
Chuxnabán Mixe Mixe-Zoque, Oaxaca /i ɨ u e o æ a/ Jany, 2007: 67
Ahi Tibeto-Burman, Yunnan constricted & unconstricted
/i ɯ u e o ɛ ɑ/
TB:8
Norwegian North Germanic,
Norway
long /iː ʉː uː eː oː æː ɑː/ &
short (8b)
U2006
Kwoma ~
Wahkuk
Middle Sepik, Papua
New Guinea
/i ɨ u e o ɛ a/ R4253
Type 7d
Language name(s) Language family /
geographic location Phonemes Reference
Hindi Indic, Uttar Pradesh long /iː uː eː oː æː aː ɔː/ &
short /ɪ ʊ ə/
R1704
Calcutta Bengali Indic, NE India /i u e o æ a ɔ/ R1721
Italian Romance, Italy /i u e o ɛ a ɔ/ R1809
Galician Romance, Spain /i u e o ɛ a ɔ/ R1832
Yoruba Volta-Niger, SW Nigeria /i u e o ɛ a ɔ/ R447
Newari Tibeto-Burman, Nepal /i u e oæ a ɔ/ TB:294
Tunica Isolate, Louisiana /i u e o ɛ ɔ a/ R4840
Ponapean Oceanic, F S Micronesia /i u e o ɛ a ɔ/ Takashi, 2009: 104
Dangla Chadic, SC Chad /i u e o ɛ a ɔ/ R1373
Vai Mande, NW Liberia /i u e o ɛ ɔ a/ R83
Kikuyu Bantoid, Kenya /i u e o ɛ ɔ a/ R933
VOWEL SPACES AND SYSTEMS
37
Eight-vowel systems
Type 8a
Language
name(s)
Language family / geographic
location Phonemes Reference
Murle E Sudanic, Ethiopia /ɪ ʋ e ə o ɛ a ɔ/ R1195
Dinik E Sudanic, Sudan /i u e ə o ɛ a ɔ/ R1210
Kenga Central Sudanic, SC Chad /i u e ə o ɛ a ɔ/ R1291
Zenaga Berber, SW Mauritania /i u e ə o ɛ a ɔ/ R1329
Jibbāli ~ Shehri South Arabian, Oman /i u e ə o ɛ a ɔ/ Simeone-Senelle, 1997: 384
Catalan Romance, Spain-Andorra-
France-Italy
/i u e ə o ɛ a ɔ/ R1822
Orok Tungusic, SE Russia long /iː uː eː əː oː ɛː aː ɔː/
& short (7b)
R1992
Thado Kukish, India-Burma /i u e ə o ɛ a ɔ/ R2231
Amwi Mon-Khmer, NE India /i u e ə o ɛ a ɔ/ R2386
Yapese Oceanic, Yap Island /i u e ɵ o ɛ a ɑ/ R2618
Javanese Sunda-Sulawesi, Indonesia /i u e ə o ɛ a ɔ/ U2409
Koma ~ Komo Nilo-Saharan, Ethiopia-Sudan /i u e ə o ɛ a ɔ/ U4220
Morwap ~ Elseng Isolate, W Papua /i u e ə o ɛ a ɔ/ R3609
Abau Sepik, Papua New Guinea /i u e ə o ɛ a ɔ/ R4236
Ulithian Oceanic, Federated States of
Micronesia
/i u e ə o æ ɐ a/ W
Konkani Indic, Mysore /i u e ə o ɛ a ɔ/ R1689
Vanimo Skou, W Papua New Guinea /i u e ɵ o ɛ a ɔ/ U8640
Iaai Austronesian, New Caledonia /i u e ɤ o æ a ɔ/ & front
rounded /y ø/
Maddieson & Anderson,
1994: 164
Ewondo Bantu, Cameroon /i u e ə o ɛ a ɔ/ U4141
Minyanka ~
Mamara ~
Mianka
Senufo, Mali /i u e ə o ɛ a ɔ/ & front
rounded /y œ/
Prost, 1964: 125
DOUG HITCH
38
Type 8b
Language
name(s)
Language family /
geographic location Phonemes Reference
Chechen NE Caucasian, Russia /i y u e ø o æ a/ R1608
Bezhta NE Caucasian, Russia /i y u e ø o æ ɑ/ R1622
Norwegian Germanic, Norway short /ɪ ʉ ʊ e ə ɒ̝ æ a/ & long (7c); U2006;
short /ɨ ʉ ʊ ɛ ə ɔ æ ɑ/ W
Fe’Fe’ Bantoid, Cameroon /i ɯ u e ɤ o a̘ ɑ/ U4148
Finnish Finno-Ugric, Finland-
Sweden-Norway-Russia
/i y u ɛ ø o æ a/; U2153;
/i y u ɛ ø o æ ɑ/ W
Selkup Samoyedic, Siberia short tense /i ɨ u e ə o æ a/ &
secondary /y ø/
Helimski, 1998: 552–553
Kashmiri Dardic, India-Pakistan /i ɨ u e ə o a ɔ/ R1685
Woleaian Oceanic, Federated States
of Micronesia
long /iː ʉː uː ɛː ɵː oː ɐː ɒː/ &
short (5a)
Ellis, 2012: 146–147
Khmer Austroasiatic, Cambodia short vowels /ɪ ɯ ʋ ɛ ɤ ɔ a̘ ɑ/15 U2306
Type 8c
Language name(s) Language family /
geographic location Phonemes Reference
Tinrin Oceanic, S New Caledonia /i ɯ u e o ɛ a ɔ/16 Osumi 1995: 13–15
Cayuvava Isolate, Bolivia short & long /i ɨ u ɛ o æ a ɔ/ U6840
15 Current descriptions give four heights for the long vowels. 16 The phonemic status of /ɛ/ is described as doubtful.
VOWEL SPACES AND SYSTEMS
39
Nine-vowel systems
Language name(s) Language family / geographic
location
Phonemes Reference
Bagirmi Central Sudanic, SW Chad /i ɨ u e ə o ε a ɔ/ R1286
Kurumba Dravidian, SW India /i ɨ u e ə o ε a ɔ/ R2043
Sgaw Karen, SE Myanmar /i ɨ u e ə o ε a ɔ/ R2095
Phlong ~ Northern
Pwo
Karen, NW Thailand /i ɨ u e ə o æ a ɔ/ R2085
Bisu Southern Loloish; N Thailand /i ɯ u e ɤ o ɛ a ɔ/ TB:49
Biyue Hani Southern Loloish, Yunnan /i y u e ə o ɛ a ɔ/ TB:132
Lahu Central Loloish, Thailand-
Burma-Yunnan
/i ɨ u e ə o ε a ɔ/ TB:192
Lawa Mon-Khmer, NW Thailand /i ɨ u e ə o ε a ɔ/ R2403
Mal ~ Thin Mon-Khmer, N Thailand /i ɨ u e ə o æ a ɔ/ R2419
Hanoi Vietnamese Vietic, N Vietnam /i ɯ u e ɤ o ɛ a ɔ/ Kirby, 2011: 384
Pacoh Katuic, Vietnam-Laos /i ɨ u e ə o ε a ɔ/, with +RTR
mid vowel plane /e̙ ə̙ o̙/
Alves, 2001: 31
Northern Temiar Aslian, Malaysia /i ʉ u e ə o ɛ a ɔ/ Matisoff, 2003: 14
Lao Tai, Laos /i ɨ u e ə o ε a ɔ/ R2573
Satawalese Micronesian, Caroline Islands short /i ʉ u ɛ ɞ o æ a ɒ/
& long (5a)
Roddy, 2007: 28–29
Saipan Carolinian Micronesian, Caroline Islands long & short /i ʉ u e ɵ o æ ɐ ɒ/ Ellis, 2012: 152–153
Polowat Micronesian, Caroline Islands long & short /i ɨ u e ɵ o æ ɐ ɒ/ Ellis, 2012: 150–151
Cham Chamic, Cambodia-Vietnam /i ɨ u e ə o ε a ɔ/ U2411
Andoke ~ Andoque Isolate, Columbia /i ɨ u e ə o æ ɜ ɒ/; U6851;
/i ɨ u e ə o a ʌ ɒ/ W
Totonpec Mixe Mixe-Zoque, Oaxaca /i ɨ u e ə o æ a ɔ/17 Elson 1992: 589
Po-ai Tai, SE Yunnan long vowels /i ɨ u e ə o ɛ a ɔ/ R2554
Gullah ~ Sea Island
Creole
English Creole, Georgia /i ɨ u e ə o ɛ a ɔ/ R5689
Akha Loloish, Laos-Burma-
Thailand-China
/i ɯ u e ɤ o ɛ a ɔ/ & a
secondary plane with /y œ/18
Lewis, 1968: 10
17 R4859 has /ʊ/ for /o/. 18 Hayashi, 2016: 75 with /ø/ for /œ/.
Recommended