Last update: 26 October 2006
This page is intended for conlangers who are looking for inspiration in choosing the vowel system of their language. It is not intended to be an exhaustive typology of every vowel system in existence, rather a description of several of those currently existing.
Of course, if you have an interesting vowel system you want me to add, just let me know.
The sound of a vowel is governed by many factors, of which the position of the tongue and rounding of the lips are the most important. The "vowel space" is thus defined as the total area over which the tongue position ranges, along the two axes of height and backness; the Language Construction Kit's section on vowels has an illustration showing how some English vowels fit into it.
Lip-rounding, for the purposes of this page, is assumed to be either "on" or "off". Other factors which affect vowel sounds, such as nasalisation and creaky voice, are not considered here. Vowel length is ignored for individual vowels, but complete systems of long vowels are sometimes considered separately where they differ from short vowels (e.g. in Sanskrit). Diphthongs are not considered; the simple vowels at the start and endpoints of diphthongs tend to be part of the vowel system anyway.
front central back
high i y i- u- w u
I Y W U
high-mid e ø 9 v o
@ 8
low-mid E Ø 3 V O
æ &
low a A D
Note the convention that vowel represented by an uppercase letter is lower (or "opener") than the vowel with the corresponding lowercase letter, with the exception of /a A/. If both /a/ and /A/ are present, /a/ is always front and /A/ back; otherwise /a/ may be considered somewhere between the two for convenience. Similarly for /e o ø v/, where the uppercase letter always denotes a lower vowel.
Except where noted, my sources are David Crystal's The Encyclopedia of Language [EofL], George Campbell's Concise Compendium of the World's Languages, and this brief look at a few vowel systems.
References to "universals" come from pages 2 and 3 of this list.
Some reconstructions of Proto-Indo-European (herafter PIE), following the "laryngeal" theory initially formulated by de Saussure, had the remarkable consequence that the PIE vowel system could be reduced to just /e/, although this is somewhat untenable; all languages have at least three vowels.
Some other reconstructions of PIE (for example, Comparitive Indo-European Linguistics by Robert Beekes) contain two vowels, /e/ and /o/. The now extinct Ubykh, a North Caucasian language, is often said to have had just two vowels /a/ and /@/ with considerable allophony, although some sources dispute this analysis.
i u a
Languages with this system include Classical Arabic, Quechua, Aleut and Greenlandic; the short vowel systems of Sanskrit, Old Norse and Icelandic are further examples.
[EofL] gives /e a o/, apparently with no fully high vowels, in Amuesha.
i u i i u
@ e o
a a e a
Rukai Navajo Akkadian
Klamath Proto-Slavic
Magalasy
i u
e o
a
Latin and some dialects of Classical Greek immortalised this, and it in turn dictated the number of "vowel letters" in the Roman and Cyrillic alphabets. Spanish, Serbo-Croat, Hebrew, Japanese, Swahili, Maori, Hausa and Basque, to name but a few, are modern-day languages with CL. (In Japanese, /u/ is typically unrounded /w/, but this doesn't really matter here.)
Two rather less common five-vowel systems, one of them square, are:
i i- u i u
e @
a e a
The first has been confirmed by correspondents as existing in Lokono Arawak, spoken in Surinamm and in "a Vanuatuan language called Big Nambas (aka Venen Taut)" respectively.
i i- u i u i u i w u i y u
e o e @ o e o e o e o
a a æ a a a
Lappish Armenian Chamorro Guaraní Inda
Bulgarian Pashto Lativian
Russian Nepali pre-umlaut Old English
%Athomine
In Bulgarian, the vowels form three heightwise pairs: /i e/, /i- a/, /u o/. In OE, they pair up frontwise instead: /i u/, /e o/, /æ a/.
Inda, by Gregory Higley, was the first conlang I encountered on the Internet;, but its page has long since disappeared. I hypothesise that /y/ is the most popular addition to CL amongst conlangers.
Tony, aka "bloodbug2", mentions "Wari' (a Chapacura-Wanham language)" which has /a e i o y ø/, with only one back vowel - effectively like the Hungarian system below without /u/.
i u iy u i i- u iy u
e o eø o e @ o e o
E O a a E
a a
Vulgar Latin Hungarian Romanian Occitian
Italian Kashmiri North Welsh
Bengali %Lemyzon
Somali %Astarien
Yoruba
%Common Dekavurian
Here are three asymmetrical 7-vowel systems.
i u iy u iy u
e @ o e @ o e o
E a æ a
a
Amharic Albanian West Saxon Old English
i u i i- u iy u iy u
e @ o e o eø o eø @ o
E O E O a A a
a a
Javanese %Liotan (long) Finnish German (at a pinch)
Catalan some Old English
Slovene
Welsh English
Turkish and Igbo also have vowel harmony; in Turkish, each front vowel pairs up with the corresponding back vowel, while in Igbo /i e o u/ alternate with /I a O U/. I am grateful to José Geraldo Gouvêa for the analysis of Portuguese oral vowels below, which he describes as "more accurate" than my nine-vowel interpretation.
iy u iy uw i u i u
I @ o I U e o
E e o E O
a eø ao a O a A
Chinese Turkish Igbo Portuguese
i u i u i u
I U I U I
e o e @ o e V o
E O E E O
a a a
Masai Berber Scottish English
Portuguese is VL extended with two central vowels. Lao and Thai are typical systems from South-East Asia with unrounded back vowels.
i u i w u i i- u
e @ o e v o e @ o
E & O E a O E a O
a
Portuguese Lao Thai
iy u i u iy i- u
eø o I U eø @ o
EØ O e @ o &
a E O a
a
Breton Panjabi Teôshi (Mark Rosenfelder)
Kadhrein Hindi
iy u i u i u-
eø @ o I U I U
EØ O e @ o e o
a E V O æ 3 O
a a A
Danish Vietnamese Australian English
French
i u iy wu i i- u iy u
I U eø vo e 9 o I
e @ o EØ AD E 3 O eø o
E V O a A D E 3
æ A a AD
Irish English %Lemyzon %Islandic %Liotan
(abandoned) (abandoned) (abandoned)
iy u iy u
I eø @ o
eØ @8 o EØ & O
E O a D
a A
Dutch Danish again
This alternative analysis of Danish comes courtesy of Routledge's The Germanic Languages; it presents long and short versions of all thirteen vowels plus /V/, total 27: this is probably a record of sorts.
i u
I @ U
e 3 o
E V O
æ aD
Geoff's idiolect of English
Universal 9 implies that a system broader than it is high, like this:
i i- u æ a A
is unlikely. In practice, the vowels /æ/ and /A/ are more likely to raise to /e/ and /o/, moving the system towards one of the first two six-vowel systems above.
Universal 10 invalidates my analyses of Portuguese, Vietnamese, Irish English, Dutch amd my own English! In all of these, what I present as /@/ should theoretically raise to /i-/; yet I certainly don't speak this way.
Universal 11 prohibits something with too many "interior" (central) vowels like this:
i i- u e @ o 3 a
while universal 12 argues against a preponderance of back vowels like this:
i u
e o
O
a