Constructed languages with only one vowel a or e, .....?

[JANKO GORENC]

Hi,

I came up with the idea to create constructed languages with only one vowel.
A family of posteiori languages could be based on natural languages.
For example, a language family with constructed languages that have only one consonant
a) a group of languages with a vowel - a
b) a group of languages with a vowel - e.
c) a group of languages with a vowel - i.
d) a group of languages with a vowel - o
e) a group of languages with a vowel - u
....
I wonder if anyone is prepared to do such a language family.
I expect your response.

[gestaltist]

Why don't you?

Personally, I think such a language would have several "surface" vowels anyway, and the single vowel would be a matter of analysis. Especially since you're saying it would be aposteriori.

[WeepingElf]

Some linguists say that an early stage of PIE had only one vowel, but this is highly controversial (personally, I don't see much merit in this, though I can see how they get at that).

[HJH]

3 vowels is usually the minimum for natlangs, but you can go for what you want.

[WeepingElf]

Apparently, Northwest Caucasian languages can be analyzed as having just two vowel phonemes, a higher and a lower one, but these at least have various allophones covering the vowel space, i.e. the high vowel is /i~ɨ~u/ and the low vowel is /ɛ~a~ɔ/ or something like that. This, however, requires positing phonemic plain/labialized/palatalized triads for most consonants, which throws up the question whether it is not better to analyze these languages as having front, central and back vowel phonemes at two degrees of vowel height. Alas, I am not an expert on these languages, and cannot tell whether my objection against the two-vowel-phoneme analysis is valid or not.

[sangi39]

3 vowels is usually the minimum for natlangs, but you can go for what you want.

Although there are languages with just two vowels, most notably the Northwest Caucasian languages (although, as gestaltist points out, these tend to have a number of phonetic realisations depending on the surrounding consonants), but my favourite language for this sort of thing is Yimas.

[sangi39]

Apparently, Northwest Caucasian languages can be analyzed as having just two vowel phonemes, a higher and a lower one, but these at least have various allophones covering the vowel space, i.e. the high vowel is /i~ɨ~u/ and the low vowel is /ɛ~a~ɔ/ or something like that. This, however, requires positing phonemic plain/labialized/palatalized triads for most consonants, which throws up the question whether it is not better to analyze these languages as having front, central and back vowel phonemes at two degrees of vowel height. Alas, I am not an expert on these languages, and cannot tell whether my objection against the two-vowel-phoneme analysis is valid or not.

From what I can remember, the reason for the two-vowel analysis comes from 1) there being "gaps" in the system (not all consonants can be palatalised or labialised, and some can't have secondary articulation at all), and 2) the addition of some consonantal affixes changes the adjacent vowel, but there don't appear to be any consonant alternations, suggesting that even if, say, /p/ and /tʷ/ derived from the same underlying *p some time in the past (at least in Ubykh, the latter sound comes from an older *pʲ), there's no alternation between /p/ and /tʷ/, caused by a vowel or otherwise.

IIRC, in Abkhaz, labials cannot be palatalised, and are treated as if they were labialised by default, meaning they can only be followed by [ u] or [o].

[Frislander]

Apparently, Northwest Caucasian languages can be analyzed as having just two vowel phonemes, a higher and a lower one, but these at least have various allophones covering the vowel space, i.e. the high vowel is /i~ɨ~u/ and the low vowel is /ɛ~a~ɔ/ or something like that. This, however, requires positing phonemic plain/labialized/palatalized triads for most consonants, which throws up the question whether it is not better to analyze these languages as having front, central and back vowel phonemes at two degrees of vowel height. Alas, I am not an expert on these languages, and cannot tell whether my objection against the two-vowel-phoneme analysis is valid or not.

From what I can remember, the reason for the two-vowel analysis comes from 1) there being "gaps" in the system (not all consonants can be palatalised or labialised, and some can't have secondary articulation at all), and 2) the addition of some consonantal affixes changes the adjacent vowel, but there don't appear to be any consonant alternations, suggesting that even if, say, /p/ and /tʷ/ derived from the same underlying *p some time in the past (at least in Ubykh, the latter sound comes from an older *pʲ), there's no alternation between /p/ and /tʷ/, caused by a vowel or otherwise.

IIRC, in Abkhaz, labials cannot be palatalised, and are treated as if they were labialised by default, meaning they can only be followed by [ u] or [o].

Additionally, from what Adyghe I've heard, there seems to be some relative free-variation between the realisations of the vowels even adjacent to the conditioning glides. Also another fun point - in both Kabardian and Adyghe (really two parts of the Circassian dialect continuum) there is a third vowel, because the "low" vowel has a length contrast, as seen in the name Adyghe /aːdəɣa/, though the shorter of the two is often more raised/fronted.

[gach]

Although there are languages with just two vowels, most notably the Northwest Caucasian languages (although, as gestaltist points out, these tend to have a number of phonetic realisations depending on the surrounding consonants), but my favourite language for this sort of thing is Yimas.

The problem with Foley's two vowel analysis for Yimas is that he has to assume that the surface vowels /u/ and /i/ are underlyingly the central vowel /ɨ/ plus a rounding or a fronting autosegment R or F. Since under this analysis /a/ + R surfaces as /aw/ and /a/ + F as /aj/, you might view /u/ and /i/ as underlyingly /ɨw/ and /ɨj/, which don't exist as surface sequences in the language. The analysis has its own academic elegance, but I'm not sure that it's actually any more economical than just assigning the R and F features directly to underlying vowels /u/ and /i/ and adding a few rules to account for the surface level feature spread, which you anyway have to describe under either analysis.

[gestaltist]

I remember reading about some African language analyzed as having no vowel phonemes - but even here, it has several vowel phones, it's just that their distribution is predictable from the consonants. I can't find the related paper anymore, I'm afraid.

[WeepingElf]

I remember reading about some African language analyzed as having no vowel phonemes - but even here, it has several vowel phones, it's just that their distribution is predictable from the consonants. I can't find the related paper anymore, I'm afraid.

I have the idea (currently on the back burner) of an Old Albic dialect with no phonemic vowels. In (Standard) Old Albic, vowels are autosegmental, like tones in African languages. There are vowel features (three of them: [+open], [+front], [+round]) that are attached to morphemes, and a phonotactic rule that inserts vowel positions into the string, which are then realized according to the attached vowel features. (There is also some umlaut and vowel harmony, i.e. vowel features spreading to neighbouring vowel positions, going on.) The idea is that in Old Ivernic (the "vowel-less" dialect), the autosegmental vowel features turn into secondary articulations (labialization and palatalization) of adjacent consonants, such that all there is on the deep level is a string of consonants, into which vowels are inserted automatically whose colours depend on the adjacent consonants.

I call a language in which vowels are inserted automatically into a consonant string a "vowel generator language", and this is IMHO the most interesting sense of the term "vowel-less language" if the vowel generator is complex enough (inserting [a] after every odd-numbered consonant obviously isn't). I mean, everybody can make a "vowel-less" conlang in which consonants like nasals and liquids function as syllable nuclei; making an interesting vowel generator is something else.

I have seen several vowel generator languages on the Net, but none of those was naturalistic, so I am trying a naturalistic vowel generator language (if there is such a thing) with Old Ivernic.

[sangi39]

Although there are languages with just two vowels, most notably the Northwest Caucasian languages (although, as gestaltist points out, these tend to have a number of phonetic realisations depending on the surrounding consonants), but my favourite language for this sort of thing is Yimas.

The problem with Foley's two vowel analysis for Yimas is that he has to assume that the surface vowels /u/ and /i/ are underlyingly the central vowel /ɨ/ plus a rounding or a fronting autosegment R or F. Since under this analysis /a/ + R surfaces as /aw/ and /a/ + F as /aj/, you might view /u/ and /i/ as underlyingly /ɨw/ and /ɨj/, which don't exist as surface sequences in the language. The analysis has its own academic elegance, but I'm not sure that it's actually any more economical than just assigning the R and F features directly to underlying vowels /u/ and /i/ and adding a few rules to account for the surface level feature spread, which you anyway have to describe under either analysis.

Ohhhhh, I think I remembered it incorrectly, then I thought, when I wrote that post, that /i/ and /u/ were epenthetic (and conditioned or only appeared in borrowings. It's a cool system either way (unlike the Northwest Caucasian languages, it looks like there's less variation in how vowels are realised phonetically and there isn't an abundance of secondary articulation on basically all consonants), but my mistake.

[gach]

Ohhhhh, I think I remembered it incorrectly, then I thought, when I wrote that post, that /i/ and /u/ were epenthetic (and conditioned or only appeared in borrowings. It's a cool system either way (unlike the Northwest Caucasian languages, it looks like there's less variation in how vowels are realised phonetically and there isn't an abundance of secondary articulation on basically all consonants), but my mistake.

Ah, typical to its region, the normal epenthetic vowel in Yimas is /ɨ/ but also /u/ and /i/ can appear if the surrounding vowels are rounded or fronted. And /u i/ are certainly quite rare in the language compared to /a ɨ/ which do have a fair bit of their own allophony.

In fact, you could nearly do away with a phonetic /ɨ/ in Yimas and Foley's grammar is full of words which he writes without any vowels. There are, however, a limited number of morphemes where /ɨ/ breaks allowed consonant clusters and so can't be seen as an automatic epenthetic vowel. Hence, you are stuck with /ɨ/ having a phonemic status.

This is a bit similar to the old analyses that were attempted for Iatmul. Before the more recent and critical descriptions, it was widely though that the typical vowel system of the languages around the mid and lower Sepik went something like /a ə ɨ/ with Northwest Caucasian style allophony on the mid and high vowels and /ɨ/ as the standard epenthetic vowel. One of the analyses of Iatmul went so war as to viewing the high vowel as fully non-phonemic and the main distinction between the two remaining vowels being length (which does seem justified when you look at the recent analysis of the Iatmul vowels). The resulting vowel system was thus /a a:/, which you can view as consisting of only a single vowel with doubling. Of course, none of this works in the light of the current understanding which gives the language a total of 12 distinct monophthongs with a total of seven different vowel qualities.

[Shemtov]

Isn't there an analysis of Mandarin that says only one or two vowels are phonemes?

[WeepingElf]

Sounds as if I have heard of that, too, but can't give any pointer.

[lsd]

create constructed languages with only one vowel

For me the question would be more the possibility to build a lexicon without multiplying the unpronounceable consonants clusters or the length of words ...

[Birdlang]

I made a conlang where there were only 2 1/2 vowels.
/i u/ and the semiphoneme was /a/ because it was a supersegmental phoneme that changed to /i u/ at the end of a word depending on the vowels in the word.

[Iyionaku]

Isn't there an analysis of Mandarin that says only one or two vowels are phonemes?

I'm not an expert on phoneme inventories, but if you simply go after minimal pair analysis, this doesn't seem right to me. For example, there is
辣 là [la˥˩] vs. 乐 lè [lɤ˥˩] vs. 力 lì [li˥˩] vs. 路 lù [lu˥˩] vs. 绿 lü4 [ly˥˩] - all of them are only distinguished by the vowel quality. Or am I overseeing something?

[Shemtov]

Isn't there an analysis of Mandarin that says only one or two vowels are phonemes?

I'm not an expert on phoneme inventories, but if you simply go after minimal pair analysis, this doesn't seem right to me. For example, there is
辣 là [la˥˩] vs. 乐 lè [lɤ˥˩] vs. 力 lì [li˥˩] vs. 路 lù [lu˥˩] vs. 绿 lü4 [ly˥˩] - all of them are only distinguished by the vowel quality. Or am I overseeing something?

IIRC, this analysis sees an /ɤ/ vs. /a/ distinction, and everything else is just allophones, as the language has more initials in this analysis, and some who specialize in phonemics take it as a serious alternative. I'm not buying it either right now, because of Occam's razor, but I'm taking an advanced class on phonemics/phonology (my college's ling department is seeing if undergrads can handle such a class) and I may change my mind.