Xonen wrote: ↑Thu 14 Dec 2017, 01:59
I think /i̯ɘ/ at least warrants its own character; it appears to be extremely common and occurs as the front-vocalic counterpart to /ɑ/ in several suffixes, so it's a question of both economy and symmetry. And if /i̯ɘ/ gets a letter of its own, then it makes sense to be systematic and give one to /y̯ɵ/ and /u̯ʊ/ as well. Incidentally, the commonness of /i̯ɘ/ also happens to make Kazakh exceptionally well suited for being written in Cyrillic, so the
most sensible option would be not to romanize it all.
I've only heard Kazakh spoken in a few videos on Youtube, briefly on TV in some documentary about something, etc. since my dedication to learning it never got to the point where I'd spend much time listening to it spoken, but I do listen to fairly large amounts of Kazakh music. That could mean my understanding of the language's phonology is skewed because it's different in singing (which I have noticed to be the case); but anyway, as far as I can tell, the vowels are actually much closer to just about any language with similar vowels than Wikipedia claims. I mean, especially the <е> sounds like [e~e̞~ë̞~ë~ɘ̟~ɘ~ə̟] or whatever much more commonly than [i̯ɘ]; consonants before it are palatalised, though, so it kind of makes sense to transcribe it as /i̯ɘ/, and maybe in real everyday speech it really does sound like that, but AFAICT it's just that it's /je/ word-initially and after vowels, and that's generally what it corresponds to in other Turkic languages.
Xonen wrote: ↑Thu 14 Dec 2017, 01:59
Other than that, it gets trickier. A part of the problem is that I don't really know Kazakh phonology that well; is that contrast between velars and uvulars actually phonemic, for instance? Seems to me like the latter only occur in back-vocalic contexts, but I could be missing something. If we assume it's not, we can ignore /q ʁ/ and just use <k g>, respectively, for those.
Yeah, that's exactly what Tatar does and why it looks so different from Bashkir even though the two form one of the most closely related language pairs considered separate languages. Personally, a huge part of why I like Bashkir more than Tatar is that its alphabet is much more badass-looking with its <ҡ>, <ғ>, <ҫ> and <ҙ>. It's also slightly closer to Kazakh, and since I fucking love Kazakh, that's automatically a plus. And let's not forget that it has /θ/ and /ð/, the latter of which is among the cutest sounds any language could dream of having.
Kinda sad that Finnish doesn't have it, especially knowing that it used to...
Xonen wrote: ↑Thu 14 Dec 2017, 01:59
Another possibility would be to use <q> (instead of <c'>) for /t͡ɕ/, but since that apparently only occurs in Russian loanwords, it's not a major concern anyway.
To get rid of the rest of the apostrophes, replace <s'> for /ʃ/ with <x>.
Using <q> for /t͡ɕ/ and <x> for /ʃ/ would be so Chinese that it would retroactively tear the Sino-Soviet split into a full-blown war.
Xonen wrote: ↑Thu 14 Dec 2017, 01:59
I can't think of a good way to distinguish between /ɑ æ/; the only graphemes that make intuitive sense are <a e>, but it's pretty much impossible to make the system work without using <e> for something else. Then again, /æ/ seems to have a fairly low functional load, and /ɑ/ only occurs in back-vocalic words, so it might not be
that bad to just use <a> for both. Not an ideal solution by any means, though, so if anyone's got any suggestions, I'm all computer screens.
Well, Uzbek uses <a> for both [æ] and [ɑ], <o> for [o] and [ø] and <u> for [u] and [y], but they're considered allophones due to the loss of vowel harmony. And, supposedly, it doesn't really have [ɑ] at all even as an allophone, but I can't not hear it in just about any clip of spoken Uzbek or Uzbek songs. Then again, the problem might be that my ears refuse to accept that [ɒ] is /o/ and as such I hear words where other Turkic languages have /ɑ/ but Uzbek has /o/ as having /ɑ/, which being [ɒ] makes me hear it as [ɑ], and Uzbek really doesn't have [ɑ] even as an allophone of anything and it's just an auditory illusion... but I kinda doubt that that's the case.
Anyway, it would be possible to do something like this:
Аа, Әә > Aa
Ее, Ыы > Ee
Оо, Өө > Oo
Ұұ, Үү > Uu
Іі, Ии > Ii
Thus, even with your consonant romanisation that uses <k> and <g> for both the velars and uvulars, it would still be possible to tell whether a word has back or front vowels by suffixes having the "wrong" letter. It could be taken a step further by romanising both <н> and <ң> as <n>.
For example,
Current Cyrillic: қол (hand), қолдың (hand's), қолдар (hands), қолдардың (of hands)
Romanisation: kol (hand), kolden (hand's), koldar (hands), koldarden (of hands)
Current Cyrillic: әйел (woman), әйелдің (woman's), әйелдер (women), әйелдердің (women's)
Romanisation: ayel (woman), ayeldin (woman's), ayelder (women), ayelderdin (women's)
I looked up if the word айыл exists in Kazakh, and it does, apparently meaning the belt in a horse's saddle or something. Even though the two would be homographs, I'm fairly certain context would clear any ambiguity in this case, and any suffix would immediately disambiguate them anyway. There probably would be homographs that wouldn't be so easily disambiguated by context, though, and I don't know how they should be dealt with. The good thing is that Kazakh's rounding harmony is not represented in writing in Cyrillic and also isn't as prominent as in other Turkic languages, so there wouldn't be confusion between words due to the back and front rounded vowels being represented by the same letter even if there was otherwise.
There are one-syllable words that could cause trouble, though, like:
тор (net, cage, grid)
төр (I don't really know what this means; this
Kazakh-Russian PDF dictionary says it's "почетное место в доме (юрте), в комнате", so some kind of home shrine or something?)
торды > torde (the net/cage/grid (accusative))
төрде > torde (in/on/at whatever төр is)
...but a cheap solution could be to romanise <ы> as <u> in such cases?