My beef about ɨ/ɯ

[linguoboy]

It's annoying that there's no convenient symbol in the Roman alphabet to represent a high, unrounded, non-front vowel. Let's review some of the candidates:

y--Used in some Slavic scripts (notably romanisations of Russian). Unfortunately, in those Western European scripts where it isn't pronounced [y], it's treated exactly the same as i[*]. Besides, it's more convenient to reserve this for a palatal semivowel anyway.

ı--A nightmare. Just what were the Turks thinking? While you're at it, why not use j for /ʤ/ and dotless j for /ʒ/ too?

î--Almost as bad; only Rumanians know what this means, everyone else just ignores the little hat.

ŭ--Used in the McCune-Reischauer romanisation of Korean. Most anywhere else, this represents [ə], occasionally [ʊ]. (Let's not even talk about the digraph eu used in the Revised Romanisation and some systems for Thai and other SEAsian languages.)

ü--Thanks to German, says [y] to anyone who doesn't just reflexively ignore diacritics.

ư--Not as misbegotten as ı, but still hardly what I'd call a robust representation. ɯ is much better in this regard but found in even fewer fonts. (The one I'm using doesn't even have an uppercase version.)



[*] With the exception of Welsh. 'Nuff said.

[roninbodhisattva]

You forgot ï. Which annoys the shit out of me. I've ranted about this to myself many a time lately, because a lot of my recent phoneme inventories include /ɨ/, or perhaps /ɯ/.

[Nortaneous]

Unfortunately, in those Western European scripts where it isn't pronounced [y], it's treated exactly the same as i[*].

So? It's common enough for a high unrounded nonfront vowel that people who know things about linguistics won't have any problem with that.

Besides, it's more convenient to reserve this for a palatal semivowel anyway.

This is what <j> is for.

Also, a lot of orthographies just use i-bar. I think there are some that use u-bar, but that is a Bad Thing and should never be done.

[Travis B.]

Despite the issues therewith, to me at least making /j/ be <j> and /ɨ/ or /ɯ/ be <y> still seems like, in the end, the most sensible choice here for anything written in Latin script. All the other choices are simply much worse.

[Bob Johnson]

What's wrong with <ɨ>? We already invented J, U, and W. Why not more letters? Yes, if it's on your keyboard it's easier to type, but a different language would have different keyboards too.

We don't have to cram new languages into exactly what English uses, or what the Romans used.

[Travis B.]

What's wrong with <ɨ>? We already invented J, U, and W. Why not more letters? Yes, if it's on your keyboard it's easier to type, but a different language would have different keyboards too.

We don't have to cram new languages into exactly what English uses, or what the Romans used.

Creating new letters for everything, though, just is not how things are typically done in Latin script (unlike, for instance, Cyrillic script). Rather the normal way things are done in Latin script is to add diacritics to letters, to repurpose existing letters, and to use digraphs and trigraphs. Hence adding new letters altogether to Latin script typically comes off as rather odd and out-of-place.

[Matt]

Are digraphs out of the question? <iy> doesn't seem as bad as some other options.

[Soap]

I use {e} in Bābākiam and {y} in all my other conlangs. Though on paper, I usually substitute a small font "3" for the e. I carried this practice over from linguistic literature in languages such as proto-Austronesian and proto-Aleut, both of which had vowel systems consisting of just /a i u ə/ (but with the precise quality of the ə uncertain enough to make me thinjk of it as a ɯ or barried i)

[Tropylium⁺]

ı--A nightmare. Just what were the Turks thinking? While you're at it, why not use j for /ʤ/ and dotless j for /ʒ/ too?

Actually, that does make some sense, certainly more than <c ç> for /dʒ tʃ/… Is there a J with overdot in Unicode yet?

You forgot ï. Which annoys the shit out of me. I've ranted about this to myself many a time lately, because a lot of my recent phoneme inventories include /ɨ/, or perhaps /ɯ/.

Logically, <ï i ı> should be /i ɨ ɯ/, I think…

Despite the issues therewith, to me at least making /j/ be <j> and /ɨ/ or /ɯ/ be <y> still seems like, in the end, the most sensible choice here for anything written in Latin script. All the other choices are simply much worse.

In the end I find <y> quite acceptable. Why does Scandinavian (and Finnish) use it rather than <ü> for /y/, anyway?

OTOH that road leads down to <ʒ ǯ> for /dz dʒ/…

[Bob Johnson]

Rather the normal way things are done in Latin script is to add diacritics to letters, to repurpose existing letters, and to use digraphs and trigraphs.

That applies for <ɯ>, I suppose -- but why is a bar through <i> different from two dots or whatever above <i>?

[finlay]

Are digraphs out of the question? <iy> doesn't seem as bad as some other options.

It's something like <eu> in a few romanisations in east asia, like thai or korean.

Polish <y> is /ɨ/...

[Travis B.]

Despite the issues therewith, to me at least making /j/ be <j> and /ɨ/ or /ɯ/ be <y> still seems like, in the end, the most sensible choice here for anything written in Latin script. All the other choices are simply much worse.

In the end I find <y> quite acceptable. Why does Scandinavian (and Finnish) use it rather than <ü> for /y/, anyway?

OTOH that road leads down to <ʒ ǯ> for /dz dʒ/…

Typically these are just written with the digraphs <dz> and <dž> in any sort of orthography that uses hačeks, Serbo-Croatian notwithstanding...

[Bristel]

I use <y> for /ɨ/ in Teskwan... it fits in weird with a vowel compliment like /a e ə i ɨ u/ <a e ă i y u>.

Since there is no <o>, that is used to indicate a rounded consonant or rounded cluster of consonants, which will later show rounded vowels.

[Shm Jay]

How about using ŷ or ÿ? It’s obviously different enough from y to indicate it is not a regular y, but to those who are know about y’s use in transliteration for ы, it can serve as a helpful reminder. I don’t think anyone uses ÿ for anything, except the French in some rare and unusual names, and then only as a sign to pronounce the vowel.

[finlay]

ÿ to me looks like it has to be /y/ because it's a compromise between ü and y, though. Soooo.... I won't be using it.

[linguoboy]

I don’t think anyone uses ÿ for anything, except the French in some rare and unusual names, and then only as a sign to pronounce the vowel.

Obsolete form of Dutch ij.

[Travis B.]

I don’t think anyone uses ÿ for anything, except the French in some rare and unusual names, and then only as a sign to pronounce the vowel.

Obsolete form of Dutch ij.

Standard Dutch ij is still written y in many names, both of people and of places, especially in Belgium. It is also used in writing many Low Franconian varieties outside of Standard Dutch, including Afrikaans.

[Zhen Lin]

One of my conlangs has, to varying phonemicity, /i 1 M u/, written i y u w. I also use ı in digraphs, ï for disambiguation, and j for /j/. Methinks, perhaps too many high vowels / pseudodiphthongs...

[Radius Solis]

One of Rory's langs (Mûtsipsa') that I've worked with a fair amount has û for /M/, in the context of having all four of /i y u M/, spelled <i y u û>. It isn't beautiful or especially intuitive, but it does its job without being too hard to tell apart from the other letters or specifically suggesting anything else. We might or might not be able to do better, but we could certainly choose worse.

[Radius Solis]

Besides, it's more convenient to reserve this for a palatal semivowel anyway.

This is what <j> is for.

For a language that has both /j/ and /y/, the latter has priority for <y> unless there is an umlaut process justifying <ü> for that. But that is the only circumstance I would ever consider using <j> for /j/. Especially when that might be needed for other things itself, such as a voiced palatal or postalveolar obstruent.

Not least because NAE speakers have wide exposure to mainly just two languages, English and Spanish, both of which use <j> for something else than /j/ - leaving the IPA feeling anomalous until you've gotten used to it from long exposure.

[Xonen]

ı--A nightmare. Just what were the Turks thinking? While you're at it, why not use j for /ʤ/ and dotless j for /ʒ/ too?

Actually, that does make some sense, certainly more than <c ç> for /dʒ tʃ/

Indeed, <c> for /dZ/ makes baby Satan cry. <ı> for /M/ on the other hand, isn't that bad; 'twas about time someone put that useless dot on top of the <i> to some actual use. Of course, it does create an inconsistency with <j>, where the tittle still occurs only in the small form and has no actual function.

Besides, it's more convenient to reserve this for a palatal semivowel anyway.

This is what <j> is for.

For a language that has both /j/ and /y/, the latter has priority for <y> unless there is an umlaut process justifying <ü> for that. But that is the only circumstance I would ever consider using <j> for /j/. Especially when that might be needed for other things itself, such as a voiced palatal or postalveolar obstruent.

What would you use for the unvoiced postalveolar obstruent, then? The most logical option would probably be <š> or maybe <sh> - but then <j> for the voiced one starts looking rather weird and incosistent. Better to use <ž> or <zh>. Which then frees <j> for /j/ and <y> for other purposes, and is exactly what is done in several eastern European orthographies.

But yes, this probably has a lot to do with what we're used to. Personally, despite having spoken English as a second language for pretty much as long as I can remember, I still don't think of <y> for /j/ as a particularly natural choice. Of course, it depends on the phoneme inventory of the language in question, but unless there are pressing reasons to get more creative, I prefer a system where <j> is used for /j/ first and the representation of other sounds is sorted out afterwards.

[Thomas Winwood]

What annoys me is there's a precomposed <ş> for /ʃ/ but no precomposed <z̧> for /ʒ/. (I do my conlanging in Notepad, and monospace fonts don't always get along with combining diacritics for some reason.)

[Radius Solis]

Besides, it's more convenient to reserve this for a palatal semivowel anyway.

This is what <j> is for.

For a language that has both /j/ and /y/, the latter has priority for <y> unless there is an umlaut process justifying <ü> for that. But that is the only circumstance I would ever consider using <j> for /j/. Especially when that might be needed for other things itself, such as a voiced palatal or postalveolar obstruent.

What would you use for the unvoiced postalveolar obstruent, then? The most logical option would probably be <š> or maybe <sh> - but then <j> for the voiced one starts looking rather weird and incosistent. Better to use <ž> or <zh>. Which then frees <j> for /j/ and <y> for other purposes, and is exactly what is done in several eastern European orthographies.

I always try to be consistent, and it definitely depends on the inventory in question. What I had in mind for <j> was palatal stops and postalveolar/etc. affricates. For voiceless versions of either, <c> is the natural partner for <j>. I might be willing to use <j> for /Z/ in some circumstances, but not where it causes consistency problems like you mention.

[masako]

/me agrees with OP

[Cedh]

What annoys me is there's a precomposed <ş> for /ʃ/ but no precomposed <z̧> for /ʒ/.

Seconded. As soon as you need diacritics on vowels and consonants, the cedilla/comma below would be the best choice aesthetically for just about every consonant in the palatal/postalveolar region, if only it was available for all the relevant letters.
(In my first conlang, which was done exclusively on paper, I had <c j s z n l> /ts dz s z n l/ and <ç j̧ ş z̧ ņ ļ> /tʃ dʒ ʃ ʒ ɲ ʎ/ -- yes, that's a <j> with cedilla for /dʒ/; that one obviously looks a lot better when handwritten.)

As for /ɨ ʉ/, I quite like using the IPA symbols for these, but unfortunately there's no capital version of either. I usually end up using <y> for /j/ even though it's my favourite choice for any non-cardinal high vowel, so I often have to resort to <ü> for /y ʉ/ and <ı> or <ï> for /ɨ ɯ/.