What I should have been asking about binary features

[alice]

I've finally worked this out, after much headscratching and subsequent hospitalisation. Where are the faults in my reasoning below?

1. Four features [high,low,front,back] are needed to allow for high/mid/low and front/central/back vowels.
2. Palatalisation can be regarded as adding the features [+high,-low,+front,-back].
3. Velarisation similarly is adding [+high,-low,-front,+back].
4, And pharyngealisation is adding [-high,+low,-front,+back].
5. For completeness, one can also add [-high,+low,+front,-back], although this is never distinctive in practice.

Now:

6. Velar consonants are [+dorsal,+high,-low,-front,+back].
7. Palatal consonants are [+dorsal,+high,-low,+front,-back], and also [+coronal].
8. Uvular consonants are [+dorsal,-high,-low,-front,+back].

So:

9. Adding palatalisation to a velar results in a palatal without [+coronal].
10. Adding velarisation to a palatal results in a velar with [+coronal].
11. Palatals and velars cannot be pharyngealised because this would mean having both [+high] and [+low].
12. Uvulars can similarly be pharyngealised, but not palatalised or velarised.

[Nortaneous]

1. Four features [high,low,front,back] are needed to allow for high/mid/low and front/central/back vowels.

Kensiu.

11. Palatals and velars cannot be pharyngealised because this would mean having both [+high] and [+low].

Abkhaz had ʕʷ > ɥ, which can apparently still be pronounced with pharyngealization. PHOIBLE reports one language with jˤ and two with kˤ.

12. Uvulars can similarly be pharyngealised, but not palatalised or velarised.

Ubykh.

[vokzhen]

I've finally worked this out, after much headscratching and subsequent hospitalisation. Where are the faults in my reasoning below?

1. Four features [high,low,front,back] are needed to allow for high/mid/low and front/central/back vowels.

Maybe I'm missing something obvious but I can't tell how you're distinguishing e/ɛ or o/ɔ in the languages that don't just have a single mid vowel. Same problem with ɪ or ʊ from their surrounding vowels.

[alice]

I've finally worked this out, after much headscratching and subsequent hospitalisation. Where are the faults in my reasoning below?

1. Four features [high,low,front,back] are needed to allow for high/mid/low and front/central/back vowels.

Maybe I'm missing something obvious but I can't tell how you're distinguishing e/ɛ or o/ɔ in the languages that don't just have a single mid vowel. Same problem with ɪ or ʊ from their surrounding vowels.

That would be with [+tense] or [+atr], which aren't relevant to this problem. For "four" read "at least four", if you must

And I don't really just want counterexamples, IWBNI a lot more useful if someone could explain why my reasoning is incorrect in not allowing them.

[Salmoneus]

...well, you're semi-arbitrarily assigning a bunch of arbitrary binary distinctions to things that are not binary, or even linear. So naturally your predictions will have no relation to reality.

[No, the difference between /e/ and /E/ is neither tensensess nor ATR. The difference is height. Tenseness and ATR are different things entirely.]

[alice]

Thank you, Salmoneus, for showing me the Light. I understand now that all of the numerous references I have consulted on this matter are fraudulent and that I am fighting a losing battle here. I also understand that, despite what I may have erroneously believed, I have actually only ever been asking one question, which is "How do I represent a potentially infinite number of vowel heights with two binary features".

[Salmoneus]

Well, you could always try sarcasm. That can probably let you surmount any obstacle you may encounter.

[zompist]

You said this was for a sound change program. If this is intended for other humans, I'd offer some advice:

-- A set of default features is great, but an ability to define one's own is better. It's less obnoxious that your system only allows 4 vowel heights if a user can add their own features.

-- Allow numerics instead of binary features for those who want them. And let them define the values. If you really don't understand why this would be useful, go ahead and write a rule that raises vowels one step using your binaries.

And Sal is quite right that the difference between [e] and [ɛ] is not tenseness cross-linguistically. (For an example see ancient Greek, which had 4 vowel heights AND a lax/tense distinction.) This is covered in the Lass reference I gave earlier.

[alice]

OK, just to clear some things up:

-- A set of default features is great, but an ability to define one's own is better. It's less obnoxious that your system only allows 4 vowel heights if a user can add their own features.

That was part of my intention all along. It just didn't seem to be worth mentioning, as it wasn't relevant to what I was originally asking.

-- Allow numerics instead of binary features for those who want them. And let them define the values. If you really don't understand why this would be useful, go ahead and write a rule that raises vowels one step using your binaries.

This is entirely reasonable, but my SCA supports other ways of doing it, and it only works if there are exactly as many vowel heights in your language as the SCA implements Again, though, that's not what I was originally asking.

And Sal is quite right that the difference between [e] and [ɛ] is not tenseness cross-linguistically. (For an example see ancient Greek, which had 4 vowel heights AND a lax/tense distinction.) This is covered in the Lass reference I gave earlier.

Well, he didn't actually say "cross-linguistically"; but yes, put that way, he is indeed right. However, at least two references I consulted equate ATR and tenseness, and I haven't actually made anything else up (except, perhaps, [+front], which makes some things much easier).

Once again, though, that's not what I was originally asking. As far as I can remember, and my first post in this thread seems to support my memory, I was wondering how (among other things) [+front] and [+back] in dorsal consonants interact with palatalisation and velarisation, although everyone else seems to think I was trying to do something clever with vowel heights.

Maybe I'll have to go without food for a week and buy a copy of Lass after all. It's not in any library around here.