Re: THEORY: Yivríndil phonology problems

[I've taken 'THEORY' out of the header, because IIRC 'THEORY' is for
general linguistic theory or theorizing about natlangs. But theorizing
about conlangs is sufficiently on-topic as to require no tag.]

Jesse S. Bangs:
> >Jesse:
> >> Since phonology's the topic of the day, I think I'll start with a
> problem
> >> that's been bothering me lately.  I've traditionally described
> >> Y(ivríndil) phonology with seven phonemes: /i I e I a o u/,
> >
> >should that second I be E?
>
> Yes.  To recap, the phonemes are traditionally /i I e E a o u/,
> orthographically {í i é e a o u}
>
> >> Here's the main arguments: /i/ and /e/ occur in complementary
> >> distribution with the dipthongs /ai oi ui/
> >
> >If that is really true, then you could or even should treat all 5 as a
> >single phoneme. Or do I misunderstand you?
>
> Oops, wrong term.  They don't have complementary distribution, like
> allophones of the same phoneme--perhaps I should have said "parallel
> distribution," meaning that they occur with the same sorts of
> restrictions and patterns of occurence, like members of the same set.
I understand.

> >>(snip)  There's also extensive neutralization
> >> between /I E/ and /i e/--the former are disallowed finally and before
> >> vowels and some consonants.
> >
> >So they contrast just before certain consonants? Which?
>
> Well, to give the allophonic rules:
> /E/ --> [e] before y, yy, and perhaps h, and before all vowels
> /I/ --> [i] in the same cases, and finally
> /e/ and /i/ are [e i] in all positions.
>
> Therein lies the ambiguity, since given final or prevocalic [e], it's
> uncertain whether the phoneme is really /E/ or /e/.
From the above, final /E/ and /e/ contrast.

> By orthographic
> convention, final [e] is always spelled {é}, but prevocalic [e] varies
> lexically.  There's even an orthographic "minimal pair":  éos "to you"
> and eos "there have been".  Both are [e.os].  The orthographic situation
> is pretty muddy, then, and exposes the problem with deciding *which*
> phoneme is actually present.
>
> >> Thus, it might be convenient to describe [i e] as underlying dipthongs
> >> /Ii Ei/,
> >
> >Why?
>
> Most importantly because it explains the vowel-lengthening forms--you
> could describe vowel-lengthening as simply "add -i- after the final
> vowel", then "add -(e)va for the 1sg possessive".  This would mean that
> [ElEd]--[EledEva] would in underlying forms be /ElEd/--/ElEidEva/, just
> like /aras/--/araisEva/.  Otherwise you need a special rule to specify
> what "long" /E/ and /I/ are.  This also eliminates the neutralization
> problem--since /e/ would no longer exist, you wouldn't need to worry
> about how to describe ambiguous cases of /E/ vs. /e/.
This is pretty knockdown argument, IMO. [= a convincing one].

> >> even though those phonetic forms
> >> never occur on the surface.  Allophonic rules would describe /I E/ -->
> [i
> >> e] for the other appropriate environments.
> >>
> >> This solution requires me to posit the existence of another phoneme
> /i/,
> >> though, which would only occur as the second element of a dipthong
> >
> >Why? Assuming that it is old /i/ that is reanalysed as a diphthong, why
> >not analyse it as /I+I/ and the other diphthongs as A+I, E+I, O+I and
> U+I,
> >which looks pleasingly elegant. That leaves you with 5 vowel phonemes,
> >A, E, I, O, U.
>
> I *could* do this, but the problem is that those vowel combinations
> aren't necessarily dipthongs.  For example, you have {anoiva}--[an'oiva]
> "my affection" vs. {ano'il}--['ano.Il] "affectionate" (the apostrophe in
> Yivríndil orthography indicates a syllable break between two vowels).  I
> need to explain why sometimes o+i becomes a diphthong and sometimes it
> doesn't, either by saying that the diphthongs contain a different phoneme
> that *always* forms a diphthong, or by supposing the existence of a
> "diaresis phoneme" which prevents diphthongization.  (For various
> reasons, this problems doesn't exist with a+i, but it does exist for
> every other vowel.)  I actually might do the latter, but in all honestly,
> neither solution is all that elegant, since they both require some ad hoc
> creation of phonemes.  Hence my continuing muddle.
I may be misunderstanding you somewhere, but it seems that introducing
an obligatorily diphthonging phoneme or an antidiphthonging diaretic
phoneme is not at all appropriate, at least from a mainstream phonological
perspective. Surely what you need is to accept that syllabification is not
fully predictable from phonemic structure, and in some cases, namely for
vowel + I, the syllabification of the sequence must be stipulated on a
word-by-word basis.

> BTW, I'm subscribed to this list in digest form, so my reply time might
> lag a little bit--sorry!
And I do my conlang mail only at weekends.

--And.