Re: Unilang: the Phonotactics

On Fri, 20 Apr 2001 18:10:06 +0000, Raymond Brown <ray.brown@...>
wrote:

>Getting your average anglophone to pronounce either final [h] (a challenge
>if ever there was one) or even [x] won't be easy.  Why do you think we are
>now saddled with all those weird -gh spellings?
>
>Scots _loch_ and German _Bach_ are normally pronounced by my fellow
>countrymen with final /k/; and Van Gogh becomes /v&ngQf/.
Actually, since I don't think Unilang will be reality anytime this century,
I'm not particularly concerned with current preferences of Anglophones, no
offence :) Allowing final /h/ (pronounceable as any dorsal fricative, such
as [x], [X], or [h]) would be in keeping with _my_ general level of
moderation between "ease" and "renderability", though I fully respect
reasonable concessions in either direction. For now, I like final /h/.

>Sounds good to me.
>
>>With
>>final /m/ added, there's a problem: so it's allowed to assimilate
>>final /n/, but not final /m/?
>
>So what do we do to avoid assimilation?  Put in an extra [p] as in _dreamt_
>/drEmpt/?  What about the sequence -mk- /mpk/ ???
I kind of solved the final /m/ thing in my last letter. I need to make some
changes to that whole scheme, though:

(I wrote)

"Reduced syllables are basically the "endings" used in the morphology; they
are unaccented, by definition. They have their own inventory of more
distinctive phonemes: the vowels /a i u/, and the finals /n s r/; if it is
a "stem syllable", i.e. part of a bisyllabic stem, it may have one onset C
from the following list: /p t k m n l r s/. So reduced syllabes are CVC,
and grammatic endings (a subset of reduced syllables) can only be VC.

A further limitation: the following medial combinations (across syllable-
boundaries: the final C of one + the onset C of the next) are legal:
any /n s l r/ + /p b t d k g/; /sm lm rm hm fl fr ns ls rs/. Note that some
of the combinations here are already ruled out by other rules above;
e.g. /lb/ won't occur because only accented syllables can have final /l/ or
initial /b/, and two accented syllables won't coexist in one morpheme."

This is kind of silly, at least as I put it. What I meant was for there to
be a system of fewer phonemic distinctions in unaccented syllables; some of
the results above are kind of strange. Unaccented stem syllables should
have /b d g/ as well, so /'alba/ would be legal. I simply didn't like the
idea of /'asga/, and still don't; here's how I solve it, and how I
rationalize it: the sequences /sb sd sg/ are disallowed, as are any other
sequences of frics + voiced plosives; reason being that the frics are
unvoiced by default (being obstruents), while the liquids and nasals aren't
(being sonorants).

Óskar