Re: Yes, another sketch for a new conlang! [very very long!]

On Fri, 14 Jan 2000 14:55:46 +0100, Christophe Grandsire
<Christophe.Grandsire@...> wrote:
(quoted in a different order)
>>-- NB: absence of /m/
>        Yes, it's an allophone of /n/.
 <...>
>>#5' + #7': It seems that these rules add new nasals which I cannot account
>>for. I guess /m/ (not in a cluster) first appears here.
>        Yes indeed. I think that it would be one of the last rules that
change the
>deep level to the surface level. Maybe even the last one.
 <...>
>>#5/5'+#7'+#5/5' may make the distribution of nasals wholly complementary.
So
>>I guess some restrictions or additional operations were meant here.
>        I want to have only one phonemic nasal with a lot of phonetic
>realisations. So it's normal that all the nasals be in complementary
>distribution, unless you mean it in another sense than me. And of course, I
>forgot a few restrictions.
Does this mean that you don't want *surface* sequences /#ma/, /#mo/, /#me/
distinct from /#na/, /#no/, /#ne/ ?

I didn't understand that, sorry. My impression was that your #5/5' implied
something like #no#a- -> #mo#a-, with further elision -> /ma/ restoring
the opposition on the surface level. Or did you mean "deep phonemic" when
you wrote "only one phonemic nasal" ?

Hmm... With other labials opposed to dentals in all environments...

>>#5' >QUESTION: What are the PoA of the vowels a, e and o? I need them to
>>>know how to pronounce the n. Can I say for instance that o is labial, e
>>>alvolar and a velar, or something else?
>>
>
>        I have to ask it again, nobody answered. Does anyone know what are
the PoA
>of the vowels a, e and o?
>
It seems to me that this does not matter, while you process the *deep*
phonemes. At this stage you need not explain what your deep vowels were
like phonetically in the language's prehistory, and can assume most weird
ways of development. Maybe, only one thing worth mentioning: of the
resulting surface nasals, /m/ and /n/ should probably be more *frequent*
than others (/N_G/, /ñ/, or whatever you planned for).

>>#0': What about CLVL? CiL and CuL (with syllabic i, u)?
>
>        I wanted to keep a parallel between V and L. If I had accepted
things like
>CLVL, I would have had to have also syllables like CVVV, which I didn't
>want. CiL and CuL with syllabic i and u are not possible simply because I
>don't want them to have a different behaviour from the other L's, and I
>just cannot see how to pronounce Cl=l differently from Cl=. So I just said
>that CLL= was possible, and not CL=L, whatever L may be.
Again, my concern was in fact about surface phonetics/phoneme distribution.
The ban for CLVL may seem unnatural (while CLV and CVL are permitted), as
well as the ban for -iL and -uL (since IMO *on the surface level* /i/ and
/u/ will work as vowels). But elision rules might fill all gaps (BTW,
helping to save some deep-level distinctions from being lost altogether).

>
>>#2: Does this also apply to iC, uC?
>
>        Yes, all liquids have the same voicing effect. That's why I didn't
put the
>syllabic versions of i and u [i, u] with the vowels, which have no voicing
>effect.
>
>>#3: Same question about Ci, Cu (with syllabic i, u).
>
>        Same answer, yes.
Again, *surface* /i/ and /u/ look more like vowels, IMO...

>        Yep. The voiced version of k_C is g_J, and the voiced version of C
is J
>(voiced palatal fricative). I'm not sure if the convention I use is used in
>any ASCII-IPA I know, but it seems best for me.
Is this /J/ distinct from the non-syllabic /i/?

>>#6 and #8-9 (if applied *after* #4a-f) yield, in theory, combinations of
all
>>new obstruents with all deep liquids. The results are far from obvious for
>>C_w/C_j + w/j, C_n/C_l/Cr + l/n/r (with n already "colored" by #5/5'?).
>>Further treatment of these may produce a set of totally new phonemes and
>>fill in the gaps in the system of sentence-initial phonemes and clusters
>>(i. e., add palatalised/labialized counterparts of C_n, C_l, L_O, Cr).
>>
>
>        They would all apply *before* #4a-f.
Actually, I meant that if you have, say, /r_j/ (not considered a cluster),
then I'd also expect that the speakers can pronounce initial /Cr_j/ and
/r_O_j/.

>
>>#3/3' produces voiced counterparts for all obstruents, but they appear
only
>>before L=.
>>
>
>        And after L and L= too.
Yes, but here I was speaking of the initial phonemes. Non-initial sounds
have to be explored separately.

>>1. Voiced obstruents appear only in combinations with L(=) and are not
>>opposed to voiceless ones, unless some additional rules are involved like
>>e. g.:
>>
>>##'n=C -> ##C_v (sentence-initial voiced)
>>L(=)'VC -> L(=)C (with voiceless obstruent preserved)
>>
>
>        That was my will to have voiced obstruents only as allophones of
voiceless
>ones. In the phonemic level, there is only one series of obstruents (just
>like in Hawaian for instance).
>
My concern was that the system of distribution bans is complex, and
therefore unstable.

E. g. you permit /adi/, /ajda/ and /ida/, but not /ada/. IMO, this is
the type of bans that tend to be violated for various reasons - in
borrowings and the like.

If a foreigner says 'My name is Dan', what do the speakers hear and
repeat? /idan/ ? /udan/? /dijan/ ? /tan/ ? Compare with the situation
in Japanese, with /F/, /t_s/, and /w/ becoming permitted before all vowels
in borrowings (and thus opposed to /h/, /t/, and zero).

Best,

Basilius

P.S. I've read your post about tones. It appears that you like complex
things (me too :) ). The Bantu influence becomes more pronounced... Was
it Rwanda ;)?