Re: Thagojian phonology (was Re: oh no, not Tech phonology again)

On 24 Feb 00, at 11:51, Daniel A. Wier wrote:

> >From: Paul Bennett <paulnkathy@...>
>
> >Since I'm currently updating and fiddling with Thagojian, here's a VERY 'in-
> >progress' updated repost of something I originally posted in October last
> >year.  This is a much reduced set of consonants, there were originally 288
> >'regular' consonants including palatalised and labialised versions, but
> >these have been absorbed into the (semi)vowel system.
>
> 288, that's exactly the same number I got for stops, affricates, fricatives
> and nasals (but not counting continuants).  And a very nice natural number,
> since it's factors are 2 × 2 × 2 × 2 × 2 × 3 × 3.
I only broke it down as far as 4 x 2 x 3 x 4 x 3 for my purposes, but yes,
it's a very pleasing number.  Factors of lots of small primes (especially
just 2 and 3) tend to be quite pretty.

> >Emphasis
> >   MN Reduced
> >   NR Normal
> >   MX Increased
> >
> >Type
> >   SP Stop
> >   FR Fricative
> >   NS Nasal
> >   GT Glottalic
> >
> >This results in the twelve MOAs listed below:
> >
> >MN-SP Voiceless Aspirated Stop
> >NR-SP Voiceless Stop
> >MX-SP Voiced Stop
> >
> >MN-FR Voiceless Fricative
> >NR-FR Voiced Fricative
> >MX-FR Voiced Affricate
> >
> >MN-NS Voiceless Nasal
> >NR-NS Voiced Nasal
> >MX-NS Prenasalised Voiced Stop
>
> So "emphasis" results in more vocalization?  Change the voiced stops to
> voiceless ejectives ("glottalized") and you get a system similar to Korean
> (just the stops/affricates though).
> Spirantization and nasalization in a similar fashion to Welsh, like I have.
> Of course Welsh has only voiceless/voiced distinction.
>
> >MN-GT Click
> >NR-GT Ejective
> >MX-GT Implosive
>
> Remember that a clicks are not glottal in origin; they are velar.  Though
> you could have an ejective click (double closure at the velum and the
> glottis; that takes A LOT of practice).  And you have them in Khoisan
> languages.
_Ejective_ clicks???  Surely the sign of a phonology pushed too far :-)

> I was thinking something along the lines of "creaky, ejective, implosive".
> Or the MN type could be released with a hard breath -- or even something
> like the Arabic "emphatics", having a secondary pharyngeal articulation.
Hmmm... I like the idea of creaky voice... I wonder if there's a
theoretical reason why (on the model of nasality in other langs) creaky
voice could not spread to following (or surrounding) syllables?  It's
certainly something you've got me thinking about.  Possibly creaky voice
remains until the next glottal sound, or word end, whichever comes first?

> By the way, your prenasalized voiced stop/affricate corresponds to Tech's
> nasalized ejective stop/affricate; it could almost be described as an
> implosive nasal
I take it you mean that this description applies to the sound in Tech, as
the Thagojian sound is definitely a regular prenasalised stop.

> -- someone on the list has these too; who was it again?
> These occur very frequently in Niger-Congo languages, including the Bantu
> group.  Vai has a ton of them (and a nasalized /h~/ incidentally).  But Yi
> has even more and it's Tibetan-Burmese!
AFAICT, Vai has 5 nasals (including /h~/) and 5 prenasalised stops, two of
the latter being implosive (the Bilabial and Alveolar), plus a syllablic
nasal that assimilates to the POA of the following consonant.

Also AFAICT, The Modern Yi syllabary seems to have just about the same
distribution plan for nasals as Thagojian, though with the addition of
prenasalised voiced affricates, and the use of slightly fewer POAs.

> >Along with the 8 POAs, these make a 12x8 grid, as illustrated below.
> >
> >Bilabial     ph  p  b    f  v  bv     mh  m  mb     p!  p'  b'
> >Linguolabial p[h p[ b[   f[ v[ b[v    m[h m[ mb[    p[! p[' b['
> >Interdental  t[h t[ d[   s[ z[ d[z    n[h n[ nd[    t[! t[' d['
> >Alveolar     th  t  d    s  z  dz     nh  n  nd     t!  t'  d'
> >Retroflex    t.h t. d.   s. z. d.z.   n.h n. n.d.   t.! t.' d.'
> >Palatal      ch  c  j    s' z' jz'    n~h n~ n~j    c!  c'  j'
> >Velar        kh  k  g    s, z, gz,    n,h n, n,g    k!  k'  g'
> >Uvular       xh  x  q    s^ z^ qz^    n^h n^ n^q    x!  x'  q'
>
> The first line makes sense.  But if you have /bv/, why not /pf/?  I've
> always had the impression that if a voiced stop, affricate or fricative
> exists, its voiceless counterpart would too.
To have more distinctions would break the symmetry, which is organised in a
way that I find pleasing.  For any given three consonants distinguished
only by Emphasis (say <1>, <2> and <3>), I have tried to organise them such
that in a language with a 2-way distinction, either the pair <1> and <2> or
the pair <2> and <3> would be (more or less) equally plausible phoneme
pairs.

> I like your linguolabials by the way.  Turns out that some Tech speakers do
> some crazy things with consonant pairs, like <pt> becoming lingualabial,
> <kt> becoming a dental click, <kp> becoming a labiovelar double stop, etc.
> And these can be voiced, aspirated, glottalized, nasalized...
On an even broader aside, what about things like <gp> or <kd> (or worse
<gwhp'>)?  Do you have clear-cut rules that determine what happens?

> Now your interdental and alveolars being distinct is realistic because it
> exists in Dravidian.  But when I think of natlangs, I think of it being
> confined to nasals, laterals and vibrants/rhotics (the three n's and l's of
> Tamil, etc.).  And Basque has dental <s> and alveolar <z>, both unvoiced;
> affricates <ts> and <tz> also occur.  However, I would add pharyngealization
> as a secondary segmental to the alveolars, just as you do to lingualabials.
> Egyptian Arabic does something similar to its "emphatics", whereas <t>, <d>,
> <s> and <z> are pure dental, <t.>, <d.>, <s.> and <z.> are
> pharyngealized/"backed" alveolar.  Rather distinct when you hear them
> pronounced side-by-side.  (Vowel allophony also helps a lot.)
To my ear, the interdental and alveolar sounds are already distinct enough,
although this is most apparent in the sibbilated consonants where I have
the set /T D dD/ compared with /s z dz/.  My pronounciation of the
interdental stops is also *very* slightly affricated.  The nasals are
fairly close, but as you suggest, there will be coloration of adjacent
(semi)vowels that should do the job of differentiating sufficiently.

> I have to take a brief aside for Tech.  I realized that when I had dentals,
> retroflexes and post-alveolars (which I called vaguely "palatals"), and
> listed their neutral, palatized and labiovelarized counterparts, I think I
> ran into some overlap.  From my study of the phonologies of Abkhaz-Adyghe
> and Chinese, I found that when retroflexes or postalveolars are palatized,
> they become alveolar-palatal (the c and z with a curl on the lower part;
> pretty much the same as [sj]/[Sj] and [zj]/[Zj] if I'm not mistaken).  Or,
> the postalveolar series are probably automatically palatized, so they
> probably have no neutral/labiovelarized variants in modern spoken Tech (in
> particular, Qotilian and Maou dialects -- incidentally the latter has no
> retroflexes at all).  For Chinese (Mandarin), see my previous post on how I
> merged zh/ch/sh with j/q/x.
> Also, your use of <s> and <z> for all non-labial fricatives might cause some
> confusion, but of course our sparse Latin representation doesn't leave us
> with much choice.  I ended up using digraphs and trigraphs that are pretty
> cliché in the linguistic world (like "sh" for /S/, "kh" for /x/ and "ng" for
> /N/).  For one thing, I just wish we had left the thorn (Þ þ) in our
> alphabet so we'd have 27 letters!
I'm very split on the exact notation.  I'm thinking about putting
diacritised {c} and {j} into service for fricatives in the back three POAs,
but that'd almost make the load on those two letters as bad.  It also
leaves me with the annoyance of some kind of double {j} being used for the
voiced palatal affricate.

I agree that *something* has to be done about it, so I'll try and puzzle
out a set of diacritics that I can use with {c} and {j}...

I've considered digging out eth and thorn for use as interdental
(af)fricat(iv)es but to my mind it creates greater aesthetic problems than
the practical problems it solves.

Maybe <h> and <h.> for the uvular fricatives?

> Oh yeah, and you use all lower-case (case-insensitive) -- great!  I hate
> mixing upper and lower cases just to mark phonetic differences.  To me, if
> you're going to use Latin, Cyrillic, Greek, Armenian or whatever, an upper
> case should represent proper names and the beginning of sentences AND THAT'S
> IT.
Doing that made the burden of the latin alphabet even harder to bear, but I
feel about as strongly as you about romanisations that resort to caps (or
smallcaps for that matter).  The worst example that springs immediately to
mind is Klingon -- ugh bleah yukk.

> >'Irregular' consonants.  These appear not to fit in the grid above.
> >
> ><s~> = <s-tilde> = voiceless bidental fricative
> ><z~> = <z-tilde> = voiced bidental fricative
> ><m'>             = simultaneous bilablial nasal and glottal stop
> ><n'>             = simultaneous palatal nasal and glottal stop
> ><tlh>            = voiceless alveolar lateral affricate
> ><dlh>            = voiced alveolar lateral affricate
> ><klh>            = voiceless velar lateral affricate
> ><glh>            = voiced velar lateral affricate
>
> The first two aren't bad, though I could've used t-tilde and l-tilde (really
> a stroke through the letters, as in Hebrew transliteration).
Hebrew??  Who whats the which what?  Why Hebrew?  Why t and l and not t and d?

> The next two,
> what are they, implosive nasals or creaky nasals?
I think they might be simultaneous pharyngeal stops instead of glottal stops.

When I hold your breath for any length of time, I tighten something in my
throat and relax my diaphram rather than relying on it.  Whatever it is
that I tighten is the same thing that gets tightened in these sounds.

I guess they're prenasalised pharyngeal/laryngeal/glottal/something stops with
nasal release, except that the tongue body and lips do not relax nor change
shape at any time.  Do this fast enough that it becomes a single sound.

> The last four are
> perfect.  (Except I just had "tl" to imply a voiceless lateral affricate,
> rather than the more accurate "tlh" -- I really hate having to use
> trigraphs.)
N-graphs are fine.  I could happily use a septograph, though I find it
hard to imagine when I'd need to.

> Some other conlanger(s) here had velar-lateral (or uvular-lateral?)
> affricates, probably alongside alveolar-laterals.  To me they sound almost
> identical,
As fricatives, I find them very hard to distinguish, but the added audio clue
of the initial consonant sound makes them seem fairly distinct to me.

> and I know of no natlang examples.  Tech doesn't have
> velar-laterals, but one or more might pop up among local variants of <kl>
> (two separate consonants, mind you), and they could even merge with <tl>
> (one phoneme here).  Also, <kr> could become <t.r> (retroflex <t> plus <r>);
> it happens in Tibetan incidentally.
> Worse yet, Tech can have <tl> as one phoneme (alveolar-lateral affricate) or
> <tl> as two (dental stop followed by lateral continuant), and theoretically,
> both can begin a word!  This would be very difficult to distinguish without
> some sort of divider -- a hyphen, a vertical bar, something.  It's just so
> ugly to look at.
Thus the use of the {h} in the trigraphs, since the sequence {lh} is otherwise
illegal.

> By the way, both <l> and <r> become voiceless before voiceless
> stops/affricates (<lt> becomes <Lt> or <lht> for example), and glottalized
> before ejectives.  But in local speech the latter might disappear especially
> word-finally, so you could end up with completely new phonemes -- <l> <lh>
> and <l'>, and <r> <rh> <r'>.  I might have to use Greek smooth
> (apostrophe/"nine" single quote) and rough ("reversed/inverted 9" single
> quote) breathing marks written above lambda and rho to clarify.
> >Semivowels.  These can be syllabic.
> >
> ><y'> = <y-acute> = velar approximant
> ><ly'>            = velar lateral approximant
> ><y>              = palatal approximant
> ><ly>             = palatal lateral approximant
> ><r>              = alveolar approximant
> ><l>              = alveolar lateral approximant
> ><w'> = <w-acute> = labio-palatal approximant
> ><w>              = labio-velar approximant
> Velar approximant, that's similar to American English/Mandarin/Armenian
> semivowel-like "r" right?  Technically, that's a "backed y" in Tech.  (No
> pun intended.)
<y'> = x-sampa /M\/ or IPA {turned-m-right-tail} -- /i/ is to /j/ as /M/ is
to /M\/  Another way to imagine it is as an 'unrounded' /w/

> And yes, I have a palatized <w>, the initial of French _huit_ in that
> by-gawd awful conlang.
<y> is to <w'> as <y'> is to <w>  It's also a 'rounded' /j/

Actually, looking at my above descriptions, it looks like a few diacritics
have gotten confuddled, maybe I need to fix that as well...

I s'pose I also need a labio-alvoelar approximant to fully round out the
set.  That's another thing I'll try to sleep on.

[snip stuff about vowels, it's late and my brane hurts, maybe tomorrow...]


---
Pb