Re: 4 phonemes, was: Another phonological extreme

Hey.

I'm noticing an interesting property of the conlangs with vanishingly
small inventories. It seems that the smaller the inventory, the more
the elements of the inventory look like features rather than segments.
Take the following examples from Vasiliy's post on a 4 phoneme
language:

On Sat, 22 Jul 2000, Vasiliy Chernov wrote:
>
> Deep level -
>
> C: k n
> V: V1, V2 (probably _a_, _i_)
>
> Syllable structure: (k/n)V(n)
>
> Then various vowel combinations contract to produce combinations /wV/
> and /jV/. Then the consonants and glides develop as follows:
(Change these around to /Vw/ and /Vj/ to avoid a problem later on; see
below.)

> kw /p/
> kj /t/
> k /k/
Okay. First off, you actually have a six-segment inventory, since you
use |w| and |j| below. But that's okay by me; six is still pretty
tiny.

First, |k| represents nothing more than "stop"; its articulation as
Dorsal is the default. When combined with |w| it is a labial stop,
hence |w| represents nothing more than Labial articulation, and |j|
represents Coronal (or perhaps more narrowly Apical) articulation.

> nw /m/
> nj /n/
> n /l/
Here |n| represents Sonorant; the default is Lateral, and with
articulatory specifications |j| and |w|, |n| also receives the
specification Nasal when combined with |j| and |w|.

> w /v/
> j /s/
This is nice; the default Labial is a voiced fricative, and the
default Apical is the sibilant /s/.

> The result is:
>
> Surface level -
>
> C: /p t k m n l v s/
> V: probably /a i u @/
This system is the result of the following articulatory
specifications (defaults are in parentheses):

Apical/Front |j|
Labial/Round |w|
(Dorsal)
Stop         |k|
Nasal        |n|
(Lateral)
(Sibilant)
High         |y| (see below on vowels)
Low          |a| ( "    "   "    "   )
(Voice)
(Fricative)

Six elements.

> The vowel contraction scheme might be something like the following:
>
> V1V1 /ji/
> V1V2 /ja/
> V2V1 /wi/
> V2V2 /wa/
>
> then,
> _in_ (+ C/#) /@/
> _an_ (+ C/#) /u/
If V1 is |a| (Low) and V2 is |y| (High), then adding the "coloring"
segments |j| and |w| gets you the following set:

|yj| = /i/
|yw| = /u/
|aj| = /e/
|aw| = /o/
|y|  = /@/
|a|  = /a/

a six vowel system.

> However, there is a problem with _nj_ and _nw_. I have to either
> prohibit some surface vowels before /s/ and /v/ (which IMO has no
> justification in terms of the surface system as such), or introduce
> more complex rules about syllabification (like _Vn.jV_ not equal to
> _V.njV_). I don't like the latter alternative either, since it
> resembles introducing a deep-level glottal stop.
>
> I still feel that some simple trick is possible that will solve
> this problem... any suggestions?
Not a problem. If |y| and |w| *follow* the vowels as well as the
consonants, there need never be a conflict.

This reminds me of Dependency Phonology; in that particular brand of
phonology, basic articulatory instructions are construed as relations
between even more primitive elements (represented by single lower-case
letters). The precise dependency relations obtaining among the
elements gives rise to full inventories and phonological processes.
Neat!

> Basilius
Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu