Re: Becoming triconsonantal, was: 'Arabiiya

Hi Eric, sorry for the delay. Everybody else, please excuse me for this
over-prolongativitudinarious post (I hope, however, that some will find it
interesting).

And another apology: the sketch is very preliminary, many details haven't
been settled. For example, I mostly operate classes of sounds rather than
individual phonemes. The idea was that I'd check several variants of sound
changes against a Latin wordlist to minimize homonymy, but I hadn't one
handy then.

BTW, this also means that the terms I use are sometimes very conventional.
For example, 'schwa' is in fact any short vowel that can later be
reduced to zero, and whose quality is irrelevant for further changes.
Similarly, 'glide' is a consonant that will work as a 'weak' radical in
the resulting system, and/or can contract with vowels. "Y" is the sign for
generic glide (and "¥" - for generic 'secondary' glide, see below).

You see, I wanted, besides, all intermediary stages to look not too
strange, typologically... :)

An important note on the terms 'word-initial', 'first syllable', etc.:
preverbs can be treated, roughly, like proclitics, i. e. not considered
part of the stem (disyllabic preverbs like _contra-_, _ante-_, _inter-_,
etc. are always treated this way). That is, this specific type of morphemic
boundaries affects phonetic development very seriously. I can give a
lengthy, but hardly absolutely convincing, explanation of why I think such
things are possible.

Finally, a tip: this message can also be read bottom-to-top... ;)


On Tue, 25 Sep 2001 11:56:29 -0400, I wrote:
[...]
>The starting point was Latin.
>
>Sound evolution goes as follows:
>
>
>1. Word-initial consonant clusters are simplified yielding new phonemes.
This starts a set of changes that lead to a stage that I called for
myself 'CVC-Romance'. Cl-clusters become lateral obstruents,  Cr-clusters
(partly) yield several retroflex phonemes. Besides, some word-initial
consonants are palatalized before front vowels (see also below).

>Non-initial consonants are modified in various ways depending on the
>quality of *preceding* vowel, also producing new phonemes.
That is, the vowel of the word's first syllable affects both the preceding
and the following consonant/cluster; other vowels affect only the
consonants/clusters that immediately follow them. By these processes, at
least some consonant phonemes are split medially into *three* different
sounds, which difference I will denote as follows: (a)_C, (i)_C, (u)_C.

For short vowels in the first syllable, I thought of the following:

a:      C   -V- (a)_C
o:      C   -V- (u)_C
u > ü:  C   -V- (i)_C, or
        C_j -V- (u)_C (depending on the consonants)
i:      C_j -V- (i)_C
e:      C_j -V- (a)_C

For long and/or medial-syllable vowels, I fixed only some details (see
below).

Another change pertaining to this stage is the simplification of clusters
'nasal + homorganeous stop' (which would be difficult to split on the stage
7, see below).

>Word-initial
>vowels are 'covered' by a glottal stop.
In fact, some of them can take on a prothetic glide ([w] or [j] - not
necessarily merging with the original Latin {v} and {j}).

The above process is not the only source for initial glottal stops. Since
there seems to be a tendency uvular -> pharyngeal -> laryngeal in this
langage's evolution, other probable sources are [r] and [g] before back
vowels.

>2. Synharmonism comes into play: all vowels in a word assimilate in
>quality with the word's *last* vowel.
To be more specific: last syllable's vowels are treated as A, I, U
(capitalization denotes their generic nature, not phonetic quality).
Then, there is some contraction: iU = iA > I, uA > U, eA > @jA, eU > A.

The idea was to group the Latin declensions as follows:

-U, pl. -Us: Lat. 2nd+4th, also 1st in -uam, mostly masc.,
    corresponding fem.: -A (partly by analogy);
-A, pl. -As: Lat. 1st, also 2nd in -eum, mostly fem.; if masc.,
    corresponding fem.: -@jA (partly by analogy);
-I, pl. -Is: Lat. 3rd+5th, also 2nd in -ium and 1st in -iam; if masc.,
    corresponding fem.: -I (same as masc.).

For this to make sense also for verbal conjugations, I assumed the
following vowel contractions and mergers in medial syllables:

iVC > i:C = e(:)C > V-(i)_C
uVC > u:C = o(:)C > V-(u)_C
eVC > @jVC

Then conjugations can be regrouped (by analogy) as follows (forms are
given in the following order: praes. 1sg, 2sg, 3sg, 1pl, 2pl, 3pl,
imperat.2sg., p.p-ple m.sg.; I am not sure if perfect will survive):

1st (except in -iare, -uare, -eare) > A-verbs:
-U, -As, -At, -a:mUs, -a:tIs, -Ant, -A, -a:tU

2nd + 4th + 3rd (except in  -uere) + 1st (in -iare) > I-verbs:
-I, -Is, -It, -i:mUs, -i:tIs, -Int, -I, -i:tU

1st (in -uare) + 3rd (in -uere) > U-verbs:
-U, -Us, -Ut, -u:mUs, -u:tIs, -Unt, -U, -u:tU

1st (in -eare) > @jA-verbs:
-A, -@jAs, -@jAt, -@ja:mUs, -@ja:tIs, -@jAnt, -@jA, -@ja:tU

After that, the last syllable's vowel changes the quality of all other
vowels in the word. In medial syllables, and in the first syllable when its
vowel is short, no former vowel distinctions survive. For initial syllables
with long vowels, I accept a somewhat artificial evolution (see below).

It is important that medial consonant phonemes present at this stage do
not include any palatalized, velarized  or labialized sounds (or [w], [j]),
which could interfere with the synharmonistic assimilation of vowel
qualities.

>3. Stress rules change: accented is the word's first syllable if it's
>heavy (ending in a clustered consonant or long vowel), otherwise the
>second syllable. In disyllabic words, the accent falls on the first
>syllable, whose vowel is lengthened if the syllable wasn't heavy.
Naturally, this yields some alternations in paradigms where both disyllabic
and polysyllabic forms are present (_puto:_ :: _puta:re, etc.). Such
alternations are then eliminated by analogy, but may partly remain as
irregularities in the resulting system.

>4. Non-initial consonants (and clusters), if not adjacent to a stressed
>vowel, are weakened (partly to zero, partly to glides, and rarely to
>other types of consonants).
This stage forms the triconsonantal skeleton: initial CV:C-, CVCC- and
CVCVC- are not affected by the consonant weakening.

Besides, intervocalic [w] and [j] show up again, being a result of the
weakening of some other consonants.

>5. Unstressed vowels are mostly reduced to schwa, sometimes to zero.
>
>6. Stress rules change again: the accent is always on the first syllable.
>When second syllable loses its stress, its vowel in certain environments
>is reduced to zero, otherwise to schwa.
These changes affect all syllables except first. The first syllable
inherits the quality of the Latin vowel in the last (inflectional)
syllable.

Reduction to zero involves the following two changes:

@Y@Y@ > @YY@
C@Y@ > CY@

There are three phonemes treated as glides here: [w], [j] and some
"laryngeal" (H). These are "primary glides".

>7. Intervocalic combinations like 'consonant cluster+glide' are broken by
>inserting a vowel whose quality corresponds to the glide's quality.
The glide itself is then dropped. That is,

-VCCjV- > -VCiCjV- > -VCiCV-,
-VCCwV- > -VCuCwV- > -VCuCV-,
-VCCHV- > -VCaCHV- > -VCaCV-.

However, word-finally -CH@ > -C@H; after that, word-final schwa's are
dropped. Then word-final 'primary' -@H, -@j, -@w > -a, -i, -u.

Combinations -VYYV- are not affected by these changes; instead, they
develop as follows:

-@wYV- > -uYV-
-@jYV- > -iYV-
-@HYV- > -aYV-

The intervocalic glides that are thus preserved coincide then with the
'secondary glides'.

The latter result from the weakening of another series of three
consonants, yielding the new w, j, H.

Main sources for the glides are provisionally taken to be the following
intervocalic consonants:

    primary:   secondary:
j   (i)_r         (a)_t (> [D] > ...)
w   (a)_b         (u)_r, (u)_l, (i)_b
H   (a)_r         (i)_n (> [ñ] > [N] > ...)

After that, C@¥@ > C¥@

>Clusters of more than two consonants are broken by inserting a schwa.
That is, CC¥V > C@C¥V

>In combinations like 'long (accented) vowel+consonant+glide',
( - A primary one; in fact, this is a subset of cluster breaking rules
formulated above; secondary glides still don't exist - )

>certain
>vowels are split into combinations 'short vowel + glide (whose quality
>depends on the original vowel) + short vowel (whose quality depends on
>the original glide, which itself is mostly dropped).
Actually, all original long vowels are treated as VY-combinations. That is,

-VYCjV- > -VYiCjV- > -VYiCV-,
-VYCwV- > -VYuCwV- > -VYuCV-,
-VYCHV- > -VYaCHV- > -VYaCV-.

(Besides, some glides here are original, with 'primary' and 'secondary'
not opposed)

This is preceded by a change that was called artificial above. The matter
is that I wanted the resulting first V's to follow the same alternation
patterns as any other vowel in first syllable (that is, to reflect the
vowel quality of the Latin final syllable). The phonetic changes themselves
don't need to be especially unnatural, but their exact matching with the
development of short vowels, indeed, looks a bit conlangish. For example:

u: {... I} > ö: > Iw
i: {... U} > ü: > Uj,

- rather than the reverse.

A special case (again, not too difficult itself) is V: > VHV. Some long
vowels acquire a specific prosodic quality (some kind of "broken accent"
à la Danish stød), and then become truly disyllabic before a CY-cluster. As
for the differentiation of long vowels, I was inspired by German dialects
of the Rhine area (with 'Rheinische Korreption', which resembles stød very
much): in some of them the accentuation maintains the opposition
"(originally) narrow" vs. "(originally) broad" in long vowels (where
"originally" means "in Old High German").

After that, all schwas in open syllables are dropped. In sequences of open
syllables with schwa, it seems that preserved are the schwas in odd
syllables, counting from the *end* of the sequence. Then all remaining
schwas merge with [a].

By that moment, the inventory of short vowels includes only three phonemes:
[a], [i], [u].

>8. Unstressed schwa's contract with adjacent glides, yielding several
>distinguishable vowel phonemes, long and short. Accented vowels also
>contract with adjacent glides, producing long vowels that merge with
>phonemes already existing in this position.
As for short, see above, treatment of 'primary' glides and schwas.

Contractions that yield new long vowels (and diphthongs) involve, in fact,
all short vowels, and work as follows:

AHC > a:C
AWC > auC
AJC > aiC
UHC > o:C
UWC > u:C
UJC > uiC
IHC > e:C
IWC > iuC
IJC > i:C
>
[...]
>Note that the last two changes, in certain forms, make the stems with
>original long vowel in first syllable indistinguishable from stems with
>first vowel followed by a glide. The analogy then destroys all difference
>between these two categories.
>
>Now, some features of the resulting system:
>
>Syllable structure is CV(C). Most stems are triconsonantal; stems
>seemingly containing only two consonants are morphologically treated
>as if they contained a contracted glide; stems of four consonants, where
>the last consonant cannot be identified with a suffix, are rare.
The above two paragraphs describe the 'Semitoid' structure of the language.

>V1 (the vowel after the word's first consonant) inherits the quality of
>the vowel in Latin inflectional ending; cannot be zero.
>
>V2 (which CAN be zero) inherits the quality of a consonant in a syllable
>that used to be unstressed (during stages 3 - 5). Mostly, this vowel
>reflects a Latin suffix or part of inflection.
That is, V2 has the quality of a primary glide that has dropped.

>V3 and all sounds further to the right reflect Latin consonants (suffixes
>or parts of inflection) that used to follow the consonant reflected by
>V2.
I. e. V3 etc. depends on a primary or secondary glide.

Just for illustration, I give here the vocalization models for some verbal
paradigms (for simplest, originally suffixless stems); verb classes are
marked as above in discussing stage 2 (the order of forms being listed is:
imperat.2sg, praes.1sg, 3sg, 2pl, inf., p.ple m.sg, m.pl, f.sg); it is
assumed that (i)_t > [z], (u)_t > [r], word-finally [s] merges with the
'secondary' j, [t] remains.

Latin endings (1st conjugation):
         -a:   -o:   -a:s   -at     -a:tis  -a:re  -a:tum  -a:to:s -a:tam

A-verbs: CaCaC CuCaC CaCCi  CaCCat  CiCaCji CiCCa  CuCCai  CuCaCji CaCCai

I-verbs: CiCaC CiCaC CiCCi  CiCCat  CiCaCzi CiCiC  CuCCaz  CuCaCzi CaCCaz

U-verbs: CuCaC CuCaC CuCCi  CuCCat  CiCaCri CiCCau CuCCar  CuCaCri CaCCar

@jA-v.:  CaCiC CaCaC CaCiCi CaCiCat CiCiCji CiCiCa CuCiCai CuCiCji CaCiCai


Basilius

Reply