Re: Old Albic Update: An autosegmental view of the vowel features

Hallo!

"David J. Peterson" wrote:
>
> Hey Jörg,
>
> Neat!
Thanks!

>       It occurs to me, looking at your phonology found here:
>
> http://listserv.brown.edu/archives/cgi-bin/wa?
> A2=ind0406c&L=conlang&F=&S=&P=47209
>
> That Albic lacks a [-open, -round, -front] vowel (i.e., [M]).  This
> leads to three questions:
>
> (1) Could this vowel possibly occur, or is it impossible?
It cannot occur because there are no roots without (positive)
vowel features attached.  "Featureless" vowels occur only
in affixes, and these borrow their vowel features from
neighbouring morphemes.  I actually considered adding /M/
to the system at some point, but decided to leave it the way
it is.

> (2) Do only positive features spread?
Yes.

> (3) Can we see a sample word with the other vowels of Albic?
Sure.  Take for example, the noun _broca_ `bear', which is
autosegmentally represented thus:

   [+open] [+open]     [+open][+open]
   [+round]  |        [+round]  / |
        |    |              |  /  |
      br°c- -°  -->        br°c- -°  --> broca

The morpheme /-a/ triggers a-umlaut which has no effect because
the vowel in /broc-/ is already open.

The plural is _brøci_:

   [+open] [+front]    [+open][+front]
   [+round]  |        [+round]  / |
        |    |              |  /  |
      br°c- -°  -->        br°c- -°  --> brøci

The morpheme /-i/ triggers i-umlaut, adding the feature [+front]
to the vowel in /broc-/.

The dual is _brocu_:

   [+open] [+round]    [+open][+round]
   [+round]  |        [+round]  / |
        |    |              |  /  |
      br°c- -°  -->        br°c- -°  --> brocu

The morpheme /-u/ triggers u-umlaut which has no effect because
the vowel in /broc-/ is already rounded.

The locative case of the latter is (the morpheme /-m/ forms the
objective stem):

   [+open] [+round]    [+open][+round]
   [+round]  |        [+round]  / |   \
        |    |              |  /  |    \
      br°c- -° -m -°l -->  br°c- -° -m -°l  --> brocumul

Here, vowel harmony occurs in the suffix /-°l/ which is realized
as [ul].  The vowelless morpheme /-m/ is skipped.

The instrumental case:

   [+open] [+round]      [+open][+round]
   [+round]  | [+front] [+round]   | [+front]
        |    |     |         |     |/    |
      br°c- -° -m -° -->    br°c- -° -m -°  --> brocymi

Here, i-umlaut affects /-u/.  Because now two features attach
to it, it does not umlaut /broc-/ (which in this case wouldn't
have any effect anyway, see above).  Again, /-m/ is skipped.

As I hope is clear, all three vowel features ([+open], [+front]
and [+round]) behave the same, namely according to the following
rules:

1. Vowel features bind to morphemes; all vowels in the morpheme
   (there is a small number of bisyllabic roots in Old Albic,
   e. g. /semel/ `wheat' or /macal/ `flesh') receive all the
   features binding to the morpheme.

2. Vowel harmony: If a morpheme contains a vowel position but
   does not carry any vowel feature, it inherits all vowel
   features from the nearest (in the direction towards the root)
   morpheme that carries vowel features.  (All roots carry
   vowel features, so there always is such a morpheme, either
   the root itself or another affix which has its own vowel
   features.)

3. Umlaut: If a vowel feature is the only feature binding to
   a particular morpheme, it spreads to the preceding morpheme.
   Morphemes without vowel positions are transparent: the vowel
   feature spreads to the nearest preceding morpheme that has a
   vowel position.  If that morpheme has two vowel positions,
   both vowels are umlauted.  (Example: the instrumental case
   of _macal_ is _meceli_.)  If hereby a morpheme receives a
   vowel feature it already carries, no change occurs.
   Precedence is from right to left: if a morpheme receives
   a second feature through umlaut, it no longer causes umlaut
   itself.

It might be elucidating to know what happened diachronically,
in the prehistory of Albic:

In Pre-Proto-Albic, there were only three vowels: *a, *i and
*u.  (Of these three vowels, *a was much more frequent than
*i and *u, possibly because several different vowels merged
into it.  This is why so many Old Albic roots have /a/ as
the root vowel.  Compare the predominance of *e in PIE.)
Accent always was on the root; affixes were unaccented.
This accent conditioned a kind of ablaut (similar to the state
of affairs in Proto-Indo-European, which might be a sister
language of Proto-Albic):

   accented   unaccented

         *a ~ *@
         *e ~ *i
         *o ~ *u

An intermediate stage probably was with *ai and *au as the
reflexes of *i and *u in stressed syllables.  The difference
between the accented and the unaccented vowels was that the
accented vowels had the feature [+open], the unaccented ones
the feature [-open].  This means that roots could only contain
the vowels *a, *e or *o, and affixes only *@, *i or *u.
(This distribution, however, was to some degree messed up by
later changes such as borrowings, but the constraint against
*@ as root vowel never fell.)

In the next stage, the positive vowel features became
autosegmental.  Proto-Albic (the ancestor of Old Albic and
further yet to be explored languages) had essentially the
same system as Old Albic, but without the umlaut rule, and
front rounded vowels did not yet occur.  The North and West
Albic languages don't have front rounded vowels until today,
but South Albic (which classical Old Albic belongs to)
innovated the umlaut rule leading to the occurence of front
rounded vowels.

> A cool system!
Thank you!

Greetings,

Jörg.