jeff_l
(Jeff_L)
#1
36 3 Phonology
The short-V merger is potentially momentous, given the importance of the
a versus ae opposition in verbal ablaut. In verbs of the shape -vCCvC-, for
example, the PerfP stem is -aCCaeC- while the Shlmpf stem is -aCCaC-, and if
the final C is a BLC this aspectual distinction is phonetically neutralized. The
stem-initial a of this and many other verb types is also subject to backing, if
the first C of the stem is a BLC. With a stem-initial short V, however, we can
test for underlying /a/ versus /as/ status by adding 3MaPl subject prefix i-,
which combines with stem-initial /a/ as i- (surfacing as [e] after
Backing/Lowering) but with stem-initial /as/ as 0-ae. In many other cases there
is no such test, but particularly in verbal morphology it is usually possible to
choose between underlying /a/ versus /as/ by comparison to forms of the same
pattern with verbs with plain C's.
In my normal transcription, I attempt to undo the effects of
Backing/Lowering, restoring original a, i, and u where possible. Most
Tamashek stems belong to recognizable patterns with characteristic vocalic
patterns (often reflecting ablaut melodies). This is true of all verb forms
(except the perfective stems of some verbs of adjectival quality). It is also true
of ablaut and mixed suffixal-ablaut nominal plurals, and of some singular
nouns. However, there are some nouns, prepositions, adverbs, and perfective
verbs of adjectival quality (these perfectives are noun-like in form), whose
underlying vocalism cannot be reliably determined on the basis of such class
analogies. If such stems have BLC's, there may be no basis for deciding
between phonemic /i/ and /e/, /u/ and /o/, or /a/ and /ae/. In such indeterminate
cases, I transcribe the vowel as it is heard phonetically, i.e. as ae, e, and o.
Examples: preposition [-vor] 'chez' transcribed ror (not rur), adjectival verb
PerfP [ka'r:oz] 'it became narrow' (where both r and ζ are BLZ's) transcribed
keerroz, though the transcriptions rür and karruz are equally compatible with
the phonetic output and are not ruled out by any morphological pattern.
Because of these indeterminacies, my transcriptions of vowels before
BLC's should be used with caution in reconstructions of proto-vocalism. The
Tamajak dialects of Niger seem to have the most conservative vocalism, since
BLC's do not seem to produce phonemic mergers in those varieties. I have not
done enough careful work on eastern Tamashek (A-grm, Gao, Kidal) to be
completely sure of the situation there. There is certainly some merging of
vowels before BLC's but the mergers are perhaps less rigorous than in T-ka.
3.1.2.3 Deletable final vowels (nouns, suffixes, clitics)
Even without comparison to other Tuareg and Berber varieties, there are
indications in the morphology that original stem- or word-final vowels have
been lost. In some cases, a good case can be made for a lexical representation
with some kind of final V that is subject to deletion in word-final position.
First, some suffixes like FeSg -t behave, for purposes of Default
Accentuation, as though they ended in a vowel: α-baembaerct 'Bambara man'
