jeff_l
(Jeff_L)
#1
7.3 Verb classes and irregular verbs 357
forms are bisyllabic, so they yield their accent when Future ad is present, as in
ad n-asks Ί will eat'. In (360.d) the verb plus suffix add up to three syllables,
so the default accent lands on the antepenult, and never shifts to a preceding
particle.
Uncontracted stem-final /i/ is realized as a surface V only in (360.d),
where it appears as a. In (360.b), /i/ contracts with suffix-initial to give as. By
contrast, the same combination contracts to a in (360.c). The difference
between (360.b) and (360.c) in the form of the contracted V also affects the
stem-initial syllable, since Short-V Harmony results in the a...3 sequence in
(360.c). The fact that we get ae not a for the stem-initial V in (360.b) shows
that the contracted V at the stem-suffix boundary is a genuine as, not merely an
underlying schwa that has been backed to ae before a BLC by a low-level
process. In (360.d), the hi occurs between two C's and so is realized as a,
again triggering Short-V Harmony to produce the a in the first syllable.
In word-final position, Shlmpf stems like -aeks 'eat' are phonologically
stable since they have acceptable word-final CC clusters. If, however, the
cluster is unacceptable, on the grounds that the final C is more sonorous than
its precedessor, resyllabification applies. This takes the form of Final-CC
Schwa-Insertion, adding epenthetic a between the two C's. In T-ka but not
other dialects, Final-CC Schwa-Insertion is accompanied by Epenthetic-Vowel
Accentuation, so the epenthetic schwa has a marked accent.
Therefore Shlmpf /-aejli-/ 'go' has the same paradigm as /-asksi-/ 'eat'
when a nonzero subject suffix is present, However, in the forms with no
subject suffix, the two paradigms differ (361).
(361) Unsuffixed Shlmpf of -vjlu- 'go' and -vksu- 'eat'
subject underlying surface
a. -vksu- 'go' (no resyllabification)
1P1 /n-aeksi/ n-seks
3MaSg /i-aeksi/ 0-aeks
3FeSg /t-aeksi/ t-seks
b. -vjlu- 'go' (resyllabification present, T-ka also shifts accent)
1P1 /n-aejli/ n-ajal
3MaSg /i-aejli/ 0-ajal
3FeSg /t-aejli/ t-ajal
[in other dialects n-ajal, etc.]
The regular Imprt Sg for -vCCu- stems of the α/ι subclass is aeCC, exactly
parallel to the Shlmpf -seCC. The MaPl Imprt is aeCC-ast, and the FePl Imprt is
aCCa-maet. The only oddity is that the MaPl Imprt, theoretically underlying
/asCCi-aet/, contracts /i + ae/ to ae rather than a. In effect, the MaPl Imprt suffix
is added directly to the surface Sg Imprt. The same resyllabification processes
