jeff_l
(Jeff_L)
#1
74 3 Phonology
bisyllabic. However, I did record a syncopated suffixal PI t-i-bkun-en
(alongside ablaut PI t-i-bakkan).
In cases like VblN ά-xbubu, note that the accent is on the surface
antepenult. However, this does not necessarily mean that Default Accentuation
applies only to the output of Stem-Initial Syncope. We can get the correct
output either by having Default Accentuation apply last, or by having it apply
first but adding a rule that accent is reassigned to the next syllable on the left
when the accented V is deleted (by Stem-Initial Syncope). This latter
interpretation would have a stage /a-xabubu/ in the derivation that ends up as
ά-xbubu (compare A-grm a-xabubi).
Assuming a Stem-Initial Syncope rule in the forms on the right in (51), the
deleted short V is clearly hi in Shlmpf -sxbuba-t (from /-axabuba-t/) and in
VblN ά-xbubu (from /a-xabubu/). This is based on the prevailing vocalic
melodies for Shlmpf stems and VblN's for stems of this general shape. Let us
see if the other cases of Syncope can be analysed in such a way that the
syncopated V is hi rather than /ae/, so that an asymmetrical Syncope rule can
be posited.
The obvious problem with this is that Stem-Initial Syncope also applies to
PerfP -aexbabae-t, which appears to have a uniform melody. The most
straightforward derivation would be from /-aexaebabae-t/ with stem-wide
melody. This would force us to adopt a symmetrical version of Stem-Initial
Syncope, applying to /ae/ as well as hi. However, there is an alternative
analysis where -aexbabae-t is instead derived from /-axababae-t/, first
syncopating to /-axbabas-t/ before surfacing (after Leftward L-Spreading, see
below) as -sxbabas-t by Short-V Harmony. In this case, we could argue that
Syncope applies specifically to hi.
The first part of the argument for this analysis is as follows. In the
perfective family of stems (and disregarding the unprefixed perfectives,
probably of nominal origin, of adjectival verbs), the surface melodies are either
of the type , as in -aexbabae-t, or , the latter expressed as surface
vocalic sequence «HL», «HHL», «HHLL», etc., depending on the number of
stem syllables. The stem-wide perfective melody occurs precisely in those
verbs that also show Syncope, while non-syncopating verbs have the
composite perfective melody . We should therefore consider the
possibility that the stem-wide perfective melody is reducible to an
underlying , which would permit us to position a schwa in the
syncopation site.
Second, in deverbal nominals of verbs, and in some nominal Sg/Pl
alternations, there is evidence that a syncopates while ae does not. For
example, the same verb xvbubu- (+ -t) seen in (51) above is related to a noun
t-ae-xaebabu-t-t 'gaping hole'. The first two syllables of the stem have L
vocalism, so we get ae rather than a after x. Whereas s in VblN /a-xabubu/
syncopates, giving ά-xbubu, the ae between χ and b in t-ae-xasbabu-t-t fails to
syncopate. Many verbs have pairs of formally similar VblN's with stem-wide
stem melody and a related noun (often agentive in sense, beginning with
