jeff_l
(Jeff_L)
#1
3.3 Accent 89
V is morphologically specialized, so is the interaction between
VV-Contraction and Default Accentuation.
When the hi is realized as a, as in aksa-naet 'they-Fe ate', of course it
counts as a syllabic nucleus for purposes of Default Accentuation. However, it
also counts for this purpose when it contracts with the initial V of a subject
suffix, as we can see in 3MaPl Future ad aksa-n 'they-Ma will eat' from /ad
aeksi-asn/. If the /i/ were not counted, the (phrasal) accent would end up on the
Future particle ad, since the 3MaPl verb form is bisyllabic on the surface.
Compare e.g. ad awat-aen 'they-Ma will hit' and ad 0-ajay 'he will tie',
which show that 3MaPl subject suffix -aen is accent-neutral, allowing
antepenultimate (i.e. default) word-accent, and that (other) bisyllabic inflected
verbs permit phrasal accent on a preverbal particle.
3MaPl subject suffix -aen and other -aeC subject (and Participial) suffixes
contract with a stem-final V, but the contracted vowel takes different forms
and has different accentual implications depending on verb class (37). At the
outset it is necessary to distinguish two major classes of V-final verbs,
augmented and unaugmented. Augmented verbs are V-final, but show a
suffixal augment -t- in inflected forms with no subject suffix or with a C-initial
subject suffix. (For more details see §7.1 and §7.3.1.6). Except when the stem-
final V is protected by an ablaut V-length element (e.g. χ-f), it is shortened
before -t-, cf. (115). An example of an augment verb is is -mvku- (+ -t)'be
extinguished': 3MaSg PerfP ϊ-mmakas-t 'it was ...', 3MaSg LoImpfP
i-t-lmaku-t 'it is ...', and 2Sg Imprt maekae-t. The PerfP and Shlmpf (including
Imprt) forms have regular default accent, and since the LoImpfP stems are all
at least trisyllabic their stem-initial ablaut accent has no effect on the output.
I assume that augment verbs are V-final, e.g. /ammaka/ for 'be
extinguished.PerfP'. Allowing for the shortening rule, this is what we see in
3MaSg 1-mmakae-t 'it was extinguished'. Now consider how the final V
contracts with V-initial subject suffixes. The 3MaPl PerfP form should be
underlying /ammaka-aen/, but appears as ammake-n (the hyphen could
arguably go before the e). Likewise we get lSg ammake-r and 2Sg
t-ammake-d. The point to note here is that regular antepenultimate accent
occurs on the surface forms, in spite of the VV-Contraction (37.d). If indeed
/ammaka-aen/ is the correct underlying form for ammake-n, the
VV-Contraction rule must precede Default Accentuation in these forms.
In the Lolmpf forms, where the stem-final V is protected by the full-V
ablaut feature of the ablaut formula, the short V of the subject suffix is deleted
(37.b): 3MaPl LoImpfP t-lmeku-n 'they are (being) extinguished'. Again note
the default antepenultimate accent (which converges with the ablaut accent in
this case).
I discussed some non-augment V-final verb stems in §3.3.1.2, just above
(-vksu- 'eat', -vbsu- 'vomit', -fVyku- 'be searched', etc.). As noted there, these
verbs lose the final V in some unsuffixed stem forms (e.g. ShlmpfP -aebs and
-aeffaeyk), but behave accentually as though the final V were still present. We
