jeff_l
(Jeff_L)
#1
64 3 Phonology
3S3W 'drink!'. A larger set of -vPQu- stems end in stable CC clusters in the Sg
Imprt and so do not undergo resyllabification: (Imprt) aebs 'vomit!', aetf 'be
poured!' (Vdf), sets 'laugh' (Vdz), aert 'be implanted!', aeks 'sprout!', aels 'get
dressed!', aelz 'shave!', aend 'collapse!', aeijY 'kill!', aens 'lie down!', sens 'be
sold!' or 'excuse!', zent 'begin!', aerv 'be on fire!', aerh 'want!', and aerz
'break!'. Note that these all involve an obstruent or h as final C. I know of no
-vPQu- stems where Ρ and Q are both sonorants and do not form a geminate.
There is no resyllabification when the cluster is geminate, even when it is a
sonorant: aekk 'go (to)', aezz 'weave!', aenn 'say!'. In the one case involving h
preceding a sonorant, the necessity for resyllabification is avoided, since the
two C's are metathesized (§3.2.2.1) when not separated by a V: 'weep'
(LoImpfP -hall- with Vhl, but PerfP -alha- and Imprt aelh with Vlh).
Resyllabification is blocked in most cases when the verb form in question
is followed by a V-initial suffix or clitic. This is regularly the case with
inflected verbs (including imperatives). Thus 2Sg Imprt assl 'hear!' but 2MaPl
Imprt aesl-aet, the latter having the same shape as the 2MaPl forms of non-
resyllabifying verbs, e.g. äebs-set 'vomit!-MaPl'. For the agentives and VblN's,
resyllabification is likewise blocked when a PI suffix (MaPl -aen, here realized
as -an) is added; see the plurals given in parentheses in (43). However, in the
T-ka VblN (and certain minor nominals) that show Stem-Final Gemination, the
geminate is retained when possessive suffixes are added, hence a-s-akann
'showing', a-s-akann-in 'my showing' (not #a-s-3kn-in as wel might expect
on the basis of unpossessed PI i-s-akn-an).
In the case of Vsw 'drink' and its derivatives, the only set where the final
C is w, the resyllabified form -ssaw is pronounced [-asu^1 ] and can be treated
as V-final at least for purposes of assigning allomorphs (postvocalic versus
postconsonantal) to a following clitic. Thus Imprt 3saw-\tt occurs dialectally
as an alternative to aesw-\e 'drink it-MaSg!'. The latter variant shows the
pattern that is obligatory in other resyllabifying verbs, e.g. aevr-\e 'read it!'.
However, in the more easterly dialects, the number of forms subject to
Stem-Final i/A-Deletion is diminished, which of course limits the scope of
resyllabification. For example, the high-frequency motion verb -vjlu- (eastern
-vglu-) 'go' has short imperfectives that undergo Stem-Final i/A-Deletion (29)
in a few dialects (A-grm T-ka T-md), but not elsewhere (Gao K-d K-f Ts), e.g.
T-ka Sg Imprt ejal 'go!' versus K-d aglu. Likewise, the long imperfectives of
verbs like -ulwu- 'be spacious' show Stem-Final i/A-Deletion (29) in some
dialects (K-d R T-ka) but keep the stem-final V in others (A-grm Gao R), e.g
T-ka 3MaSg LoImpfP i-t-ilsw versus Gao i-t-llwa.
The most systematic dialectal difference is in the VblN's of prefixally
derived stems. For example, T-ka α-s-ilaww 'making spacious' (PI
l-s-alw-αη) corresponds to A-grm a-s-alwi, and T-ka a-s-akann 'showing' (PI
l-s-skn-an) corresponds to A-grm a-s-akni (PI i-s-akni-taen).
On occasion I have recorded Imprt and Shlmpf verb forms ending in
Augment -t, although the verb in question is not otherwise of the augment
class. This tends to happen in elicitation of paradigms of less common heavy
