3.4 Ablaut 107quite different paradigms. Moreover, both the 'begin' and 'destroy' classes
behave as light verb stems (e.g. for purposes of VblN formation), while the
-CvCvC- type is heavy (more specifically, middleweight). I therefore prefer
the reconfiguration analysis given in (81). However, the effect of
reconfiguration is similar to the effect of templatic ablaut morphology.
3.4.2 Gemination and degemination in ablaut
3.4.2.1 Τ and Γ-ε2 in long imperfectives
Stem-Initial Gemination occurs in the perfectives and inflectable Shlmpf of
most non-syncopating verbs (§3.4.8): -bvrvj- 'be boastful' (Imprt baeraej,
VblN α-baraj and variants), but PerfP PerfP -abbaraej-. In verbal ablaut, we
get gemination of a medial C (Γ-m) in the perfective of some adjectival verbs
(§7.3.1.10). A more important stem-internal gemination in verb morphology is
considered here.
The long imperfective system of verbs (§7.2.5) displays either internal
gemination or prefixation of -t-, but (with one exception) not both. For
example, -vknvs- 'fight' (PerfP -aknaes-) has geminated LoImpfP -kdnnaes-,
while -mvtvllu- 'be confused' has LoImpfP -t-imtalli- with -t-. There is one
stem-shape that requires internal gemination, and optionally adds prefix -t-.
This is the -vCvC- type, e.g. -vwvt- 'hit' (PerfP -awast-), LoImpfP -awwdt- or
-t-awwdt- (also -aggat-, -t-aggat-).
One might argue that the -t- prefix, which is used with all heavy verb
stems and with those light stems that already have a medial geminate, is the
same phonological entity as ablaut-induced medial gemination. One could
imagine, for example, a formative consisting of underlying -t- that targets a
medial ungeminated C where possible (with /tP/ then realized as geminate PP);
if this targeting is unsuccessful, it materializes as a prefix. However, there is
no direct support for such an analysis in Tamashek morphology or phonology,
and nongeminate surface clusters like tk are common medially in stems. The
fact that -t- and medial gemination occasionally co-occur, as in the variant
LoImpfP type -t-awwdt- just mentioned, is further evidence against the
unifying analysis.
I therefore recognize two consonantal ablaut formatives in long
imperfectives, Τ (i.e. -t- prefix) and T-c2 (gemination of second stem C).
As argued in the preceding section, for the stem-shapes -vPQvC- and
-vPQu-, i.e. light verb stems whose first C position is filled by a nongeminate
cluster, a pre-ablaut reconfiguration to -PvQvC- and -PvQu- (as inputs to long
imperfective ablaut) is useful. This will simplify the process whereby T-c2 is
implemented, converting Q to QQ. Example: stem -vknvs- 'fight', long
imperfective pre-ablaut reconfiguration -kvnvs-, LoImpfP -kdnnaes- (also
LoImpfN -kannas-, Prohib -ksennass-).