3.4 Ablaut 105shape. If the input noun does not have four C's, the output makes use of filler
semivowels, e.g. sabun 'soap', noun of profession swabn-i 'soap seller' with
w in C 2 position.
Tamashek has no productive ablaut of a templatic nature, and is clearly in
the componential camp. However, some long imperfectives approach templatic
status. The best example is an output shape -CaPP®(C)- associated with the
(derived) long imperfective stem of several types of light verbs. The various
ways in which different verbs achieve this output are somewhat reminiscent of
Arabic ablaut. Examples: -vjjvs- 'enter' and LoImpfP -t-djjaes-, -vknvs-
'fight' and LoImpfP -kannaes-, -vbsu- 'vomit' and LoImpfP -bdss-
(underlying /-bassA-/).
Nevertheless, even this -CdPPas(C)- is less rigid than a typical Arabic
ablaut. It subsumes two variants, -CaPP- (for /-CaPPae-/) and -CaPPaeC-, and
these are valid (as such) for only certain types of verbs (including no heavy
verbs). In this larger context, it is possible to account for the same outputs
using a componential model, dismembering the output shape into smaller
ingredients such as consonantal gemination, a
formative χ-pcl, and an accent formative (χ-pcl). So Tamashek can be said to
flirt with a templatic ablaut pattern, but it probably doesn't get there.
Perhaps a more promising area to look for templatic targets is in infinitives
(verbal nouns), where verbs of a given canonical shape tend to have infinitives
of a particular shape. For example, -vPQvC- verbs have VblN a-PaeQaC,
-vPQvC- verbs have uPsC, and -vCvC- verbs have e-CeC. The various VblN
shapes are idiosyncratic and cannot be produced by applying the same set of
VblN ablaut components to the respective input stems. This suggests that each
of these stem-shapes is associated with a VblN template whose vocalism is
pre-specified, and gets its consonants filled by mapping from the input stem.
The VblN patterns are therefore templatic, but (unlike the prototypical Arabic
cases) they do not force inputs of highly variable shapes into the same output
shape. Rather, the templates are in a one-to-one relationship to input shapes.
In nominal plurals, there is one pattern that does seem to force inputs with
a modest range of shapes into a single prefabricated output shape (80). This is
the PI type l-CaCC-αη (masculine) or t-i-CaCC-en (feminine). These are
mixed suffixal-ablaut plurals, where the PI suffix (notably MaPl -aen) is
included in the domain of ablaut, so the α of -an in this case is affected by the
L part of the PI
-CsCC- shape of the core stem is the templatic feature, since we get -CaPQ-
when the Sg input has geminated PP (e.g. 'acacia pod'), and -CaPw- (several
examples) or in one case -CsPy- ('vein') when the input has only two C's
(even if the second one is geminated, i.e. PP, in the Sg).