A Grammar of Tamashek (Tuareg of Mali)

298 7 Verbal morphology

There is a partial phonological basis for the distinction between augment

and non-augment stems. First, the quality of the stem-final V shows more

variability in non-augment (306.a) than in augmented (306.b). The PerfP in

(306.b) always ends in ...ae-t, while we have examples of final α and u in

(306.a); likewise, the imperfectives for non-augment verbs can end in i, u, a,

or zero (which depending on the particular form can be analysed as III or /A/),

while augment verbs have only high V's.

Second, a glance at (306) shows that all light stems are non-augment.

There are middleweight and superheavy verbs of both augmented and non-

augment types. However, among middleweight verbs, my data suggest that all

-CvCCu- stems are non-augment, all -CvCu- stems are augmented, and only

-CuCCu- and -CuCu- stems are split between the two.

Among superheavy stems, -CvCvCCu- and -CvCuCu- stems can be

augmented or non-augment, but all other underived superheavy verbs are

strictly augmented. So -CuCu-, -CuCCu-, -CvCvCCu-, and -CvCuCu- are the

four shapes where either choice can be made. For -CuCCu-, note non-augment

-jujju- 'load' and augmented -jukku- 'fine'. For -CvCvCCu- note non-

augment -mvtvllu- 'be confused' and augmented -hvrvkku- 'obtain'.

The situation in A-grm dialect is rather different. Augmented verbs with

-CuCCu- shape (full u medially) in T-ka and other dialects usually have

cognates in A-grm with medial short V, i.e. -CvCCu- (+ -t). For this dialect,

non-augment and augmented classes compete broadly across middleweight

V-final verbs.

7.1.2 Alternative segmentations of Augment -t-

As with some other suffixes, there is some ambiguity as to the precise form of

the -t- augment. I segment it as -t- and assign the preceding as or a to the verb

stem. Certainly there is evidence that the stems in question end in a full V (it

appears overtly in the VblN, for example), so we need a V-Shortening rule that

applies to the V when it is followed by the augment. The closest parallel to this

is the shortening of a stem-final V (in unaugmented verbs) before C-initial

subject and Participial suffixes, e.g. PI Partpl -nen (§8.5.1). This analysis

allows us to take the final full V (i or u) that appears overtly in the VblN and

in the LoImpfP as lexical (except where a final u is attributable to spreading

from another medial u in the stem).

However, other segmentations of the augment are also possible.

Alternative #1 is to segment the augment as -vt- (v = short vowel), realized as

-aet- or -at-. This strategy would require application of VV-Contraction,

effectively deleting the stem-final full V before the short V of the suffix. This

would entail hyphenating 'burst.PerfP' as -abbuqq-ast (instead of my

-abbuqqae-t), and its Sg Imprt as büqq-at (instead of buqqa-t). To account for

the alternation of -set- and -at- in this analysis, one could simply include the

augment in the scope of the respective stem melodies. Alternatively, we could