A Grammar of Tamashek (Tuareg of Mali)

3.4 Ablaut 121

3.4.6 From to pure melody (PerfP verbs)

In §3.4.3.2, above, we saw that while a great many verbs have <H L> melody

in perfectives, another large set of verbs has a stem-wide <L> perfective

melody. The issue to be discussed here is whether the <L> melody can be

reduced to <H L>. Since the stems in question have at least two syllables, and

usually three or more, there is no obvious reason why the Η component of

<H L> should be unable to associate audibly with at least the stem-initial

syllable. However, the <H> melody is in a one-to-one relationship (in T-ka and

other dialects, except A-grm). Given this correlation, the possibility of a

phonological motivation for the <H> perfective melody should be considered.

I will exclude from this discussion the unusual perfectives of adjectival

verbs. These adjectival perfectives are typically C-initial (§7.3.1.11), but

nonetheless fail to accept subject prefixes, including 3MaSg i-. Many of these

adjectival perfectives have stem-wide (usually bisyllabic) <H> vocalism, but

others have <L> vocalism. The perfective stems of these adjectival verbs often

diverge sharply in shape from imperfectives, and generally behave as

lexicalized, idiosyncratic perfective stems. I choose not to decompose them

into ablaut components.

With the adjectival stems excluded, we turn to examples like those in (94).

(94) Perfective Verb (Surface <L> Melody, Stem-Initial Syncope)

lexical form PerfP gloss

a. underived

-xvbubu- (+ -t)

-tvruru-

-jvmvnjvmu- (+ -t)

-aexbabae-t

-aetrara-

-aejmaenjaemae-t

'(hole) gape'

'go down'

'nibble'

b. derived

-m-ukvr-

-nvm-vdvd-

-s-vjvn-

-s-vm-vku- (+ -t)

-s-vnvm-vhvl-

-aem-ekaer-

-aenm-asdaed-

-aes-jaen

-aes-m-askae-t

-ass-naem-aehasl-

'be stolen'

'bite each other'

'make kneel'

'extinguish'

'adjust'

The idea that an underlying melody would become in
connection with Stem-Initial Syncope was sketched in §3.2.7.1, where it helps
reconcile the loss, by Stem-Initial Syncope, of the short V following χ in PerfP
-aexbabas-t with the observation that the syncope rule elsewhere specifically
targets a and does not affect ae. This ceases to be a problem if the syncopated
V is schwa at the point when Stem-Initial Syncope applies, as in the derivation
(95).