A Grammar of Tamashek (Tuareg of Mali)

3.2 Local assimilations and syllabification rules 57

There are some additional cases where we can posit an underlying stem-
final V that is deleted from the unsuffixed MaSg form, based on morphological
analogy, vocalism, and/or accent (39.b). For fuller data and discussion see
§4.1.2.13.
There are fewer cases of VV-Contraction involving FePl suffix -en, since
most V-final feminine nouns have an inner Fe suffix -t- (before FeSg -t or PI
-en), or else (if the FeSg is unsuffixed) use FePl allomorph -ten. In the rare
case where FePl -en is added directly to a V-final stem, the output is
predictably -en since the suffix already has a full V (39.c).

VV-Contraction with V-Final Noun Stem and MaPl -aen or FePl -en

input output: example (Sg) PI gloss

a. /a + as/ α ae-hara i-har-an 'saltlick'

/i + ae/ α e-daehi i-daeh-an 'sand'

/u + ae/ —» α ae-s-aru i-s-ur-an 'pretext'

/ο + ae/ —> α ά-laesso 1-laess-an 'turban cloth'

b. /V + ae/ —> α a-s-ant "i-s-ant-an 'beginning'

["V" = deletable stem-final vowel, here /i/ in /a-ssnti/]

c. fa + e/ e t-a-kands-t-t t-l-kand-en 'lily tuber'

Comparison of the verb and noun data show that no simple phonological

analysis can account for VV-Contraction in the two cases. Nominal MaPl

suffix -asn imposes its quality on the contraction V, resulting in a. By contrast,

3MaPl suffix -aen on verbs basically loses its /ae/ after a stem-final V. Even

within verb morphology, there are important differences between e.g. 3MaPl

-asn, lSg -aer, and 2Sg -aed. As usual in Tamashek, the "phonology" is

morphologically specialised.

V-initial clitics are directionals (Centripetal -\sdd or -\idd, Centrifugal

-\ln), dative pronominals (beginning in -\a-), and certain object pronominals

(beginning in i or e). All of the V-initial clitics have allomorphs beginning in

h, or else allow a homorganic semivowel to be inserted after a preceding high

V, either of which obviates the need for VV-Contraction. A further complexity

is that several object pronominals have structurally different allomorphs

depending on whether the preceding stem ends in a V or in a C (e.g. 3MaSg

-\tt and -\e), so we do not always have a non-contracted version to clarify the

"underlying" form of the postvocalic allomorph. There is also a fair amount of

dialectal variation, especially in the use of h-initial allomorphs, but also in the

vowel of directional clitics (e.g. Centripetal -\add or -\ldd).

The inventory of VV-Contractions for clitics is given in (40).

Combinations requiring intervening h or a homorganic semivowel are omitted.