A Grammar of Tamashek (Tuareg of Mali)

(Jeff_L) #1
100 3 Phonology

melody, rather than from the lexical representation of the stem. In this analysis,

therefore, the basic representation would be something like -vjjvs- 'enter' or

-usvl- 'run', where ν is an unspecified short vowel and υ is an unspecified full

V. 'Enter', for example, will actually appear in melodically specified stems

such as PerfP -ajjass- and Imprt ajjas.

The alternative is to represent e.g. 'enter' as -ajjas- with lexical schwas.

Then, instead of applying stem-wide melodies, we would take the Imprt (and

Shlmpf) as simply retaining the underlying vocalism, and have a more local

melodic fragment that converts the second schwa of -ajjas- to ae in the

perfective system.

There are too many difficulties in implementing the latter solution, not the

least of which is the fact that for some other types of verb it is not so easy to

derive the perfectives from the Imprt and other imperfectives. For example,

'run' has Imprt asal and PerfP -osael-. In nearly every case, the vocalism is

predictable from the skeletal stem shape, defined as a sequence whose

elements are (actual) consonants with interspersed ν and υ. Thus all -uCvC-

verbs have Imprt aCaC, PerfP -oCaeC-. and LoImpfP -t-iCaC- (A-grm dialect

-t-dCasC-), while all -vPPvC- verbs where PP is a geminate consonant have

Imprt aPPaC, PerfP -aPPaeC-. and LoImpfP -t-aPPasC-. (For some syllabic

shapes it is necessary to specify a cluster as geminated PP or as non-geminate

PQ.)

I will use formulae like -vjjvs- 'enter' and'-usvl- 'run' as basic

representations of verbs, with ν and υ marking the short and full vowel

positions. I will use schemata like -vPPvC- and -uCvC- to designate the

relevant stem-shape classes in a more general way. It should be emphasized

that "C" (and its substitutes P, Q, R) is not part of the basic representation of

any individual stem. Therefore "C" is used quite differently from ν and υ.

Capital "V" denotes any vowel (unless otherwise specified).

In some verb types (those with a medial full V, and augmented V-final

verbs), it seems necessary to specify lexical full-V features rather than relying

on vocalic melodies. This is because the melodies consist of L and Η elements,

which suffice to distinguish the low full V α from the high V's, but are not

precise enough to predict which of the two full Η vowels will appear, i or u. In

verbs, u is more common than i, but the choice is lexical. An example with u is

-durhvn- 'desire', which appears as PerfP -addurhaen- (dialectally

-aeddurhasn-), Imprt dürhan, LoImpfP -t-ldarhun- (dialectally -t-odasrhan-),

and VblN α-durhan. The apparent jumping of the u into the final sylalble in

-t-idarhun- is due to Medial V-Shortening and u-Spreading (§3.4.9). An

example with i is -kvrikvw- 'perform sorcery', with PerfP -sekrakasw-, Imprt

karikaw, LoImpfP -t-lkrikiw- and VblN α-krlkaw. There are some difficulties

in analysing the relevant paradigms, especially since stems with i and u are

often not exactly symmetrical. However, on the whole it does seem advisable

to specify at least some of the lexical υ positions as u or i.

The stem -kvrikvw- 'perform sorcery' just mentioned raises another issue.

Many of the longer stems have an extra initial short V in the perfectives and
Free download pdf