A Grammar of Tamashek (Tuareg of Mali)

3.2 Local assimilations and syllabification rules 71

Thus aeqqim-3Y-\3dd Ί sat (=stayed) here'. Because γ is a BLC, it is often
difficult to distinguish as from 3 in this combination. In cases like Shlmpf
aeks-aer Ί (will) eat', where -aer combines with the hi of /asksi/ to give as
instead of the usual 3, and therefore does not trigger Short-V Harmony to
#sks3-Y (parallel to 3MaPl Skss-n), the low-level optional raising of -aeY to
-3Y- before the Centripetal does not affect the first syllable, hence
aeks-aeY-\sdd varying with asks-svAsdd, not #sks-SY-\sdd. Likewise, cases
like Imprt aenY-Vsdd 'kill!' with Centrip clitic do not raise the ae of the stem to
3.
Nouns borrowed from Arabic often incorporate Arabic Definite prefix
al-, and some of these cases require harmonization in T-ka. An example is
T-ka sddslil 'reason' versus aeddaelil in other dialects. The important noun
meaning 'people' is of Arabic origin too and shows the same dialectal
variation: T-ka sddinaet, other dialects' aeddinast.
The harmonic rule is basically asymmetrical, converting ae to 3 but not
vice-versa. At least I know of no cases (in T-ka) where an underlying 3 is
converted to ae due to harmony with a low V in the following syllable. There
are composite vocalic melodies in both verbal and nominal ablaut,
which have the effect of producing vowel sequences (in adjacent syllables) of
the type schwa plus low V (PerfP -sjjaes- 'enter', PI t-i-dskkal 'handfuls').
Note that the schwa's are not converted to ae here, even for T-ka.

3.2.6.2 Symmetrical version

The analysis of (asymmetrical) Short-V Harmony given above, especially as it

interacts with Final-CC Schwa-Insertion, is debatable. I have presented it as a

more or less pure phonological process, but at least for some dialects it is more

highly morphologized than this suggests. Consider, for example, verbs with

PerfP shape -sCaeC- like -swaet- (dialectally also monosyllabic -wset-) 'hit'

(§7.3.1.1). The Shlmpf stem in most dialects is invariant -swst-, with the

Shlmpf <H> melody found for other similar verb types (compare Shlmpf

-sPQsC- and -sPPsC- for stems with a medial CC cluster).

However, some dialects around Kidal have Shlmpf -aewt- (instead of, or

varying with, -swst-) when followed by a V-initial suffix, hence Future lSg

ad aewt-aeY (note the accent) instead of or varying with the more widespread

ad swst-aeY. Likewise, Kidal-area 2Sg Future ad t-sewt-aed and 3MaPl ad

aewt-sen. In all dialects the word-final or preconsonantal form is -swst-, as in

Future 3FeSg ad t-swst and 1P1 ad n-swst.

In Kidal-type dialects with prevocalic -aewt- alternating with word-final or

preconsonantal -swst-, one should take -swst- as basic in the light of the

overall nature of the ablaut system. This means deriving -aewt- by some

(morpho-)phonological process sensitive to prevocalic position. The only

reasonable implementation of this strategy is to allow the medial /s/ of -swat-

to be syncopated, giving /-swt-/. (For Syncope see §3.2.7.2, below) Actually,