A Grammar of Tamashek (Tuareg of Mali)

72 3 Phonology

the penultimate accent in lSg ad aewt-aer suggests that Default Accentuation
precedes Syncope in this case (whereupon the accent becomes "marked"). This
leaves us with the problem of how to account for the shift from expected #swt-
to -aswt-. Perhaps the shift is just another case of Short-V Harmony, since the
V-initial subject suffixes all happen to begin with ae, and since postconsonantal
3MaSgO allomorph -\e has a mid-height vowel (which functions as "low" in
the environment for harmony). A possible derivation is given in (49).

(49) Derivation of Shlmpf -aewt- (Kidal)

If, with whatever morphological restrictions, underlying hi as well as /as/
is subject to harmony with the V of the following syllable, we would have to
formulate Short-V Harmony as symmetrical, converting vowel sequence
<o as» to «ae ae», as well as converting «ae 3» to <o 3». The symmetrical
version of the rule is given as (50).

(50) Short-V Harmony (symmetrical version, applying to both /a/ and /as/)

However, objections may be lodged to the derivation (49) and in general to
the symmetrical version of Short-V Harmony (50). In other morphological
contexts (e.g. many PerfP stems) the vocalic sequence «3 ae» is stable, so there
is no basis for a broadly applicable harmonic rule converting <o ae» into «as as»
in any dialect. As a result, Short-V Harmony could only apply to <o ae» under
very tight morphological restrictions even in the Kidal dialects. Therefore even
in this version the rule is not fully symmetrical.
Moreover, it seems odd that, in the Shlmpf paradigm of -vCvC- verbs like
-vwvt- 'hit', the (Kidal) -aeCC- form is derived (via syncope) from the -sCsC-
form, whereas in the Shlmpf of -vCCu- verbs like -vknx> 'make' (see
beginning of this section) resyllabified -sCsC is derived from /-aeCCi/, via
/-aeCC/ (after Stem-Final i/A-Deletion (29), by Final-CC Schwa-Insertion (44).
In both cases, -aeCC- is associated with prevocalic position, while -sCsC-
occurs word-finally or before a C. It may well be that speakers in some dialects
are trying to reconcile the alternations in the two paradigms, resulting in an

/swst-asr/

/awst-aer/

/awt-aer/

aewt-aer

underlying

Default Accentuation

Syncope (accent becomes marked)

Short-V Harmony (symmetrical version)

as a before a syllable with a high V {i 3 u}

[morphological range as in §3.2.6.2]

3 —» as ; before a syllable with a low V {a ae}

[morphologically restricted contexts]