A Grammar of Tamashek (Tuareg of Mali)

120 3 Phonology

light V-final stems, like (PerfP) -aCCa-, it has an audible effect only in
certain subject-suffix combinations where we end up (in the PerfP) with an
ae in the right position, e.g. 3MaPl -aCC-aen (from /-aCCa-aen/). Thus 1-bsa
'he vomited' and war i-bsa 'he didn't vomit' (PerfP = PerfN), but absae-n
'they-Ma vomited' and war abse-n 'they didn't vomit' (PerfN distinct from
PerfP). Note also war abse-naet 'they-Fe didn't vomit' from PerfP absse-naet,
where the aet of the 3FePl suffix is disregarded but where the suffix's η helps
create the shape [absae-n] that allows e-pclf to appear audibly. With C-final
verbs, the addition of a subject suffix such as 3MaPl -aen has no effect on the
PerfN stem, so the relevant brackeing is confined to the stem proper.

Allowing for the selective bracketing of subject suffixes (or parts thereof)
with V-final stems only, the descriptive analysis on the targeting of e-pclf is
now complete. While there is no way to tell, it is reasonable to assume that
e-pclf is "present" in all PerfN stems, but can only be realized audibly under
the conditions noted. When these conditions are not satisfied ("pel" does not
converge with "f," or they do converge but the relevant V is not ae), e-pclf
misses its target and disappears into empty space.

3.4.5.5 Penultimate V

The penultimate V of the stem is targeted by χ-pen, which forms part of the
ablaut of VblN's of non-augment heavy stems. Consider (93).

(93) Final-Syllable Vowel (Heavy VblN)

gloss stem

a. 'shine'

'be rude'

-mvlvwlvw-

-budvr-

b. 'be confused' -mvtvllu-

c. 'scatter' -wvsu- (+ -t)

VblN

a-mlawlaw

a-büdar

a-mtall

ά-wasi

formative

χ-pen

χ-pen

χ-pen

The forms in (93.a) show a clear marked accent on the penult. (I omit
variants with α rather than a in the final syllable, like a-mlawlaw, since this
change does not affect accent.)
In (93.b), we have apparent final-syllable accent, but this is due to deletion
of a stem-final V, compare LoImpfP -t-lmtalli- and PerfP -aemtaella-. If we
order the association of χ-pen to the stem before Stem-Final i/A-Deletion (29),
the form will come out correctly.
(93.c) shows that augment verbs, i.e. verbs that take Augment -t in most
inflected forms, do not have χ-pen in the VblN, even though the Augment is
not part of the VblN (or any other nominal).