A Grammar of Tamashek (Tuareg of Mali)

156 3 Phonology

3.5.3.3 Lexical Accent Erasure andχ-Erasure

Consider now the (definite) object relatives in (134).

(134) a. w-ά

Ma-Sg/Dem

'what she eats'

0-taett

3FeSgS-eat.LoImpfP

b. w-ά

Ma-Sg/Dem

'what he eats'

i-taett

3MaSgS-eat.LoImpfP

c. w-ά

Ma-Sg/Dem

'what you-Sg eat'

0-taettae-d

2SgS-eat.LoImpfP-2SgS

d. w-ά

Ma-Sg/Dem

'what they-Ma eat'

taettae-n

eat.Lo!mpfP-3MaPlS

The accents on the verb in (134.b) for 3MaSg subject, and in (134.c-d) for

2Sg and 3MaPl subject, are those predicted from Rightward Accent Shift. This

rule does not apply to (134.b) because of its audible subject prefix, but does

apply to (134.c-d) because they have no audible subject prefix (either because

there is no prefix, or because a t- prefix for 3FeSg or 2nd person has been

deleted before a C). However, the verb in (134.a) has no accent at all,

resulting in phrasal accent on the demonstrative.

The verb in (134.a) is 0-tastt, from /t-tattA/ before χ-pcl Erasure and the

other rules apply. When the 3FeSg prefix /t-/ is deleted (as usual in verbal

morphology before a C-initial stem), resulting in a (surface) unprefixed verb,

we might have expected Rightward Accent Shift to apply, shifting the accent

onto the stem-final /A/. Indeed, there is no prohibition on shifting the accent to

a word-final V (135).

(135) w-ά 0-raddii

Ma-Sg/Dem 3FeSgS-expect.LoImpfP

'what she expects'

However, the /A/ of LoImpfP /-tattA/ is one of the underspecified vowels

that is deleted word-finally by Stem-Final i/A-Deletion (29) (§3.1.2.4). Since

accentuation is orthogonal to this deletion (and most other segmental

phonological rules), being accented after Rightward Accent Shift does not save

the /A/ from deletion in (134.a). Having nowhere to land, the ablaut-induced

accent of /t-tattA/ simply vanishes, and we end up with unaccented 0-taett as

seen in w-ά 0-taett.