A Grammar of Tamashek (Tuareg of Mali)

3.2 Local assimilations and syllabification rules 59

preceding V to shift to i), or else they do have an initial V but undergo

VV-Contraction prior to Default Accentuation.

3.2.3.4 Possible W- Contraction with PI prefix i-

In the analysis I prefer, such Sg/Pl nominal prefixal alternations as seen in Sg

ae-jola 'stepchild', PI i-jdl-an are interpreted as alternations of a Sg prefix

(here ae-) and a corresponding PI prefix i-. However, there are two alternative

analyses. One, not involving VV-Contraction, is that as- (or any other prefixal

V) becomes i by ablaut (i.e. by having the Sg prefixal V targeted by both a

<H> vocalic melody and a V-lengthening ablaut component χ). I reject this

analysis, since the Sg/Pl prefixal alternation takes place not only when the PI

shows stem ablaut, but also when the stem is unchanged from Sg to PI so the

PI is expressed solely by affixes.

There remains a possible third analysis whereby PI prefix i- is added to,

rather than replacing, the initial V of the Sg, which is therefore better

considered to be part of the stem (rather than a Sg prefix α-, e-, ae-, or a-). This

would entail a VV-Contraction rule, e.g. /i-asjola-aen/—»i-jo... I reject this

analysis too. There are other reasons to segment Sg prefixes, specifically those

with a full V (a- or e-), since these reduce to 'as- or "a- in certain syntactic

positions by Prefix Reduction, as in a-baembasra 'Bambara man' (reduced

nae-baembaera, cf. PI i-baembaera-taen). Such stems contrast with others that

begin with a true stem-final V that shows no reductions or Sg/Pl prefixal shifts,

e.g. arab 'Arab (man)' and PI arab-aen 'Arabs'.

In short, while mildly tempted by the alternatives, I consider PI i- to be a

simple prefix, replacing Sg vocalic prefixes.

3.2.3.5 Summary of W-Contraction processes

The treatment of underlying VV combinations in the preceding sections is
summarized in (41), showing contractions, and (42), showing intervening h or
homorganic semivowel. The summary is valid for T-ka and excludes some
minor eastern and northern dialectal peculiarities mentioned above. Also
excluded are the analyses of nominal PI i- that were just considered but
rejected in (§3.2.3.4, above), and cases of suffixal α instead of ae that are not
due to VV-Contraction as such, rather to ablaut-induced lengthening from /ae/
to a.