3.2 Local assimilations and syllabification rules 51'buy, sell' cited above, I have a hard time finding a single example (not a
recent borrowing) that does not respect sibilant harmony, as opposed to
hundreds of stems with harmonized sibilants.3.2.3 VV-Contraction
When two vowels come together at a morpheme boundary, contraction occurs
unless the combination can be converted into a VCV sequence (by inserting h
or a homorganic semivowel). See (41), below, for a summary of the various
outputs.
3.2.3.1 W-Contraction with 3MaSg subject prefix i-
One case of VV-Contraction is when 3MaPl subject prefix i- is attached to a
verb beginning in a vowel. Inflectable verb stems may begin with a C (see
below) or with one of the V's {a as α i ο u}, with u the rarest (-ujaj- 'be
distant').
When the verb-initial vowel is a, the contracted output is i-, as in
/i-abdasd/> 1-bdasd 'he stood up'. This is true even when a verb-initial a
appears on the surface as as because of the shift of a to ae before a BLC like q
or d (§3.1.2.2). However, the BLC does force the contracted I'll to lower to e,
phonetic [ε]. An example is 'he consented', with 3MaSgS li-l added to PerfP
stem -aqbael- to produce 1-qbael, phonetic [e'qbasl].
However, when the 3MaPl prefix is added to a verb beginning with a true
ae, the output is ae, hence 0-aess-omaem 'he sucked'. Because many PerfP
verbs stems begin in a in T-ka but in ae in other dialects, there are many
conspicuous dialectal differences, e.g. between T-ka 1-mmu-t 'he died' and
other dialects' 0-aemmu-t.
The 3MaSg prefix is likewise realized as zero before a full vowel {α i ο u)
(there are no verb forms beginning in e), as in (PerfP) 0-ogdaeh 'he was equal'
and (Resit) 0-ujaj 'he is distant'.
The 3MaSg subject prefix that I interpret as i- for Tamashek may well
have been *y- originally. In this event, Tamashek contracted *y-a to i, and
deleted *y before other vowels. In Tayert dialects of Niger, we still get 3MaSg
y- before a and before the full vowels (LTF2.419).3.2.3.2 W- Contraction with C(3)- subject prefixesThe C-initial subject prefixes (§7.4.1), e.g. 1P1 n(a)-, 2nd person t(a)-, and
3FeSg t(a)-, could be represented either as C- or Ca-. If we choose C-, at least
in some dialects a Schwa-Epenthesis rule would have to apply (before a stem-
initial C) to account for a Ca- allomorph, as in 1P1 LoImpfP na-bass 'we