3.1 Segments 41realization of /a/, harmonic to i, before a BLC, but for other dialects it is a true
initial ze). The Shlmpf is dialectally variable, e.g. Sg Imprt ram (qam) versus
Yayam. This variation suggests competing analyses of the perfectives, as
either -aqqima- or -sqqayma- (plus variants with initial ae). The LoImpfP is
usually -t-arasyma- but -t-arama- is attested dialectally. See §7.3.2.9 for
more on 'sit', and on the other stem of the same type (Vjyh or Vjh 'witness').
3.1.2.7 Medial ...Cy/...Cay/...Ciy,...Cw/...Caw/...Cuw before V
A noun meaning 'preaching' is pronounced [talu'lija]. Just from its
pronunciation, it is difficult to determine whether the correct phonemic
transcription of the stem is -lulya, -lulaya, or -luliya. This uncertainty is
reflected in the accents of variant suffixal plurals.
The main T-ka informant gave the PI as phonetic [tilu'l(i)jawen]. Since the
accent is never to the left of the antepenult, there is clearly no structurally
recognized V between the second 1 and the y of the Sg, which I therefore
represent as t-s-lulya. However, the same speaker also gave a variant PI
[tiluli'jawen], where the V between 1 and y ([j]) is accented and so is
obviously structurally recognized. This variant PI points to a Sg t-a-lütaya.
Another case of variation, this time dialectal, is the Arabic loanword
aelrafayaet or aelrafyaet 'peace' (among other variants). The difference is best
seen in the PI, aslrafayaet-asn (T-md) or aelvaiy-aet-en (A-grm).
The same uncertainty can apply to ...Cw/...Caw/...Cuw before a V. Thus
'kinship', pronounced [tat:i'r(u)wa], arguably phonemic t-attirwa or
t-atfirawa.
In many cases one can decide on a "correct" transcription by recognizing
the morphological pattern involved, and transcribing on the model of the
clearer instances of that pattern. This is always the case in verbs, which fall
into fixed syllabic types, each of which has its own paradigm. For example,
PerfP -ajyass- 'vaccinate' and -ajwaes- 'trim', regardless of any low-level
phonetic variation, clearly belong to the -vPQvC- verb type, as confirmed by
their various inflectable stems and derivatives. In the case of 'kinship'
(preceding paragraph), the transcription t-atfirawa seems best, by comparison
with similar abstractives like t-awwlmada 'humanity' (§8.6.5). As a result, the
only real uncertainties occur in a modest number of nouns like [talu'l(i)ja]
'preaching' that do not fit transparently into a stem class.