198 4 Nominal and pronominal morphologyf. feminine, initial α/e alternation
t-a-zsf-t t-ezf-en (A-grm) 'axe'g. feminine, initial ae, with Degemination
t-aellaem-t t-£elm-en (T-ka) 'she-camel'
[several regional variants of Sg and PI]In the masculine cases (174.a-c), and in the feminine noun in (174.g), the
Sg has lexical penultimate accent (hence e.g. daer "'ae-hasrr 'in the lion') while
the PI has no marked accent (hence e.g. daer ehr-αη 'in the lions' with phrasal
accent on the preposition). The feminines in (174.d-f) are compatible with this,
although we cannot tell whether the Sg forms have a lexical accent (FeSg
suffix -t forces penultimate accent anyway). All forms in (174.b-g) show
Syncope of a schwa (less often ae, or in 'hobbles' Q), and where applicable
Degemination.
The two stems in (174.a) show Desyllabification of u (in the Sg) to w
before the PI suffix. Since stem-final u sometimes functions as the equivalent
of aw, these alternations are not far removed from the Syncope cases in
(174.b-c). Alternatively, one could argue that the w is lexical (/elw/, /esw/)
and vocalizes when word-final.
Of interest is the fact that all plurals in (174a-e) are unaccented, even
where the Sg has a clear lexical accent (174.a-c). This indicates that Lexical
Accent Erasure applies to the plurals. The erasure rule elsewhere accompanies
(and appears to be triggered by) the loss of a stem vowel. In (174.b-c), we
arguably have not one but two vocalic deletions: the syncopated schwa, and
the underlying stem-final /i/ that we might posit in order to account for the -an
allomorph in (174.b-c).
In (174.a), we could consider Desyllabification of u to w to be an effective
equivalent of these vocalic deletions, since the desyllabified semivowel is no
longer relevant to accentuation (i.e. it is no longer syllabic or moraic). If that
doesn't convince, we could use a Syncope analysis (/aw/ syncopates
prevocalically to w) or take w as lexical. Furthermore, the MaPl -an allomorph
could be taken as evidence for an additional stem-final vocalic segment (hence
e.g. /elwi/ 'elephant').
In fairness, it should be noted that there is an alternative interpretation for
most of the Sg/Pl alternations in (174), using ablaut rather than
VV-Contraction to account for MaPl -an. The idea would be to have PI ablaut
apply to the combination of the stem and MaPl suffix /-aen/, as I suggest (with
much greater confidence) for the Sg/Pl cases covered in §4.1.2.14, below. This
alternative analysis would take PI elw-αη (174.a), elr-αη (174.b), etc., as
mixed ablaut-suffixal plurals, e.g. suffixed /elar-aen/ plus PI melody
plus lengthening formative χ-f. The derivation would require the L melodic
component and χ-f to target the suffixal vowel, the medial schwa to be
syncopated, and the stem-initial e (like stem-initial i, see below) to escape
melodic modification to u.