56 3 Phonology/a-ae/ —» ae d. light V-final non-augment verbs: 2MaPl and 3MaPl
(but not lSg or 2Sg) subject, and V-initial
Participial suffixes
e. heavy non-augment verbs: all subject and Participial
suffixesSome ViV 2 combinations escape contraction by adding h, or by converting
a high V, into V, plus homorganic semivowel. One could explain the
alternations of h and zero as due to allomorphy, to h-Insertion, or to
h-deletion. Simple allomorphy works for MaPl Imprt C)-aet, Hortative C)-et,
and Future e, which occur in very restricted environments. The h-Insertion rule
is formulated as (653) in §10.2.1.1, but note the fine print there. With dative
pronominal clitics, the h-form occurs in several postconsonantal environments,
making h-Insertion dubious. h-Insertion and homorganic semivowels are
absent from nominal morphology, and where they do occur there is much
variation. Thus -vsu- 'cough' has MaPl Imprt asü-haet (T-ka), asüw-aet (R), or
aess-aet with VV-Contraction (A-grm).
In T-ka, the Shlmpf of augmented verbs like -buffu- 'be in large quantity'
undergoes (among other things) u-Spreading (119, §3.4.9) in forms lacking the
Augment -t-, as in 3MaPl Shlmpf abbaffu-n 'they (will) be in large quantity'
from /-abbuffi-asn/ (via /-abbaffu-aen/ with u-Spreading and Medial
V-Shortening (120)). If the u of -abbaffu-n is considered to be contracted from
/u-as/ (after u-Spreading), it fits (37.b). Alternatively, we could order
VV-Contraction before u-Spreading, e.g. /-abbuffi-asn/ —> /-abbuffi-n/ —>
/-abbaffu-n/.
However, in some eastern dialects, u-Spreading and Medial V-Shortening
do not occur in this Shlmpf paradigm. Instead of abbaffu-n, the 3MaPl Shlmpf
appears (e.g. in R dialect) as abbuffe-n 'they (will) be numerous'. Here the
medial u has not been shortened, and has not transmitted its rounding feature
to a subsequent V. Since the augmented forms in the Shlmpf paradigm have a
(i.e. a high V), as in 3MaSg 1-bbuffa-t, it can be inferred that the 3MaPl is
from /abbuffi-aen/. If so, we would have a case of Γι + ae/ realized as e (rather
than as i). This aberrant output e may reflect analogical interference from the
perfective paradigms of the same verbs, where output e results from stem-final
/a/ plus suffix-initial /ae/, e.g. PerfP abbuffe-n 'they were (or became)
numerous'.
For nouns, VV-Contraction occurs in a subset of combinations of V-final
stem plus MaPl suffix -aen. The majority of such cases avoid VV-Contraction
by using a postvocalic MaPl allomorph -taen, but there is a significant subset of
cases where the MaPl allomorph -aen is used (§4.1.2.13), and these instances
do require VV-Contraction. The output of /...V-aen/ is -an after contraction.
The input stem-final V is α or i in the vast majority of instances, but there is
one case each of ο and u (39.a). Though I cannot cite a full set of stem-final
V's, there is no counterexample to the generalization that stem-final V
combines with MaPl -aen to produce -an.