jeff_l
(Jeff_L)
#1
32 3 Phonology
There are some (sub-)dialectal variants such as e-kew and e-ke for 'root',
and afrew versus afaerr 'wing', where one variant shows final w. In a-laesso
and variant ά-laessaew 'black turban cloth', asw varies with o. Likewise aggu
'griot' (regional word) varies dialectally with dggiw.
Among verbs, note such pairings as -olaeh-'resemble' and related noun
m-llhaw 'resemblance', and the intraparadigmatic variation seen in PerfP
-seffo- (dialectally -aeffew-) and Imprtlfaw for '(day) break'. For more detail
on '(day) break' see §7.3.2.15.
The largest set of alternations of final zero with semivowel are nominal
plurals that involve both a suffix (MaPl -aen or -an, FePl -en) and a stem
change. One pattern typical of certain types of VblN, especially Sg type uCaC,
and some other nouns, is exemplified by lidaf 'holding', PI udfaw-aen (see
§4.1.2.10). Here one could argue that the w serves to separate the α of the stem
from the suffix-initial V. However, the productive device for avoiding a vowel
cluster in this position is to use a postvocalic allomorph of the PI suffix, e.g.
MaPl -taen instead of -aen. The addition of w just before a PI suffix like MaPl
-asn is often accompanied by other idiosyncratic stem changes, such as the
extra α in üdfaw-asn (compare Sg udaf), or the broader ablaut seem in e.g.
e-baeje 'horse', PI i-bajw-αη. I therefore take the w in üdfaw-aen (and
l-bajw-αη) to be part of the stem rather than part of the suffix or an epenthetic
linker. As a result, I recognize more cases here of stem-final w confined to the
plural.
There are many nouns that end in phonetic [i] or [u] in the unsuffixed
masculine singular, [i] can represent phonemic /ay/ or /i/, and [u] can represent
/aw/ or /u/. The following tests for distinguishing diphthongal from
monophthongal representations are available: a) adding a lSg possessive
suffix, which is -in after C but -nin after V; b) adding a V-initial suffix like
MaPl suffix -sen, which is normally extended to -taen after V but not after C;
and c) if there is an unsuffixed ablaut plural, /i/ and /u/ should be replaced by
a, /ay/ by ay, and /aw/ by aw. As it happens, the data from these tests are often
inconsistent, even for single speakers. Most often, singular nouns ending in [i]
or [u] are treated as V-final when a pronominal possessor suffix is added, so
we usually get postvocalic lSg allomorph -nin, though postconsonantal
allomorph -in is occasionally attested. Some of these same nouns consistently
form ablaut or suffixal plurals that presuppose final /ay/ or /a w/ in the
singular. All the verbal nouns of the heavy verb stems ending in ...ay and ...aw
are of this type, e.g. VblN a-flawfalaw [aflufa'lu] from (PerfP)
-aeflsewfselasw- 'flicker' (lSg possessve α-flawfalü-nin, much less often
α-flawfalaw-in, but plural consistently a-flawfaliw-aen). There are also a
number of simple nouns like ae-bori 'stick' of this type, though for these nouns
the singulars are more consistently treated as V-final (lSg possessive regularly
ae-bori-nin, but plural i-boray-aen).
One way to analyse the differential effect of a pronominal possessor suffix
and a PI formation (ablaut or suffixal) is to argue that plurals take shape at an
