3.2 Local assimilations and syllabification rules 49(36) Cases of Consonantal Metathesis
a. hS Sh (S = a sonorant or fricative)
b. rwr wnr rv
c. rs sr
d. dj jd
e. dj jd jjI begin with instances involving h, mainly underlying /hS/ metathesizing
to Sh when not separated by a V, where S is a sonorant or fricative.
The verb 'see' appears with consonantal sequence Vnhy or Vhny. The
latter is used when a V intervenes between the first two stem C's, the former
when there is no V in this position: LoImpfP -hannaey- and VblN a-haenay,
but Shlmpf -anhay-, PerfP -anhaey-, and nominal m-anhuy 'something visible,
(a) sight'. I recognize a lexical stem /-vhnvy-/ that metathesizes /hn/ to nh
when the two are not separated by a V.
The verb 'weep' has a similar alternation: LoImpfP -hdll- and VblN
t-a-haela, but Shlmpf -aelh (/-aelhi-/) and PerfP -alha-. I recognize lexical
/-vhlu-/ with /hi/ metathesizing when the two C's are not separated.
Metathesis does not occur when the C's in question are the first two in a
superheavy stem subject to Stem-Initial Syncope. Examples: 'break loose',
with Sg Imprt halabba-t, PerfP -aehlaebbae-t (not metathesized to #-selh...), and
(inflected) LoImpfP -t-lhlabbi-t; 'moan' with Sg Imprt hanaffa-t and PerfP
-shnaeffae-t. If we take the metathesis in 'see' and 'weep' to be a genuine
phonological rule, we would have to order it before Stem-Initial Syncope,
which seems reasonable on other grounds (in some dialects Stem-Initial
Syncope either fails to apply [A-grm], or does apply but does not feed even
routine CC-cluster assimilations such as Nasal Harmony [K-d]).
The LoImpfP corresponding to PerfP -ohaer- 'snatch' is -t-ihar- in most
dialects. For Im, R, and some Kidal-area dialects, LoImpfP verbs of the type
-t-ICaC- syncopate the schwa before a V-initial subject suffix (an Im example
is t-lwr-an 'they-Ma dance'; for Κ I can cite tihz-aen 'they approach'). In R
but not the other dialects, this syncope leads to metathesis of /hr/ to /rh/:
metathesized t-irh-an 'they-Ma snatched' corresponding to K-d t-ihr-an (cf.
T-ka t-lhav-aen). No metathesis occurs in R in this morphological context
when the cluster is /hr/ or /hz/ instead of /hr], so there is no general rule even
in this dialect that IhCI metathesizes to /Ch/: wasr t-ihz-ad 'you-Sg do not
approach'. T-ka and some other dialects do not syncopate -t-iCaC- stems, so
the issue of metathesis does not arise here.
Another dialectal case is the term for 'scrub acacia', i.e. Acacia tortilis
(dominant species in thorn scrub): ά-haks (Κ, T, Ts) but ά-hask (R), also ά-sak
without the *h in the east (A-grm, Gao, Im).
I now turn to cases not involving h. The stem '(be) yellow, light brown'
has a C-sequence Vwnr (Imprt Iwrar 'be yellow!', adjectival noun ά-wrar
'yellow one') alternating intraparadigmatically with Vrr (PerfP aerav 'it