8.1 Causative 443"elicition-ese," and in several cases when the form was repeated later it did
respect Sibilant Harmony. All causatives that are in common use respect
harmony (in T-ka). Harmony seems to be less reliable in R and A-grm, though
this comment is based on elicited examples. I recorded PerfP
-aes-raerzaezzaew- and later -aez-vaerzaszzaew- from the R speaker for 'fell
(tree)'. Another R form is PerfP -aes-kaerzaszzaew- 'cluck (tongue)'. A-grm
PerfP -s-aeffaenasssae-t 'break (melon) open with hand' is another non-
harmonic example.
For -vnsu- 'be sold (or bought)', the causative is -s-vnso- for R and T-ka
but -z-vnsu- for other dialects (A-grm I K-d). Here we get positional but not
voicing harmony. The causative is much more common than the intransitive;
with different directional clitics it is the basic verb for 'buy', 'sell', and
'barter'.
A-grm causative -s-vzvd- and K-d unsegmentable -svzzvd- 'prostrate
oneself (in prayer' are borrowed from Classical Arabic root Vsjd (j is
pronounced ζ in Hassaniya Arabic). T-ka causative -s-vjvd- (same gloss) does
not violate Sibilant Harmony.
The simple verb 'laugh' has C's d...z when separated by a V, but ts
varying with ts when clustered, e.g. PerfP -stsa- (-atsa-) and LoImpfP -ddzz-.
In the causative, the C's in question are always clustered. There is a (faint)
possibility of deriving ts from underlying /dz/, and if native speakers agreed
we could, in theory, have -z- as Causative prefix. In fact, we get -s- (e.g. PerfP
-sss-atsa-), showing that native speakers take the stem-internal sibilant to be s
(or possibly s) instead of z.8.1.3 "Double" causatives
In addition to simple causatives with prefix -s- and its variants, there are also
some morphological double causatives with two adjacent -s- prefixes (or
their harmonized variants) that are separated in non-perfective stems by a V. In
the perfective stems, the difference between simple and double causative is not
immediately obvious to the untrained eye, since simple s in the prefix is
geminated to ss by C,-Gemination, while the type with double prefixation has
s-s with no separating V. However, the imperfective stems and the VblN have
separating V's in the double-causative derivatives. To add to the confusion, a
third category is that of simple causatives of underlying verbs that already
happen to begin in a sibilant.
'Wash' (453.a) is a simple causative (note the secondary gemination of the
prefixal sibilant in the PerfP; 'make curdle' (453.b) is a simple causative of a
sibilant-initial stem (Vs-sly); and 'make similar' (453.c) is a double causative
(Vs-s-lh). The patterning of initial sibilants is as follows (omitting hyphens):
(453.a) has s alternating with ss ; (453.b) has ssvs alternating with sus and svs
("v" = short V, "υ" = full V); and (453.c) has ss alternating with sus and with
svss.