--- In
cybalist@yahoogroups.com, "fournet.arnaud" <fournet.arnaud@...>
wrote:
>> I haven't found any definite examples, but the Ethiopic languages
look
>> quite promising. Geez and Tigrinya look quite possible for a contrast
>> between /kW/ (labiovelar) and cluster /kw/.
>> A phonological difference
>> from the clusters, even if there be no minimal pair, is that while
>> Tigrinya quinquiliterals are extremely rare (reportedly just one true
>> quinquiliteral, namely g-r-n-g-r 'start to form pod (of legume)'),
>> roots with two labiovelars are not, e.g. gW-n-kW, kW-r-kW-m, kW-r-kW
>> and gW-r-gW.
> It looks like reduplication / infixation of root : R n/r/ R.
Wiki has an article on the Tigrinya verb -
http://en.wikipedia.org/wiki/Tigrinya_verbs . It's a bit limited -
it's like a Hebrew grammar that gives you the seven basic verbs stems,
and leaves you to discover the hishtaph`el on your own.
Digging around further, I can't find any evidence of an 'n' or 'r'
infix. There are three consonants that can occur after the tä- and
'a- prefixes, yielding for example tan-, tas- and taS (where I have to
leave the nature of the sibilant unclarified).
> Are there roots like g _n_w / g_r_w / g_l_w ?
Probably - I've no dictionary. I think you mean g-w-n etc.
According to Eugene Buckley
(ftp://ling.upenn.edu/papers/working-papers/volume4.3/buckley.ps - I
suggest you Google the URL and then take the conversion to plain test
if you don't have a PostScript reader), the root consonant sequences
actually seen are:
Biliterals:
C1 C2 C2
C1 C2 C2 C2
C1 C2 C1 C2
Triliterals:
C1 C2 C3
C1 C2 C2 C3
C1 C2 C3 C3
C1 C2 C3 C2 C3
Quadriliterals four-slot
C1 C2 C3 C4
C1 C2 C3 C4 C4
He must be counting the reciprocal-causative prefix as "'at" - thouhg
I've only seen examples in which the /t/ assimilates to the following
consonant.
None of these patterns will give you g-w-n-g-w-n.
Please ignore my claims of gW-n-kW, kW-r-kW, gW-r-gW. They arise from
trying to interpret non-Unicode text. The roots with multiple
labiovelars are all quadriliterals with labiovelars for C1 and C3.
It has been suggested that the kW-r-kW-m type roots actually do derive
from sequences C1 C2 C1 C2 (one of the patterns above) or C1 C2 C1 C3
(a formation known from Arabic).
> or g_n_g / g_r_g / g_l_g ?
> or k instead of g ?
No, this is a striking difference between velars and labiovelars. The
only root with two velars is g-r-n-g-r, but 10 roots with two
labiovelars. It is claimed that labiovelars do not go back to
Proto-Semitic.
There is a possibility of finding a constrast between /kW/ and /kw/ in
the verbal inflexions. The imperfect active forms with a vocalic
suffix may yield the cluster. For example, the 3m.pl. imperfect
active of fätäwä 'like' can be y&fattu or y&fatwu. Even a form in
-kku could be evidence that /kW/ and /kw/ have been difference. The
feminine endings may be better evidence, for the Ethiopic script has
no way of representing /kWu/. (Can any Semiticists help us out on this?)
Richard.