From: dubbelax
Message: 63688
Date: 2009-03-29
>At least Arabic has quite a number of such roots. It seems that C1=C3 roots are a result of simplification of reduplicated roots, i.e. of C1-C2-C1-C2. Another thing is that such trimmed reduplications show a tendency for dissimilation, cf. Ar. a'a'a "fail to conceive or be fructified"; a'isa "be hard and firm", "be hard and rugged"; a'iza "be hard and rugged"; assa "be dry, withered"; possibly also as.s.a "have little milk" and az.z.a "be hard (i.e. toilsome)". It looks like full reduplications hold the original value, while trimmings, where the awareness of reduplication is weaker, feel free to deviate. Which is quite natural. This is also why I am inclined to see the Semitic root for "three" (and also for "sun") as a C1=C3 root, which may have gone through some dissimilative changes here and there, especially if we take into account the proneness of numerals to (sometimes pretty dramatic) simplification, alliteration, rhyming etc. It is perhaps worth noting, however, that Aramaic and Syriac forms cannot be counted among such dissimilations, because their t_ is just a relatively young allophonic variation of t and, as such, has nothing to do with the reconstructed Semitic t_. Concerning South Semitic, I have nothing to say, for I do not see very deep into it and all my books on those languages are unreachable for me at the moment. Nevertheless, I can draw your attention to a possible supportive evidence for the original Semitic t_ in "three", namely the Arabic word t_allatun "flock of sheep", which may fit into the idea of counting 1-2-MANY, expressed here by Rick McCallister. However, take the last sentence with reserve :-)
> C1=C3 is far from unknown in Semitic - we have Hebrew/Aramaic n-t-n
> 'give' (also Geez, though it has been sugested this is a loan), and
> Akkadian n-d-n. What's the Proto-Semitic 'sun' root currently
> considered to be? s^-m-s^ or s^-m-s? s^-C-s^ is actually quite
> common in Hebrew, but that may just be a feature of Hebrew.
>
> Richard.
>