________________
INTERNAL RECONSTRUCTION AND COMPARATIVE METHOD
287
say jd developed allophones in the word final position which were similar to k and ultimately merged with it thus giving us aszk, while ja in the same position developed allophones which were similar to ļ and hence merged with that phoneme to give us sgt. In the position before the vowel, both of them merged to give a simgle sound j, and hence we have astjà and srja. It is customary to denote these two reconstructed sounds as IE W and g(labiovelar and palatal).
This kind of non-unique alternation is not restricted to j alone in Sanskrit. The sound $ also shows such double alteration. For example, the sof dis alternates with k, hence dik, disā, but ś of vis alternates with t in indentical environments, hence vit, visā. Applying the argument of the above case of twofold alternation of j, we should say that Sanskrit s also represents a merger of two proto-phonemes, whatever might have been the phonetic values of these reconstructed phonemes. We may represent them provisionally as k'i and k'. As in the above case, we would say that k'' developed allophones in the word final position which were similor to k and hence merged with it to give us dik, while k'in the same position developed allophones which were similar to t and hence merged with it to give us vit. Before vowel, however, both of them merged into an identical sound & to give us disa and visa. However, we know that the comparativists have reconstructed only one sound to account for Sanskrit &, and it is represented as k. It is suggested that the comparative method here should be corrected by internal reconstruction and we should assume a twofold origin for Sanskrit s.
We may now ask, do we have any traces of the two-fold origin of Sanskrit & in some other language? The answer to the question, it seems, can be in the affirmative. A Sanskrit & corresponds normally to Avestan & (Skt. sapha = Av. safa, Skt. pašu = Av. pasu etc.). But there are a few words where Skt. $ does not correspond to s in. Avesta but to the dental un voiced fricative th' Thus we have Skt. Vsam = Av. tham, Skt. šūra =Av. thüra (besides Av. süra), Skt. Või = Av. V this in aiwithyo 'over-sleeping', and Avestan sakhta and thakhta from Vsac for which there is no Skt. Vlac. Normally we would have set aside these few cases as residual forms which cannot be explained. But since we are led by internal reconstruction to a two-fold IE origin for Skt. &, we may well say that these few aberrant cases in Avesta also point to a two-fold origin for Avestan and th' viz. IE kl and k's. We conclude by saying that IE k" became sin Avestan and thus merged
Madhu Vidya/372
Jain Education International
For Private & Personal Use Only
www.jainelibrary.org