________________
SEPTEMBER, 1918]
E AND O IN MARWARI AND GUJARATI
231
It will be noticed that in all the above examples the ya or va which combines with preceding a to form é or ó, is initial in the second member of a nominal or verbal compound. This is a very important circumstance, because it contains in itself the reason why the ya or va in all these cases did not undergo sam prasarana. I have said above that sam prasarana is one of the most marked features of the O. W. Rajasthani, and that every ava of the Apabhramša is changed into aü in the former language. But when va is initial in a word or comes immediately after a prefix, sam prasdrara does not take place. Thus Ap. vaïra remains vaïra in O. W. Rajasthani, and similarly a-vasa remains a-vasa. but ravamaü becomes naümaü. Mr. Divatia has made the mistake of overlooking the fact that nitial va cannot undergo samprasdrana and has given a series of etymologies in which he presupposes two conditions incompatible with one another : the existence of a stress or accent on a va or ya initial in the second member of a compound, and the weakening of this va or ya into u, i. It is obvious that so long as the va in kasa-vart retains the stress or emphasis which naturally falls on the initial syllable of every word, it can never undergo samprasarana. The form kasaü!f can only be possible, if at all, when the word kasa-vati has ceased to be considered as a compound and the va has lost its stress or has transferred it to another syllable. 14 As a matter of fact, this has happened in the case of kasavali, and we have of this word two parallel developments: (a) kasa-vart > kasdit, and (6) kasavati > kasaüļi 15 > kasoţi.
In all the examples of the change aya > é, ava > which have been given above, the ya and va are initial, a condition which is essential in 0. W. Rajasthani for the production of the narrow sound. But if we step out of the boundaries of the 0. W. Rajasthani into the domains of Apabhram sa and Prakrit, we find that here the change aya > e, ava >o is not confined to cases when ya and va are initial, but extends to other cases as well. Thus we find lena (< Skt. layana), lona (< Skt. lavaņa), ohi ( < Skt. avadhi), etc. I need not give more examples of this kind because the reader can see them for himself in Pischel's Grammatik der PrakritSprachen, $$ 153-4, but what I wish to remark here is that the change aya > é, ava >ó is not a peculiarity of the 0. W. Rajasthani, but rather the continuation of a process which was already in operation in the early Prakrit-Apabhramša stage.
But to revert to the è, ò sound. Certainly I need not waste time to emphasize the absurdity of Mr. Divatia's suggestion that this sound was probably matured under the influence of certain Arabic and Persian words. Mr. Divatia must be very little persuaded of the plausibility of his own derivation of this sound, if he finds it necessary to supplement it with such hypotheses. No, Arabic and Persian could not be responsible for the birth of è, o in the least, and to accuse them of sharing the paternity of these sounds is ridiculous, just as ridiculous as it would be to impeach English, because it possesses words, like "hat", and "hot", whereof the vowels are pronounced much like the è, o of modern Gujarati! But why, instead of going so far in search of foreign influences, why not lay greater stress on the analogy of the è, d of Marwari-Gujarati with the ai, au of Hindi? The latter sounds are identical with the former, except that they represent a slightly earlier stage, the very same
14 Cases of this kind are not unheard of. Cfr. taruara (staru-vara), and homarahomvara (< haya. vara).
15 I have found the form kasaufa unod once in Somasundarasuri's balduabodha to Dharmadasa's "Uvacsamála", 473. Ita derivative kaadet is used in Hindi (Bates, p. 111). Mr. Divatia cannot possibly claim this form kasaüfi as evidence in support of his theory that an a followed by an accented - gives ó, boonuso, as I have remarked above, the win karaütf cannot be accented.
