Madhuvidya

Author(s): S D Laddu, T N Dharmadhikari, Madhvi Kolhatkar, Pratibha Pingle
Publisher: L D Indology Ahmedabad

Page 376

INTERNAL AND COMPARATIVE RECONSTRUCTION
45
yield facts which, at a time, are removed only one or two stages from the attested one, it may be regarded to be in a better position to give phonetic details of the reconstructed stage. Therefore it seems reasonable to uphold Chafe's contention that internal reconstruction should precede the comparative method. This in fact will help us to eliminate some of the problems of comparative reconstruction. For example if on the basis of the alternation t~0 in Germanic, the third person sg. present is first internally reconstructed as 0,*is would replace ist'is'. In that case while attempting comparative method, Hoenigswald's set 1 (cf. Principal Step, Lg. 26.358) t/t, extracted from Skt. asti : Gmc ist, will disappear. Instead we will have t/0 which will be identical with his sét 3. This will lead to economy, as we have been able to reduce the number of sets and also eliminate the difficulty encountered by Hoenigswald about grouping the set 1 with 2 and 3 or with 5 (Language, 26.360).
If it is agreed to attempt internal reconstruction first before proceeding to comparative niethod, then it may be suggested to take one more step in between and apply the information made available by internal reconstruction to some specific items in the data. For instance, we have in Sanskrit an alternation t~t in the past passive participle morpheme-ta, e.g gata : tusta. Internally t is reconstructed as *t in this morpheme since in Skt. t never occurs after s. We now make use of this information and reconstruct t as *t after s also in other items where no alternation between t and t is available, e. g. Skt. asta < *aşta 'eight'. This is done because while comparing Skt. with Avesta it helps us to get rid of a set t/t which we would have extracted from cognates Skt. asta: Av. ašta. To take another example, it has already been shown that word-final -t can be reconstructed as *-d in German /Bunt/ and Sanskrit /sarat/. We may now proceed further from this step and also reconstruct -t as *-d in some other items where no paradigms are avainable, e.g. German /unt/, and Sanskrit prefix /ut/, indeclinables like /işat/ and /yugapat/, and the abl. ending /-āt/ (even when the latter had not shown final /-d/ in external sandhi). It is, of course, not advisable to do this reconstruction if none of the related languages shows final /-d/ in the cognates.
A typical problem in comparative reconstruction has been posed by W. S. Allen (TPS 1953, p. 82 ff.) and its solution suggested by A. M. Ghatage (Indian Linguistics 22. 82-85, 1961 and Historical Linguistics and Indo-Aryan Languages, Bombay 1962, pp. 65 ff.). Allen has raised the problem about
7
Chafe observes "In general ... ... ... internally reconstructed features tend to be more recent than those reconstructed by the comparative method" (Lg. 35.495)
Madhu Vidya/351