________________
42
M. A. MEHENDALE
t/t, in the above instance, occurring in the same environment as the set t/d. However, it cannot be assumed that such a set in a language would always occur, and if it does not, then there is no set with which the set showing the alternation t/d could be compared. It will remain in complementary distribution with both t/t and d/d which do not occur in the environment of t/d.
Chafe's feeling that Hoenigswald has left vague this point of the procedure which helps one to make the choice of the reconstructed phoneme is not correct. Hoenigswald has unfortunately made one statement regarding the reconstruction of Gmc. t, on the basis of the alternation between t and 0, which has led to this impression. It appears that Hoenigswald has been influenced by his knowledge of comparative Indo-European when he lebels his reconstruction as *t. Hoenigswald probably did not feel it necessary to test his statement by applying the procedure of internal reconstruction outlined by him because, in the light of our present knowledge, the reconstruction as such (*t) is the correct one. But if he had done so he would have found that the evidence leads to *0 (and not to *t). That this * is after all to be replaced by *t is shown by comparative evidence and cannot be obtained by internal one. This is not the fault of the procedure but inerely constitutes a limitation of internal reconstruction But for the above statement, Hoenigswald has clearly stated the procedure regarding the choice of the reconstructed phoneme: “Compulsory alternation between restricted /x/ and free /y/ in a paradigm indicates a previous conditioned sound change from /x/ to /y/ in the position from which it is
3 Word final t, as in [unt] (und), would help Hoenigswald but not Chafe
as it does not yield any paradigmatic alternation. 4 See A. M. Ghatage, Indian Linguistics 21.88 (1960) and M. A. Mehendale,
Limitations of the Method of Internal Reconstruction, Indian Linguistics 21.101-103, 1960. Hoenigswald's defence against Chafe's objection (Phonetic Similarity in Internal Reconstruction, Language, 36.191-92, (1960) is not a very happy one, because it forces him to bring in a statement like "Furthermore, we know-whether from internal reconstruction or in other ways-that a Gothic t may also have another antecedent /*z/, (e.g. in the word which corresponds to E nest)." (p. 191) How one can bring forward this as an argument when one has set oneself the task of internal reconstruction is not clear. His footnote 6"It would have been better to punctuate: one orignal morpheme, "-to"!" is also not very helpful. It is intended to suggest that the reconstruction pertains to "one morpheme' and that the label' to'attached to it is of no significance. All this has come in because one hesitates to admit the limitation of the internal reconstruction.
Madhu Vidya/348
Jain Education International
For Private & Personal Use Only
www.jainelibrary.org
