Grammatical Vowel Length and the Classification of Quechua Dialects

Abstract

0. Introduction.1 Torero (1964:458-61) proposed the existence of phonemic vowel length in Proto-Quechua, based on the occurrence of long vowels in a number of modern Quechua dialects. Long vowels are widely found in Quechua I, one of the two main groups into which Torero divides the Quechua dialects.2 The second main group in Torero's classification, Quechua II, is further subdivided into Quechua IIA, IIB, and IIC. Long vowels occur in several Quechua IIA dialects (Pacaraos, Lincha, Laraos), or indications of their existence at an earlier stage can be found (Ferreinafe). In the remaining Quechua II dialects, including all dialects of the subgroups IIB and IIC, vowellength distinctions are lacking (Torero 1974:22). The question of why no trace of long vowels can be found in such a large number of Quechua dialects, while they occur frequently in others, is still to be answered. Most, but not all, of the long vowels in Quechua I can easily be interpreted as the result of innovations. The remaining cases support Torero's view that vowel length may have been phonemic in Proto-Quechua. In this article I intend to examine two instances of vowel length used for grammatical purposes. I shall argue that both cases are due to innovations, a conclusion which would further reduce the number of occurrences of long vowels to be reconstructed for Proto-Quechua. The cases of vowel length discussed are suffix-final and root-final a: in verb forms, and the indication of the first-person subject or possessor with verbs and