Comparative and internal reconstruction of the Tupi-Guarani language family

Charles Schleicher

Ph.D. Dissertation, 1998

Abstract

This study examines the Tupi-Guarani (TG) language family a language family comprising languages spoken in Brazil, Paraguay, Argentina, Bolivia, and French Guiana. This study applies the comparative method of reconstruction to attested TG languages to arrive at a reconstruction of Proto-Tupi-Guarani (PTG). The protolanguage, in turn, is subjected to internal reconstruction to arrive at a Pre-TG phonology and morphology. The proposed reconstruction of PTG is significant in that it differs in a number of significant ways from earlier attempts at reconstruction by Jensen (1984). Points of difference are as follows: (1) The PTG phonemic inventory reconstructed here is analyzed as having a heretofore unrecognized phonological categorization, called fortis and lenis (which can be identified in other Amarindian languages). Fortes are voiceless stops and lenes are voiced continuants homorganic with the fortes. This captures the fact that, in PTG and many TG languages, voicing and continuance are redundancy features. This leads to an internal reconstruction of PTG. (2) PTG is seen here to have three series of fortes: plain, labialized, and palatalized. This also contributes to the internal reconstruction. (3) PTG is here found to have had three autosegmental phonemes, two at the morpheme level and one at the level of the syllable. Additionally, the phonological reconstruction of Pre-TG is shown to contribute positively to the ability to identify cognates between TG and non-TG languages in the Tupian stock. These discoveries, and the resultant reconstruction of Pre-TG, lead in turn to important conclusions about the morphology of PTG: (1) Agentive, circumstantial and abstract nominalizers are found to be related morphemes. (2) Form classes of nouns and verbs are found to be originally phonologically conditioned. (3) A more accurate reconstruction of the pronominal system, both formally and functionally, is possible. From these morphological conclusions, the following syntactic conclusions are also made: (1) PTG was not ergative, as suggested by Jensen (1990). (2) Subordination was expressed by nominalized verb phrases, rather than by separate clauses. (3) The distinction between modifiers and nouns was essentially non-existent in PTG and the distinction between nouns and verbs was minimal and may in fact have been more pragmatic than syntactic. The study concludes with a review of the question of internal classification within PTG and a proposal of a PTG homeland.