LINGUISTIC VARIATION IN BRAZILIAN PORTUGUESE: ASPECTS OF THE PHONOLOGY, SYNTAX, AND LANGUAGE HISTORY

Abstract

This dissertation presents a quantitative analysis of a group of linguistic variables in Brazilian Portuguese, and uses the results to attack several problems in the history of the language, and in the theory of linguistic variation. The variables in question are two cases where phonology and syntax interact. First, there is variation in nominal plural marking, constrained mainly by position of a word in the NP, so that the first word is almost always marked, but subsequent words are rarely marked. Interacting with this is a variable phonological rule deleting final S (producing, e.g. meno for menos)--which happens to be the main nominal plural marker. Second, there is variation in subject-verb number agreement, so that plural subjects often co-occur with singular verbs. This is constrained mainly by subject position (immediately pre-verbal subjects favor agreement more than distant or postponed subjects) and by morphological class of the verb (verbs with a more salient difference between singular and plural--e.g. fez-fizeram--are more likely to be plural-marked than those with less alient differences--e.g. come-comem). Interacting with this is a variable phonological rule denasalizing final vowels (ontem (--->) onte). (Nasalization of a final vowel is the most general third person verbal plural marker.) The analytical problems in each case are to separate the effects of the syntactic and phonological rules, and specify the constraints on each. The data analyzed (over 30,000 tokens for the four rules) were drawn from a recorded corpus of 140 hours of sociolinguistic interviews with 20 principle informants. The informants are all working-class illiterate residents of the city of Rio de Janeiro. Hence the variety of Brazilian Portuguese (BP) investigated is the popular speech, rather than the spoken or written standards of the middle and upper classes. The quantitative analysis uses the variable rule approach of Labov and Sankoff, and in particular the Sankoff VARBRUL 2 program for estimating constraint effects. An investigation of the social distribution of the variables showed small stylistic differences, regular (and sometimes powerful) sex differences with women more closely approximating the standard than men, and no age grading nor any evidence of change in progress. The results of the quantitative analysis are used in an investigation of the origins of the popular dialect. Is popular BP just an ordinary dialect of Portuguese, or might it have a creole history? Evidence from social history, such as the vast numbers of African slaves taken to Brazil and the economic impact and long duration of the institution of slavery in Brazil, make a creole history very plausible, indeed likely. It is suggested that some of the linguistic facts are also better explained by a creole or creole-like history than by a natural dialect history. For example, the pattern of indicating plurality by the first word in an NP is found in other creole varieties of Portuguese (and Spanish) and may be a calque on an African pattern, and the tendency towards higher rates of subject-verb agreement in verbs whose plurals are more morphologically salient could be the result of the kind of "borrowing" or language-learning processes that go on during decreolization. The case is not conclusive, but the weight of evidence tends to support a creole-like origin for popular BP. Also touched on are such issues as locating variation in a grammatical description, and functional constraints in phonology and syntax. In regard to the latter, it is suggested that functional constraints operate only when expressible in formal terms.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 206 citations exist upstream.