Reanalyzing Variable Agreement with tu Using an Online Megacorpus of Brazilian Portuguese

Abstract

We reanalyze the phenomenon of verbal (non)agreement with the 2SG tu in a megacorpus of Brazilian Portuguese compiled from the web. Unlike previous research, which has analyzed sociolinguistic interview data and regional differences, we examine these data with a focus on the internal linguistic factors that constrain the variability. Our analysis of 4860 tokens of tu + verb reveals that non-agreement with the 3SG verb form is by far the most common pattern, 2SG agreement being relatively infrequent. Individual verb lexemes show highly distinct rates of (non)agreement. In addition, the specific tense/aspect/mood forms and main/auxiliary status are likewise significant factors affecting the variation. We conclude that future studies of this phenomenon should not ignore these internal linguistic factors. We situate our study within a group of other recent studies in Romance linguistics, which have found that individual verbal and constructional patterns can have diverse effects on morphosyntactic variation.