Frequency of Use, Form Retention, and Real Time Effects
Michael S. Gradoville · SSRN Electronic Journal · 2008
Modern Portuguese possesses two grammatical constructions that have the principal function of expressing future time reference, the Morphological Future (MF; example 1) and the Periphrastic Future (PF; example 2):(1) no inicio o criterio de avaliacao... PODERA ser bastante flexivel... 'In the beginning the criterion for evaluation... can (future) be quite flexible...'(2) voce nao VAI ASSISTIR? 'You aren't gonna attend?'Following Bybee's (1985) proposal that irregular patterns in a language remain irregular as a result of high frequency while low frequency leads to regularization, the purpose of this paper is to show that the frequency of use of the verbs used in MF is responsible for the retention of the form where it remains, in light of the highly productive, emerging PF. This study uses transcripts from approximately 240,000 words of the NURC-Rio de Janeiro corpus (Callou 1992; Callou & Lopes 1993, 1994, w/o date). All tokens of MF and PF were extracted from the corpus, yielding the 897 tokens used in this study with 12% being of MF and 88% being of PF. Each token was coded for numerous linguistic and extralinguistic variables. The tokens were subjected to a multivariate analysis using the GoldVarb X (Sankoff, Tagliamonte, & Smith 2005). In line with Zipf's (1965[1935]) observation that high frequency words tend to be shorter than low frequency words, frequency of use and the length of the verb stem interact heavily and as a result frequency of use and stem length had to be considered a single variable in the statistical analysis. GoldVarb X selected the combined frequency/length variable as significant with the greatest magnitude of effect. Shorter, more frequent verbs favor retention in MF, while longer, less frequent verbs disfavor such retention. This result supports Bybee's (1985) hypothesis that frequent items remain in a language in the face of language change while less frequent items tend to occur with the more productive patterns. Other factor groups selected as significant include the stem (ir)regularity in MF, type of multi-verb construction, and speaker gender. A subset of the interviews in the corpus of study were repeated twenty years later on the same speakers, allowing for the examination of language change in real time. Results from the comparison between the two time periods suggest that while the rate of use of MF diminishes through time, the overall distribution of the verbs used with MF remains relatively stable.