Dealing with speaking rate as a factor group in variation studies

Jeff Tennant

Fri. 2:00-4:05 A

Speaking rate has often been hypothesized to be a contributing factor in some cases of linguistic variation, particularly when the dependent variable involves retention or deletion of a phonological segment (Harris 1969; Dressler 1972; Zwicky 1972; Rudes 1977; Santerre et al. 1977). However, the relation between this prosodic variable (as well as other temporal variables of speech) and pronunciation forms appears to have been subjected to empirical investigation in only a handful of studies (Leon 1987; Shockey 1987, Omozuwa 1989; Trepanier & Archambault 1993; Tennant 1993, 1995). These studies tend to confirm the hypothesis that a faster rate of speech leads to a higher rate of surface reductions.

The investigation of speaking rate effects in a variationist framework must take into account a number of methodological problems. How do we isolate speech samples which accurately reflect the speaking rate corresponding to each occurrence of the linguistic variable? What type of normalization procedures are required to allow interspeaker comparability? How is speaking rate dealt with as a factor group in variable rule analysis?

In this paper, a number of solutions to these problems will be discussed. The discussion will be based on data obtained from an analysis of putative fast-speech rules in Canadian French, using a corpus of 36 adolescent speakers from Northern Ontario.