CLS Speaker Series - Grant Berry (Penn State University) Shrinking Down Sound Change: Dual Mechanisms of Cognitive Control and Phonological Adaptation
Shrinking Down Sound Change: Dual Mechanisms of Cognitive Control and Phonological Adaptation
Language users readily adapt to linguistic variation, both in the short term—after exposure to another’s speech—and in the long term—as linguistic structures change in the surrounding community. How is an individual’s interaction with linguistic variation influenced by the general cognitive strategies they use to resolve competition inherent to their surrounding environment? Under a dual mechanisms framework (e.g., Braver et al., 2007; Braver, 2012), cognitive control—the mechanisms engaged to resolve conflict and regulate expectations—is divided into two interrelated strategies: proactive and reactive control. Less habitual engagement of these strategies is hypothesized to facilitate the integration of phonological variation (Berry, 2016; see also Darcy et al., 2016; Lev-Ari and Peperkamp, 2013, 2014; Lev-Ari and Keysar, 2014). In this talk, I discuss two studies examining the relationship between dual mechanisms of cognitive control and phonological adaptation. Study 1 investigates how proactive and reactive control modulate participants’ adaptation to distributional changes in their speech input, using a controlled laboratory setting to simulate sound change (lowering of /ɪ/ to /ɛ/ preceding voiceless coronal codas). In this paradigm, participants alternated between listening to a model talker produce 80 controlled mono- and bi-syllabic words in isolation and blocks where they spoke those words aloud themselves (cf. Maye et al., 2008; Kittredge and Dell, 2016). Gradually, a sound change was embedded in the exposure blocks, such that the relative frequency of a lowered variant in a pre-specified phonetic context increased by 25% in each block. The degree to which this change was integrated was calculated by measuring participants’ log-mean normalized F1 from the production block for words in the pre-specified environment as a function of block number and the variant heard in the preceding listening block, and these were correlated to composite indices of proactive and reactive control. Results of linear mixed effects modeling (cf. Barr et al., 2013; Bates et al., 2015) indicate that individuals with weaker reactive control integrated the simulated sound change into their own production gradually as the relative proportion of the novel, lowered variant increased in the exposure stimuli. This finding suggests that cognitive strategies engaged to resolve conflict correlate to one’s tendency to resolve distributional changes in one’s linguistic input. Study 2 explores some long-term implications of these findings in community-based research examining individual participation in three socially-stratified sound changes-in-progress in Philadelphia. An example is EY-raising (when ‘wait’ sounds more like ‘wheat’), which is rapidly advancing in that community (cf. Labov et al., 2013). The hypothesis is that—modulo the influence of community-level social valuation of a given variable—findings from Study 1 will also correlate to findings from Study 2, i.e., that those who are most advanced in the target sound changes-in-progress are also those who demonstrate less habitual engagement of cognitive control. Connections between laboratory and field approaches to this question are then discussed, which together motivate the inclusion of cognitive control strategies in models of language variation and change.