Employ a wider range of theories and methods

Collaboration and cross-pollination between research subfields facilitate the employment of new tools and concepts in panel studies. Where the amount and type of data support it, analysts should follow the lead of some contributors to this volume in using and adapting statistical techniques such as GAMMs (Baayen et al. 2021; Beaman and Tomaschek 2021), and mixed effects regression (Buchstaller et al. 2021; Harrington and Reubold 2021). Methodological ingenuity pays dividends: Beaman and Tomaschek (2021) develop a measure of diphthong trajectory similarity, called TEDS, that can benefit future researchers who need to track nonlinear acoustic movement over time. But quantitative tools do not need to be high-level statistical techniques. Indices, long a staple of the language change researcher’s toolkit, also have helpful roles to play in panel studies. They can aggregate across complex sets of features or characteristics that are too highly correlated, or too numerically sparse, to be analyzed separately in a meaningful way. Van Hofwegen and Wolfram (2018), for example, argued that an overall Dialect Density Measure was the best way to handle their longitudinal child language data, for which they had only small amounts of speech per child in some time slices. In the present volume, Buchstaller et al. (2021) establish an overall instability index to determine which of their panel speakers exhibits the most lifespan linguistic change across multiple heuristics. Sundgren and colleagues (2021), and Beaman and Tomaschek (2021) employ, to good effect, indices of local orientation and community integration as a way to capture the constellation of interleaving environmental, socioeconomic and attitudinal factors that motivate speakers to retain or abandon vernacular language over their lifetimes.

Rickford (2021) points out that there are further attitudinal variables, such as mood and personality, that are routinely considered by social psychologists but rarely included in sociolinguistic research and which could be of value in panel studies. Even so, untangling these variables from the personal histories of panel speakers will be challenging. Is Foxy Boston, for example, a “stylistic chameleon” because of her personality or because of some crucial aspects of her life story, including those that may be unknown to the analyst?

Nonetheless, it could be worth the struggle to integrate mood, memory, personality and other socio-cognitive factors into panel research, so long as there are enough data for this to be interpretable. Bowie (2021) thoughtfully notes that he excluded Mormon preachers who had experienced brain injuries, because his sample size did not support including cognitive decline as a predictor of lifespan linguistic change. Certainly, as sociolinguists integrate an aging global population into studies of language change (Pichler et al. 2018), the imperative to grapple with age-associated cognitive shifts will increase. Harrington and Reubold (2021) address the issue head- on in their follow-up analysis of Queen Elizabeth II. They suggest that the Queen’s post-1990s return to more conservative vowel realizations cannot be attributed to fewer interactions with progressive (younger, less upper-class) speakers. Rather, the Queen may be exhibiting a decline in her episodic memory, brain plasticity, and/or hearing acuity, any and all of which could have affected her ability to perceive, process, and produce more contemporary phonetic variants. Harrington and Reubold (2021) further harness exemplar theory to propose that episodic memory may be especially relevant, since its deterioration could lead the Queen to rely upon her more deeply entrenched phonetic exemplars from earlier in her life. Yet, as they also discuss, the predictions made by exemplar theory for lifespan linguistic change are not always borne out, and it will require a great deal of further research to untangle contradictory results. For now, at least, there is intuitive appeal in using lexical frequency as a proxy for episodic memory function (see also Baayen et al. 2021; Beaman and Tomaschek 2021), as well as in more abstractly theorizing about robust- versus less-robust cognitive representations (e.g., Standing and Petre’s use of Construction Grammar 2021).

< Prev   CONTENTS   Source   Next >