The systematicity of style : investigating the full range of variation in everyday speech
- Language variationists of all stripes recognize that linguistic features vary not only between individual speakers, but also within them. Many view this intra-speaker variation to be stylistic in nature. In particular, third-wave theorists (Eckert 2012) see it as indicative of how speakers use variables to enact various identities and convey different social meanings. The conventional sociolinguistic interview has proven useful for illuminating community-wide patterns, but falls short for eliciting a wide stylistic range. Accordingly, this dissertation investigates the full range of intra-speaker stylistic production via data that are truly multi-dimensional, representative, and quantitatively rich: self-recordings of everyday speech. The self-recordings come from "Pearl, " a young female lifelong Californian and California Vowel Shift (CVS) participant. With a high-quality audio recorder and microphone affixed under her clothing, Pearl recorded every waking minute over a two-week period. Because of the unobtrusive set-up, these self-recordings comprise a corpus largely untouched by the observer effect. Specifically, this dissertation analyzes two full days of Pearl's self-recorded speech, the days differing from each other in terms of age of interlocutor (one day Pearl speaks with adults only, the other day with adults and small children). I first descriptively compare these two self-recorded days with each other and with a sociolinguistic interview conducted before self-recording commenced. I demonstrate that the two self-recorded days are fundamentally different from each other (due to the effect of child-directed speech on day two) and the sociolinguistic interview across a variety of linguistic measures. I then analyze the full range of variation of three vowel classes: trap, toe (post-coronal goat), and goat, produced by Pearl in these three data samples. All of these vowels are implicated in the CVS and known to carry salient regional and stylistic meanings (D'Onofrio 2016; Podesva 2011). As a CVS speaker, Pearl generally retracts trap and fronts toe/goat, but not all of her vowel articulations pattern so uniformly. Normal mixtures model-based cluster analysis conducted for each vowel class revealed significant and consistent clustering patterns in the vowel space, resulting in three main cluster types: reduced, baseline, and stylized. Reduced clusters contain phonologically reduced articulations of the vowels. Baseline clusters comprise the majority of tokens for each vowel class, which hover tightly around a consistent articulatory target. Finally, smaller stylized clusters exhibit extreme formant measurements and longer durations. I argue that stylized clusters are such not only because of their extreme F1/F2/duration values but also because they correlate significantly with factors known to be perceptually and stylistically salient, including prosodic prominence (Mo 2008; Cole et al. 2010) and non-modal voice quality (e.g., Podesva 2007), among other measures. Baseline clusters, on the other hand, are stylistically unmarked, having no significant perceptually/stylistically salient characteristics. These clustering patterns suggest that Pearl systematically uses acoustically distinct articulations of vowels to both present a consistent baseline style and to deviate from it for stylistic effect. These findings provide some support for the vernacular principle (Labov 1972), in that the baseline style resembles something of the "style which is most regular in its structure" (p. 112), but they also demonstrate that stylization can be just as quantitatively systematic as the baseline, contra Labov's contention that intra-speaker variation is largely unpredictable. Finally, I used generalized additive models to examine how the stylized tokens in Pearl's repertoire unfold over the course of a day—to see if there are moments in which she is significantly likely to use them. I qualitatively examine these moments to see how and with whom they occur. In doing so, I show how stylization unfolds in real time interaction and gain insight into the social meanings these stylized vowels convey.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Van Hofwegen, Janneke
|Stanford University, Department of Linguistics.
|Statement of responsibility
|Janneke Van Hofwegen.
|Submitted to the Department of Linguistics.
|Thesis (Ph.D.)--Stanford University, 2017.
- © 2017 by Janneke Marie Van Hofwegen
Also listed in
Loading usage metrics...