On October 21, My advisor Dr. Annette D’Onofrio and I presented our Politicians Project research at NWAV 2018 at New York University. Below is our submitted abstract:
Studies of sociolinguistic style illustrate that individuals use packages of linguistic features to project personae interactionally. However, less attention has been paid to the ways that an individual’s overall variability or consistency in the use of composite features can construct a socially-meaningful image of that speaker. This study explores how two well-studied variables — ING and t-release — are recruited by presidential candidates to project personae during campaign rallies. We examine the variability a given candidate shows in deploying these features toward stylistic ends, both across and within rallies. Findings suggest that the degree of sociolinguistic flexibility a speaker demonstrates, even within a single speaking context, can itself take on social meaning that contributes to the holistic construction of a public self.
We analyze recordings of 32 campaign rallies for the major party nominees in the 2008 and 2012 U.S. presidential elections (Obama, McCain, and Romney), eight per speaker per election year. Geographical location of rallies was balanced by region. Every viable instance of variable word-final ING (velar v. alveolar) and word-final /t/ (released v. unreleased) was hand-coded auditorily. Variant choice for these features was assessed via logistic mixed-effects models, including linguistic predictors which have been previously shown to influence these variants, and extra-linguistic predictors of speaker and audience-related factors. Further, to examine intra-rally variation, rallies were coded by phrase for eight argument/topic categories (e.g. policy, “ordinary Americans”), which were also tested as predictors.
Differences emerged in the degree and nature of variance a given candidate exhibited in the use of these features. Obama’s production of velar ING showed wider standard deviation than other candidates across rallies in both campaign years (2008=11%; 2012=10%; Romney=4%; McCain=3%). While Obama showed similar variance to Romney in the use of released /t/ (Obama SD=4%; Romney=5%; McCain=0.8%), Obama is the only speaker who showed variability that was structured by audience-level and argument-level factors. Controlling for linguistic environment, Obama’s use of both velar ING (p<0.01) and released /t/ (p<0.05) was positively correlated with average amount of formal education in the rally’s locale. Further, Obama’s velar ING was significantly more likely to occur in phrases about policy or abstract concepts than those about “ordinary Americans” (p<0.05) or directives to the audience (p<0.001). While Romney showed linguistically-structured variance, no meaningful correspondence with extra-linguistic factors emerged. Finally, McCain showed negligible variance in his use of both features.
We argue that the degree of socially-meaningful linguistic variability a candidate exhibits can contribute to an ideological public image, and more broadly, serve as a resource in stylistic practice. Obama’s variability belies a layering of both “qualified” and “relatable” personae, drawing upon clusters of features to index these social meanings in particular contexts of identity construction. By contrast, McCain’s invariance is linkable to the “straight talk” persona he invokes, by which his inflexibility is potentially interpretable as, even iconic of, “consistency/authenticity.” Overall, we suggest that speakers not only construct personae through a bricolage of various socially meaningful features, but that the amount of variability a speaker exhibits is itself a dimension of sociolinguistic style.
We are looking forward to writing up our study beginning January 2019. Please feel free to reach out to either researcher for additional information!
Annette D’Onofrio: donofrio@northwestern.edu
Amelia Stecker: astecker@u.northwestern.edu