“How Social Meaning Impacts Linguistic Borrowing”
Dr. Betsy Sneller, Assistant Professor of Linguistics at Michigan State University
Friday, April 14, 9:00–10:30 a.m. EDT, in 127 Moore Building and virtually via Zoom
Naturalistic data – whether focused on rich ethnographic data or large-scale corpus data – have been the cornerstone of modern sociolinguistic research, providing the primary data for sociolinguistic theory. But more recent advances in computational and experimental approaches also offer additional insight into hypotheses drawn from naturalistic data.
In this talk, Betsy Sneller begins with an analysis of ethnographic research conducted in a neighborhood in South Philadelphia,demonstrating that white men who are in regular hostile contact with their African American neighbors have borrowed the African American English (AAE) feature of (TH)-fronting (pronouncing /θ/ as /f/, as in “bofe” for “both”). This result is in itself somewhat surprising, as white speakers using features of AAE typically also exhibit a positive affiliation towards Black folk or Black culture (e.g., Cutler 1997, Sweetland 2002, Fix 2010), while her participants exhibit overtly hostile attitudes toward their Black neighbors. She argues that, for these white speakers, (TH)-fronting is underspecified for ethnolect and is therefore available to be borrowed as an index of “toughness”.
In the second half of her talk, she presents the results of a laboratory experiment designed to test the central hypotheses drawn from this ethnographic work. Using an artificial language game, she demonstrates that borrowing across dialects propagates at a higher rate when a linguistic feature is a higher-order index than when it is a first-order index. In other words, a feature was borrowed more readily when it was both alienable (when it was associated with “tough” players rather than the species of player) and when it was socially relevant (in this case, when fighting was an option in game play). Variants that were either alienable or socially relevant – but not both – were treated like first-order variants, suggesting that the difference between first-order indexicality and higher-order indexicality is categorical rather than gradient.