& Language Processing Conference (SVLP)
The Inn at Virginia Tech and Skelton Conference Center , Blacksburg, Virginia
A conference on Sociolinguistic Variation and Language Processing (SVALP) was held at the Virginia Polytechnic Institute and State University (Virginia Tech), March 31 - April 2, 2016.
Sociolinguistic Variation and Language Processing
When we listen to a speaker, we're doing two things: we're comprehending the content of their message, and we're also extracting social information about the speaker (their age, their gender, their mood, their trustworthiness, etc.). There has been a tradition of assuming that these two processes are independent of each other, but this conference showcased work that suggests that in fact, the two processes are dependent on each other. That is, how we understand the content of a message is influenced by who we think is talking, and our social judgments about a speaker are influenced by what their message is and how they produce it.
An example of how social information affects language processing can be seen in the work of invited speaker Okim Kang (Applied Linguistics, Northern Arizona), who shows that if listeners believe they are listening to an Asian vs. White teacher, they not only report the former voice as being more accented, but they appear to learn less from the pre-recorded lecture. Critically, they hear the same voice in both conditions, so the fact that they perform worse with the Asian photo cannot be about the pronunciation of a speaker, but must be about their expectations based on their ethnolinguistic beliefs and/or their previous experiences.
An example of how linguistic factors influence social judgments can be seen in Erez Levon's (Linguistics, Queen Mary University of London) plenary talk. Looking at British English, he showed that listeners associate a fronted /s/ sound (pronouncing the /s/ sound fronter in your mouth) with gay males, and associate TH-fronting (saying "fanks" instead of "thanks") with a more likeable speaker. However, when the two features are both present in a sentence, the likeability of TH-fronting appears to be inhibited: combined with the "gay" /s/, the "fanks" no longer makes the speaker sound likeable.
Most studies like these are conducted with lots of listeners, and the significant results reflect the dominant pattern in the data. But this doesn't mean that every individual listener shows the same effects. A number of talks and posters at SVALP focused on the interaction between performance in linguistic tasks and various measures of individual differences in listeners. For example, Kang found that students who had taken a linguistics class weren't as affected by the Asian/White photo manipulation. Meredith Tamminga's (Linguistics, UPenn) findings suggest a relationship between how sensitive a listener is to the social meaning of variants (i.e., judging /s/ as gay, of "fanks" as friendly), and how much they participate in a sound change in progress in their dialect. Miranda Marques' (English, Virginia Tech) provides preliminary evidence that listeners who are part of the queer community are faster at processing speaker-known "they" (i.e., "Rebecca loves themself") than people outside the community.
The fact that there are differences across listeners raises the possibility that listeners can be trained to use or ignore social information in ways that aid in effective cross-dialectal communication. Simply put, the burden of good communication does not always need to fall on the speaker – it appears that listeners can work to better understand unfamiliar dialects or speakers. We hope future work in this area will explore ways to facilitate such efforts.
To explore the work presented at this conference in more detail, check out the Program page for pdfs of some of the talks and posters!