Biases and Understanding

This week's readings focus heavily on the dual purpose of speech as a linguistic and social agent. Through various studies that observe the effects of speakers' different dialects on phenomena like free association and semantic priming, the authors attempt to uncover how individuals respond to both stereotypical and rare forms of language. Professor Sumner and company explore the companion topics of frequency and social weight, utilizing speakers from varying dialects like British English, General American English, and New York City English to highlight linguistic trends.

On class in Thursday, we discussed an experiment conducted by Lev-Ari and Keysar in 2010. They conclude that comprehension difficulty leads to unreliability and untrustworthiness. Additionally, they state that the "voice characteristics of a talker do not influence comprehension" and that "attributes of reliability are made post-comprehension." These conclusions conflict with the general results presented in the works presented by Sumner, King, and Kataoka. In Sumner's piece that discusses the social weight of spoken words, she addresses the fact that social representations trigger almost immediately when an individual begins to speak. The social aspect of speech contains a host of different identifiers that help us categorize a speaker’s gender and age; these characteristics can be picked out almost instantaneously and have proven to generate great differences in comprehension. In fact, as Johnson pointed out in our first set of readings, test subjects have greater difficulty understanding women and men with atypical voices; likewise, Sumner discusses how aspects like age and emotion can affect the understanding of words when these characteristics can be deemed "unexpected". Unlike Lev-Ari and Keysar's study, this week's readings show that the voice characteristics of a speaker immediately influence our ability to understand the speaker's words. Attributes of reliability probably are not made post-comprehension but instead stem from a set of past experiences.

Continuing the topic of stereotypes and bias from last week’s readings, I found the comparisons of differently perceived dialects to be very applicable to the real world. The example involving rental advertisements and the clear preference for potential renters speaking Standard English provided an obvious example of the biases that can be formed simply from an individual’s speech. However, I found the idea of social weight to be an equally disturbing and subtler source of bias. Looking at Sumner and Kataoka’s discussion of the experiments exploring false memories, we see that British English and General American both lead to more accurate accounts than the regional dialect of New York City. Though British English and NYC English both are non-GA tongues, we dedicate more of our cognitive resources to British English because we see it as more prestigious; in other words, we give more social weight to the words spoken in this dialect because we unconsciously deem the speaker to be more important.

If this bias affects us so early in our comprehension process, how can we study any linguistic material fairly? It seems difficult to reconcile the fact that there will always be a most common stereotype (ie. Standard English for English as a whole) and that we believe we can make unprejudiced conclusions about linguistic topics.