Intercultural Address Differences, and Egocentric Listeners: 

An Anchoring and Adjustment Heuristic Moderates Computer-Mediated Dialogues, and an Availability Heuristic Moderates Their Overhearers

Roxanne Benoit Raine & Marlene Burke
Institute of Cognitive Science, University of Louisiana at Lafayette

Horton and Keysar's monitoring and adjustment hypothesis (MAH) is based on two theories. First is an availability heuristic: one's belief in something's probability is related to how available it is to him/herself. Second is an anchoring and adjustment heuristic: people anchor to concepts and adjust as a repair. These two hypotheses are applied to language use in the MAH: Speakers assume others' perspectives are overly similar to their own (availability), and they anchor models of their listeners in their own perspectives, repairing only when necessary (anchoring and adjustment). 

We test the monitoring and adjustment hypothesis in novel ways. First, we examine whether interlocutors hold to egocentric biases in conversations conducted in a computer-mediated environment (CMC), as opposed to a verbal framework. Second, we examine the endurance of overhearer effects for organic lab conversations. Although a few other studies have simulated text overhearer experiments of the MAH (eg., Keysar, 1994; Gibbs, O'Brien & Doolittle, 2003; Keysar & Henly, 2002), those studies were comprised of sentences composed by the experimenters, as opposed to dialogues between people. Furthermore, this is the first study to examine the components of the MAH (availability heuristic and anchoring heuristic) separately. 

Local participants were told that their interlocutors were locals or from a different culture (determined to be unfamiliar from pretest surveys). Previous studies found speakers to consider cultural backgrounds of addressees (e.g., Isaacs & Clark's speakers described NYC landmarks differently to native/non-native New Yorkers, 1987); however, this is the first study wherein the cultural difference is not pertinent to the topic of conversation. Participants spoke with each other over the computer to complete a task together. When the transcripts were later given to overhearers, there was a significant effect in attribution errors for dialogues but not for monologues. Overhearers tended to err based on an availability heuristic. This result supports the monitoring and adjustment hypothesis of language use in respect to overhearers and CMC. Measures of word counts, turns taken, and accuracy did not reach significance, but their trends paralleled those found in the earlier study examining monologues (where those measures were significant). It is believed that the dialogue measurements were not as robust as the monologues' because interlocutor feedback normalized the results, which is in accord with the anchoring and adjustment heuristic. These effects' endurance in CMC overhearers provides compelling support for the MAH, because effects should be weaker in these situations.