Experiments in dialogue reconstruction
In an research article called Conversation, Slugoski and Hilton desrcibes the 'methodology of reconstruction' created by Clarke:
a multifaceted research instrument by Clarke (1975; 1983), who predicted and found that people would be able to reassemble randomized turns in an unfamiliar conversation with greater than chance accuracy. (...) Interestingly, sorters’ success at the task was not tied to the presence of syntactic cues to turn position in the floor holdings, as Clarke (1983, Experiment 2) found that dialogues made devoid of such cues were, if anything, better reconstructed that those with syntactic cues present.
What is interesting is now how this can be apply to people who are or are not familiar to each other:
Kent, Davis and Shapiro (1981) found that the shared personal knowledge, or common ground, between friends made their conversations less accurately reconstructed by an outsider than were dialogues produced between two strangers, who did not share common knowledge and hence had to be more explicit in their utterances.
And of course some people tried even weirder variables:
Another relevant finding on individual differences in the generation of reconstructable speech patterns is that utterances produced by schizophrenic individuals are less accurately sorted than those produced by psychiatrically normal subjects (Rutter, 1979), a difference presumably due to schizophrenics’ frequent breaches of the Gricean (1975) maxim of relation; that is, the utterances were not connected in any relevant or meaningful way to one another (see Rochester, Martin, & Thurstone, 1977). It turns out, however, that simply telling sorters that the dialogue they are about to reconstruct involves a schizophrenic participant results in significantly poorer reconstruction than among those not so informed (Slugoski & Turnbull, 1987)
Why do I blog this? this is interesting for our project about mutual inferences in collaboration.