| |
Abstract:
The Gricean maxim of quantity states that speakers should be
only as informative as necessary. According to Referential Theory
(Altmann & Steedman, 1988), listeners expect definite NPs used
with modifiers to presuppose a contrast within a contextually
supplied set, and use such information to resolve ambiguity during
sentence interpretation.
However, listeners must take into account not only the visual
context of the scene, but also the discourse history of the
conversation. Off-line studies have shown that speakers balance the
Gricean maxim of quantity against a tendency to use the same form
of reference repeatedly, even if it is over-informative. For
example, something referred to as "the tall cup" earlier in the
conversation may continue to be called "the tall cup", even when a
second cup is no longer present (Brennan & Clark, 1996).
Using a head mounted eye-tracker, we measure RT latencies to
target objects. Subjects were instructed to touch various objects
with a pointer
(e.g., Touch the tall cup).
Replicating previous studies (Sedivy et al., submitted), we find
that listeners are guided by such Gricean expectations when
interpreting sentences containing adjectival modifiers: They more
rapidly identify the referent for the tall cup in an environment
where there are two cups contrasting in size, rather than a single
cup. In addition, we find that listeners are faster to look at the
correct target (e.g., the tall cup) when they have previously heard
an identical description of the target
(the tall cup)
rather than a different description
(the styrofoam cup),
or no previous description.
This result sheds light on what qualifies as a felicitous use of
a modifying adjective in on-line processing. Listeners evidently
have several criteria for evaluating adjectival use, with one being
how the item was labeled in the course of conversation. Referential
shifts, of the type used in the different name condition, actually
exert a processing cost in situations where there is no contextual
support for the shift. Overall, our results suggest that the
alternative items present in the visual scene and the specific form
of reference to the target items in the discourse interact to
determine how on-line processing decisions proceed.
References
Altmann, G., & Steedman, M. (1988). Interaction with context
during human sentence processing.
Cognition,
30, 191-238.
Brennan, S. E., & Clark, H. H. (1996). Conceptual pacts and
lexical choice in conversation.
Journal of Experimental Psychology: Learning, Memory and
Cognition,
22, 482-1493.
Sedivy, J., Tanenhaus, M., Chambers, C., & Carlson, G.
(submitted). Achieving incremental semantic interpretation through
contextual representation.
|
