Abstract
Accessibility theory associates more complex referring expressions with less accessible referents. Felicitous referring expressions should reflect accessibility from the addressee's perspective, which may be difficult for speakers to assess incrementally. If mechanisms shared by perception and production help interlocutors align internal representations, then dyads with different roles and different things to say should profit less from alignment. We examined introductory mentions of on-screen shapes within a joint task for effects of access to the addressee's attention, of players’ actions and of speakers’ roles. Only speakers’ actions affected the form of referring expression and only different role dyads made egocentric use of actions hidden from listeners. Analysis of players’ gaze around referring expressions confirmed this pattern; only same role dyads coordinated attention as the accessibility theory predicts. The results are discussed within a model distributing collaborative effort under the constraints of joint tasks.
Acknowledgements
The authors are grateful to Tim Taylor, Joseph Eddy, Jonathan Kilgour, Craig Nicol and Jean Carletta for assistance with programming and to Jan Peter de Ruiter and Mark Steedman for helpful discussions. The faults remaining are entirely ours. Dr Foster is now at the Department of Mathematical and Computing Sciences, Heriot Watt University, Scotland. Dr Arai is now at the Department of Language and Computing Sciences, University of Tokyo, Japan.
Funding
This work was supported by the EU FP6 IST Cognitive Systems Program under Integrated Project ‘JAST: Joint Action for Science and Technology’ [FP6-003747-IP]. The University of Edinburgh and research sponsors are authorised to reproduce and distribute online copies for their purposes notwithstanding any copyright annotation hereon.