Models and empirical data for the production of referring expressions

Abstract

This paper introduces a special issue of Language, Cognition and Neuroscience dedicated to Production of Referring Expressions: Models and Empirical Data, focusing on models of reference production that make empirically testable predictions, as well as on empirical work that can inform the design of such models. In addition to introducing the volume, this paper also gives an overview of recent experimental and modelling work, focusing on two principal aspects of reference production, namely, choice of anaphoric referential expression and choice of semantic content for referential noun phrases. It also addresses the distinction between dialogue and non-dialogue settings, focussing especially on the impact of a dialogue setting on referential choice and the evidence for audience design in the choices speakers make.

Keywords:

• language production
• computational modelling
• referring expressions

Acknowledgements

The authors gratefully acknowledge the support of the Cognitive Science Society for the organisation of the Workshop on Production of Referring Expressions: Bridging the Gap between Cognitive and Computational Approaches to Reference, from which this special issue originated.

Funding

Emiel Krahmer and Albert Gatt thank The Netherlands Organisation for Scientific Research (NWO) for VICI grant Bridging the Gap between Computational Linguistics and Psycholinguistics: The Case of Referring Expressions (grant number 277-70-007).

Notes

1. In what follows, we will sometimes use the term cognitive modelling, with the understanding that the models under consideration have been explicitly developed with a view to model the results of experimental findings related to human speech and reference production and to characterise or explain the underlying cognitive processes.

2. A classic example – one among many application areas in NLG – is the generation of a report that summarises raw data (such as meteorological or clinical data) to facilitate human access to relevant information (e.g. Gatt et al., 2009; Goldberg et al., 1994; Portet et al., 2009; Reiter et al., 2005). Such systems incorporate REG algorithms to generate references to domain entities.

3. The extent to which Grice intended his maxims as ‘rules’ governing conversation is, of course, debatable. What follows should not be read as a suggestion that speakers somehow ‘violate’ Gricean rules; rather, we are interested here in the predictions that stem from a specific interpretation of the Maxim of Quantity, applied to simple referential tasks.

4. For example, once the IA selects colour in Figure 1(b), only distractors with a different colour remain to be excluded and this motivates the choice of the next property to be included in the description.

Additional information

Funding

Funding: Emiel Krahmer and Albert Gatt thank The Netherlands Organisation for Scientific Research (NWO) for VICI grant Bridging the Gap between Computational Linguistics and Psycholinguistics: The Case of Referring Expressions (grant number 277-70-007).