Special Session: The role of informativity in language production and comprehension

The topic of this year's special session is Exploring the (un)expected: The role of informativity in language production and comprehension. This special session takes as its starting point the notion of informativity, broadly construed. It is designed to evaluate and extend our understanding of how informativity impacts language production and comprehension, and what this means for theories of language processing.

Invited speakers:

Jennifer Arnold, UNC
Ann Bradlow, Northwestern
Susanne Gahl, Berkeley
Florian Jaeger, Rochester
Emiel Krahmer, Tilburg
Roger Levy, UC San Diego

Language offers a powerful means to share information. However, not all parts of an utterance are equally informative. Some parts may be low in informativity because they are highly predictable or refer to already-mentioned information. Other parts may be more informative because they are unexpected or introduce new entities. Speakers have to make choices (largely unconsciously) about how to structure their utterances -- e.g., what word order and referring expressions to use. Comprehenders are faced with the task of extracting the intended structure and meaning from a signal with fluctuating levels of informativity.

The intuitive observation that the linguistic signal can fluctuate in its informativity raises fundamental questions about the extent and manner in which these differences influence (i) how speakers structure their utterances, (ii) how easily comprehenders can understand these utterances, and (iii) what the relation is between the two.

A growing body of research suggests that informativity has intriguing effects on many levels (phonological, lexical, syntactic, discourse) on both comprehension and production, leading us to ask how these effects can be captured by theories of sentence processing, whether they can be unified, and what they tell us about effects of (or lack thereof) communicative pressures on language processing and grammar.  In addition, further discussion is needed because different traditions have approached the notion of informativity in different ways that do not map directly onto each other:  E.g., information-theoretic research focuses largely on statistical measures of frequency and predictability, but pragmatic work focuses on information-structural notions such as given/new and topic/focus.

The special session is designed to address these issues and to offer an opportunity to evaluate our knowledge of this phenomenon. It aims to explore commonalities and divergences in different lines of research, as well the implications of this work for issues including communicative efficiency, speaker- vs. listener-oriented accounts, and gradience in language.

We are very grateful for funding from the National Science Foundation for the Special Session.