The communicative nature of language has been explored for centuries and serves as a solid base for concepts in modern linguistic theory, particularly regarding the shape that linguistic processes take in order to maximize the effective transmission of a message from one person to another. The development of a mathematical theory of communication in the 1940s (e.g., Shannon & Weaver 1949) spurred a number of linguists to consider analyzing linguistic phenomena in terms of mathematically explicit concepts of this theory. While information theoretic concepts are well established in the area of computational linguistics, the probabilistic nature of information-theoretic approaches was essentially abandoned in mainstream linguistics. Recently, however, the field of linguistics has witnessed a re-emergence of interest in information theory as researchers explore ways in which tools from information theory can help advance our understanding of linguistic systems.
The objective of this proposal is to provide funding for a symposium held during the Linguistic Society of America's biennial Summer Institute. The symposium will allow scholars interested in the application of information theory to language to identify phenomena in linguistic research that may fruitfully be investigated and modeled using information theory, and to share and appraise methods by which information theory can be applied to such phenomena. There will also be a tutorial for scholars not familiar with information theory. The event is intended as a springboard for the development of a research community within linguistics investigating the many roles that information theory plays as an organizing principle of linguistic phenomena. An edited, peer-reviewed collection of the papers will be produced by the organizers of the symposium.
" was held on July 16 and 17, 2011, at the University of Colorado-Boulder, as part of the Linguistic Society of America's Summer Institute. A central goal of linguistic research is the development of a rigorous theory able to predict the wide range of observations associated with language by means of a simple set of theoretical constructs. Shannon's mathematical theory of communication (also known as Information Theory) developed in the 1940s (see Shannon & Weaver 1949) was considered by some linguists at the time as having the potential to move the field closer toward achieving this goal (e.g. Cherry, Halle & Jakobson 1953; Hockett 1955). However, with the advent of Generative Linguistics, information- theoretic approaches to language were criticized (e.g., Chomsky 1956) as being incapable of expressing linguistic competence. Information theory thus took a backseat, as Generative approaches to language became mainstream. Intervening years have seen advances in computation, probability theory, and corpus building, leading to renewed interest in information theory and a reconsideration of its usefulness in informing linguistic theory. Concurrent with these developments is research in linguistics suggesting that the concepts of information, entropy, and information content can shed light on a number of longstanding issues. These range from predicting the variation in targets and outcomes of phonological, morphological and syntactic processes, to explaining the cognitive bases for these processes, to helping analysts objectively choose the best model for their data. Thus, the goal of the grant was to hold a symposium where scholars interested in the application of Information Theory to language would be able to share knowledge, tools, insights, and specific research findings. It started with John Goldsmith (U. of Chicago), one of the leaders of this area of research, giving a tutorial on the basics of the approach for participants who were not already familiar with it. There were then eight invited talks, twelve submitted posters, and a roundtable discussion, which allowed for a wide range of uses of Information Theory to be shared and evaluated. The workshop was successful in familiarizing participants with well-established mathematical tools that can be used to analyze language. The workshop also provided a venue for researchers to share their findings and to get to know others with similar research interests. Feedback during and after the symposium was quite positive, with considerable interest expressed in holding workshops building on this event every other year. John Goldsmith and Jason Riggle volunteered to organize the second such workshop to be held in 2013.
