Discourse and Information Structure. Special Issue of the Journal of Logic, Language and Information, Vol. 12, No. 3, 2003. Ivana Kruijff-Korbayová (Saarland University) and Mark Steedman (University of Edinburgh) (eds.)
More details here
There were between 60 and 80 ESSLLI participants
attending the workshop every day!
Information structure (IS) is construed broadly here as comprising structural and semantic properties of utterances relating to the discourse status of their content, the actual and attributed attentional states of the discourse participants, and the participants' prior and changing attitudes (knowledge, beliefs, intentions, expectations, etc.). This broad view of IS is meant to subsume notions like focus, presupposition, given vs. new, theme vs. rheme and the various dichotomies such as topic vs. comment or focus, ground or background vs. focus etc.
While discourse structure (DS) is more difficult to define, there is at least agreement that coherent discourse (multi-sentence dialogue or monologic text) is more than a sequence of propositions, just as sentences are more than sequences of words. In discourse, both explicit and implicit devices signify links between sentences, between groups of sentences, and between elements within sentences, and in turn, carry additional elements of discourse semantics. We are thus taking DS broadly, to cover all aspects of the internal organisational structure of a discourse. DS thus subsumes notions such as segmentation, relations between segments (informational and intentional), anaphoric relations, modal subordination, discourse topic, thematic progression, etc.
Understanding IS in light of DS and vice versa is not only justified on theoretical grounds: Experience with applications such as translating telephony and interactive query-answering makes it painfully clear that a theory relating IS and DS is essential for accurate Natural Language Processing. Fortunately, formal accounts addressing these issues have started to emerge and some, to be embodied in computational models of discourse processing. Further development and adaptation into practical systems will require expertise from linguistics, logic and computation. ESSLLI provides an appropriate forum for fostering collaboration between interested researchers across the relevant areas.
The workshop aims to attract researchers in computational and theoretical linguistics and logic who are investigating IS-sensitive discourse processing and to facilitate contact, exchange of ideas and cooperation between them.
We hope that the participants of the workshop will transcend the difficulties caused by proliferating terminologies, and concentrate instead on investigating the interactions between IS and DS, with the goal of understanding how sentence-level semantic devices that make up IS symbiotically serve the needs of discourse cohesion and coherence. We seek contributions advancing beyond descriptive frameworks towards an explanatory account of how IS and DS, in whatever framework, interact to refer to and update a dynamically evolving representation of discourse context.
To achieve these goals, the workshop will consist of five sessions, each containing two to three contributed paper presentations, and discussion. In the introductory session, we will attempt to lay out the evolution of IS in various approaches and in the closing session, to summarize the implications of the contributions presented at the workshop. There will also be a key note speaker.
We aim to assemble a coherent set of papers, addressing closely related issues from various perspectives, such that the exchange of ideas among the workshop participants can be maximized. The submitted papers will be refereed by the PC.
Mon Aug 20: Intro + 2 contributed
Tue Aug 21: Invited talk + 1 contributed talk
Wed Aug 22: 3 contributed talks
Thu Aug 23: Invited talk + 1 contributed talk
Fri Aug 24: 2 contributed talks + Closing
Each contributed talk will have 25 min for presentation plus 10 min for discussion, the Introduction and Closing will have 30-35 min, and the invited talks will have 1 hour for presentation plus 10 minutes for discussion.
Please note the 15 minutes extension in comparison to the normal ESSLLI schedule. This is in order to give enough time to all the talks.
Note also that all workshop participants must register as particpants of ESSLLI, including the speakers/authors. The early registration fees apply to speakers/authors of accepted papers.
Livia Polanyi (Fuji-Xerox Palo Alto Laboratory)
The Relationship of Information Strcuture to Discourse Structure