Back

The computation of presuppositions and their justification in discourse

Hans Kamp and Peter Krause
Section: Language and Logic
Level: Introductory

Description

The account of presupposition to be presented in this course involves two stages, just as in the theory of van der Sandt (1992).

In the first stage, as part of the computation of a semantic representation, all presuppositions are explicitly represented. During the second stage, the presuppositions are justified through binding, accommodation, or a combination of both.

However, unlike van der Sandt, and most of the linguistic literature on presupposition of the past thirty years, we will concentrate not so much on the problem of presupposition projection, but on the problem of the simultaneous solution of multiple presuppositional contraints.

This course shares some important features with the theory of van der Sandt (1992), in that two stages are posited for the identification and justification of presuppositions. During the first stage, the presuppositions of a sentence/utterance are represented, as part of the computation of an explicit semantic representation (i.e. a Discourse Representation Structure (DRS), or an (Underspecified) DRS) of the sentence.

During the second stage, the represented presuppositions are justified Justification can take the form of binding, of accommodation, or of a combination of both.

The novel feature of the course is that, unlike van der Sandt and most of the linguistic literature on presupposition of the past thirty years, our focus is on not the problem of presupposition projection, but, instead, on the problem of simultaneous solution of multiple presuppositional constraints, a problem which arises as much for presuppositions generated in the highest possible position, as for presuppositions in subordinate positions, (which give rise to projection questions).

This problem is theoretically interesting, and is of central importance for computational linguistics, since even perfectly ordinary sentences of moderate length often generate clusters consisting of several presuppositions. These clusters typically put heavy constraints on the possible readings of the sentence, and delimit the way its interpretation can be fitted into the discourse of which it is a part.

As an example of the "multiple presupposition problem" consider the sentence: "Again, the rector's wife brought another escort, too." This sentence generates presuppositions triggered by: (i)-(ii) the words "again" and "too"; (iii) the definite NP "the rector", (iv) the definite NP "the rector's wife", (v) the word "another", (vi)-(vii) the implicit goal argument, and the prestate presupposition associated with "brought". Sentences like these present non-trivial problems both at the level of presupposition computation (especially where the scope relations between different triggers are concerned), and at the level of presupposition justification.

Literature

David Beaver.  Presupposition.  In Johan van Benthem and Alice ter
    Meulen (eds.):  Handbook of Logic and Language.  Elsevier, 1996.
Bart Geurts.  Pronouns and Presuppositions.  Elsevier, 2000.
Hans Kamp and Uwe Reyle.  From Discourse to Logic.  Kluwer, 1993.
Rob van der Sandt.  Presupposition as Anaphora Resolution.  Journal of
  Semantics, 1992.

Prerequisites

Some basic knowledge of Discourse Representation Theory.

Materials

Lecturers

Hans Kamp & Peter Krause
Institut für Maschinelle Sprachverarbeitung
University of Stuttgart
Azenbergstr. 12, D-70174 Stuttgart, Germany.
{hans,peter}@ims.uni-stuttgart.de.
Fax: (+49)-711-121-1366.