Computational Linguistics Colloquium

Tuesday, December 04, 11:15, Building 17, Seminar Room
NOTE UNUSUAL DAY, TIME, AND PLACE

A Proof Theoretical Account of Polarity Items and Monotonic Inference

Raffaella Bernardi
Utrecht Institute of Linguistics OTS
University of Utrecht

Natural language offers several phenomena which depend on semantic motivations for grammaticality. Polarity items (PIs) are a case in point: their distribution depends on the monotonicity properties of their licenser. The same properties justify (some) natural reasoning inferences as well. In the talk I will present a system able to:

1. encode this semantic information,
2. take advantage of them to account for PIs distribution, and
3. give a proof theoretical account of natural reasoning inferences.

The analysis presented results from the embedding of the natural logic proposed in [van Benthem 1986] and [Sanchez 1991] into a Lambek calculus extended with unary operators [Moortgat 1997]. The first part of the talk will be used to give the necessary background, introducing the concept of natural logic, the algorithm used by Sanchez and the extended Lambek calculus.

If you would like to meet with the speaker, please contact Alexander Koller.