Formal Semantics Meets Experimental Results

There has been considerable increase of interest recently among theoretical linguists in empirical data that go beyond intuitive judgements on grammaticality and meaning of linguistic expressions. This concerns, among other areas, data resulting from experiments during which subjects carry out various linguistic or non-linguistic tasks that are somehow connected with the production or comprehension of utterances. Such experiments provide us with information about language processing in the first instance, and only indirectly about linguistic knowledge – while it has been linguistic knowledge, and not linguistic processing, that has been in the focus of theoretical linguistics for most of the last half century. So how can the evidence collected in such experiments be used to confirm or disconfirm claims made in theoretical linguistics about a speaker's knowledge of the language? Or, from another perspective, how can linguistic theory be of use for a theory of language processing? After some general considerations I will discuss this issue with respect to the meaning of the definite determiner. 1 Linguistic knowlege and language processing Theoretical linguistics investigates the native speaker's implicit knowledge of the language. This includes not only phonology, morphology, syntax, and compositional semantics, but also the systematic and equally implicit knowledge the speaker has of using linguistic expressions appropriately in a context or situation. Linguistic knowledge in this sense is characterized in abstract algebraic terms, very much as in the theory of formal languages. Questions relating to how linguistic knowledge is implemented in human behaviour or in the human brain are not part of theoretical linguistics as it is normally understood, but belong to a theory of linguistic processing, i.e., to neurolinguistics or psycholinguistics. But what is the relation between linguistic knowledge and linguistic processing? I don't think that anybody really has a good answer yet. But one fairly simple way of relating linguistic knowledge to linguistic 1 The experiments reviewed in this paper were conducted at the labs of the Institute of Cognitive Science in Osnabrück and involve work by Nadine Hartmann, Anke Karabanov, Graham Katz, and Peter König. I am very grateful to all of them for their cooperation. I am also grateful for the opportunity to present these ideas at the Seventh International Symposium on Language, Logic and Information, 1-5 October 2007, in Tbilisi and for comments that I received.