In the course, two main approaches to the study of language were introduced and contrasted: cognitive linguistics (represented by Raymond Gibbs and Eve Sweetser) and formal semantics (represented by Jan Tore Lønning and Helge Dyvik).
During the course, these two areas could be seen as different in a number of ways. Cognitive linguists study how language is processed within the cognitive system. They emphasise the importance of embodied experience for understanding language. Their models are rarely formalised. Formal semanticists, on the other hand, usually study language as a sign-system, abstracted from any individual language users. They describe the conventions of language using formal systems (thereby not assuming that language itself is a formal system). However, it should be noted that it is not formalisation per se that distinguishes the approaches from each other. Cognitive semantics models can be and are being formalised. Philosophically, there are also differences. The formal semanticists subscribe to an instrumental view of theories, while cognitive linguists often claim that what is in their theories are also in the actual cognitive system (although this was not true of the proponents of cognitive linguistics present at this course). Cognitive models are falsifiable in a traditional sense, while formal models are so in a different sense.
The two approaches focus on different questions, and in the end, a comparison between them boils down to the subjective matter of which questions are the most interesting to ask about language.
Raymond Gibbs talked about psychological experiments on meaning and cognition. By predicting language behaviour from language-independent studies, he argued that people understand metaphor through embodied experience. He also questioned that there were any such thing as 'the literal meaning' of an utterance.
Eve Sweetser talked about how mental spaces theory can account for the meaning of metaphors and metonyms. "No formalising until it's time." was a quote embraced by Sweetser. She also talked about the embodiment of metaphor through the study of metaphorical gestures.
Jan Tore Lønning introduced formal semantics and showed how it could handle entailment and compositionality.
Helge Dyvik gave the philosophical foundation for formal semantics as well as showed a practical example of how formal semantics could be used computationally within translation.
The two approaches have different views on meaning and cognition, although this was rarely explicitly stated by the speakers.
Cognitive linguists take meaning to be what is constructed by an individual language user through cognitive processing. Cognition is thereby seen as something central to the creation of meaning.
Formal semanticists, on the other hand, regard cognition as largely irrelevant for their study of meaning. In their sense, meaning is something of concern to the scientist, not the language user, and is crystallised from the conventions of language of some culture. Further, a vital part of meaning, that can be formalised, are the truth-conditions for some part of language, such as an utterance. In a formal semanticist's view, knowing the meaning of an utterance is partly knowing what has to be true about the world in order for the utterance to be true.
Here I will present a loose collection of observations, notes, etc. I made during the course that relates to my own research.
A question that was discussed was what type of linguistic processing model one adopts.
Distinct from Chomskian approaches, cognitive linguistics also study the contents of language.
According to Lønning, one needs to take into account the cognitive constraints in order to formalise meaning, in any system. For instance, when studying bees' dances, the analysis must not posit any meanings which cannot be communicated among the bees (for instance, if the bees were blind, no visual signals could be communicated). I asked how these constraints come into play when studying human language. One answer I received was that the formal semanticist should not posit any mechanisms that demand an infinite amount of resources, for instance.
When building language technology, such as computerised ticket-ordering systems, Lønning argued that we do not want to construct these systems with a basis in human cognition (which is often "incorrect"). Instead, the core of language technology should be a formal system, which then at the upmost layer can be extended with a pragmatic layer.
Lønning said that possible-world semantics had headed into trouble. An alternative is situation semantics.