I am working on a PhD relating to cognitive science and linguistics, using computational modelling of the visual system in the hope of shedding some light on language.
Linguists and psychologists have traditionally had a strained relationship. Linguists study the abstract symbolic rules of language, while psychologists probe the mental mechanisms that let humans use language. Though there is overlap, these strategies were, until recently, largely independent. This separation is now breaking down. One result of this is cognitive linguistics, which seeks to explain language in terms of our knowledge of psychology. Another is the increasing interest in the origins and evolution of language, which requires interdisciplinary cooperation. It is in this spirit that I am investigating a possible relationship between a purely linguistic phenomenon, the noun phrase, and psychological phenomena related to object perception, categorisation and attention in the sensorimotor system.
Eh?Okay, here's an example. Here's a picture of two simple noun phrases, with the important bits highlighted. The important bits in this case are the determiner ("a" or "the") and the noun itself ("cat" here). Nothing surprising here, but now think about the difference between these two linguistic snatches. They could both easily refer to the same thing (Tibbles sitting under the table over there), so why use two different determiners?
The answer isn't simple, but it seems to have something to do with introducing things into discourse and referring back to things introduced earlier. So
Now take a look at this picture of object files at work. Object files were suggested by Kahneman, Treisman and Gibbs in 1992 as a model for part of human visual attention. The idea is that each of us has a limited number of object files (probably less than half a dozen), each of which acts a mental pointer to an object in the visual field. Object files stay tied to their objects even if they move or temporarily disappear from sight. Object files are theoretical, but there is evidence that supports their existence.
Now, two object file operations of interest are creation and reference. An object file is created when it is pointed to a new object in the visual field. Reference follows an existing object file to its object. Does this sound familiar? Let me draw a picture.
So here, perhaps the use of different determiners in language is just the same as the creation and reference of object files in the sensorimotor system.
The search for the mechanisms of language is actually a search for the aspects of human biology and culture that constrain variation in language. My proposal links noun phrases to sensorimotor mechanisms. If the constraints on noun phrases are biological in nature this would imply that some of the brain wiring used for language derives from brain wiring originally used by the sensorimotor system. If the constraints are the result of cultural evolution then the sensorimotor system should refect the way language divides up the world, because language obtains its meaning by interaction with the outside world.
So is that what you're working on?
Well, no, not exactly, but I like the little example above because it neatly shows the flavour of what I'm doing. I am currently investigating a link between visual classification and attention on one side, and linguistic number (singulars and plurals) on the other. You can find out more about this, and a more detailed justification of this approach in my publications..
- Walles, H. (2006). Psychobabble: Grounding language in the brain. Junctures, (6): 45-61.
- Forked Tongue: Pathways to Language in the Brain. A presentation delivered on 28 July 2006 as part of the 2006 Computer and Information Science Seminar Series at the University of Otago.