Parsing, fast and slow: a two-system approach to some problems in linguistics

Daniel Kahneman's influential book Thinking, fast and slow brought to popular attention the so-called 'dual process' theory from psychology. This posits that there are two distinct kinds of thought process: a fast, intuitive, automatic kind (System 1), and a slow, deliberative, controlled kind (System 2). It also suggests that the two can be in conflict, and that, despite what economists might like to believe about 'rational actors', we rely on the former kind of thinking rather more than the latter. Parsing, i.e. the processing of linguistic information, is generally seen as being of the automatic, System 1 type, but in this talk I explore the idea that the same distinction, between fast/automatic and slow/deliberate processes, can equally well be drawn inside the domain of language processing, and that it offers a useful explanatory lens through which to re-contextualise some well-studied linguistic phenomena (such as the nature of literary language), as well as the recent rise of Large Language Models. It also further contributes to the debate around experimental research methodologies in theoretical linguistics -- namely whether we really know what processes such experiments are in fact probing.