Back to the future: Linguistic emergence in usage and the psychological mechanism of learning

25 Apr 2022

It is commonly assumed that the modern days of psychology of language started with Chomsky’s (1959) vitriolic review of Skinner’s (1957) book Verbal Behavior (cf., Harley, 2005). The term psycholinguistics appeared almost a decade and a half earlier in a book with the same title: Psycholinguistics: A survey of theory and research problems, edited by Osgood and Sebeok (1954). The research questions and the methodology described there were, however, known for much longer. Later, Osgood himself reflected on this “formative period” as being “characterized by extremely good relations between psychologists and linguists, [... having] a common commitment to an operationalist philosophy of science, and a division of labor” (1975, p. 17).

The new era started with a major upset and a rather vocal division of domains rather than with a division of labour to achieve common goals: linguists were told to focus on competence while psychologists were allowed to scrutinize performance only (cf., Greene, 1972). Interestingly enough, it appears as if Chomsky’s take on modularity spilt over to the research culture itself, allowing for very little, if any at all, cross-fertilization between interested disciplines. The last man of older traditions still standing – Osgood, accepted being “a dinosaur”, detached from the discipline but at peace with his position. For him, the fact that Chomsky also completely disunited form and meaning, showing no interest in the latter, was unsurmountable, in principle (see Julia, 1983 as well).

Another major branch of linguistics, usage-based linguistics (cf., Divjak, 2019), took a position profoundly different from that of generativism. In particular, within usage-based linguistics, language is seen as a complex dynamic system shaped by usage, of both language (as structure and/or data) and other general cognitive capacities (as hypothesized principal mechanisms). That form and meaning are inseparable was never even doubted. With this in mind, it seems meaningful to re-introduce linguistics and psychology to one another, once again. With common research interests and complementary methodologies, they have much to offer for the advancement of both disciplines into interdisciplinary language sciences.
In this talk, I will present some of the work we are doing in the Out of our Minds team [], which operationalizes the usage-based linguistic notion of emergence from use through psychological principles of learning. I will pay special attention to the hybrid methodology we are developing and using, which combines corpus analysis, computational modelling, and experimentation. I will argue that corpus analysis and computational modelling of learning can constrain one another to provide insights for both, the necessary and sufficient linguistic abstractions as well as the necessary and sufficient learning functionality, to meet the requirement of cognitive commitment and plausibility that is finally tested in the lab. We believe that our approach to language is in perfect harmony with Poggio’s (2012) addition to Marr’s (1982) three level of analysis, in which learning represents a self-sufficient level of explanation, and where the dynamic pressure of language-in-use constrains what emerges and how it gets learned.