Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
Moving away from lexicalism in psycho- and neuro-linguistics: A non-lexicalist model of language production
Poster C31 in Poster Session C, Friday, October 7, 10:15 am - 12:00 pm EDT, Millennium Hall
Also presenting in Poster Slam C, Friday, October 7, 10:00 - 10:15 am EDT, Regency Ballroom
Alexandra Krauska1, Ellen Lau1; 1University of Maryland
In standard models of language production, the units which are retrieved and combined into a syntactic structure are ‘lemmas’ or ‘lexical items’. Such models often implicitly assume ‘lexicalism’, the idea that syntactic structure does not extend below the word level. However, across the last several decades, linguistic research examining a wider array of languages has provided strong evidence against this assumption. This has significant implications for models of language processing as well as understanding of aphasia and other language disorders. Here we provide an overview of the arguments against lexicalism, discuss how lexicalist assumptions have influenced models of language production, and propose a non-lexicalist model of language production as an alternative. Lexicalist approaches are characterized by some or all of the following assumptions: firstly, that sub-word and supra-word processes are different in kind, such that the rules that apply in the syntax cannot apply within lexical items; secondly, that wordhood is the domain of listedness; and thirdly, that lexical items necessarily include triads of sound, meaning, and syntax. In standard models of language production, lemmas - and other varieties of ‘lexical’ representations - act as triads of meaning, syntax, and form, also serving to codify the correspondence between wordhood and listedness, as well as a distinction between morphology and syntax. These models also exhibit a division between syntactic and ‘lexical’ processes, or treat morphology as a separate operation from syntax. For early models of language production, which were based largely on Dutch and English, these seem like intuitive conclusions; however, these assumptions become problematic when applied to other languages. For example, in polysynthetic languages, where sentences are regularly formed by a single word composed of multiple highly productive morphemes, lemmas would need to include either massive inflectional paradigms, or there would need to be thousands of redundant lemmas that could be created spontaneously. In addition, several languages exhibit verb suppletion based on the syntactic properties of the object. This indicates that form must be determined based on syntactic context, and that lemma selection cannot be done in isolation from syntactic structure building. In order to move away from lexicalism in models of language, it is not enough to simply update the syntactic representations; the algorithms also need to be reconsidered. Here we propose a non-lexicalist model of language production which does not rely on a lemma representation, but instead represents that knowledge as mappings between (a) meaning and syntax, and (b) syntax and form. We do not assume any architectural restriction on the size of the units that participate in stored mappings; as in the case of so-called ‘idioms’, syntactic complexes can have stored mappings to meaning or to form. The model has a single integrated stage for morpheme retrieval and syntactic structure building. Our model also emphasizes the role of cognitive control mechanisms in linearizing speech. By moving away from lexicalist assumptions, this model provides better cross-linguistic coverage, and also aligns better with contemporary syntactic theory.
Topic Areas: Language Production, Syntax