Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Multilingualism WithOut Words: A non-lexicalist model of bilingualism
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Alexandra Krauska1, Danielle Fahey2; 1University of Massachusetts Amherst, 2University of Alabama
Standard models of bilingualism are primarily concerned with how words/lexical representations in multiple languages are stored and associated in the mind (Multilink: Dijkstra et al., 2018; Revised Hierarchical Model: Kroll & Stewart, 1994; etc.). However, there is a growing consensus in linguistic theory that language cannot be organized around words, given that: a) structure above and below the word level must be part of the same system, and b) linguistic units (meaning, syntax, and form) do not necessarily align at the ‘word’ level (Krauska & Lau, 2023; Bruening, 2018; Haspelmath, 2017; etc.). We propose a non-lexicalist model of bilingualism, the Multilingualism WithOut Words (M-WOW) model, in order to realign theories of bilingualism with current morphosyntactic theory and psycholinguistic models, and account for a broader typology of languages. This is an extension of the WithOut Words (WOW) model of language production (Krauska, 2024), in which independent representations of meaning, syntax, and form are linked together by separate mapping mechanisms, one between meaning and syntax, and another between syntax and form. This model also integrates prosody and linearization in the process of language production. In contrast to other models of bilingualism, M-WOW allows a single term in the first language (L1) to correspond to a (phonological) word, morpheme, or complex phrase in the second language (L2), allowing much more flexibility in how different features are realized across languages and also expanding the possibilities for code-switching and code-mixing. M-WOW also allows for L2 acquisition to progress non-uniformly across language mechanisms; features of the L2 that the learner is able to identify at one level (e.g., form) can be incorporated into the language network even if they do not have sufficient evidence to develop corresponding representations at other levels (e.g., syntax and meaning). For example, at early stages of L2 acquisition, a learner can quickly gather evidence for the phonological forms of the language, and associate those forms with counterparts in their L1 (e.g., associating ‘gato’ (L2Spanish) with ‘cat’ (L1English)), but they may piggyback on the syntactic representations of their L1 without positing distinct representations for the L2 (e.g., L1English-L2Spanish learners may initially rely on the phonological cue to perform adjective or determiner agreement in noun phrases, rather than a feature of grammatical gender; L2Spanish: ‘/o/ is masculine’, not ‘gato is masculine’). This enables the learner to maximize learning from directly observable input, while respecting cognitive economy. As the learner progresses, they can extract information about other types of representations from input based on the components of the L2 that they have already learned (e.g, a learner may use agreement patterns and distributional information to develop a syntactic representation for L2Spanish gato that does include a gender feature, in order facilitate more efficient processing in the language). Because the M-WOW model is non-lexicalist, it can accommodate a broader set of languages and linguistic phenomena, especially those that are not well-suited to lexicalist analyses, and provide new insights into the dynamic and variable process of second language acquisition.
Topic Areas: Multilingualism, Language Production