The analogical modeling of linguistic categories

Abstract

In recent years proponents of usage-based linguistics have singled out ‘categorization’ as possibly the fundamental cognitive operation underlying the acquisition and use of language. Despite this increasing appeal to the importance of categorization, few researchers have yet offered explicit interpretations of how linguistic categories might be represented in the brain other than vague allusions to prototype theory, especially as implemented in connectionist-like frameworks. In this paper, I discuss in some detail the implications of superimposing the theoretical representations of linguistic structures onto domain-general models of categorization. I first review the evidence that instance-based, or exemplar-based, models of categorization provide empirically and theoretically better models of both domain-general categorization and of linguistic categorization than do the most commonly cited alternative models. I then argue that of the three exemplar-based models currently being applied to linguistic data, Skousen’s Analogical Model (AM) appears to provide the simplest, most straightforward account of the data and that it appears to be fully compatible with our current understanding of the psychological capabilities and operations that underlie categorization behavior.

Against stored abstractions: A radical exemplar model of language acquisition

“…Skousen’s, 1989, Analogical Model). Chandler (2017, p. 67) argues against feature weightings on the basis that they ‘must be determined through separate experimental and computational operations applied to a set of stimuli ahead of time’ and ‘cannot be interpreted theoretically as resident components of the mental representations of the linguistic forms being modelled’. However, neither of these criticisms would seem to straightforwardly apply to feature weightings that emerge over the course of learning, as for discriminative learning models.…”

“…First, while connectionist models of the English past tense (and inflectional morphology more generally) are well known amongst child language acquisition researchers, most – at least in my experience – seem to be largely unaware of exemplar models and how they work. Second, the details matter; not so much the choice of GCM, TiMBL or AM per se , none of which consistently outperforms the others (see Chandler, 2017, for a review), but the various implementational decisions that must be taken regarding feature representation, feature weighting, decay functions, the use of types versus tokens, and so on. When given the task of predicting the judgement and production data from Albright and Hayes’ (2003) novel verb study, each of these exemplar models – depending on the particular instantiation – equals or betters both a state-of-the-art connectionist model (cf.…”

Against Stored Abstractions: A Radical Exemplar Model of Language Acquisition

“…complicate matters -does affect performance (Eddington, 2004) .For example spling and swim do not share a final feature in a phoneme representation, but do so in a phonetic-feature representation which represents both /ŋ/ and /m/ as nasals). Importantly many versions of the GCM use feature weighting (like the attention weights in Regier's, 2005, LEX model) Chandler, 2017, for a review), but the various implementational decisions that must be taken regarding feature representation, feature weighting, decay functions, the use of types versus tokens, and so on. When given the task of predicting the judgment and production data from Albright and Hayes' (2003) novel verb study, each of these exemplar modelsdepending on the particular instantiation -equals or betters both a state-of-the-art connectionist model (c.f., Chandler, 2010: Table 1, Kirov & Cotterell, 2018; Table 5; though see Corkery, Matusevych & Goldwater, submitted, for concerns regarding the stability of these simulations) and Albright and Hayes' (2003) own model which constructs an explicit micro-rule for each and every sub regularity; an approach which shows no regard for psychological plausibility, in contrast to many exemplar models which have their origins in models and findings from the nonlinguistic categorization literature.…”

Chaining and the growth of linguistic categories

“…Many discussions of chaining within cognitive linguistics are heavily influenced by Rosch and her prototype theory of categorization (e.g., Geeraerts, 1997), but this literature has been largely separate from the psychological literature on computational models of categorization [7,8]. The modeling literature includes many comparisons between exemplar models and prototype models of categorization, and the question of whether categories have a central core lies at the heart of the difference between the two approaches.…”

“…Third, as mentioned already our space of models includes exemplar-based approaches that have not been explored in depth by previous computational accounts of chaining. Previous scholars have given exemplar-based accounts of several aspects of language including phonetics, phonology, morphology, word senses, and constructions [22,23,24,8,25,26], and our approach builds on and contributes to this tradition.…”