The Mental Lexicon - Volume 15, Issue 3, 2020

This study applies the computational theory of the ‘discriminative lexicon’ (Baayen, Chuang, and Blevins, 2019) to the modeling of the production of English verbs in aphasic speech. Under semantic impairment, speakers have been reported to have greater difficulties with irregular verbs, whereas speakers with phonological impairment are described as having greater problems with regulars. Joanisse and Seidenberg (1999) were able to model this dissociation, but only by adding noise to the semantic units of their model. We report two simulation studies in which topographically coherent regions of phonological and semantic networks were selectively damaged. Our model replicated the main findings, including the high variability in the consequences of brain lesions for speech production. Importantly, our model generated these results without having to lesion the semantic system more than the phonological system. The model’s success hinges on the use of a corpus-based distributional vector space for representing verbs’ meanings. Irregular verbs have denser semantic neighborhoods than do regular verbs (Baayen and Moscoso del Prado Martín, 2005). In our model this renders irregular verbs more fragile under semantic impairment. These results provide further support for the central idea underlying the discriminative lexicon: that behavioral patterns can, to a considerable extent, be understood as emerging from the distributional properties of a language and basic principles of human learning.

We examine unintentional spelling errors on verb homophones in informal online chat conversations of Flemish adolescents. In experiments, these verb forms yielded an effect of homophone dominance, i.e., most errors occurred on the lower-frequency form (Sandra et al., 1999). Verb homophones are argued to require the conscious application of a spelling rule, which may cause a temporary overload of working memory resources and trigger automatic retrieval of the higher-frequency spelling from the mental lexicon. Unlike most previous research, we investigate homophone intrusions in a natural writing context. Thus, we test the ‘ecological validity’ of psycholinguistic experiments. Importantly, this study relates these psycholinguistic constructs to different social variables in social media writing to test a prediction that directly follows from Sandra et al.’s account. Whereas social factors likely affect the error rates, they should not affect the error pattern: the number of working memory failures occurs at another processing level than the homophone intrusions. Hence, the focus is on the interaction between homophone dominance and the social variables. The errors for two types of verb homophones reveal (a) an impact of all social variables, (b) an effect of homophone dominance, and (c) no interaction between this effect and the social factors.

Although several studies have investigated the influence of the grammatical class of cue words on response patterns in the word association task, relatively little is known about the influence of more fine-grained distinctions such as cue transitivity. The present study tests two predictions, made in existing studies, of the influence of this variable. The first is that cue transitivity would influence the grammatical class of responses; the second, that it would affect the directionality of position-based, or syntagmatic, responses. English language associative responses to 49 transitive and 49 intransitive cues were gathered from 53 English L1 respondents. These responses were then analysed according to their grammatical class and categorical designation. Results suggested that cue transitivity influences both of these measurements: transitive cues yielded more noun responses and more associations classified as likely to follow the cue in text than did intransitives, while transitives received more verb responses and more associations likely to precede the cue. These results are discussed in the light of contiguity-based and semantic theories of the determinants of word association.

This study investigates the effect of stress change during compound processing in Modern Greek. Twenty-five native speakers were tested in a cross-modal lexical decision task and a naming task in order to test for performance differences across stress-change vs. non-stress-change compounds. No statistically significant difference was found for the lexical decision task. However, the naming task showed a significant effect of stress change in compound processing, with the production of non-stress-change compounds showing facilitation. These results indicate that stress change is reflected in compound processing in Greek and underscore the importance of considering the interplay between specific tasks and the computational role of linguistic features.

In this study we examined uses of the number 3017 as a neologism by members of an online forum. 3017 has a number of factors working against its success as a neologism, but its use grew dramatically over the course of six years. Statistical analyses showed that the growth data were very well modeled by both a quadratic and a sigmoid curve. The form was used primarily as an adjective and to a lesser extent as a noun over the first 500 days, before verbal forms came to dominate. To understand the structure of the 3017 concept in the mental lexicons of users, we examine attempts to define the term, and disagreements and negotiations about what the term does and does not include. Finally, we include examples of users’ creativity and productivity with the form, including readily-understood jokes.