Coherence in Swabian time and space: a cognitive-computational perspective

Karen V. Beaman, R. Harald Baayen, Konstantin Sering

Beaman and Guy (2022) quantify the coherence of a dialect by assessing the extent to which a series of dialect features co-vary. The more that dialect features are jointly present or jointly absent, the more coherent a given variety is taken to be. A more comprehensive measure of coherence can be obtained by evaluating how well speakers can learn the relations between word forms and their meanings. To investigate coherence in Swabian across time and space, this project operationalizes the concept of “systemic lexical coherence” with the help of the Discriminative Lexicon Model (Baayen et al., 2019; Chuang and Baayen, 2021; Heitmeier et al., 2025), an error-driven cognitive computational modeling framework for the mental lexicon. Our approach is based on the linguistic systems of monolingual and balanced bilingual (or bidialectal) speakers. Monolingual speakers handle one single system and thus have considerable exposure to this system. In contrast, bilingual/bidialectal speakers must juggle two systems, with half the exposure to each system. Furthermore, the bilingual/bidialectal speaker needs to know two words for any given concept as opposed to the one word that the monolingual speaker must know. Unavoidably, the bilingual/bidialectal speaker has a more complex learning task; in addition, because they have a larger lexicon, lexical access is slower than for monolinguals (cf. Lemhoefer et al., 2008). We expect that when bidialectal speakers use both varieties together, the coherence of their mental lexicon depends on the extent to which the two are distinct or mixed. Thus, speakers who use more dialect variants, as well as those who use more standard variants, will have more coherent lexicons than speakers who exhibit a mixture of both dialect and standard language variants. In other words, we approximate coherence by means of intelligibility in comprehension and the consequences of the relative homogeneity of the spoken word forms. We expect a more coherent mental lexicon to be simpler to learn and more accurate than a less coherent lexicon.