In this work, we use language modeling to investigate the factors that influence insertional code-switching. Code-switching occurs when a speaker alternates between one language variety (the primary language) and another (the secondary language), and is widely observed in multilingual contexts. Recent work has shown that code-switching is often correlated with areas of low predictability in the primary language, but it is unclear whether low primary language predictability only makes the secondary language relatively easier to produce at code-switching points – that is, purely speaker-driven code-switching – or whether code-switching is additionally used by speakers for other purposes, for instance to signal the need for greater attention on the part of listeners. In this paper, we use bilingual Chinese–English online forum posts and transcripts of spontaneous Chinese–English speech to replicate prior findings that low primary language (Chinese) predictability is correlated with insertional switches to the secondary language (English). We then demonstrate that the predictability of the English productions is even lower than that of meaning-equivalent Chinese alternatives, and these are therefore not easier to produce, rejecting the purely speaker-driven theory of code-switching in both writing and speech.

This study investigates code-switching (CS) within the noun phrase in Portuguese–German bilingual children and adolescents (aged 8–16) in German-speaking Switzerland. Using an elicited imitation task with 49 participants, we examine how linguistic and extralinguistic factors shape CS behaviour. The experiment manipulated matrix language (German vs. Portuguese), insertion type (adjective vs. noun), and adjective position (prenominal vs. postnominal). The results show that CS strategies vary depending on the grammatical properties of the matrix language. In German, prenominal adjective position—regardless of the language of the inserted adjective—was the strongest predictor of repetition accuracy. In Portuguese, the language of the adjective played a central role. We propose the Constraint Integration Model to account for the interaction between matrix-language properties and lexical features. Additionally, older age and more positive attitudes towards German increased the likelihood of producing switched utterances.

What motivates a fluent bilingual speaker to switch languages within a single utterance? We propose a novel discourse-functional motivation: less predictable, high information-content meanings are encoded in one language, and more predictable, lower information-content meanings are encoded in another language. Switches to a speaker's less frequently used, and hence more salient, language offer a distinct encoding that highlights information-rich material that comprehenders should attend to especially carefully. Using a corpus of natural Czech-English bilingual discourse, we test this hypothesis against an extensive set of control factors from sociolinguistic, psycholinguistic, and discourse-functional lines of research using mixed-effects logistic regression, in the first such quantitative multifactorial investigation of code-switching in discourse. We find, using a Shannon guessing game to quantify predictability of meanings in conversation, that words with difficult-to-guess meanings are indeed more likely to be code-switch sites, and that this is in fact one of the most highly explanatory factors in predicting the occurrence of code-switching in our data. We argue that choice of language thus serves as a formal marker of information content in discourse, along with familiar means such as prosody and syntax. We further argue for the utility of rigorous, multifactorial approaches to sociolinguistic speaker-choice phenomena in natural conversation.

A bilingual's two languages can interact in their mind, but the mechanism of this interaction is still open to debate. In this article we employ a variant of GRADIENT SYMBOLIC COMPUTATION (GSC; Smolensky et al. 2014) to model the code-switched utterances of unbalanced Dutch-English bilinguals. We aimed to evaluate GSC as an appropriate architecture to model bilingual code-switching grammars, and to explore the extent of variability within and across individual bilingual speakers. The results indicate that the structure of individual grammars can vary widely from the structure of the grammar that emerges when the population is studied as a whole. We interpret these results as evidence that individual variation characterizes not only language processing (e.g. Fricke et al. 2019, Kidd et al. 2018), but also the structure of bilingual grammar itself.

Learning to map novel words onto their intended referents is a complex challenge, and one that becomes even harder when acquiring multiple languages. We investigated how label mixing affected learning novel words in one versus two languages. In a cross-situational word learning study, 80 adult participants learned either one-to-one word–object mappings, or two-to-one mappings, reflecting different challenges in learning one or two languages. We manipulated whether mappings co-occurred locally, where repetitions were prevalent, or whether co-occurrences were more distributed throughout exposure. Learners acquired two-to-one mappings better when they did not occur in local co-occurrences, but there was no effect of learning conditions for one-to-one mappings. Whether participants were proficient or not in an additional language did not have an observable effect on the learning. We suggest that local co-occurrences of multiple labels, as in language mixing environments, increase the challenge of learning words, though this effect may be only short-lived.

Most people are multilingual, and most multilinguals code-switch, yet the characteristics of code-switched language are not fully understood. We developed a chatbot capable of completing a Map Task with human participants using code-switched Spanish and English. In two experiments, we prompted the bot to code-switch according to different strategies, examining (1) the feasibility of such experiments for investigating bilingual language use and (2) whether participants would be sensitive to variations in discourse and grammatical patterns. Participants generally enjoyed code-switching with our bot as long as it produced predictable code-switching behavior; when code-switching was random or ungrammatical (as when producing unattested incongruent mixed-language noun phrases, such as ‘la fork’), participants enjoyed the task less and were less successful at completing it. These results underscore the potential downsides of deploying insufficiently developed multilingual language technology, while also illustrating the promise of such technology for conducting research on bilingual language use.

How does the bilingual experience affect online processing? The distribution of lexical items shared between monolinguals and bilinguals can differ greatly. One critical difference is how code-switching allows more variability in the relative co-occurrence of words. The current study uses a visual world paradigm to test whether the relative distribution between Spanish gender-marked determiners (“el,” “la”) and the non-marked English determiner (“the”) predict the Spanish–English bilingual’s ability to predict and/or integrate an incoming noun. While we replicate a previously observed asymmetry among Spanish–English bilinguals between the masculine “el” and feminine “la,” our cluster permutation test results reveal differences in how bilinguals predict and integrate nouns when preceded by “el” versus “la” or “the.” Comparing our results to existing corpus data, we argue that bilinguals rely on the distributional norms they experience across both single-language and code-switched contexts to facilitate online processing.

Code-blending is the simultaneous expression of utterances using both a sign language and a spoken language. We expect that like code-switching, code-blending is linguistically constrained and thus we investigate two hypothesized constraints using an acceptability judgment task. Participants rated the acceptability of code-blended utterances designed to be consistent or inconsistent with these hypothesized constraints. We find strong support for the proposed constraint that each modality of code-blended utterances contributes content to a single proposition. We also find support for the proposed constraint that – at least for American Sign Language (ASL) and English – code-blended utterances make use of a single derivation which is realized using surface forms in the two languages, rather than two simultaneous derivations, one for each language. While this study was limited to ASL/English code-blending and further investigation is needed, we hope that this novel study will encourage future research comparing linguistic constraints on code-blending and code-switching.

This study explores lexical borrowing and loanword nativization from a neuro-cognitive perspective testing bi-dialectal speakers of Standard Chinese and Shanghainese Chinese. We created holistic and morpheme-based cross-dialectal loanwords for auditory sentence processing and compared them with Shanghainese-specific words, code-switches, and pre-existing etymologically related words. Participants rated their acceptance of each word, indicating Shanghainese-specific lexical nativeness. GAM analysis of EEG signals revealed that reduced acceptance correlated with frontal positive shifts in ERPs. Holistic loanwords triggered P300-like shifts associated with form-mismatch, whereas morpheme-based loanwords produced LPC-like shifts, suggesting sentence-level re-analysis, and N400-like early frontal negative shifts, indicating lexical integration challenges. Our results indicate that both lexical acceptance and adaptation strategies are pivotal in the cognitive integration of loanwords, revealing distinct neuropsychological stages and pathways in loanword nativization.