Abstract Previous research has found a correlation between code-switching and filler words in corpus analyses of bilingual speech. To examine this correlation, 22 high-proficiency English-Spanish bilinguals completed a picture-based oral narrative task with one of three conditions — using only English or Spanish or switching freely (Mixed condition). In the Mixed condition, phrases containing a code-switch were significantly more likely to contain filler words. Speakers in the Mixed condition used significantly more filler words than the Spanish-only condition, despite using majority-Spanish utterances. These results support the correlation between filler words and code-switching, demonstrate that code-switching itself does not increase filler word use, and offer evidence that speakers’ filler word patterns are influenced by both languages when permitted to code-switch.
Abstract This study investigated the interpretations of Japanese negated disjunctions and conjunctions by L1-English and L1-Mandarin learners of L2-Japanese. Negated disjunctions and conjunctions have different interpretations in English as opposed to Japanese and Mandarin. The results of a Truth Value Judgment Task show that L1-Mandarin L2-Japanese learners had more target-like interpretations than L1-English L2-Japanese learners, suggesting L1 transfer. L1-English L2-Japanese learners had more difficulty with Japanese negated disjunctions than conjunctions, contrary to predictions based on learnability. L1-English L2-Japanese learners also had difficulty with another form of Japanese negated disjunction. From these results, I argue that the cognitive complexity of negated disjunctions is a potential delaying factor in the acquisition of their correct interpretations.
Abstract This study explores the subset — superset relation between English and Turkish through the L2 acquisition of reflexives in both languages. The objective is to investigate whether L2 learners can achieve target-like interpretations of reflexives regardless of their L1 structures or whether they exhibit a universal preference for a specific antecedent type. Unlike English reflexives, Turkish reflexives kendi and kendisi allow both local and non-local binding. The study involved two native speaker groups, and two learner groups: 24 English learners of Turkish, and 24 Turkish learners of English. Results from a Truth Value Judgment Task and a Picture Description Task reveal that both learner groups prefer the local antecedent over the nonlocal one, suggesting that the local readings serve as the default option in the L2 acquisition of reflexives.
Abstract This study investigates how second language (L2) learners and native speakers interpret pragmatically underinformative responses involving the quantifier some , and how their interpretations are influenced by the preceding question under discussion (QUD). Specifically, it examines how QUD cues containing all or any shape expectations for scalar implicatures. While all primes an upper-bound (not all) interpretation, any supports a lower-bound (at least some) reading. Both L2 learners and native speakers showed sensitivity to these contextual cues, adjusting their interpretation of some accordingly. However, only L2 learners’ sensitivity to the QUD was modulated by individual differences in inhibitory control. These findings suggest that aspects of executive function facilitate learners’ ability to integrate contextual information, thereby supporting L2 comprehension of pragmatic inferences.
Abstract This study examines whether Japanese learners of English (JLEs) can acquire universal constraints on syntactic movement (e.g., anti-locality), focusing on the that -trace effect and its amelioration effects through adverbial intervention (i.e., the adverb effect). Two acceptability judgment experiments were conducted to investigate JLEs’ sensitivity to these effects across proficiency levels. The results showed that JLEs are sensitive to that -trace violations and adverb effects with increased English proficiency. In addition, more proficient learners showed a preference for null complementizers over overt that , a pattern consistent with English native speakers. These findings suggest that the anti-locality constraint is operative in the L2 grammar of JLEs, supporting the Full Access view of second language acquisition.
Abstract This study tests the predictions of several third language (L3) wholesale and property-by-property transfer models regarding the use of morphosyntactic cues and language dominance/use for determining transfer source. L1 German-L2 English and L1 English-L2 German bilinguals were exposed to one of two Korean-based input conditions that varied in morphosyntactic similarity to each of the learners’ previously acquired grammars. Participants then completed production and judgment tasks in Korean for properties not evidenced in the input during the familiarization phase. Results suggest a primary role for morphosyntactic cues and a secondary role for language use. Additionally, more evidence of ‘mixed’ responses in the judgment than the production task suggests support for a ‘Grappling Period’. Results are most compatible with property-by-property transfer.
Abstract English doubly-quantified sentences such as A boy climbed every tree allow both a surface reading (one particular boy climbed every tree) and an inverse reading (every tree was climbed by a (different) boy); the latter reading, by contrast, is not allowed in Mandarin analogues and has repeatedly been found to be unavailable in English to L1-Mandarin learners of English. This study investigated whether stories that — in light of real-world knowledge — support only the inverse reading of doubly-quantified sentences can stimulate acquisition of that reading. We found no evidence for acquisition: While L1-Mandarin learners of English accepted the inverse reading when it was the only plausible interpretation, they did not extend this acceptance to sentences in contexts supporting both surface and inverse readings.
Abstract I consider the potential research value of Big Educational Language Learning (BELL) data, that is, data generated in teaching and assessment environments as part of the everyday operations of relevant organizations and institutions. Online language learning and assessment has created unprecedented opportunities for the collection of large and diverse samples of learner data across proficiency, for a variety of tasks and from learners with variable linguistic and educational backgrounds. I argue that, thanks to their size and diverse samples, BELL data can support multi-factor empirical investigations that are not easily feasible through lab-based or small-scale field studies. Importantly, they enable the investigation of several abstract features in a single design, contributing essential empirical evidence for feature-based hypotheses and accounts of L2 acquisition, complementing experimental investigations.
Abstract We examine pronoun interpretation in an overt subject language (English) by L2 learners from a null subject L1 (Italian), using a task where participants choose potential antecedents for pronouns in bi-clausal sentences which manipulate presence or absence of stress on the pronoun and presence or absence of a pause between clauses. Results show that learners are sensitive to the overturning effect of stress, resulting in a shift away from an antecedent prominent in the discourse, but that native speakers are not. Sensitivity for the learners may arise from prosodic properties of their null subject L1; lack of sensitivity for the native speakers may suggest that prosody alone is insufficient to influence antecedent choice for speakers from overt subject languages and that richer contextual cues are required.
We investigate two superficially similar constructions in Spanish: modal periphrases as in podemos salir (we can leave) and nominal infinitival clauses as in deseamos salir (we wish to leave). The absence of a superficial structural contrast (inflected verb form followed by an infinitive), together with the semantic complexity conveyed by modality and the grammatical constraints which differentiate these constructions, could interfere with their acquisition. To address whether non-Spanish dominant bilinguals can distinguish between them, we administered a Grammaticality Judgment Task to three groups of adult English–Spanish bilinguals with different linguistic backgrounds. Results showed no significant differences between the groups, which leads us to propose that this outcome constitutes evidence in favor of the poverty of the stimulus argument.
Abstract This study examines how adult native and nonnative Korean speakers interpret negated disjunction, inspired by Grüter, Lieberman, and Gualmini (2010) . Using a novel methodology with controlled prosody and the prediction mode of the truth-value judgment task, we tested Korean native speakers and two nonnative Korean groups — L1-English and L1-Japanese learners of Korean. Results confirm that Korean native speakers predominantly accept a conjunctive reading, with limited acceptance of a disjunctive reading. Similarly, nonnative speakers prefer the conjunctive reading, with no strong evidence of L1 transfer or proficiency effects, though some advanced L1-Japanese learners of Korean display an L1-like behavior. These findings suggest that while both interpretations are available in Korean, their acceptance differs, with a clear preference for the conjunctive reading.
Abstract The present study investigates bilingual compound verbs (hereafter BCVs), verbs in which the components of a verbal compound come from the two languages of a bilingual speaker. We investigate to what extent grammatical category constrains the items occurring in the production of BCVs by Persian-Dutch bilinguals. Previous studies on BCVs reported that nominal constituents are often replaced by a verb from the other language, contrary to the grammatical category constraint (see Purmohammad, Vorwerg, & Abutaleb, 2022 ). Our study investigates whether these observations also apply to BCVs in Persian-Dutch bilinguals. Our results suggest that the grammatical category constraint is not violated, as subjects are sensitive to the nominal features of non-finite verbs, and that the grammatical category constraint should be refined rather than abandoned.
Abstract This chapter explores the L2/L3 acquisition of Mayan ejectives from multiple languages. Proceeding from the dimensional framework of Avery and Idsardi (2001) , we lay out a unified analysis of the phonological acquisition of Yucatec Maya by Spanish speakers, Kaqchikel by Spanish-English speakers, and Q’eqchi’ by English speakers. Both Spanish and English phonologies provide for a path to acquisition of glottalized consonants through their respective contrastive dimensions. Through phonetic completion and enhancement, learners redeploy knowledge of other laryngeal dimensions to successfully acquire sounds which L1 users of the target languages represent with Larynx Height, a dimension not utilized in the learners’ known language(s). Thus, language learners develop non-targetlike grammars that adequately map novel contrasts through the constraints provided by Universal Grammar.
Abstract We re-examine a recent contribution by Mitrofanova, Leivada, and Westergaard (2023) to the debate on wholesale vs. property-by-property transfer in L3 acquisition. They argued that results from their artificial-language experiment support the Linguistic Proximity Model (LPM) because in three of four conditions, the L3-Russian-Norwegian group’s mean acceptance fell ‘between’ those of the L2-Russian and L2-Norwegian groups. Our reanalysis of their original data reveals that those ‘between’ results stem from bimodal/continuous distributions, not from unimodal distributions centering around a ‘between’ result. We next inspected each individual’s cross-conditional response patterns, grouping them into response profiles, to test three logically possible versions of LPM-generated predictions. Qualitative analyses show all three fail. We end with a methodological discussion of testing for transfer in L3 models.
Abstract This study examines the acquisition of singular generic expressions in L3 Norwegian. Using a contextualized Acceptability Judgment Task, we tested generic markers, known for their polysemy. Norwegian generics align partially with both Polish and English, allowing the target speakers to master genericity in L3 Norwegian via their previously acquired languages. Two L1-user groups were tested: English and Polish. The Polish learners were further split into immersive (residing in Norway) and instructed (residing in Poland) groups. No acquisition was observed in the instructed group. Evidence of cross-linguistic influence from English was established in the immersive group. We conclude that both the participants’ native language (Polish vs. English) and the degree of exposure (immersive vs. instructed) influenced the L3 acquisition process.
The present study investigates the interaction of gender with (declension) class in the acquisition data of one bilingual child (from two to five years old) who develops Italian as a weak language in combination with German in Germany. As reported in the literature, the Italian child acquires gender with ease, reflected in the nearly exceptionless target-like gender marking on determiners. Of the two possible errors, omission and commission, the Italian child vastly omits determiners. Nouns are inflected according to (declension) classes in adult Italian, most of which correspond to one gender. If the gender feature can be derived by class, as proposed by Lowenstamm (2007) for adult French, a different and integrated approach to gender acquisition is possible.
We present a case study of a heritage speaker of German, Luisa, who is growing up in an English-speaking part of Canada, focussing on the acquisition of grammatical gender in German. While German has cues to gender assignment, the acquisition of gender in this setting is compromised by the magnitude of gender cues and form syncretism, and the absence of gender in English. We present longitudinal, naturalistic data from three periods: age 1-2, age 4- 5, and age 7. We ask whether Luisa develops grammatical gender akin to monolingual children or whether there are indications of delay, stagnation, or attrition, as observed for heritage speakers of other languages. The results show monolingual-like development despite a shift in dominance from German to English.
Our objective is to shed light on recursion through an exploration of the L1 development of four DP structures in English. Since recursion is in narrow syntax, there is no reason to expect asymmetries between constructions but specific formal differences between structures might make acquisition of certain forms more difficult, and there may be differences according to meaning classes. We report a study targeting possessives, comitatives, locatives, and part-whole expressions, each with 2-level embedding. The results reveal statistically significant effects of age and condition, with no interaction. Possessives and comitatives develop earlier than locatives and relatives, and are also more productive for adults. These results help us to delineate the domain of recursion and formulate a credible developmental scenario.
How language learners segment (recognise and store words) in the speech stream has typically been explored with children (Jusczyk 1997). Researchers have only recently begun to examine how adults segment an unfamiliar natural language after first exposure without instruction (Gullberg et al. 2010; Gullberg et al. 2012; Carroll 2012, 2013, 2014; Shoemaker & Rast 2013). We report on a study of how 28 English-speaking adults begin to segment words after hearing them in fluent Russian during four sessions. The results showed that segmentation improved significantly over time. Segmentation patterns reflected the influence of English phonotactics and sensitivity to weak-strong stress. We conclude that beyond native language bias, adults deploy the segmentation mechanisms similar to those children use.
