ABSTRACT. Developing students’ multilingual competences through learning several languages in school has become a key priority of European Union initiatives and policies (Council of Europe, 2022). However, collaboration among language teachers, which is essential for promoting these competences, is often limited or absent (e.g., Forbes, 2019; Gunning et al., 2016; Haukås, 2025). Therefore, investigating how language teachers perceive and implement cross-linguistic collaboration is both relevant and urgent. To date, research on language teachers’ mindsets regarding such collaboration is scarce (Haukås & Drange, 2024). The LATECO project (Fostering Multilingualism through LAnguage TEacher Collaboration), funded by the European Commission under the Marie Skłodowska-Curie Actions, explores language teachers’ mindsets and the effects of teacher collaboration across three language subjects in schools, aiming to enhance students’ multilingualism and teachers’ professional development. Mindsets, defined as beliefs about the malleability of one’s abilities, perceived control over improving them, and approaches to managing failure (Dweck, 2006), are crucial to investigate because they influence language teachers’ behaviour, motivation, and engagement in professional development (Leis et al., 2025), including cross-linguistic collaboration. This presentation reports the first phase of LATECO, focusing on teachers’ mindsets in the Norwegian upper secondary school context. Inspired by Q methodology (Haukås & Mercer, 2022), data were collected through background questionnaires, a sorting activity, and follow-up interviews with a group of teachers of Norwegian, English, and other second foreign languages (e.g., German, Spanish, French) from two schools. The results provide an innovative overview of teachers’ mindsets regarding collaboration to teach multilingually.

ABSTRACT. Teachers are called on to enact multilingual pedagogies to support inclusive education that promotes language learning, social justice, and the well-being of learners (Li, 2023, Drachmann et al., 2023, Council of Europe, 2022). However, the systematic enactment of such principles in local classrooms is uncommon (Burner & Carlson, 2023). Through researching the multilingual beliefs and practices of English teachers in linguistically and culturally diverse schools in Norway, and what influences these, this study seeks to explore how to foster multilingual pedagogies. The theoretical underpinnings include an ecological view (van Lier, 2004) on language teacher cognition (Borg 2006), and a multilingualism as a resource orientation (de Jong et al. 2016, 2019) to multilingual pedagogies, including pedagogical translanguaging ideals (Garcia et al. 2017). An explanatory sequential mixed-methods design was used (Creswell & Clark, 2011). First, in a quantitative strand, we mapped teachers’ beliefs and reported practices with a survey (N=110). The results indicate tensions with both monolingual and multilingual orientations, and that younger teachers and those in years five and above were more likely to have pro-multilingual beliefs and practices. To explore these results in-depth, we engaged in a qualitative strand of research inspired by appreciative inquiry (Whitney & Trosten-Bloom). First, we engaged in classroom observations followed by individual interviews with six teachers in grades 3-8 to uncover spaces used and described to enact multilingualism as a resource. Nine types of spaces were found functioning to acknowledge and support learners’ multilingual development, learning and well-being. Next, a collaborative classroom intervention was undertaken by one teacher and researcher, where we explored influences on enacting multilingual pedagogies. The study indicated opportunities for co-learning between the teacher, researcher and learners were supported by equitable dialogue. The overarching results suggest multilingual pedagogical literacy may be needed by various stakeholders to foster multilingual pedagogies in schools systematically.

ABSTRACT. In recent years, teaching additional languages has expanded considerably worldwide, with English emerging as the dominant option under globalization, while European policies continue to promote multilingualism beyond the L1 (Cenoz & Gorter, 2025; Kirsch et al., 2020). Current school of thought conceptualizes multilingualism as the functional, communicative use of multiple linguistic resources rather than unattainable native-like proficiency, challenging rigid hierarchies and language separation. Despite this shift, monolingual norms remain prevalent in classroom practices, making teachers’ beliefs a key factor in the adoption of multilingual pedagogies (Chik & Melo-Pfeifer, 2024; Galante & dela Cruz, 2024). Addressing this educational landscape, this research examines the beliefs of 438 in-service teachers in the Southern Basque Country as part of the CAMELTA-BCQ project. This initiative investigates multilingual settings in Quebec and the Basque Country through mixed methods, exploring how teachers conceptualize multilingualism in a sociolinguistically sensitive minority-language context. Findings indicate strong alignment with contemporary perspectives. Teachers display a clearly positive, functional view of multilingualism, agreeing on its benefits for communicative competence, deep language learning, academic success, and overall development. Similarly, they explicitly reject native-speaker-oriented conceptions, including the view of multilingualism as the sum of monolinguals. Interestingly, multilingualism is not perceived as a threat, provided language policies remain firmly oriented toward the revitalization of Basque. Nevertheless, results reveal ideological tensions regarding translanguaging. Teachers express reservations about its implementation in contexts marked by sociolinguistic imbalance, voicing concerns that such practices could undermine local language revitalization efforts. Overall, the study extends previous research in the Basque Country (Gartziarena, 2021; Gartziarena et al., 2023) and contributes to international debates on teacher cognition in minority-language education.

ABSTRACT. This presentation reports on an ongoing longitudinal study of an exceptionally talented multilingual child born in 2017 and raised in the Czech Republic by Israeli parents. At eight years old, the child actively uses nine languages: Hebrew as her mother tongue, Czech as her second language, and English, French, Spanish, German, Japanese, Mandarin, and most recently Modern Greek as additional languages. Earlier stages of this research documented her plurilingual strategies at the age of four, when she already operated in five languages (Doleží, Müllerová 2024). The current phase of the study examines the dynamic nature of her linguistic repertoire (Companion Volume, 2018), with particular emphasis on the acquisition of Modern Greek, her newest asset. We capture her linguistic competence in this language with the help of the CEFR descriptors for young learners aged 7-10 (Szabó and Goodier, 2018) and argue that CEFR descriptors for young learners, where available, offer a promising tool for capturing the complexity of multilingual development. At the same time we underscore the limitations of the task as well as the tool itself. Nevertheless, we believe that this approach can complement existing diagnostic instruments, many of which are not easily accessible or user-friendly for teachers, speech therapists, or parents. While the growth of heteroglossic approaches in multiple language learning is apparent, there is a contrast in missing assessment practices (Saville and Seed, 2022). By tracing the child’s progression across multiple languages over several years, we highlight the non‑linear, dynamic nature of multilingual development and show how new languages are integrated into an already complex repertoire. Ultimately, this presentation contributes to the underresearched field of multilingual child assessment and provides insights into how individual multilingual minds develop and operate.

ABSTRACT. Cognitive models of multilingualism, such as the Adaptive Control Hypothesis (ACH; Green & Abutalebi, 2013), accurately map the mechanisms of language selection but often overlook the affective dimensions that constrain these processes. This oversight is critical for L3 research, where affective alignment across languages (e.g., L2 vs. L3) may differentially impact processing efficiency (Dewaele, 2015a; 2015b). To determine how affective factors interact with cognitive control demands, we examined whether an affective variable – language attitudes (Li & Wei, 2022) – modulates switch costs in a controlled bilingual baseline.

Thirty-four late Dutch–English bilinguals completed a cued picture-naming task adapted from Broersma et al. (2016), alternating between L1 and L2 while naming cognate and non-cognate words. We assessed individual differences in cognitive control (L2 proficiency, exposure) and affective alignment across languages (language attitudes).

Linear mixed-effects models revealed significant asymmetrical switch costs (larger L2→L1) that were modulated by higher L2 proficiency. Notably, speakers with more positive alignment towards their L2 (i.e., positive language attitudes) showed attenuated switch costs specifically during non-cognate to cognate transitions. This suggests that affective alignment with the L2 may play an important role in buffering against the cognitive load of switching, facilitating reactive control when lexical overlap facilitates activation.

These findings indicate that multilingual control is not purely a product of executive function but is dynamically shaped by affective factors. By integrating language attitudes into the Adaptive Control framework, we offer a more holistic predictor of performance – one that is essential for future studies on trilingual populations (e.g., Frisian–Dutch–English), where affective relationships with the L3 often differ distinctively from the L2 (Bosma & Blom, 2019).

ABSTRACT. This study examines whether multilingual proficiency accounts for variation in personality traits. Participants were 157 young adults (aged 20–29; M ≈ 22.6, SD ≈ 1.8), all L1 Polish speakers with L2 English and functional command of multiple additional languages. The 40-item Multicultural Personality Questionnaire Short Form (MPQ-SF; Van der Zee et al., 2013) was administered to measure five traits relevant to functioning in culturally diverse contexts: Cultural Empathy, Flexibility, Social Initiative, Openmindedness, and Emotional Stability. Internal consistency was acceptable to strong for four scales (α = .79–.84) and lower for Openmindedness (α = .61).

Multilingual proficiency was measured as lexical proficiency with DIALANG vocabulary placement tests (Alderson, 2005), administered in L2 English and additional languages reported by participants (Spanish, French, German, Italian, Norwegian, Danish, Swedish, Portuguese, and Greek). Lexical proficiency scores were corrected for false alarms following standard yes/no procedures (Meara & Buxton, 1987; Read, 2007) and summed across languages to yield a composite proficiency index.

The relationship between multilingual proficiency and personality was analysed using a linear mixed-effects model with MPQ trait scores as the outcome variable, multilingual proficiency as a fixed effect, and random intercepts for participants. A significant main effect of multilingual proficiency was observed (p = .005), indicating that higher lexical proficiency accounted for higher personality trait scores across dimensions. While mean trait scores differed significantly (p < .001), no interaction between proficiency and trait type was found (p = .390). The model explained 56.4% of variance in trait scores at the fixed-effects level, increasing to 64.9% when individual variability was included.

Overall, the findings indicate that multilingual lexical proficiency accounts for meaningful variation in personality traits, with effects observed consistently across dimensions—Cultural Empathy, Flexibility, Social Initiative, Openmindedness, and Emotional Stability—highlighting a systematic link between accumulated linguistic proficiency and stable dispositional characteristics.

ABSTRACT. Although recent L3 acquisition research suggests that cross-linguistic transfer works on a property-by-property basis rather than wholly, we still know relatively little about how these specific syntactic properties develop over time. This study looks at English Wh-questions among Moroccan Arabic-French bilinguals to see if different aspects of Wh-questions follow different developmental paths as learners move from beginner to advanced levels. Drawing on Westergaard et al.'s (2017) Linguistic Proximity Model and using Optimality Theory as an analytical framework, we examine the acquisition of four properties that relate to Wh-Questions: Wh-fronting, Do-support, Subject-Auxiliary inversion, and Wh-in-situ restrictions. The study involves 60 participants split between beginners and advanced, who have completed a grammaticality judgment task and a translation task. A mixed ANOVA tests for the predicted Proficiency × Property interaction, which would indicate that different properties show varied patterns of development. The OT compares constraint rankings between the two proficiency groups for each property. Based on the structural characteristics of the source languages, we expect Wh-fronting to improve considerably since French provides a good structural match. Do-support, on the other hand, will likely remain problematic at both levels because neither Arabic nor French has this feature. Inversion may show some improvement given that French has inversion, while Wh-in-situ restrictions should improve more slowly as learners work against interference from source languages. In OT terms, properties that improve should show evidence of constraint re-ranking (like Q-SCOPE outranking STAY for Wh-fronting), whereas persistent errors suggest learners maintain their non-target rankings (FULL-INT still dominating OB-HD for do-support). The findings have implications beyond this particular language combination. They provide evidence that property-specific models explain L3 development better than models predicting wholesale transfer, show how OT can capture differential development through constraint re-ranking, and offer implications for syntax, morphosyntax and practical insights into structured intervention for teaching Wh-questions.

ABSTRACT. This study focuses on the initial state (IS) of third language acquisition (L3A). It investigates how ab initio third language (L3) learners, whose first language (L1) is Chinese and whose second language (L2) is proficient English, produce sentences in two semi-artificial L3s during their first exposure.

Although morphosyntactic transfer in early L3A has been widely examined, most prior studies involved some degree of training or exposure to the target languages (TL). As noted by Puig-Mayenco and Rothman (2019), even minimal exposure can substantially affect developmental patterns and theoretical interpretations. Moreover, earlier studies have often relied on receptive tasks, such as grammaticality judgments. The present study, however, adopts a stricter definition of the IS (cf. Zhang, 2025) and employs a production-based task to test ab initio learners at first exposure.

The target structure is the use of dummy subjects in weather expressions and existential constructions, which are required in English but absent in Chinese. The TLs are written in either Swedish or Japanese vocabulary, which share formal similarities with Chinese and English, respectively. In each task, participants viewed an image designed to elicit the target structure, accompanied by a set of cards in one of the TLs, containing essential words for describing the image, a potential dummy subject, and fillers. Each card included a TL lexical item with Chinese and English translations. Participants selected and arranged the cards into what they considered grammatical in the TL.

Results suggest that learners begin forming syntactic hypotheses at the onset of exposure by observing formal similarities. Both L1 and L2 are activated, but the formally similar language appears to be more activated, contradicting the wholesale transfer suggested by the Typological Primacy Model (Rothman, 2015) and supporting property-by-property transfer and co-activation proposed by the Linguistic Proximity Model (Westergaard et al., 2017).

ABSTRACT. Crosslinguistic influence (CLI) in multilingual speakers has attracted increasing attention in recent decades. However, existing research has primarily focused on CLI from L1/L2 to L3 (De Angelis, 2007; Rothman, 2011; Bardel & Falk, 2012; Eibensteiner, 2024), with comparatively little work examining the influence of a later-learned language on an earlier one. To address this gap, the current study investigates the potential influence of Portuguese L3 on Chinese learners’ sensitivity to word order in English L2, focusing on the placement of manner adverbs. An acceptability judgement task was administered to two groups: 30 Chinese learners with B2+ proficiency in Portuguese L3 and B1+ in English L2, and 30 Chinese adults with B1+ proficiency in English as their only foreign language. Participants evaluated English sentences containing manner adverbs in three positions defined by the linear order of a transitive verb and its object: Adv-V-O (e.g., he clearly explained the task); V-Adv-O (e.g., ?he explained clearly the task); and V-O-Adv (e.g., he explained the task clearly). The V-Adv-O position reflects a Portuguese pattern (Raposo, 2013) and is considered unnatural in English (Huddleston & Pullum, 2002). Preliminary quantitative analyses revealed that the Portuguese L3 learners showed significantly higher acceptability ratings for the V-Adv-O sequence compared to the participants without Portuguese, and they also demonstrated greater individual variability. These findings suggest that L3 Portuguese may influence learners’ sensitivity to word order constraints in L2 English. More broadly, they indicate that CLI is multidirectional rather than one-way, showing that multilingual grammars can be reorganised when new languages are acquired.

ABSTRACT. Attitudes towards minority languages are complex and contradictory; they may be valued for their identity and solidarity functions (Woodlard, 1985) but not as linguistic capital (Bourdieu, 1991). Many such languages are subject to ongoing revitalisation processes, in which it is crucial to understand speakers’ attitudes towards them, as well as any influence this may have on the languages that exist alongside them. We know that media presence can be a driver of language attitudes (Dragojevic et al., 2021), and research has shown a positive affective relationship between media use and minority languages (O’Hanlon & Paterson, 2019; Smirnov, 2015). Nevertheless, there remains a notable gap in terms of studies that focus on digital media use (Fitzgerald, 2022; Hulsen, de Bot, & Weltans, 2002), which is essential to cover given the prominence of digital media in the literacy practices of young minority language speakers. The scarce work that exists in this vein points towards a positive relationship between extramural digital media use and attitudes towards minority and foreign languages (Nightingale, 2016; 2025), a result we find very encouraging. With this in mind, the current study attempts to consolidate research on the relationships between the use of modern digital media among young multilinguals and their attitudes towards languages. We focus on Valencian and Scots Gaelic, two minority languages with distinct sociolinguistic statuses and relationships to the dominant and foreign languages of their respective contexts. In doing so, we compare the survey results of 93 high-school students from Glasgow and 93 from Castelló. Results corroborate existing research regarding minority and foreign languages and also indicate some interesting interactions between languages, as well as implications for language attitudes and language use. We believe these findings make a humble contribution to deepening our understanding of digital representation and language revitalisation processes in multilingual contexts.

ABSTRACT. We investigate the peer learner networks of 251 Ukrainian refugees enrolled in an intensive course of L2/L3 Polish. The forced migration context is particularly noteworthy given the close typological similarity between the languages spoken and being acquired, and the participants’ unique language constellation profile, with most being functionally bilingual in Ukrainian and Russian, but exhibiting different degrees of dominance in each language and complicated attitudes toward the latter. Using a specially devised name-interpreter survey combined with a pre-/post-test design, we employ computational social network analysis to examine i) patterns of participants’ out-of-class communication, with a focus on their co-national cohort, and ii) the relationship between these patterns, background factors, and linguistic gains. Our findings reveal that while dominance in Ukrainian vs Russian did not affect progress in the target language, speaking Ukrainian correlated with centrality in the contact network: a reconstruction of the learner networks revealed higher weighted degree centrality among participants declaring Ukrainian as their L1. Russian-dominant speakers in turn often concealed their use of this language, possibly because they were frequently found at the network periphery, suggesting linguistic segregation and marginalisation. The strongest significant predictors of self-perceived progress in both overall proficiency and grammar were level of course enjoyment and two social network measures: the degree of being indicated as interlocutors by well-connected peers (PageRank) and the degree of interaction with Russian-speaking friends. Objectively measured progress, however, instead showed a strong negative correlation with length of residence in Poland. The findings fill a crucial gap in refugee language learning scholarship and have far-reaching implications for the design of language courses tailored to forced migrant groups with varied background dynamics and ideologies.

ABSTRACT. This talk examines how uneven literacy trajectories among migrants from Portuguese-speaking African countries (PALOP) shape their engagement with Portuguese language certification in Portugal. Portuguese, the official language in several African nations, is embedded in diverse sociolinguistic ecologies, often marked by limited access to formal education (Alexandre & Swolkien, 2024; Hagemeijer, 2016). As a result, many migrants have had fewer opportunities to develop academic and assessment-related literacies. These asymmetries influence how candidates interact with language tests and how their linguistic abilities are interpreted. In Portugal, language proficiency is certified by CAPLE–ULisboa through examinations aligned with the CEFR (2020). While the CEFR provides a widely recognized framework for describing language competence, its descriptors assume a homogeneous adult learner profile and familiarity with literacy-based assessment practices. This assumption does not reflect the heterogeneous linguistic and educational backgrounds of PALOP candidates, who may demonstrate strong oral communicative competence in Portuguese regardless the variety while facing difficulties with the literacy demands embedded in test design, written tasks, and instructions. Recent migration flows from countries such as Guinea-Bissau and Cape Verde further expose the mismatch between legal expectations of linguistic readiness and migrants’ lived literacy experiences. In high-stakes exams, this gap raises concerns about fairness and ethical assessment, particularly when literacy constraints are conflated with limited communicative ability. The talk argues for a more context-sensitive interpretation of CEFR descriptors. Drawing on corpus-based analyses of oral (Masotti, 2024) and written productions from the CRPC-África corpus, it identifies characteristic linguistic features and literacy needs of PALOP speakers, with a view to informing more equitable assessment practices. Finally, the talk highlights that a certification of communicative competences tailored to this public may not only improve fairness but also motivate learners to further engage with Portuguese, encouraging learning that values links between Portuguese and their mother tongues.

ABSTRACT. This study examines teachers’ perspectives on pupils’ multilingual repertoires in lower-secondary classrooms in the Czech Republic, drawing on research that conceptualises multilingualism as a dynamic and flexible repertoire of practices (Grosjean, 2019; Cenoz & Gorter, 2015). It focuses on pupils who use three or more languages in daily life, reflecting growing linguistic diversity in Czech lower-secondary education, where approximately 7.4% of pupils use a home language other than Czech (Czech Statistical Office, 2025). The study investigates how teachers perceive and engage with pupils’ multilingual repertoires and how these perceptions inform instructional practices.

Methodologically, pupils created language portraits – visual representations of their linguistic repertoires and the contexts in which different languages are used (Busch, 2012). This method was selected because it enables access to pupils’ subjective and implicit experiences of multilingualism, which are difficult to capture through verbal data alone, while also offering a holistic representation of language use. The portraits functioned as reflective prompts, encouraging teachers to engage with pupils’ linguistic resources in pedagogical reflection.

Semi-structured teacher interviews, using the language portraits as prompts, were conducted to connect pupils’ self-representations with teachers’ interpretations, enabling a contextualised understanding of classroom multilingualism. This methodological design was chosen to triangulate perspectives and to better capture the relationship between pupils’ multilingual practices and teachers’ pedagogical reasoning. Interviews explored teachers’ perceptions of pupils’ language use, perceived benefits of multilingual repertoires, and strategies for integration into teaching. Data were analysed using qualitative content analysis (Mayring, 2014).

Preliminary results indicate that while teachers value pupils’ multilingual resources, they often lack practical strategies for incorporating them into instruction. Language portraits proved useful as both a diagnostic and reflective tool, fostering dialogue between pupils’ perspectives and teachers’ pedagogical decision-making. The study contributes to understanding inclusive classroom practices and highlights ways to support linguistically responsive learning environments (Kramsch, 2006).

ABSTRACT. German multilingual classrooms increasingly require teachers to move beyond monolingual pedagogical norms (García & Wei, 2014), as approximately 29% of students have an immigrant background, while only about 11% of teachers do (Destatis, 2024). While teachers’ beliefs (Burner & Carlsen, 2023) and multilingual identities (De Angelis, 2011) are known to influence multilingual pedagogy, research on specific demographic and classroom-related predictors remains limited.

Addressing this gap, this study investigates:

(1) which demographic and classroom factors associate with teachers’ attitudes and practices

(2) how these factors relate to teachers’ own multilingual identity beliefs.

The quantitative study draws on data from a survey of N = 308 secondary EFL teachers in Lower Saxony, Germany. Multilingual attitudes and identity beliefs were operationalized as separate scales. Multilingual practices were analyzed via exploratory factor analysis, resulting in four dimensions: (1) language awareness, (2) assessment/learner autonomy, (3) multimodal tasks, and (4) language policy. Data were analyzed using non-parametric correlations and group comparisons (Mann–Whitney/Kruskal–Wallis) with Bonferroni-adjusted post-hoc tests.

Preliminary findings indicate that teachers’ multilingual backgrounds - specifically the number of languages spoken and self-perceived multilingualism - show significant associations with identity beliefs (H(6)=16.331, p=.012, ε²=.054) and specific practice dimensions, such as language awareness (H(6)=17.195, p=.009, ε²=.092). Perceived classroom multilingualism also correlates with multilingual attitudes (H(4)=10.455, p=.033, ε²=.035) and language awareness practices. In contrast, gender and formal subject background showed no significant associations.

While many pairwise differences vanished after Bonferroni correction, stable differences remained for self-perceived multilingualism regarding identity beliefs (adj. p=.046, r=.20) and classroom perception regarding cognitive attitudes (adj. p=.023, r=.25).

Overall, results suggest that personal multilingual biographies and classroom perceptions are the strongest facilitators for multilingual pedagogy. The presentation will discuss patterns in relation to existing models (Krulatz, Neokleous & Lorenz, 2023) and offer implications for teacher professional development.

ABSTRACT. This study investigates monitoring behaviour in multilingual writers composing texts in three languages. It examines how text monitoring unfolds in L1, L2, and L3—each mastered at an advanced level—and whether observed differences reflect individual rather than language-specific patterns. Participants were Polish university students enrolled in a Swedish language and literature programme. Their L1 was Polish, L2 was English or German, and L3 was Swedish. Each participant wrote three essays, one in each language. Writing processes were recorded using keystroke logging software, followed by retrospective interviews about pauses and revisions. Results show that multilinguals most often paused during lexical retrieval and conceptual planning, indicating that even advanced writers plan their texts continuously. Fewer linguistic revisions occurred, with greater focus on content modification. Linguistic monitoring—such as checking lexical appropriateness and grammatical accuracy—appeared mainly before transcription. Overall, advanced multilinguals approached text creation similarly across languages, though linguistic-level monitoring was most frequent in Swedish (L3) and least frequent in Polish (L1). Considerable individual variability was observed across most aspects of monitoring behaviour.

ABSTRACT. As populations age and language learning extends beyond formal education and working life, it is vital to understand how motivation for multilingual learning develops and persists in later adulthood (Darnault et al., 2024). Building on this line of inquiry, this paper examines retired language learners’ motivation for developing their multilingual repertoire. Based on language portraits and qualitative interviews with 12 participants in Japan and Norway, the study adopts a life-course perspective on multilingualism, foregrounding learners’ long-term engagement with languages across changing social, educational, and personal contexts. The use of language portraits as visual narratives complements interview data by showing how participants conceptualize and value their multilingual resources.

The thematic analysis identifies three interconnected motivational orientations linked to sustained multilingual learning. The first concerns an interest in linguistic systems and structures, where languages are approached as objects of analysis and intellectual exploration. The second reflects a desire to access and understand the mentalities, values, and lived experiences associated with different linguistic and cultural communities. The third highlights language learning as a social practice, driven by interaction, belonging, and shared meaning-making.

Taken together, the findings point to diverse motivational trajectories among retired language learners and challenge deficit-oriented views of ageing and learning. The study shows how engagement with languages remains meaningful well beyond working life and offers insights into how language education can support lifelong multilingual development.

Darnault, B., Tragant, E., & Lasagabaster, D. (2024). Tracing the dynamics of motivational self-constructs across the lifespan: Tales from senior learners of English. System, 126, 103487. https://doi.org/10.1016/j.system.2024.103487

ABSTRACT. The presence of multiple languages in European classrooms is now an undeniable reality. The shift from monolingual teaching methods to multilingual approaches presents significant challenges for teachers. Recent research in language education has emphasised the importance of multilingual teaching (Cenoz, 2013; Cenoz & Gorter, 2015; Duarte & Meij, 2018; Safont Jordá, 2017). These studies argue that integrating all languages from learners’ linguistic repertoire can facilitate learning the target language. At the same time, the concept of the “Dominant Language Constellation” (DLC), proposed by Aronin (2006), focuses only on the languages that are most salient and functional for the individual (Aronin, 2016; Aronin & Singleton, 2012). In recognition of the pedagogical relevance of DLCs in the teaching of Portuguese as a foreign language (PFL), a study was conducted to characterise the DLCs present in the classroom and to understand how teachers use these configurations as a methodological resource. The study also considered the role of English as a dominant lingua franca and its potential impact on students’ use of other languages. The study was based on an analysis of the linguistic profiles of 100 PFL learners of different levels, as well as a survey of 16 teachers. The research questions are: (i) What are the similarities and differences between teachers’ DLCs and students’ DLCs? (ii) What strategies do teachers use to capitalize students’ DLCs in learning PFL? (iii) How does English as a lingua franca affect the opportunities available to students and teachers to use DLCs in the classroom? The results suggest that teachers’ DLCs tend to be relatively homogeneous, whereas students’ DLCs reveal greater diversity. Given this diversity, teachers believe that using English as a language of mediation between Portuguese and students’ other languages can facilitate learning rather than hinder the promotion of multilingualism in the classroom.

ABSTRACT. Despite growing interest in multilingualism, research on multilingual adolescents remains limited, particularly studies integrating outcomes with the perspectives of educational stakeholders. This cross-sectional mixed-methods study had three main aims: (1) to compare Hebrew (L2) and English (L3) language and literacy skills among immigrant youth in Grades 10, 11, and 12; (2) to examine the contribution of demographic, contextual, and socio- emotional variables to variation in learners’ L2 and L3 outcomes; and (3) to investigate the attributions of educational stakeholders. A parallel test battery was administered to assess receptive vocabulary, morpho-orthographic knowledge, syntactic skills, and writing in Hebrew and English. Students also completed a demographic questionnaire and a psychological-emotional well-being scale. To contextualize these data, semi-structured interviews were conducted with school-based stakeholders examining school practices, beliefs, and attributions related to immigrant students’ linguistic and academic integration. Results revealed a non-linear and language-specific developmental pattern. Hebrew gains were modest and limited primarily to orthography between Grades 10 and 11, whereas English proficiency showed a marked decline among 12th graders across all measures. Length of Residency strongly predicted Hebrew outcomes but had no effect on English. L1 typology influenced Hebrew development, with Latin- and English-background students outperforming Slavic L1 speakers. Hierarchical regressions showed that Hebrew proficiency was driven mainly by exposure variables, while English performance was significantly shaped by emotional engagement and parental aspiration. Qualitative analyses further highlighted the central roles of emotional adjustment, school support, instructional beliefs, and linguistic distance in shaping students’ L2 and L3 trajectories. Integrated analyses illuminate how institutional priorities, exposure conditions, emotional experiences, and stakeholder expectations interact to shape multilingual immigrant adolescents’ language development. Findings underscore the need for differentiated pedagogical approaches, strengthened teacher preparation, and school policies that more effectively support the academic and psychological-emotional needs of trilingual immigrant high-school students.

ABSTRACT. Across Slavic languages, verbal aspect (perfective vs. imperfective) is typically used to distinguish between ongoing from completed events, but its use in habitual contexts (e.g., with adverbials like every day) varies. West Slavic languages (e.g., Czech) allow both aspects in habitual contexts, while Polish, Ukrainian, and Russian only allow the imperfective (Dickey 2000). We conducted an acceptability judgment task paired with pupillometry (Schmidtke 2018, Demberg & Sayeed 2016) to explore children’s sensitivity to this grammatical manipulation. Children assessed sentences produced by a fictional language learner, judging whether they sounded good or bad. Test sentences paired habitual or episodic adverbials (every evening vs. yesterday evening) with telic events using either imperfective or perfective verbs (grandma baked Imp/Pf a sweet cake). Participants were Ukrainian refugee children (ages 7–14) raised bilingually in Ukrainian and Russian in Ukraine, with subsequent immersion in L3 Polish and L3 Czech (n=42 and n=43, respectively). Monolingual Polish and Czech-speaking children were tested as controls (n=15 and n=18, respectively). We hypothesized that Ukrainian children in Czechia, influenced by the societal language, would show weakened sensitivity to ungrammatical habitual+perfective sentences compared to those in Poland. This would manifest as a) higher acceptance of such sentences in offline judgments and b) reduced pupil dilation during processing. Results confirmed our predictions. Czech children accepted habitual+perfective sentences, while Polish children rejected them. In Ukrainian, the results revealed higher acceptance of ungrammatical habitual+perfective structures among Ukrainian children in Czechia vs. Poland (Fig. 1). Age significantly interacted with group and aspect, with a stronger effect of Age in the critical condition (habitual + perfective) for the group in Poland vs Czechia (see Fig.2). Pupil dilation results will be presented at the symposium. These findings suggest first-language grammatical adaptations under immersion, contributing to theories of multilingual acquisition and attrition (Flores 2020; Schmid et al. 2013).

ABSTRACT. This study investigates the acquisition of English as a foreign language (EFL) among child heritage language (HL) bilinguals — a population rarely examined in third language acquisition (L3A) research. We examine whether cross-linguistic influence (CLI) in L3A depends on typological similarity, structural overlap, or further factors. Participants were 118 HL bilinguals in Norway and 87 Norwegian monolingual children aged 8–12. English was an L3 for the HL group and an L2 for the monolingual group. The HLs included Russian (N=30), German (N=25), Hazaragi (N=12), Persian (N=10), Urdu (N=10), and additional languages (N=31). The children completed a written sentence ordering task in English and Norwegian targeting two non-subject-initial declaratives (e.g., Now the boy plays the guitar) and adverb placement (e.g., The girl often builds the castle), two domains for which Norwegian differs from English. The HLs either pattern with English (XSVO, SAVO) or Norwegian (XVSO, SVAO), or are verb final languages (XSOV, SAOV/SOAV). English lexical proficiency was measured separately using the British Picture Vocabulary Scales (Dunn, 2019), and background variables were collected via the Q-BEx questionnaire (De Cat et al., 2025). All groups performed target-like in Norwegian. In English, L2 learners demonstrated high accuracy, but L3 learners differed significantly depending on their HL, even when controlling for age and English proficiency. Specifically, non-facilitative CLI was observed only for HLs that did not overlap with English and were not verb-final. These findings suggest that the HL can be a stronger source of non-facilitative CLI than the societal language in L3 English acquisition. They further indicate a potential role for typological distance. The results refine current L3A models by demonstrating that prior languages do not exert uniform influence and underscore the importance of considering learners’ full linguistic repertoires in pedagogy.

ABSTRACT. Despite a growing interest in morphosyntactic cross-linguistic influence (CLI) in third language (L3) acquisition, there is a scarcity of online (e.g., eye-tracking, event-related potential) studies focusing on incremental processing of features similar to participants' first (L1) and/or second (L2) language. In order to determine which of the previously acquired languages influences the processing of morphosyntax in the acquisition of L3 ab initio, we conducted an eye-tracking study using the miniature language learning paradigm.

Forty-one highly proficient Polish-English bilinguals with no prior knowledge of any Scandinavian language were first taught a few words in Norwegian, and then familiarised with sentences testing adjectival number agreement and indefinite articles. During the main task, they made acceptability judgements on correct and incorrect sentences in Norwegian while their eye movements were recorded. Drawing on evidence from L3 learners with higher proficiency levels (e.g., Abbas et al., 2021), we hypothesised that participants would show CLI from both previously acquired languages.

Reading times (RTs) were analysed as a function of condition and sentence grammaticality using Bayesian mixed effects modelling. The results showed facilitative and non-facilitative CLI from L1 and L2, varying as a function of condition. Participants had shorter RTs for Norwegian correct sentences sharing morphosyntax with Polish or English in comparison to incorrect sentences dissimilar from either language, indicating facilitation from both languages. Yet, when one sentence from a correct-incorrect pair was Polish-like, and the other, English-like, shorter RTs were associated with the Polish-like sentence, irrespective of its grammaticality in Norwegian. This suggests that CLI from L1 overrides that from L2 when both languages are potential sources of CLI.

Overall, the results are compatible with models positing that CLI operates in a property-by-property manner based on structural similarity to both previously acquired languages (e.g., Slabakova, 2017; Westergaard et al., 2017).

ABSTRACT. This study explores how speakers of a linguistic system with an ongoing phonemic merger produce phonological contrasts in their second and third linguistic systems. It examines how predictions of SLM(-r) [4] [5], PAM(-L2) [1] [2] and L2LP [3] [8] can be extended to a context of ongoing sound change in multilingual phonological acquisition. In the context of Chongqing, a city in southwestern China, people acquire Chongqing Mandarin as their first linguistic system (L1), followed by Standard Mandarin (L2), and then English (L3). In Chongqing Mandarin, there is an ongoing merger between /l/ and /n/ [6], but these two sounds are phonemic in Standard Mandarin and English.

40 L1 Chongqing Mandarin speakers participated in /l/-/n/ production experiments. Stimuli contained word-initial near-minimal /l/-/n/ pairs controlled for vowel and tone in Chongqing Mandarin and Standard Mandarin, and controlled for vowel in English. Production materials included 112 monosyllabic words in three linguistic systems. 4474 production tokens were analysed acoustically and modelled with linear mixed-effects modelling and linear discriminant analysis. Pearson’s correlation coefficient was used to explore the correlations between /l/-/n/ production in different linguistics systems.

Results show that Chongqing Mandarin speakers produced almost no difference between /l/ and /n/ in their L1, greater distinction in L2 Standard Mandarin, and the greatest distinction in L3 English. The /l/-/n/ merger is more likely to occur in high-vowel or nasal-coda contexts ([7]). Their L1 and L2 production was biased towards /l/ while their L3 production was not biased towards /l/ or /n/. No significant correlation of the production accuracy among L1, L2, and L3 was found. We argue that the context of multilingual phonological acquisition from a merger in progress is in line with the NEW scenario in L2LP or the Category Goodness assimilation in PAM(-L2). It is also in line with the equivalence classification by SLM(-r).

ABSTRACT. We investigate the sources of phonological crosslinguistic influence in 19 L1 German-L2 English-L3 Spanish and 10 L1 German-L2 English speakers acquiring an artificial language (AL). The goal was to determine whether initial state learners show influence from multiple previously acquired languages, and how an additional spoken language (L3 Spanish) facilitates acquisition. Participants completed picture-naming tasks (Part 1) in German, English, and Spanish (if they spoke Spanish). The results were used to identify sources of influence when analyzing the AL production. Part 2 involved learning the AL over three days. The AL target segments /ɾ-/, /VɹV/, /-r/, /x/ after front vowels, /y/) were based largely on Seediq [1]. Participants learned the AL during 30-minute online sessions on days one and two. They were exposed to pictures and corresponding audios (no orthography) during training phases. These sessions, which included production practice, taught learners (implicitly) the grammar and vocabulary (33 words) of copula sentences in the AL. On Day 3, in a lab, participants produced 38 target sentences in the AL corresponding to images (e.g., /ɡe styɡən nɡanor syn/ the women sick is ‘the woman is sick’). We identified the type of sound produced for each target production in Parts 1 and 2. Each speaker’s Part 2 results were compared to their L1/L2/L3 productions (from Part 1), to identify sources of influence (e.g., [2]). Results revealed facilitative and non-facilitative transfer from all of participants’ previously acquired languages, suggesting that initial state learners use previously acquired structures from multiple languages. While facilitative transfer could reflect rapid learning, the bilingual group had more difficulty with the /x/, /ɾ/, and /r/ (Spanish) targets, indicating a facilitative role of the L3 Spanish in the trilingual group. These findings are best explained by models that assume property-by-property transfer (i.e., [3, 4]).

ABSTRACT. Models of third language (L3) acquisition seek to explain the sources and mechanisms of cross-linguistic influence (CLI), which research has shown to be selective and feature-specific rather than categorical (Rothman, 2010b). Yet dominant L3 models typically assume internally uniform source languages, an assumption rarely tested in multilingual ecologies where the first language (L1) itself exhibits contact-induced phonological stratification (i.e., distinct native and borrowed subsystems). This study addresses this gap by examining L3 English speech production in Santali–Bangla–English trilinguals through the lens of psycholinguistic models of speech production and inhibitory control (Green, 1998).

Santali, an Austro-Asiatic language spoken by an indigenous community in Bangladesh and long in contact with Indo-European languages, exhibits extensive phonological convergence alongside typologically distinctive native features, including word-final checked (often glottalized) stops (Donegan & Stampe, 2004). Descriptive and sociolinguistic accounts identify native and borrowed phonological subsystems ( Choksi, 2020), making Santali a particularly revealing test case for L3 phonological transfer.

The study analyzes acoustic data from word reading, sentence reading, and semi-spontaneous narrative tasks designed to manipulate monitoring and processing demands. Speech data were collected from approximately 24 Santali–Bangla–English trilinguals, with comparison groups of about 12 Santali–Bangla bilinguals and 12 Bangla–English bilinguals. Word-final and word-initial stop consonants in L3 English were analyzed using Praat, with mixed-effects models incorporating task type and lexical frequency as indices of processing load.

Expected outcomes indicate that deeply entrenched native Santali features surface more robustly under reduced monitoring and increased processing load, whereas borrowed features pattern more closely with Bangla-mediated realizations. These findings challenge categorical transfer accounts (e.g., the L2 Status Factor; Bardel & Falk, 2007) and support a psycholinguistic, usage-based view of L3 CLI as emerging from online competition among weighted phonological representations, motivating a refinement of current L3 acquisition models.

ABSTRACT. The increasing linguistic diversity of language classrooms has brought about the need for a paradigm shift in the teaching of English that acknowledges the multilingual repertoire of students as a resource. In this new scenario, there is a demand for research-informed teacher education that prepares teachers to be active agents in transforming course syllabi, classroom practices and learning materials accordingly. Even though previous studies have focused on how teachers interpret, adapt and implement language teaching materials (for example, Harwood, 2014), not much focus has been given to the agentive role of teachers as material designers (Li et al., 2025), particularly in the teaching of English to multilingual learners. This paper addresses such research gap by looking into a case study of an in-service English teacher participating in a teacher training course at a Norwegian university. Following a processual approach to the design of pedagogical materials, the paper investigates how teacher agency emerges while the participant creatively appropriates pre-existing visual digital tools designed to explore multilingualism among schoolchildren (Storto, 2022), adapting and recasting them in a lesson plan that addresses the needs of real students. The data discussed in this paper comprises two commented drafts of the lesson plan, the pedagogical materials therein and a follow-up interview with the participant. The findings illustrate how past experiences, present contingencies, and future aspirations intersect and interact in shaping the teacher’s agentive choices (Priestley, 2015) in the successful design of the pedagogical materials and their integration in the lesson plan. The study contributes to ongoing debates on how teacher agency can support the enactment of multilingual pedagogies in the English classroom and highlights the importance of empowering teachers as designers of innovation, a fundamental yet undertheorized aspect of multilingual pedagogies.

ABSTRACT. Collaboration between language teachers across the school’s language subjects is uncommon (Haukås, 2016, 2025; Haukås & Drange, 2024). This presentation reports on a research project carried out in three Norwegian upper secondary schools, examining how collaboration across language subjects, Norwegian, English, and foreign languages (French, German, and Spanish), can be organised and enacted in practice. The study explored different strategies for strengthening students’ multilingual awareness and for creating greater coherence across the teaching of language subjects.

Two of the schools piloted an intensive collaboration format concentrated within a single week, while the third school chose a longitudinal approach, integrating multiple forms across the entire school year.

The presentation outlines how these models were implemented and compares their respective advantages and challenges. Drawing on multiple data sources, we analyse how the different approaches influenced students’ multilingual identities, their views on multilingualism, and their engagement in language learning, alongside practical considerations related to organisation, time, and resource use. Furthermore, we draw on teachers’ reflections regarding collaborative work, professional learning, and the possibilities for establishing enduring structures for collaboration across language subjects in secondary education. Finally, we reflect on our role as teacher educators and researchers in the process of implementing new collaboration practices.

Drawing on insights from the three cases, we present a conceptual model offering concrete guidance for schools and supporting teacher educators that wish to strengthen collaboration across language subjects in ways that are flexible, educationally meaningful, and sustainable in the long term.

ABSTRACT. Linguistic scholars have been promoting the integration of students’ diverse linguistic and cultural repertoires into the language classroom for decades. Such multilingual language teaching (MLT) approaches are believed to foster effective development of students’ target language proficiency (Cenoz & Gorter, 2021, 2022), alongside their language awareness (Cenoz & Gorter, 2017) and cultural awareness (García, 2017). Nevertheless, the application of MLT approaches is not widely established in secondary educational systems (Jaspers, 2024). One reason for this discrepancy is that language teachers still lack concrete and ecologically valid pedagogical guidelines for tailoring their teaching to the needs of diverse classrooms (Van Beuningen & Polišenská, 2019).

To tackle this issue, the present study aims to formulate theory- and practice-informed design propositions for multilingual language teaching (MLT) practices that facilitate the development of students’ target language proficiency as well as their language awareness, and cultural awareness. Using an educational design research approach (McKenney & Reeves, 2019), a professional learning community existing of language teachers (N=6), teacher educators (N=4) and researchers (N=4) iteratively developed, implemented, evaluated, and re-conceptualized MLT activities and the underlying design principles. In a subsequent validation phase, language teachers (N=12) adapted the resultant MLT activities to their own teaching practice and reflected upon their lessons in guided teacher logs. Student (N=30) interviews additionally captured learners’ experiences with the linguistically and culturally responsive approach. Drawing on these data, we will present the refined design propositions for MLT, and examples of in-class activities embodying these principles. Furthermore, results will illustrate the perceived effectiveness of MLT practices for target language proficiency, language awareness, and cultural awareness development according to teachers and students. In doing so, we offer ecologically valid insights into how teachers and students experience MLT, informing the sustainable integration of linguistically and culturally responsive pedagogy in everyday classroom contexts.

ABSTRACT. Crosslinguistic influence (CLI) in third language (L3) acquisition shows great variability across grammatical properties and learning contexts. While much research has focused on identifying the source of CLI, less focus has been given to learner-related variables that may shape how CLI manifests (Foryś-Nogala et al., 2023). In particular, empirical evidence of the role of L3 metalinguistic knowledge (MTK) remains limited. This study investigates whether property-specific MTK influences accuracy in L3 Norwegian and contributes to CLI patterns. Participants were 49 adults with a Romance first language (L1; Spanish or Brazilian Portuguese) and English as a second language (L2). CLI was assessed through an acceptability task targeting six conditions across syntax, morphology, and semantics: three aligned with Romance structures (bare predicate nominals, gender agreement, and abstract genericity), and three aligned with English structures (expletives, comparatives, and mass nouns). MTK was measured with a multiple-choice task testing explicit knowledge of Norwegian grammatical rules relevant to each condition. Results revealed that L3 MTK was associated with higher accuracy, although its effects varied by grammatical condition and item type. MTK effects were stronger for ungrammatical items, showing that explicit L3 knowledge was particularly relevant when detecting violations. Norwegian proficiency was associated with higher MTK scores across most conditions, indicating that MTK and L3 development jointly shaped performance. Moreover, analytic L2 English proficiency showed negative associations with L3 MTK in Romance-aligned conditions but positive associations with English-aligned conditions, suggesting that L3 MTK was shaped not only by L3 development but also by crosslinguistic similarity patterns. Overall, L3 MTK did not uniformly enhance L3 performance but selectively modulated accuracy across conditions. By showing that MTK effects were strongest in the detection of ungrammaticality and tied to crosslinguistic similarities, the study provides new evidence that explicit L3 rule knowledge contributes to variability in CLI patterns in L3 acquisition.

ABSTRACT. Learning a gendered foreign language can influence conceptual representations of speakers with genderless first languages (L1s) by introducing grammatical gender (GG) distinctions (Athanasopoulos & Boutonnet, 2016). However, when speakers of gendered L1 acquire and are tested in a genderless L2, GG effects are inconsistent across studies (Samuel et al., 2019), often leading to the interpretation that GG does not modulate perception altogether. We argue instead that such inconsistency may stem from previously unaccounted-for differences in foreign language proficiency and dominance. The current study examined whether English, a genderless language (L3), modulated previously found GG effects rooted in two gendered first languages (L1s), Ukrainian and Russian, with partially overlapping GG systems.

In Experiment 1 (categorisation), sixty-four Ukrainian-Russian-English multilinguals rated how similar object-character pairs were (e.g., an object shown alongside a male/female figure). Objects had either matching (e.g., candle - feminine in both) or mismatching GG across L1s (e.g., beetroot - masculine in Ukrainian, feminine in Russian). In Experiment 2 (an associative learning task), a separate sample of ninety-four multilinguals learned and recalled object-name pairs (e.g., Alexander/Alexandra-candle), analogously varying in GG congruency. Analyses available at OSF: https://osf.io/efwt4/?view_only=e22c481aa47e426faba6f0f8398dcad6.

Mixed-effects models revealed that categorisation patterns were modulated by L3 proficiency (Cambridge University Press, 2024) and self-rated dominance (Gertken et al., 2014): greater English scores predicted weaker GG effects of gendered L1s. Individual-difference analyses revealed that L3 variables such as years of education, length of residence, and use in family settings best predicted restructuring outcomes. No corresponding effects were found in recall, suggesting domain-related restructuring.

Overall, our findings suggest that variability observed in previous GG studies may reflect differences in multilingual proficiency and dominance rather than the complete absence of GG effects. Importantly, they highlight that L3 experience can attenuate conceptual patterns even from closely related first language(s), revealing how cumulative multilingual experiences reshape cognition.

ABSTRACT. This Study examines how Chinese learners of German construe and verbalize motion events when acquiring German as a third language (L3) after Chinese (L1) and English (L2). Motion events descriptions constitute a highly sensitive domain of linguistic conceptualization, as they reflect typological preferences in event encoding and therefore provide a privileged testing ground for cross-linguistic influence in multilingual acquisition. The study addresses three interconnected research research dimensions. First, it describes the patterns of motion event encoding produced by Chinese learner of German across proficiency levels (RQ1). Second, it compares these patterns with those of native speakers of German in order to identify learner-specific preferences and divergences (RQ2). Third, it investigates the sources of these divergences by systematically examining (a) learners’ multilingual language background (Chinese–English–German), (b) proficiency in L2 and L3, and (c) event-internal properties such as motion type and path type (RQ3). The theoretical framework integrates Talmy’s typological model of motion events with insights from cognitive construction grammar, allowing for a principled analysis of how motion-related semantic components are distributed across constructions in use. Empirically, the study adopts an experimental elicitation design in which participants produce controlled motion descriptions. The data are annotated for semantic, syntactic, and information-structural features and analyzed using a combination of qualitative and quantitative approaches to capture both systematic patterns and learner-specific variation. Standardized language proficiency tests are used to ensure comparability of learners’ proficiency levels both in L2 (English) and L3 (German), while language background questionnaires provide additional information on learners’ multilingual experience. By integrating the conceptual structure and semantic dimensions of motion events from a cognitive-typological perspective with a construction-grammatical analysis, the study offers a more fine-grained account for cross-linguistic influence and event conceptualization in third language acquisition.

ABSTRACT. Research on third language acquisition has produced a substantial empirical literature, yet there is no consensus on the role of transfer in L3 acquisition. Competing models disagree on whether the L1, L2, or typologically closest language underlies L3 grammars (e.g. Angelovska&Hahn, 2017; Puig-Mayenco et al., 2020), often postulating incompatible interpretations of comparable data. Although more recent approaches allow for selective contributions from multiple prior languages (e.g. Slabakova, 2017; Westegaard et al., 2017) they continue to frame development in terms of transfer. Here we argue that the ongoing disagreement reflects a conceptual rather than an empirical problem or flawed data. Multilingual acquisition has been framed primarily as a problem of source attribution, an approach that tends to obscure the mechanisms of grammatical development. We propose that L3 acquisition requires a theory of development that goes beyond surface resemblance to prior languages. Building on the Grammatical Mapping (GM) paradigm (Lust, 2012; Fernández-Berkes&Flynn, 2023), we argue for a UG-constrained, structure-dependent process of grammatical computation that is continuous with L1 and L2 acquisition. Under this view, previously acquired languages constitute repositories of grammatical options available to the learner. These options are neither ranked nor copied wholesale into the L3 grammar. We claim that cross-linguistic influence is a by-product of development, rather than its driving force. It is interpreted as emergent outcomes of GM rather than transfer of individual properties. This approach is believed to explain convergent developmental trajectories across typologically distinct learners (Flynn et al., 2004), the staged emergence of complex syntactic structures (e.g. CP-related phenomena: Berkes&Flynn, 2012), and systematic error patterns (Flynn&Berkes, 2017). We conclude that reframing prior linguistic knowledge as a developmental resource rather than a transfer source yields a more explanatory and unified account of L3 acquisition. Furthermore, it provides a principled basis for reinterpreting cross-linguistic influence in multilingual research.

ABSTRACT. While contemporary research on L3/Ln acquisition has primarily focused on learning processes, comparatively little attention has been paid to real-time language processing (see Lago et al., 2019, for discussion). This gap is particularly evident in the domain of (morpho)syntax, where questions remain regarding how previously acquired languages influence the online computation of grammatical features in L3/Ln.

The present study investigates whether prior knowledge of L2 German affects gender agreement computation in Danish as an L3/Ln by examining the production of bare nouns and indefinite article-noun sequences. Background information was collected via a detailed questionnaire assessing self-reported German proficiency, exposure, and use on Likert scales.

Participants were 49 Polish university students learning Danish as an L3/Ln (with intermediate to advanced proficiency, as assessed via Dialang) who had previously learnt German as an L2 in school. They completed a picture naming task with 96 non-cognate Danish nouns, producing bare nouns in the first block and indefinite article-noun sequences in the second. Items were controlled for length, frequency, concreteness, and imageability. Response times for valid trials were analysed using linear mixed-effects models in R, testing the effects of German-related variables and their interactions with Phrase Type.

Results showed significantly faster response times for bare nouns than for article–noun sequences. Crucially, Phrase Type interacted with German Exposure (see Figure 1), but not German Proficiency. Participants with higher German exposure showed faster response times when producing article-noun sequences (p = .014), but not bare nouns (p = .314).

These findings are interpreted as evidence for facilitated gender agreement computation in L3/Ln Danish as a result of exposure to L2 German, a language that marks agreement within the NP using formally similar devices.

References Lago, S., Mosca, M., & Stutter Garcia, A. (2021). The role of crosslinguistic influence in multilingual processing: Lexicon versus syntax. Language Learning.

ABSTRACT. Using Visual World eye-tracking, we examine how bilingual German-Russian and Greek-German children process English aspectual forms, comparing them to age- and proficiency-matched controls. The languages involved exhibit distinct aspectual systems. Russian and Greek differentiate between perfective and imperfective aspects, associating them with completed and ongoing events, respectively. English uses past progressive for ongoing events in the past [1] while present perfect of accomplishment verbs in English are strongly associated with completed event interpretations [2]. German lacks grammaticalized aspect and the semantics of its present perfect forms has been argued to be broader, allowing for both perfective and imperfective readings [3]. We adapted the paradigm from [1] to test whether the aspectual systems of the participants' previously known languages influence their processing of L3 English present perfect forms. Data collection with German-Russian bilingual children residing in Germany (n=39) and German-Greek bilinguals residing in Greece (n=32)has been analyzed (Age=8-13y.). The results revealed significantly preference for the Completed event interpretation in the Greek-German group vs German-Russian group (82% vs 73%, controlled for overall English proficiency and age in modelling). Gaze patterns align with offline preferences (see Fig. 1). We hypothesize that the difference between the groups can be connected to the role of dominance: thus, non-facilitative influence from the non-aspectual language German is less strong in the group where German in the heritage language (Greek-German group in Greece) vs when it is the societal language (German-Russian group in Germany). Data collection has been completed with age-matched controls without the knowledge of German and is ongoing with L1 German controls. Full data analysis will be presented at the conference. We discuss our findings in relation to contemporary L3 acquisition models [4, 5], compare them to previous studies [6-9] and propose an possible model of how language dominance fits with similarity-driven co-activation of grammatical representations.

ABSTRACT. While research on L3 phonological acquisition has expanded considerably over the past decade, studies examining the relationship between perception and production remain limited. This study investigates the development of the perception and production of final obstruents in L2 English and L3 Polish in three adult L1 German learners (aged 22–32) over the first year of instructed L3 learning. The three languages differ in allowing voiced coda obstruents (English), or having coda obstruent devoicing (German, Polish). Grounded in Complex Dynamic Systems Theory (CDST), the study conceptualises multilingual phonological development as a non-linear and individualised process. Data were collected at eight testing points (T1-T8: 4-40 weeks) for L2 English and seven for L3 Polish, with T1 excluded for Polish due to the early stage of L3 learning. Perception was tested using a forced-choice goodness task, while production was elicited through delayed repetition, picture naming, and storytelling. The accuracy analyses of the perception task revealed striking individual developmental differences: one learner improved in L3 perception while declining in L2, another showed decreasing accuracy in both languages, and the third maintained stable L2 perception while developing more accurate yet increasingly variable L3 perception. Analyses of the production data, based on auditory assessment of accuracy by three phonetically trained listeners, are ongoing. The final analyses will employ within-subject effect sizes for perception and production differences in mean scores, alongside moving correlations. Gut et al. (2023) analysed L2 and L3 rhotic perception-production development in the same dataset, reporting aligned development in the two skills for this feature and both languages. Given the differing perceptual salience and articulatory demands of final obstruent (de)voicing compared to rhotics, distinct perception-production relationships are expected. The findings will offer new insights into the complexity of individual L3 speech development and the dynamic interplay between multilinguals’ linguistic (sub)systems.

ABSTRACT. In third language phonetics and phonology, voice onset time (VOT) has been a popular choice to examine cross-linguistic influence. In studies examining progressive transfer, the general finding of speakers whose L1 and L2 belong to two distinct language classes has been that the third language falls between average L1 and L2 VOT values at the group level. This finding has been reported in a variety of language combinations (Wrembel, 2011, 2014; author, 2024)

One potential issue with this approach is that it often does not focus on individual data, despite evidence that individual differences play an important role in the acquisition and outcomes of late-learned languages (e.g. Dewaele, 2009). It is a possibility that the L3 falls between the L1 and the L2 due to a 50:50 ratio of participants whose L3 is either L1 or L2-like, but this cannot be directly examined with existing data due to low within-participant statistical power.

The present work is a simulation in which synthetic data sets for hypothetical L3 VOT experiments were generated. The purpose of this work was to investigate how distinct combinations of underlying participant-types (truly intermediate L3, L1-like or L2-like L3) influenced the group model, which could change narrative conclusions. Additionally, an analysis of the optimal post-hoc test and quantity of data points per person per language was investigated. Initial results show an overwhelming probability of intermediate models, in line with previous studies, but that this narrative conclusion is compatible with a variety of underlying participant types (Figure 1). The optimal post-hoc test and quantity of observations depended on effect size and the number of total participants in the data set (Figure 2). Taken together, these findings suggest that fewer participants with more data per language can uncover whether and to what degree individuals differ in phonetic influence.

ABSTRACT. Cognate facilitation is established for lexical learning and access (e.g., Ecke & Hall, 2021), yet evidence for phonological learning is mixed, learners often shifting toward the non-target language; for example, English-speaking students may produce the t in Spanish taco with more aspiration than in a non-cognate word (Flege et al. 1998). This talk has two objectives: we examine cognate effects and cross-linguistic influence in a postlexical phonological process in advanced multilinguals and evaluate how different perception paradigms capture sensitivity to a non-contrastive phonetic cue. First, we report a production study with fourteen advanced multilingual speakers (L1 Hungarian, L2 English, and L3 Spanish) targeting two voicing-related phenomena: Spanish voiced stop lenition (LEN), which requires learners to acquire a new allophonic pattern, and English word-initial voiceless stop aspiration. Test words are non-cognates; double cognates (English–Spanish); and triple cognates (English–Spanish–Hungarian). Results reveal a complex picture, especially compared to previous studies (Amengual, 2021). In the English interlanguage of participants, the facilitative effect of L3 Spanish could not counterbalance the inhibitory effect of L1 Hungarian, while the L3 Spanish data show no cognate effects. We hypothesise that this can be explained by typological markedness and sociophonetic factors. Second, we offer a methodological contribution to perception research by examining whether different experimental paradigms converge when testing sensitivity to aspiration (VOT) in English /p/. Building on SLM/SLM-r, we ask whether Hungarian short-lag VOT promotes equivalence classification that impedes formation of an English long-lag category, while also considering individual differences in auditory precision. We compare three complementary measures within the same learner group: (1) VOT-focused accentedness evaluations, (2) category-goodness ratings along a VOT continuum to test internal category structure, and (3) an oddity (AXB) task assessing fine-grained discrimination. By comparing these tasks, we show how multiple methods lead to clearer conclusions about learners’ perception. Data analysis is ongoing.