ABSTRACT. A long-standing question in bilingual word recognition is whether processing a word in one language automatically activates lexical information in another. The Bilingual Interactive Activation Plus (BIA+) model (Dijkstra & Van Heuven, 2002) posits that such cross-language activation occurs automatically through a fully interconnected architecture, in which sub-lexical and lexical orthographic, phonological, and semantic representations are simultaneously activated across languages. Empirical studies with bilinguals have consistently demonstrated cross-language co-activation. However, this research has focused almost exclusively on bilinguals typically those whose languages share a common writing system. Therefore, it remains unclear whether the principles of non-selective lexical access extend to speakers navigating three typologically distinct writing systems, leaving a critical gap in our understanding of trilingual lexical processing. To address these gaps, the present study investigates whether, and to what extent, Tibetan–Chinese–English (TCE) trilinguals activate Tibetan (L1) and/or Chinese (L2) orthographic and phonological representations during English (L3) word processing, and how such activation is modulated by L2 and L3 proficiency. This population offers a compelling testing ground because Tibetan (abugida), Chinese (logographic), and English (alphabetic) employ fundamentally different grapheme–phoneme correspondence rules. Sixty-two TCE trilinguals completed a translation recognition task in which L1 or L2 primes and L3 targets were either translation equivalents, orthographically related, phonologically related, or unrelated. Reaction times (RTs) and accuracy rates (ACC) were analyzed using linear and generalized linear mixed-effects regression models in R software. The results revealed two primary findings: First, L1 orthographic information was automatically activated and interfered with L3 word recognition, with the interference diminishing as L3 proficiency increased; second, L2 phonological information began to influence L3 processing as L3 proficiency rose. These findings contribute to the understanding of cross-language orthographic and phonological activation within a cross-script trilingual context, and highlight the moderating role of L3 proficiency on cross-language lexical activation.

ABSTRACT. This study examines the L3 initial acquisition of Spanish alveolar taps and trills among Mandarin speakers with L1 and English as their L2. We explore how perception accuracy, phonetic environments, and speech styles (clear vs. casual) influence the production of these consonants. Three primary questions are addressed: (1) How does perception accuracy predict production? (2) How does speech style affect both production and perception? (3) How do phonetic environments impact acquisition difficulty?

Theoretically, we test Flege and Bohn’s (1995, 2021) Speech Learning Model (SLM & SLM-r), which suggests a relationship between perception and production. However, the findings by Rose (2010) indicate a misalignment, suggesting that more variables are involved in speech learning. Regarding phonetic environment, we test Recasens’ (1991) claim that high front vowels [i] hinder trill production. Finally, we hypothesize that clear speech aids perception but may impair production due to aerodynamic interference.

The methodology employs AXB discrimination tasks for perception and imitation tasks for production, using non-word stimuli across both speech styles. Data were acoustically analyzed via Praat and modeled using linear and ordinal Bayesian Regression.

Results indicate a misalignment between domains; participants achieved high accuracy in perception but struggled to produce the required occlusions. Phonetic environment analysis revealed that the high back vowel /u/, rather than [i], significantly lowered both perception and production accuracy, particularly in intervocalic positions. This contradicts prior claims regarding the incompatibility of high front vowels with trill dynamics. Speech style effects were primarily observed through complex interactions with other variables, though substantial individual differences suggest the need for larger sample sizes. Ultimately, these findings highlight the complexity of L3 phonology and suggest that pedagogy should integrate perception and production training across diverse phonetic environments and speech style contexts.

ABSTRACT. Studies on L2 speech acquisition show that influence from L1 can make bilinguals sound different from monolinguals (Mennen et al., 2020). According to the Speech Learning Model (Flege, 1995), L2 sounds that are acoustically similar tend to merge with existing L1 categories, resulting in cross-linguistic influence (CLI). However, research on trilingual speech remains limited, particularly across typologically distant languages (Wrembel, 2015). This study investigates the sources of CLI in L3 English vowel production by comparing Uyghur-Mandarin-English trilinguals (n=25) and Mandarin-English bilinguals (n=20). Participants completed read-aloud tasks targeting monophthongs in controlled bilabial and alveolar contexts, as well as an elicited imitation task to evaluate their language proficiency. The first and second formants of target vowels were extracted, normalized using the Lobanov (1971) z-score method, and analyzed via Linear Mixed-Effects Models (LMMs). Additionally, Mahalanobis distances between L3 English vowels and corresponding L1/L2 vowels were calculated to assess acoustic distance. The LMMs revealed a highly significant Group × Vowel interaction across both the F1 and F2 dimensions. Post-hoc analyses showed distinct, dimension-specific patterns of acoustic divergence between the groups. For example, vowels /æ/ and /ɔ/ shows a significant difference between groups in F1 but no significant difference in F2. Conversely, the vowels /ɪ/, /ʊ/, and /u/ show a significant difference in F2 but no significant difference in F1. Vowels /ɑ/ /ɑ˞/ /i/ /e/ /ʊ/ /ʌ/ are significanlty different between groups in both F1 and F2 dimensions. These results suggest that L1 Uyghur exerts a significantly stronger cross-linguistic influence (CLI) on L3 English vowel production than L2 Mandarin, and that this phonetic transfer operates on a feature-by-feature basis. Furthermore, Mahalanobis distance measurements revealed that L1 Uyghur vowels exhibit a much shorter acoustic distance to the L3 targets than L2 Mandarin vowels, which suggests that stronger CLI stems from the acoustically closest L1 Uyghur vowels.

ABSTRACT. In the last two decades, the traditional monolingual ideology of foreign language teaching is being challenged as growing societal and educational multilingualism has received increasing attention from applied linguistics (Jiang & Zhang, 2023). As a result, traditional English Medium Instruction (EMI) programs are being reassessed based on new insights into multilingual learners' language development. Pedagogical translanguaging (Cenoz & Gorter, 2021), as a relatively new multilingual paradigm that emphasizes multilingual speakers' comprehensive use of their linguistic resources, may provide new perspectives for foreign language teaching in settings that were traditionally monolingual. This paper focuses on writing in a foreign language in Chinese higher education, as it is a complex productive skill that challenges students (Fang et al., 2023). To explore teaching and acquisition of these complex skills, we compared translanguaging and EMI pedagogies in 5 classes of learners following an English writing class at a Chinese university. With an experimental design including questionnaires and students' compositions, we explored the differences between the translanguaging and EMI groups in terms of their writing performance, motivation and anxiety. The results indicated that: 1) the translanguaging group scored higher in writing than the EMI group; 2) students were more motivated and less anxious post-intervention; 3) linear regression analysis indicated that higher motivation and lower anxiety were associated with better performance; 4) the effects of EMI and translanguaging teaching methods on performance were fully mediated by motivation and anxiety. The study illustrates how EMI writing classes can be enriched by a pedagogical translanguaging perspective.

ABSTRACT. Recent studies show that the degree of cross-linguistic orthographic and phonological similarity, as well as modality, can modulate cognate word recognition (see, e.g., Frances et al., 2021). However, most research has focused on the recognition of L1–L2 cognates, and cross-linguistic influence from one non-native language to another remains relatively underexplored. The present study examines how the degree of orthographic and phonological similarity, modality, and language proficiency jointly influence the recognition of L2–L3 cognates and non-cognates in an L3.

The participants in the present study are late learners of L3 French across different proficiency levels, with L1 Finnish and knowledge of L2 English. They will perform visual and auditory lexical decision tasks in French with English–French cognates and non-cognates. To gain a more comprehensive understanding of the role of form similarity, continuous orthographic and phonological cross-linguistic similarity scores between English and French are included for both cognate and non-cognate words. Accuracy rates and reaction times from the lexical decision tasks will be analyzed using linear mixed models.

The study is positioned within current theories and models of bi- and multilingual word recognition, such as Multilink (Dijkstra et al., 2019), and may motivate further developments in current theoretical accounts. It aims to contribute novel empirical evidence on less well-studied L3 learners by examining the effects of proficiency, modality, and the degree of cross-linguistic similarity together.

ABSTRACT. Self-efficacy beliefs, understood as learners’ convictions about their ability to meet task-related demands (Mills, 2014), constitute a central predictor of success in language learning. Additionally, self-efficacy is closely intertwined with learners’ heightened behavioral, cognitive, affective, and social task engagement (Hiver et al., 2024). In this regard, plurilingual approaches (Candelier et al., 2012), which systematically draw on learners’ entire plurilingual repertoires to promote language learning, can be expected to create new substantial opportunities for the development and enactment of learners’ self-efficacy than the still predominantly monolingual routines of traditional language teaching. Game-based language learning (Reinhard, 2019), which draws on the motivational affordances of games, offers additional potential to support such processes.

Building on these theoretical considerations, this paper for the PhD workshop reports initial findings from a doctoral project examining a physical intercomprehension-based mystery escape game for German learners of French. The study investigates (1) the extent to which a gameful, plurilingual intervention can foster intercomprehension self-efficacy, and (2) how plurilingual learning and interaction unfold through task engagement. While research on intercomprehension, as a specific form of plurilingual practice, has documented positive effects of drawing on related languages on students’ motivation (Bär 2009), its potential to foster self-efficacy—especially in dynamic, game-based learning settings—remains scarce.

A mixed-methods design was employed to capture both learner perceptions and observable interaction. Quantitative data stem from pre- and post-test questionnaires (N=61) assessing learner characteristics, intercomprehension self-efficacy, and task engagement. Qualitative data include audio-recorded game interactions (N=30) and semi-structured interviews (N=10). Interactional data were transcribed and deductively coded using a task engagement framework (Iwashita et al., 2025), while interview data were analyzed through structuring content analysis.

Initial analyses indicate that the intercomprehension-based, game-based intervention created a highly engaging plurilingual learning environment: It was associated with increases in learners’ intercomprehension self-efficacy, accompanied by sustained task engagement.

ABSTRACT. Whether there exists inhibitory control in multilingual comprehension as in production? “Modality-specific” hypothesis posits a fundamentally different control mechanisms for production and comprehension. However, the absence of inhibition in comprehension may be due to the low task demands—typical language-switching comprehension tasks (e.g., lexical decision) lack the active language selection required in production tasks (e.g., color-cued picture naming).

To test this, we introduced a novel comprehension task with active language selection. 64 trilingual (Tibetan-Chinese-English) participants judged if auditory digit words (in Tibetan, Chinese, or English) match on-screen digits, in both language (cued by color) and concept (Experiment 1). A control condition used a low-demand task judging only conceptual matches (Experiment 2). Both experiments employed the n-2 repetition paradigm.

In Experiment 1, a significant n-2 repetition cost (an indicator of inhibition) was observed overall, with post-hoc tests confirming this inhibitory effect for both L1 Tibetan and L2 Chinese, though not for L3 English. In contrast, Experiment 2 showed no significant n-2 repetition cost overall. Interestingly, a reversed n-2 repetition benefit (i.e., faster RTs in repetition trials) emerged specifically for L3 English, potentially indicating language activation rather than inhibition.

The findings demonstrate that inhibitory control operates during multilingual comprehension and is recruited flexibly based on task demands. This pattern challenges the “modality-specific” hypothesis and supports a “modality-general” account of language control.

ABSTRACT. While the generalizability of cognitive advantages of bilinguals is debated (Von Bastian et al., 2016), Bialystok & Craik (2022) argue that bilinguals differ in resource allocation in complicated attention tasks. To date, bi- vs. trilingual comparisons yielded no behavioral differences (Hayakawa & Marian, 2019), however, previous studies defined multilingualism differently. Further, although bilinguals show higher cue attentiveness (Cristia et al., 2012), which fades in noise (Morini & Newman, 2019), the impact of multilingualism on accented speech comprehension remains understudied. To address these research gaps, this project explores the extent to which the degree of multilingualism impacts both attention performance and the comprehension of accented speech. Specifically, we aim to employ demanding attention and speech comprehension testing tasks, measuring multilingualism as a continuous variable using the Multilingual Language Diversity score (MLD; Li et al., 2020; Berthele, 2022). The study involves 120 participants, differing in the number of languages they use daily, controlled for L2/L3 proficiency (LexTale; Lemhöfer & Broersma, 2012), and working memory (backward digit span). The battery of tests features a lexical decision task, a sentence comprehension task, and the task-switching tests using visual and audio stimuli (Stoet et al., 2013; Mora-Plaza et al., 2022), designed using the PsyToolkit platform (Stoet, 2017). The recordings for the linguistic tasks were normed for comparability in accentedness and comprehensibility. The experimental procedure was refined based on a pilot study. The data collection is ongoing and currently focused on participants who use one or two languages daily. Participants with higher MLD scores are predicted to show lower switch and mixing costs in the task switch paradigm. The difference is also expected in accented speech processing. If the accent acts as "noise," multilingual participants may perform worse, however, if the accent is treated as novel linguistic input, they may show a multilingual advantage.

ABSTRACT. Recent research on multilingualism has called for moving beyond monolingual and English-centred perspectives to explore how motivation, emotion, and identity develop across multiple languages and sociocultural contexts. However, longitudinal studies adopting a Complex Dynamic Systems Theory (CDST) perspective remain limited, particularly among ethnic minority students in China. This doctoral project addresses this gap by investigating the dynamic development of multilingual motivation and emotion among Tibetan undergraduates who study English as an L3 alongside an additional language other than English (LOTE), such as German or Japanese, as an L4. Guided by CDST (Larsen-Freeman, 2017), the study draws on multilingual adaptations of the L2 Motivational Self System (Dörnyei, 2005; Henry, 2017) and existing research emphasising the role of emotion in language learning (Dewaele & MacIntyre, 2014). Rather than focusing on predefined emotions, the study adopts an exploratory perspective on emotional experience in multilingual learning. It examines how motivation and emotion develop, interact across languages, and contribute to multilingual identity formation. A longitudinal mixed-methods design will be employed with second- and third-year Tibetan students majoring in English at an inland Chinese university. Data will be collected at three time points during one academic semester. The study combines Q methodology, language portraits, and semi-structured interviews to explore changes in motivational and emotional experiences. Q methodology will identify motivation profiles and developmental changes, while language portraits and interviews will provide deeper insights into learners’ emotional experiences and interpretations of change. Within a CDST perspective, integrating these methods aims to capture the dynamic and evolving nature of multilingual development. Feedback is particularly welcomed on the feasibility of the longitudinal design, the integration of motivation and emotion within a CDST-informed framework, and the suitability of combining Q methodology, language portraits, and semi-structured interviews to investigate multilingual development over time.

ABSTRACT. Current multilingualism literature has focused primarily on the influence of the first language (L1) on the second (L2), or on cognitive transfer from L1/L2 ​​to L3 (e.g., Rothman’s Typological Primacy Model, 2011). However, the retrospective transfer mechanism has not been sufficiently addressed. According to the Cumulative Enhancement Model (Flynn et al., 2004), all previously and currently learned languages ​​contribute to a person's linguistic repertoire. In this context, we suggest that learning a language with an article system that has high morpho-syntactic transparency (like der/die/das in German) increases one's metalinguistic awareness. Since Turkish does not have a distinct article system, positive transfer from a third language to a second language (English) in terms of articles is possible. Therefore, learning can transform article usage in English from a meaningless, language-specific rule into a universal requirement. In this study, participants will be divided into two groups. The first group will consist of bilingual individuals who can speak Turkish and English. The second group will consist of multilingual individuals who are proficient in a language with more than one definite article, such as German, in addition to Turkish and English. The groups will first be assessed in terms of their English proficiency. The aim is to prevent the difference between the groups from stemming solely from their English knowledge. Participants ultimately will be presented with sentence completion and grammar assessment questions. The questions will focus on the use of definite articles in English. Accuracy between the groups will be the dependent variable. The results of the experiment could be helpful in understanding the mechanisms of cognitive transfer between languages. The project may broaden discussions on typological and metalinguistic awareness related to the dynamic expansion of linguistic repertoire.

ABSTRACT. This study examines how students in a trilingual educational context (Basque, Spanish, and English) experience learning English across key educational stages, with a particular focus on the transition from primary to secondary education. Situated within the Basque Country's multilingual schooling system, where students navigate multiple languages simultaneously, this research asks: How does the transition from primary to secondary school shape the linguistic self-perceptions, motivation, and anxiety of students learning English as an L3 in a multilingual Basque context?

Adopting a multimodal narrative framework (Kalaja & Melo-Pfeifer, 2019; Lingard de la Vega et al., 2023; Mäntylä et al., 2022), we collected 66 visual and written narratives from students in their final year of baccalaureate studies. Participants were asked to reflect retrospectively on their English learning journey, with particular attention to how their multilingual background intersected with their experience of changing educational contexts. Data were analyzed inductively through thematic analysis, following an iterative coding process that resulted in four thematic categories: space, emotions, teacher roles, and methodologies.

The findings reveal that the primary-to-secondary transition constitutes a critical juncture in students' multilingual trajectories: participants frequently described shifts in their emotional engagement with English, changes in their sense of belonging in the language classroom and contrasting pedagogical approaches across educational stages. These findings highlight how the management—or mismanagement—of educational transitions can reinforce or disrupt students' developing multilingual identities.

We discuss the pedagogical implications for supporting learners during transitional periods in multilingual contexts, and argue for teaching practices that foster belonging, engagement, and motivation across educational stages. This study contributes to current discussions on inclusive multilingual education by foregrounding student voices as essential to promoting equity and well-being in the language classroom.