WBW G/552 and online (please email olivia.knapton@kcl.ac.uk to receive Teams link)

Speaker 1: Pau Sitja Marquez

Intonational acquisition of Catalan polar questions by Italian learners: challenges and victories

This study examines the acquisition of Catalan polar questions by Italian speakers, with the aim of assessing the influence of L1 dialectal background (Romanesco and Sicilian) and L2 proficiency level (A and B). This research is especially timely given the recent surge of interest in Catalan studies across Italy, as well as the large Italian expatriate community currently residing in Barcelona who are in daily contact with the language. The corpus consists of 285 spontaneous yes-no questions produced by 38 learners and a control group of 62 native speakers. The Melodic Analysis of Speech framework is used to describe the parameters of first peak, body, final inflection and tonal range, and the data are tested using linear mixed-effects models. The results reveal systematic divergences in the interlanguages, including a lower frequency of the first peak, flat bodies with less prominent tonal inflections—often shifted towards pretonic vowels—compression of the tonal range and insufficient final rises. However, the Sicilian group (B) shows a systematic convergence towards the native pattern, suggesting a possible influence of dialectal background and historical factors of linguistic contact. Overall, these findings contribute to a better understanding of intonational development in phonetic acquisition at diverse stages of the interlanguage and confirm that it constitutes a structured and evolving system, sensitive to both dialectal background and level of proficiency. Since the phonetic features identified constitute a distinct foreign accent, the results of this study offer direct practical value. They are highly applicable for training Catalan teachers and developing targeted didactic materials.

 

Speaker 2: Weiman Chen

Individual versus Paired Interaction with a Voice-Based GenAI Chatbot: Effects on Chinese EFL Learners’ Oral Proficiency, Affective Responses, and Language-Related Episodes

The rapid proliferation of GenAI chatbots in EFL education has generated a growing body of empirical research, with studies consistently demonstrating positive effects on learners’ linguistic outcomes, affective states, and behavioural engagement. The majority of this research, however, has focused on individual learner-AI interaction, leaving the paired configuration, in which two learners collaborate alongside a shared AI interlocutor, comparatively underexplored. Whether the two configurations produce meaningfully different learning outcomes remains an open empirical question with direct pedagogical implications.

Drawing on Vygotsky’s sociocultural theory, the AI chatbot is positioned as a cognitive mediating tool that extends learners’ Zone of Proximal Development, providing the kind of assisted performance that supports language development. The individual-AI configuration leverages this mediating function directly, while the paired-AI configuration additionally incorporates peer collaboration, grounded in Swain’s collaborative dialogue framework, which holds that knowledge is actively co-constructed through joint problem-solving and explicit attention to language form. Comparing the two configurations thus allows the study to examine not only whether AI-mediated interaction promotes oral language learning, but whether the social dimension of peer collaboration introduces a cognitive added value that individual AI interaction alone cannot replicate.

The study adopts a quasi-experimental mixed-methods design involving three intact classes of Chinese EFL undergraduates enrolled in a Business English Listening and Speaking course at a university in Guangzhou, China. Over a 16-week intervention, one class engages in individual one-to-one voice-based interaction with the GenAI chatbot Doubao (IND condition), a second class completes the same tasks collaboratively with a partner alongside the chatbot (PAIR condition), and a third class serves as the control group (CG), completing equivalent tasks through traditional peer-based pair work without chatbot involvement. Three research questions guide the investigation: the differential effects of the two AI-mediated configurations on oral proficiency development; their differential effects on foreign language anxiety and enjoyment; and how the two configurations differ in terms of language-focused interactional engagement during oral task performance.