Abstract We investigated priming of Dutch prepositional phrase (PP)‐final and ‐medial passives in a within‐Dutch structural priming experiment in French–Dutch bilinguals. Unlike Dutch, French allows only the PP‐final passive. We asked whether L2 learners can be primed to produce structures that are dissimilar to their L1 (i.e., PP‐medial passive) and whether this depends on L2 proficiency. We observed an unexpected production preference for the PP‐medial passive, which we traced to explicit classroom instruction form preferences. This instructional history, interacting with learners' age of L2 acquisition and evolving implicit expectations about the two passive variants, shaped both production biases and priming effects: Lower proficiency learners mostly produced PP‐medial passives and showed strong PP‐final priming, whereas higher proficiency learners produced both forms, creating room for PP‐medial priming. We conclude that proficiency effects in L2 structural priming likely reflect the interplay between acquisition history, instruction, and implicit expectation adaptation rather than proficiency alone.
We investigated the influence of orthographic transparency, and learners' awareness of it, on the second language (L2) phonolexical encoding of Brazilian Portuguese (BP) mid-vowel contrasts. In BP, accent marks indicate vowel quality (mid-closed vs. mid-open), resulting in transparent grapheme-phoneme correspondences, whereas words without accents are opaque. To test whether transparency benefits phonolexical encoding of vowel quality, 20 native speakers of American English learning BP completed a perceptual categorization, a lexical decision, and an orthographic knowledge task. Learners perceived the difference between the vowel quality contrasts but confused the vowels lexically. The learner group more easily rejected nonwords created from real words carrying accent marks than nonwords based on words without accents. However, when separating learners into low- and high-awareness groups, the effect remained only for the high-awareness group. The results suggest that orthographic transparency can result in more precise phonolexical encoding, but only for those who are orthographically aware.
Abstract As language assessment increasingly recognizes interaction as socially co‐constructed, interactional competence (IC) has emerged as a central construct for understanding learners’ ability to manage communication across social and cultural contexts. At the same time, digitally mediated communication environments and artificial intelligence (AI)‐powered conversational agents are transforming both interaction itself and the ways in which it can be assessed. These technologies also create new opportunities for second language research more broadly by facilitating data collection from geographically dispersed and historically underrepresented learner populations. This introduction reviews key challenges in IC assessment, including tensions among construct validity, interactional authenticity, reliability, and practicality, and discusses how digital technologies offer new opportunities for scalable, standardized, and pedagogically meaningful assessment. It synthesizes the six empirical and conceptual studies and one commentary included in the special issue and concludes by outlining future directions for research, including expanding linguistic and sociocultural diversity, integrating discourse analytic approaches, reconceptualizing IC for AI‐mediated interaction, and exploring automated scoring methods.
Interactional competence (IC) is a core component of oral proficiency; however, opportunities to practice and develop IC in second language (L2) classrooms are often limited. Generative artificial intelligence chatbots provide easily accessible avenues for communicative practice. Employing a quasiexperimental design, this study compared chatbot interactions with face-to-face peer interactions to investigate the potential of both modes for eliciting and developing English as a foreign language (EFL) learners' IC. The participants were 88 Chinese college EFL learners: 48 in the chatbot group and 40 in the peer group. Learners' IC performance was evaluated through a pretest and a posttest in conjunction with conversation analysis. Findings revealed that both groups improved their IC, but the peer group excelled in interactive listening. Further questionnaire and interview data indicated learners' concerns about the authenticity of chatbot interactions; yet they appreciated the chatbot's flexibility and anxiety-reducing effect. They also expressed a willingness for long-term engagement.
Do immersion and nonimmersion learners' English grammaticality judgment test (GJT) scores reflect the same underlying processes in language learning? Drawing on data from Chen and Hartshorne's (2021) study, we argue that they do not. Using generalized additive mixed models (GAMMs), we found that age of onset was the strongest predictor of GJT performance among immersion learners, whereas linguistic distance was the most influential variable for nonimmersion learners. Moreover, the nonlinear effects of age of onset, length of experience, education, and linguistic distance differed substantially between the two groups. Female learners consistently outperformed male learners across both contexts. Similarly, learners residing in an English-speaking environment scored higher on the GJT than those living in non-English-speaking countries.
