Statistical regularities can be acquired from usage. To examine language speakers’ statistical metacognition about multiword expressions (MWEs), we collected ratings for frequency, dispersion, and directional association strength of English binomials from L1, advanced and intermediate L2 speakers. Mixed-effects modeling showed all speakers had limited speaker-to-corpus consistency but significant sensitivity to statistical regularities of language, supporting usage-based (Gries & Ellis, 2015) and statistical learning theories (Christiansen, 2019). Their statistical metacognition was also shaped by word-level cues, consistent with dual-route model (Carrol & Conklin, 2014). Despite similarities, frequency metacognition showed the strongest speaker-to-corpus consistency, while dispersion metacognition was the hardest to develop. Advanced L2 speakers showed the greatest speaker-to-corpus consistency and sensitivity, while lower-proficiency speakers relied more on word-level cues in metacognitive judgments, supporting the shallow-structure hypothesis (Clahsen & Felser, 2006). Overall, L1 and L2 speakers develop diverse statistical metacognition, with L2 speakers not necessarily inferior, suggesting that statistical metacognition is not solely shaped by usage-based experience.

Fluency is an essential aspect of second language (L2) oral proficiency. Recent studies have demonstrated that L1 individual speaking style is connected to L2 fluency, suggesting that L2 speech fluency does not solely represent L2-specific skills. Furthermore, task mode (monologue vs. dialogue) has been shown to influence fluency. The present study examines the extent to which these two factors (L1 speaking style and task mode) can predict L2 speech fluency, and how such connections are modified by the learners’ L2 proficiency level. The data consist of monologic and dialogic speech samples from 50 advanced students of English in their L1 (Finnish) and L2 (English). The samples were analyzed for speed, breakdown, repair, and composite fluency. The results of multiple linear regressions demonstrated high predictive power for speed, breakdown, and composite fluency dimensions, while the model for repair fluency showed weak predictive power. The results have implications for L2 fluency research.

The love factor in the field of second language acquisition has gained considerable traction since the turn of the century. This article is the first to take a variationist perspective to investigate how multilingual coupledom affects sociolinguistic development in the second language (L2). Participants were 76 users of L2 German living in Austria, all of whom were in a romantic relationship with an Austrian partner. We analyzed the effects of multilingual coupledom on self-reported changes in learners’ use of, attitudes toward, and proficiency in standard German, the Austrian dialect variety, and first language(s), and whether (psycho-)social variables moderate this relationship. Individual differences in psychological and social variables (e.g., adaptability, Open-mindedness, length of residence, orientation toward the Austrian dialect) predicted reported changes in the sociolinguistic repertoire. Qualitative analysis revealed a blended operation of socioaffective and exposure-related factors, which helped explain why, how, and for whom multilingual coupledom affects (socio-)linguistic development.

This paper introduces the Chinese Learner English Corpus (CLEC), comprising argumentative texts written by Chinese lower and upper secondary school students. CLEC expands learner corpus research by including texts from intermediate-level learners and rich metadata on their backgrounds, including engagement with self-initiated, so-called Extramural English (EE) activities outside the classroom. To illustrate potential uses, two case studies are presented. The first uses a keyword analysis to reveal thematic and stylistic differences between CLEC and its Swedish counterpart, SLEC, highlighting linguistic priorities related to distinct learning contexts. The second investigates lexical bundles associated with gaming, demonstrating how EE engagement might influence learners’ use of multiword units. Freely available online, CLEC facilitates contrastive interlanguage analysis and supports further research into L2 learning and use, particularly regarding the role of language exposure. The corpus is also a valuable resource for teacher trainees aiming to deepen their understanding of SLA processes.

While statistical learning of adjacent constructions is well-documented in SLA, our knowledge of this cognitive mechanism concerning nonadjacent constructions remains limited. To address this, we investigated the acquisition of Mandarin predicate-argument constructions containing the preposition duì. Specifically, via a corpus-based approach, we probed whether learners’ core predicate use within these nonadjacent constructions mirrors the patterns of frequency and contingency in their natural language input. Our findings show that learners’ usage aligns with target language distributional regularities, which is consistent with statistical learning. However, our study underscores the necessity of going beyond a sole focus on distributional factors within learners’ input to more fully comprehend L2 production choices and the intricacies of statistical learning. This includes examining variables that shape learners’ exposure to input, such as input accessibility, proficiency, and prototypicality. Finally, we demonstrate the suitability of mixed-effects negative binomial regression to effectively address non-normality and overdispersion in linguistic data.

This study revisits the relationship between second-language (L2) learners’ ability to distinguish sounds in non-native phonological contrasts and to recognize spoken words when recognition depends on these sounds, while addressing the role of methodological similarity. Bilingual Catalan/Spanish learners of English were tested on the identification of two vowel contrasts (VI) of diverging difficulty, /i/-/ɪ/ (difficult) and /ɛ/-/æ/ (easy), in monosyllabic minimal pairs, and on their recognition of the same pairs in a word-picture matching task (WPM). Learners performed substantially better with /i/-/ɪ/ in VI than in WPM, and individual scores were only weakly correlated. By replicating previous findings through a more symmetrical design, we show that an account of prior work rooted in methodological dissimilarity is improbable and provide additional support for the claim that accuracy in sound identification does not guarantee improvements in word recognition. This has implications for our understanding of L2-speech acquisition and L2 pronunciation training.

Non-native languages tend to be acquired through a combination of explicit and implicit learning, where implicit learning requires coordination of language information with referents in the environment. In this study, we examined how learners use both language input and environmental cues to acquire vocabulary and morphology in a novel language and how their language background influences this process. We trained 105 adults with native languages (L1s) varying in morphological richness (English, German, Mandarin) on an artificial language comprising nouns and verbs with morphological features (number, tense, and subject-verb [SV] agreement) appearing alongside referential visual scenes. Participants were able to learn both word stems and morphological features from cross-situational statistical correspondences between language and the environment, without any instruction. German-speakers learned SV agreement worse than other morphological features, which were acquired equally effectively by English or Mandarin speakers, indicating the subtle and varied influence of L1 morphological richness on implicit non-native language learning.

While much discussion has focused on what researchers do and should do in second language proficiency assessment, less attention has been given to why persistent trends continue. This study investigated second language acquisition (SLA) researchers’ beliefs, reported practices, and decision-making rationales regarding proficiency assessment. Using an online survey, we collected responses from 111 SLA researchers. Findings revealed that while researchers generally endorsed recommended methodological standards, practical constraints—such as time, accessibility, and ease of administration—frequently influenced their reported practices. A consistent belief–practice gap emerged across several key areas. Notably, reduced redundancy tests were rated favorably for both validity and practicality, reflecting a growing shift toward efficient, validated tools. These findings suggest that although methodological awareness is high, practical barriers continue to challenge the adoption of more rigorous proficiency assessment practices in SLA research.

The present study investigated if/how individual differences in heritage language (HL) experience modulate gender agreement processing among Spanish heritage speakers (HSs). We reanalyzed the data from Luque and colleagues (2023), which reported an aggregate biphasic N400–P600. The present analysis revealed that sensitivity to morphological markedness was positively modulated by HL proficiency and exposure/use. Higher proficiency led to increased P600 across markedness conditions—the typical signature of L1-dominant processing—while increased Spanish exposure/use resulted in increased N400 for Default Errors—a signature attested only in HSs in this domain. Formal instruction led to increased N400 but reduced P600 for Feature Clash Errors. We interpret these results to suggest that the N400 reflects a morphophonological pattern-matching strategy with some HSs relying (more) on this mechanism as Spanish exposure and use increases. Markedness also modulated the relative engagement of pattern-matching (N400) versus automatic grammatical processing (P600), depending on the transparency/saliency of morphophonological patterns.

Selecting appropriate texts for second language (L2) learners is essential for effective education. However, current text difficulty models often inadequately classify materials for L2 learners by proficiency levels. This study addresses this deficiency by employing the Common European Framework of Reference for Languages (CEFR) as its foundational framework. A cohort of expert English-L2 educators classified 1,181 texts from the CommonLit Ease of Readability corpus into CEFR levels. A random forest model was then trained using 24 linguistic complexity features to predict the CEFR levels of English texts for L2 learners. The model achieved 62.6% exact-level accuracy across the six granular CEFR levels and 82.6% across the three overarching levels, outperforming a baseline model based on three existing readability formulas. Additionally, it identified shared and unique linguistic features across different CEFR levels, highlighting the necessity to adjust text classification models to accommodate the distinct linguistic profiles of low- and high-proficiency readers.

Al-Hoorie et al. (2024) have described L2 Motivational Self System research as having a “validation crisis.” In this response, I offer a brief history of recent motivation research to contextualize how we reached this point. I then discuss measurement invariance testing, the purpose of which is to ensure that the proposed measurement model holds across groups or time. Crucially, invariance testing is a necessary precursor to subsequent analyses because if invariance is not upheld, it is impossible to know whether differences reflect true underlying differences in the latent construct or are simply the byproduct of a changing measurement model. I argue that by combining measurement invariance testing with open science practices, we can create and validate better measurement models while cultivating a better understanding of their scope of application.

In recent years, the L2 Motivational Self System has faced increasing scrutiny over its theoretical clarity and empirical rigor. One element of this model, the L2 Learning Experience, remains ambiguously defined and theoretically underdeveloped. This study examined the content validity of the L2 Learning Experience scale and its potential overlap with intrinsic motivation, a cornerstone of self-determination theory. Using a panel of experts, we assessed to what extent items traditionally associated with the L2 Learning Experience scale align with their intended construct. Findings revealed that the items were predominantly identified as intrinsic motivation, not the L2 Learning Experience. These results suggest a significant overlap between the two constructs and raise concerns about a potential jangle fallacy. Our results also underscore the need for greater theoretical and terminological clarity in the field. Aligning language learning motivation research with broader psychological frameworks could lead to more parsimonious and robust theoretical models.

Al-Hoorie, Hiver, and In’nami (2024) challenged the validity and corresponding validation processes of L2 Motivational Self System (L2MSS) research. A component of this challenge included claims of weak discriminant validity due to high correlations among L2MSS constructs. Papi and Teimouri (2024) countered by using semi-partial correlations to control for other L2MSS constructs, finding weak-to-moderate associations, which they claimed mollified potential discriminant validity concerns. In this methods-oriented response paper, we present a historical case that semi-partial correlations should be viewed within the context of multiple regression analysis, not as a standalone bivariate metric. Challenging Papi and Teimouri’s approach, we suggest that their method does not adequately address discriminant validity issues. Furthermore, when their semi-partial correlations are treated as multiple regression models, Al-Hoorie et al.’s concerns remain valid. Finally, we demonstrate that L2MSS literature does not support the assignment of outcome and predictor variables in Papi and Teimouri’s semi-partial correlations when correctly considered as multiple regression models.

To investigate the effects of vocabulary practice timing on lexical learning and neural processing during communication tasks, we engaged 80 Japanese learners of English (40 pairs) in information-gap tasks with vocabulary practice in the pre-task or post-task phase. Learning of verb-noun combinations was orally assessed through translation and narrative tasks one week later. To quantify neural coupling between participants during task-based interaction, hyperscanning with fNIRS was used to measure inter-brain synchronization (IBS). Results showed that pre-task vocabulary practice led to greater learning, while post-task practice resulted in higher IBS in the brain region underlying language processing. Better vocabulary practice performance predicted more automatic use of multiword expressions in the post-task condition. IBS in the brain regions underlying social cognition and language processing predicted multiword learning. These findings reveal how practice timing influences neural synchronization and language acquisition, highlighting the importance of balancing lexical learning with communication processes in task-based language teaching.

In this rejoinder, we provide a historical overview of the emerging critiques of the L2 Motivational Self System and examine the structural and conceptual factors that have perpetuated these unresolved issues. As our analysis shows, a core concern is that the L2 Motivational Self System lacks clear falsifiability criteria, making it difficult to evaluate or revise in light of contradictory evidence. Despite numerous inconsistent or null findings, there appears to be no threshold at which core assumptions are reconsidered. We argue that advancing the field requires a renewed commitment to falsifiability, where constructs are subjected to empirical scrutiny and can, in principle, be shown to be wrong. Beyond technical matters, we acknowledge the emotional and professional challenges involved in confronting evidence that undermines familiar frameworks. We advocate for a shift toward greater theoretical precision, methodological transparency, and openness to critique.

Word processing during reading is known to be influenced by lexical features, especially word length, frequency, and predictability. This study examined the relative importance of these features in word processing during second language (L2) English reading. We used data from an eye-tracking corpus and applied a machine-learning approach to model word-level eye-tracking measures and identify key predictors. Predictors comprised several lexical features, including length, frequency, and predictability (e.g., surprisal). Additionally, sentence, passage, and reader characteristics were considered for comparison. The analysis found that word length was the most important variable across several eye-tracking measures. However, for certain measures, word frequency and predictability were more important than length, and in some cases, reader characteristics such as proficiency were more significant than lexical features. These findings highlight the complexity of word processing during reading, the shared processes between first language (L1) and L2 reading, and their potential to refine models of eye-movement control.

A large body of literature has examined perceptual training, especially using the high variability phonetic training (HVPT) technique, where multiple talkers are included in the training set to help learners develop more accurate additional (second) language (L2) speech sound categories. Yet, most experimental studies focus on relatively short-term gains using a pre-post–delayed design, providing limited insight into longer-term training effects and how the timing of training might regulate its effectiveness. To begin addressing this gap, we implemented HVPT at two contextually relevant windows of opportunity during a university study program. Thirty-six first (native) language Spanish students participated in this study. Students were randomly assigned to two groups. One group (G1) received training at the beginning of their study program, which coincided with the onset of intensive L2 exposure; the second group (G2) received training in the second year, while enrolled in an English phonetics and phonology course. Both groups completed four HVPT sessions (identification tasks) focusing on a set of challenging L2 English vowels (/iː ɪ æ ʌ ɜː e ɒ ɔː/). Perception was measured at four testing times (in years 1 and 2, before and after HVPT) with identification tasks. The results showed that HVPT had a positive impact regardless of the timing of its implementation. However, students also improved outside of training, which suggests that intensive language study can facilitate some perceptual learning.

Al-Hoorie, Hiver, and In’nami (2024) offer compelling arguments for why L2 motivational self-system research is currently in a state of validation crisis. Seeking a constructive resolution to the crisis, in this response we argue that two fundamental conditions are needed for the field to emerge stronger: psychological readiness and methodological maturity. For psychological readiness, we call for a reframing of the “crisis” narrative. We highlight the need to value controversy, to normalize failure and (self-)correction, and to resist the allure of novelty. For methodological maturity, we suggest that an argument-based approach to validation can provide a constructive solution to current controversies. We present an integrated framework that can guide systematic validation efforts, and we demonstrate its application using a recent validation study as an example.

In a close replication study of Darcy et al., (2016), Huensch (2024) reported a lack of clear relationships between inhibitory control (IC) and phonological processing, contrary to the initial findings. Given the general unreliability of response-time differences, which are often the basis of IC measures and could potentially mask small effects, we performed secondary analyses on Huensch’s (2024) open data set to investigate (a) the extent to which the reliability of IC measures could be improved using model-based approaches (Hui & Wu, 2024), (b) the correlations between the different IC tasks, and (c) their predictive power for phonological processing, based on the more reliable indices. Results showed that model-based approaches generally improved reliability, and particularly for the Stroop and Simon tasks to acceptable levels. Yet, correlations between IC tasks remained low, and partial correlation and hierarchical regression still failed to reveal significant relationships between IC and phonological processing, further confirming Huensch’s (2024) findings.