A large dataset of word recognition behavior from nonnative speakers (NNS) of English was collected using an online crowdsourced lexical decision task. Lexical features were used to predict NNS lexical decision latencies and accuracies. Predictors of NNS latencies and accuracy included contextual diversity, age of acquisition, and contextual distinctiveness, while length moderated the impact of contextual diversity and neighborhood size on accuracy. Results have implications for second language word recognition and demonstrate that NNS behavioral data collected through large crowdsourcing projects can afford a rich source for SLA research.

The current study examined the occurrence and benefits of linguistic alignment in two modalities, face-to-face (FTF) and synchronous computer-mediated communication (SCMC), focusing on stranded prepositions in relative clauses. It further examined how learner characteristics (i.e., working memory, language proficiency, previous knowledge of the target structure) mediate the effects of linguistic alignment. Ninety-four Korean students were assigned to one of the following groups: FTF alignment, SCMC alignment, FTF control, and SCMC control. The alignment experimental groups completed two alignment sessions, finished three stranded preposition tests, and carried out a running span test and cloze test over three weeks. Results indicated not only that linguistic alignment occurred in both FTF and SCMC modes but also that alignment was facilitated significantly more in the SCMC than FTF interactions. Furthermore, the findings suggest immediate and delayed learning effects in both modalities, and that learners’ prior knowledge of the target structure was significantly associated with the occurrence of alignment.

We test the interpretation of pronominal forms in L2 speakers of English whose L1 is Spanish. Previous research on learners of nonnull subject languages has shown conflicting results. The aim of the present study is to reconcile previous evidence and shed light on the factors that determine learners’ difficulty to interpret pronominal forms in the L2. In six comprehension experiments, we found that intermediate L2 speakers did not show increased difficulty compared to native speakers in integrating multiple sources of information (syntactic, discourse, pragmatic) to resolve ambiguous pronouns in intrasentential anaphora and cataphora conditions. However, we also found that when two referents with equal prominence are introduced using a conjoined noun phrase in the preceding context, the learner’s performance is significantly different than the performance of the native speakers, both in intrasentential and intersentential anaphora. We suggest that L2 speakers may encounter difficulties evaluating the salience of the antecedents during pronoun resolution.

This study explored the impact of contextualized practice on second language (L2) learners’ production of wh-questions in the L2 classroom. It examined the quality of practice (correct vs. incorrect production) and the contribution of declarative knowledge to proceduralization. Thirty-four university-level English as a foreign language learners first completed a declarative knowledge test. Then, they engaged in various communicative activities over five weeks. Their production of wh-questions was coded for accuracy (absence of errors) and fluency (speech rate, mean length of pauses, and repair phenomena). Improvement was measured as the difference between the first and last practice sessions. The results showed that accuracy, speech rate, and pauses improved but with distinct patterns. Regression models showed that declarative knowledge did not predict accuracy or fluency; however, declarative knowledge assisted the learners to engage in targetlike behaviors at the initial stage of proceduralization. Furthermore, whereas production of accurate wh-questions predicted accuracy improvement, it had no impact on fluency.

Elicited imitation tasks (EITs) have been shown to be a valid, reliable, and practical method for establishing second-language (L2) oral proficiency across a variety of languages. Nevertheless, research that has validated existing EITs has most often explored a narrow range of learner proficiencies or has examined how well the EIT distinguishes between low- and high-level learners. Bowden’s (2016) validation study of the Spanish EIT was the first to expand the range of learner proficiencies by including low, advanced, and very advanced L2 learners; her results suggested that, in its current state, the EIT may not be equipped to distinguish finer-grained levels of proficiency, especially at more advanced levels. The present study revisits the Spanish EIT to examine (a) its ability to distinguish between higher level learners and (b) whether including additional, more challenging EIT items allows for finer-grained distinctions at higher levels. Results suggest that the addition of longer, more challenging EIT items can improve the discriminability of the test among more advanced L2 learners.

Despite extensive theoretical and empirical research, we do not have estimations of the magnitude of sensitivity to grammatical information during L2 online processing. This is largely due to reliance on null hypothesis significance testing (Plonsky, 2015). The current meta-analysis draws on data from one elicitation technique, self-paced reading, across 57 studies (N = 3,052), to estimate sensitivity to L2 morphosyntax and how far L1 background moderates this. Overall, we found a reliable sensitivity to L2 morphosyntax at advanced proficiencies (d = .20, 95% CIs .15, .25), with some evidence that this was reliably lower than for native speakers (NSs). These patterns were not generally moderated by linguistic feature or sentence region. However, effects for anomaly detection were larger among NSs than L2 learners and the effects among L2 learners appeared to show a trend toward L1 influence. Finding smaller effects than in other subdomains, we provide an initial framework of reference for L2 reading time effect sizes.

Based on writing produced by second language learners at different proficiency levels (CEFR A1 to C1), we adopted a usage-based approach (Ellis, Römer, & O’Donnell, 2016; Tyler & Ortega, 2018) to investigate how German and Spanish learner knowledge of 19 English verb-argument constructions (VACs; e.g., “V with n,” illustrated by he always agrees with her) develops. We extracted VACs from subsets of the Education First-Cambridge Open Language Database, altogether comprising more than 68,000 texts and 6 million words. For each VAC, L1 learner group, and proficiency level, we determined type and token frequencies, as well as the most dominant verb-VAC associations. To study effects of proficiency and L1 on VAC production, we carried out correlation analyses to compare verb-VAC associations of learners at different levels and different L1 backgrounds. We also correlated each learner dataset with comparable data from a large reference corpus of native English usage. Results indicate that with increasing proficiency, learners expand their VAC repertoire and productivity, and verb-VAC associations move closer to native usage.

Language-analytic ability, or the ability to treat language as an object of analysis and arrive at linguistic generalizations, is at the core of the constructs of language learning aptitude and metalinguistic awareness, which are implicated in our ability to learn explicitly. In the context of child second language (L2) learning, it has been argued that children learn primarily implicitly and that the most important component of aptitude may be memory ability. However, no empirical research to date has investigated the relationship and development of aptitude and metalinguistic awareness longitudinally as well as examined their predictive power for children’s L2 achievement in the classroom. In a study with English-speaking learners aged 8–9 (N = 111), we found that although aptitude and metalinguistic awareness were (still) dynamic, they significantly predicted children’s achievement in L2 French. Moreover, language-analytic ability proved to be the component with the strongest predictive power. This finding suggests that it may not be level of cognitive maturity alone that determines children’s approach to L2 learning; experiencing explicit, form-focused instruction may foster the role of language-analytic ability even in children as young as 8–9 years.

The current study examines how second language (L2) users differentially assess the comprehensibility (i.e., ease of understanding) of foreign-accented speech according to a range of background variables, including first language (L1) profiles, L2 proficiency, age, experience, familiarity, and metacognition. A total of 110 L2 listeners first evaluated the global comprehensibility of 50 spontaneous speech samples produced by low-, mid-, and high-proficiency Japanese speakers of English. The listeners were categorized into two subgroups according to a cluster analysis of their rating scores: lenient and strict. Results showed that while the lenient listeners appeared to rely equally on many linguistic areas of speech during their judgments, the strict listeners were strongly attuned to phonological accuracy. Analysis of the background questionnaire data revealed that more lenient listeners likely had higher levels of awareness of the importance of comprehensibility for communication (metacognition); regularly used L2 English in professional settings (experience); and had L1s more linguistically close to the target speech samples, Japanese-accented English (L1-L2 distance).

This research report examines the occurrence of listener visual cues during nonunderstanding episodes and investigates raters’ sensitivity to those cues. Nonunderstanding episodes (n = 21) and length-matched understanding episodes (n = 21) were taken from a larger dataset of video-recorded conversations between second language (L2) English speakers and a bilingual French-English interlocutor (McDonough, Trofimovich, Dao, & Abashidze, 2018). Episode videos were analyzed for the occurrence of listener visual cues, such as head nods, blinks, facial expressions, and holds. Videos of the listener’s face were manipulated to create three rating conditions: clear voice/clear face, distorted voice/clear face, and clear voice/blurred face. Raters in the same speech community (N = 66) were assigned to a video condition to assess the listener’s comprehension. Results revealed differences in the occurrence of listener visual cues between the understanding and nonunderstanding episodes. In addition, raters gave lower ratings of listener comprehension when they had access to the listener’s visual cues.