Listeners whose first language has fixed stress often struggle to perceive word stress in a second language (L2), a phenomenon known as "stress deafness." This study investigates how L2 proficiency and musical perception skills jointly influence Czech learners' perception of English lexical stress. Fifty-four Czech adults, representing a wide range of English proficiency (measured by LexTALE) and musical perception skills (measured by PROMS-S), completed an AXB discrimination task with English pseudowords differing segmentally and in stress placement. Mixed-effects logistic regression revealed main effects of both L2 vocabulary size and musical perception ability, as well as their interaction: highly accurate stress discrimination required high scores for both measures. Unlike much previous research, which often found no effect of L2 proficiency, the present study did find one, possibly because the effect was disentangled from individual differences in auditory acuity by including musical aptitude measures. Analysis of the musical subtests showed that temporal (tempo, rhythm) and dynamic (beat) perception predicted stress discrimination, whereas tonal and spectral skills did not. These results demonstrate that temporal auditory acuity enhances the perception of L2 stress, particularly when combined with robust L2 lexical knowledge, highlighting the interplay of linguistic and domain-general auditory factors in L2 prosodic learning.

Traffic time-series forecasting faces significant challenges from varying data complexity and domain-specific temporal patterns that existing transformer approaches fail to address through fixed architectural configurations. This article introduces the adaptive complexity-aware traffic transformer (ACTFormer), a novel framework that establishes adaptive processing strategies based on data characteristics. ACTFormer contributes four key innovations: entropy-based complexity analysis for traffic pattern quantification, differentiable adaptive vocabulary selection via Gumbel-Softmax relaxation, enabling end-to-end optimization, traffic-aware contextual encoding capturing domain-specific dependencies, and comprehensive experimental validation demonstrating effectiveness across diverse scenarios. The adaptive mechanism intelligently matches computational resources to problem difficulty, with high-complexity scenarios benefiting from large vocabularies (1024 tokens), achieving 15.6% performance gains. Comprehensive experiments across six benchmark tasks (PeMS04, PeMS07, PeMS08, METR-LA, NYCTaxi Drop-off, and NYCTaxi Pick-up) demonstrate superior performance against 34 baseline methods; on the PeMS and NYCTaxi tasks, ACTFormer achieves consistent 8.7%-14.6% mean absolute error (MAE) improvements over the strongest transformer (STGAFormer) baseline, with strong correlation (0.84) between complexity scores and selected vocabulary sizes. ACTFormer's computational efficiency (1.181 M parameters and 29.49 s training time) enables immediate deployment in intelligent transportation systems (ITSs) while establishing adaptive complexity-aware processing as a fundamental principle for transformer architectures.

This article presents the results of a pedagogical experiment on the development and testing of a comprehensive model for fostering foreign language academic speaking skills in non-linguistic major students. The relevance of the study is driven by the growing demands of academic and professional communication, which require modern specialists to confidently master the skills of publicly presenting scientific concepts in a foreign language. The theoretical foundation of the research is the theory of the cognitotype, understood as a mental-linguistic schema that organizes a text according to a frame structure and reflects the genre features of an utterance. Based on this theory, the author proposes an original five-level methodology aimed at the sequential formation of key competencies: conceptual (working with and transforming scientific information), genre-compositional (assimilating structural models of scientific texts), linguistic-stylistic (mastering the linguistic means and clichés of the academic style), metacognitive (conscious use of knowledge about scientific discourse for text production), and pragmatic-communicative (effective application of the developed skills in real situations of professional communication, including the use of visual aids and argumentation). The pedagogical experiment, conducted at the Novosibirsk State Technical University (NSTU), confirmed the high effectiveness of this integrative approach. A comparative analysis of the results from the control and experimental groups showed that students trained according to the proposed model demonstrated significant improvement in their skills of composing scientific presentations in a foreign language, increased adherence to genre criteria in their speeches, and enhanced the quality of visual support usage. The most notable progress was observed in aspects such as compliance with the norms of the academic style and the logical structure of the presentation. At the same time, expanding their active vocabulary, despite the overall positive outcome, required students to undertake additional targeted work. In conclusion, it is argued that the developed model provides a targeted and effective means of preparing non-linguistic major students for successful oral scientific presentations in a foreign language.

ABSTRACTIncreasing evidence suggests that interactive feeding is an important variable in language development, with early disruptions in feeding potentially having long‐term consequences. This longitudinal experiment characterized early patterns of feeding difficulties and language development in full‐term, typically‐developing infants across the developmental milestones of 8, 12, 18, 24, and 54 months. Parent‐directed assessments were used to assess feeding difficulties and language abilities between 8 and 24 months and a clinical language assessment was administered at 54 months. Results revealed that certain feeding difficulties, such as poor saliva control at 18 months, correlate negatively with expressive vocabulary size and may signal co‐occurring and later occurring less developmentally advanced language skills at 18 and 24 months. Others, such as food selectivity, are characteristic of children with less developmentally advanced language skills at 24 months. Together, these findings suggest that feeding difficulties may reduce opportunities for language stimulation and/or reflect underlying oral‐motor developmental vulnerabilities that also affect language development.

The advancement of digital intelligence technologies presents new opportunities for business English education. Their integration not only enriches course content but also introduces novel teaching methods such as online learning platforms and virtual reality. To further enhance teaching effectiveness, this paper designs a deep learning-based student behavior recognition and teaching effectiveness evaluation mechanism (LBREM). This mechanism first employs deep learning algorithms to analyze student behavior, then evaluates teaching effectiveness based on this analysis. Instructors adjust teaching approaches according to the evaluation outcomes, ultimately improving business English learning outcomes. The object detection model (YOLOv5s-Ghost-D4) and human keypoint detection model (AlphaPose) used in this mechanism demonstrate high recognition performance. When applying the LBREM method for learning effectiveness evaluation, the results show minimal deviation from manual evaluations. Based on the digital-intelligent teaching model and LBREM method, student learning outcomes showed significant improvement: business vocabulary size increased by 155.2%, oral fluency by 26.2%, listening comprehension rate by 29%, writing proficiency by 18.7%, and business adaptability by 20.8%.

Importance. Generative artificial intelligence tools are actively replacing tools that we have been accustomed to using for decades in all areas of life. For example, the issue of using neural networks as a teacher’s aid, and in particular their ability to check written foreign language papers, has been raised in the educational environment for several years. Despite a fairly large number of studies devoted to this problem, teaching students of non-linguistic specialties to write professional essays based on practice with AI was not the subject of separate works. The purpose of the study: to develop a methodology for teaching students of an agricultural university to write professional essays based on practice using generative AI and to test its effectiveness. Materials and Methods. Students ( N = 94) of Voronezh State Agrarian University named after Emperor Peter the Great participated in this study in the fields of study 03.21.02 “Land Management and Cadastre” and 03.35.04 “Agronomy”. To test the effectiveness of the author’s methodology, eight stages of learning are developed within the framework of the study: 1) the formative stage; 2) working with active vocabulary; 3) discussing the structure of the essay; 4) goal setting; 5) writing an essay plan; 6) interaction with generative AI; 7) revision of the essay text and 8) discussion and reflection. Results and Discussion. In the framework of this study, it is noted that the author’s method of teaching agricultural university students writing professional essays based on practice using generative artificial intelligence is effective in mastering learning skills.: a) composition of the essay structure ( t = 4.19 at p = 0.0005); b) use of grammatical means; c) use of active vocabulary ( t = 3.3 at p = 0.0001). When students mastered the skills to argue their position ( t = 1.77 at p = 0.041) and achieve text coherence ( t = 1.77 at p = 0.041), there is minimal difference with the traditional teaching methodology. Conclusion. The novelty of this research lies in the development of a methodology for teaching students of an agricultural university writing professional essays based on practice using generative artificial intelligence (Deepseek). The learning algorithm proposed by the authors has been tested and proved its effectiveness in comparison with the traditional teaching method, which may serve as a prospect for its use in other non-linguistic areas.

Importance. The active introduction of artificial intelligence tools and technologies into all spheres of human life is significantly transforming the traditional methods of teaching foreign languages. There is an emerging pedagogical triad “teacher – artificial intelligence – learner”, where neural networks act as an active third party in the process. Artificial intelligence automates routine operations, helps with the search for materials, data processing, visualization, formatting of sources and evaluation of texts, thereby freeing up time for tasks requiring deeper human thinking. Neural networks provide instant feedback from educational and social to conditionally creative. Their integration is particularly effective for developing students’ writing skills, including writing essays in a foreign language. Materials and Methods. The study involved students ( N = 48) of the Russian Presidential Academy of National Economy and Public Administration in the field of preparation 41.03.01 “Foreign Regional Studies”. To test the effectiveness of the author’ methodology, seven stages of learning are developed as part of the work: 1) setting goals and objectives; 2) explaining the rules of author ethics; 3) interacting with an artificial intelligence tool; 4) checking the text with a neural network; 5) finalizing the text; 6) discussing the results of interaction in small groups; 7) verifying the text of the teacher’s work). Results and Discussion. According to the experiment results, it is revealed that the author’s method of developing students’ writing skills based on technological solutions based on artificial intelligence had a positive effect on mastering active vocabulary ( t = 2.3 at p = 0.015), the structure of essay types ( t = 2.46 at p = 0.01) and the content of the work ( t = 3.71 at p = 0.0005). In terms of the grammatical design of the text, a comparative analysis of the data did not reveal any differences between the samples ( t = 1 at p = 0.16), which is explained by the students' possession of the necessary grammar knowledge at a high level before the experiment. Conclusion. The novelty of this research lies in of a step-by-step methodology development for improving writing skills in students of higher educational institutions. The results obtained can be used in future research in the formation of individual writing skills such as grammar, spelling, argumentation, work plan creation, conclusions using artificial intelligence tools.

Infants can compare large (>3) and small (≤3) numerosities but face difficulty with cross-boundary comparisons (e.g., 1 vs. 4). This study examined 312 Chinese infants (11-24 months; 141 girls, data collection: 2022-2025) by using a forced-choice task to investigate when and how they overcome this challenge. Success emerged at 15 months for 1 vs. 4 (1:4 ratio). Older infants succeeded with the same (2 vs. 8) and smaller ratios (2 vs. 6; 1:3), but even at 24 months struggled with 2 vs. 4 (1:2). Performance was unrelated to vocabulary size, suggesting that cross-boundary success is independent from language development. These findings highlight the engagement of nonverbal approximate number processing for both small and large numerosities.

The global spread of English and its role as a lingua franca coincide with its dominance in the media, creating new opportunities for informal exposure. While extramural English learning has been studied in Northern and Central Europe, Italy remains relatively underexplored. This study examines the informal acquisition of English slang among 148 Italian undergraduate students. Participants completed a questionnaire on media habits, a Vocabulary Size Test, a Slang Test assessing general and social-media-influenced slang, and an Oxford Placement Test to measure proficiency. Analyses in SPSS (v. 31) show that students are more familiar with social-media slang than general slang, and that slang knowledge correlates positively with both proficiency and vocabulary size. Findings underscore the role of digital media exposure and personal interest in supporting variety-specific L2 vocabulary acquisition, highlighting the potential of informal, media-driven learning in the Italian context.

This study examines Italian university students’ knowledge of English for Sports as a specialised subset of ESP, exploring how general proficiency, personal engagement with sport and exposure to English-language sports media shape both general and domain-specific lexical competence. 87 students (MA Languages; BA Economics) completed a questionnaire on media habits, a General Vocabulary Size Test, and a 60-item Sports VST. Moreover, a longitudinal sub-study tracked 12 MA students over one year through written production tasks and targeted retrieval tests. Results show that general vocabulary breadth is strongly associated with CEFR proficiency, whereas sports-specific vocabulary is largely predicted by exposure to English sports media and, to a lesser extent, by engagement in sport. BA Economics students, despite lower overall proficiency, outperform MA Language students on the Sports VST due to higher media-based exposure. A longitudinal analysis of language students’ performance reveals improvements in grammatical accuracy but minimal growth in specialised vocabulary, indicating that English for Sports does not develop incidentally in the absence of sustained, domain-rich input. These findings highlight the need for ESP pedagogies that integrate learners’ informal media practices to foster specialised lexical development.

Over the last decade, Italian university students have come to encounter English increasingly outside formal instruction, through a broad and fast-changing media ecology that includes traditional and digital formats, platforms and everyday communicative practices. This introduction frames the monographic issue Informal Pathways to English Learning within the national PRIN project on media-driven informalisation of English learning in Italy, which first established a large-scale baseline by pairing the IECoL questionnaire on informal contact with English and a receptive vocabulary test. Building on that map of behaviours and correlates, the present issue pursues a second, more fine-grained stage of inquiry by asking what kinds of English are acquired informally, under which conditions and with what limits. Across contributions, learning outcomes are investigated through targeted designs that combine general and domain-specific vocabulary tests (often inspired by Vocabulary Size Test formats), reaction-time measures, corpus-based analyses of learner production and, in one case, longitudinal follow-up. The studies cover socio-political lexis, sports and cooking terminology, false friends, slang and pragmatic features such as vague language. Overall, they show that informal learning effects are substantial but uneven across domains, strongly mediated by proficiency and shaped by the quality and relevance of engagement, with clear implications for how instruction can build strategically on students’ extramural practices.

This study investigates whether informal engagement with special-interest content can foster the acquisition of specialised English vocabulary and how such learning is shaped by individual and experiential factors. One hundred and forty-six Italian university students, who had received only general-purpose English instruction, completed a background questionnaire, a general Vocabulary Size Test (VST), a specialised VST targeting cooking terminology (with accuracy and reaction times), and a figurative-language interpretation task. The specialised test contained 120 cooking-related items (verbs, nouns and adjectives) sampled from B1–C2 levels and presented in a four-option multiple-choice format. Results show substantial knowledge of specialised culinary vocabulary despite the absence of formal instruction, with a clear proficiency gradient and a verb advantage across levels. At lower proficiency levels, specialised scores were strongly associated with general vocabulary size and with self-reported intensity of exposure to English-language videos. Regression analyses indicated that specialised knowledge was best predicted by proficiency, reading in English and domain-specific habits (e.g., language of recipes), whereas the same variables explained little variance in general vocabulary. Reaction-time data suggested partial “automatisation” of specialised items, particularly for intermediate learners. Figurative-language performance revealed that, for many participants, specialised lexical representations were sufficiently deep to support interpretation in novel, non-literal contexts, especially when learners frequently followed recipes or watched cooking videos in English. The findings highlight both the potential and the limits of learning specialised vocabulary through informal digital practices and point to ways in which formal instruction can systematically build on learners’ extramural experiences.

ABSTRACT Children growing up in structurally disadvantaged contexts face increased risks of delayed language development, yet evidence from the Latin American Caribbean remains limited. This study examined how personal, prelinguistic, and contextual factors jointly shape expressive language among young children in Caribbean Colombia, a setting marked by sociocultural diversity, poverty, and inequality. A cross-sectional sample of 174 children aged 8–30 months attending early childhood development centers together with their primary caregivers, who completed caregiver-reported versions of the Colombian adaptation of the MacArthur – Bates Communicative Development Inventories. As an exploratory approach, Random Forest regression models were estimated separately for children aged 8–16 months (expressive vocabulary) and 16–30 months (vocabulary size and phrase complexity), including child characteristics, prelinguistic abilities, family composition, and socioeconomic indicators as candidate predictors. In the 8–16-month group, prelinguistic skills (gesture use, imitation, phrase comprehension), age, and caregiver education emerged as key contributors to expressive vocabulary, alongside markers of lower economic burden. In the 16–30-month group, model performance was weaker and less stable; exploratory patterns suggested a growing influence of socioeconomic disadvantage and caregiver education on vocabulary and early syntax, which requires cautious interpretation. Overall, findings highlight how early communicative environments, prelinguistic abilities, and structural constraints intersect in vulnerable communities. Strengthening context-sensitive early stimulation, supporting caregivers living in poverty, and improving access to linguistically rich services may help prevent the consolidation of language and learning inequalities in early childhood.

This study intended to investigate the difficulties encountered by EFL learners in using four derivational adjective-noun suffixes (-al, -ous, -ic, and -ful) and their factors. Although derivational morphology can contribute to vocabulary development and academic skills, the use of suffixes that form adjectives is still a difficulty for many EFL learners. This research used a non-experimental quantitative descriptive method. The population and sample for this study comprised 33 students in the fifth semester of the Morphology course in the English Education Study Program. Data were collected using a test and a questionnaire to assess conditions influencing students' abilities. Test results showed that the four endings varied in difficulty. The endings -ous and -ic were the most difficult for students to use correctly, while -al was the easiest. Student use of -ful was not entirely principled. However, the level of accuracy for the base noun seemed to be a defining factor in suffix use accuracy. The results of the questionnaire also showed that a lack of morphological awareness, limited vocabulary size and pedagogical limitations were the three primary variables influencing students' capacity to generate adjectives. There was no appreciable impact from other factors such L1 interference, overgeneralization, lack of productive practice, and limited exposure to authentic texts. These results emphasize the need to provide more direct instruction in derivational morphology, broaden the range of activities, and increase exposure to scholarly terminology. Improvements in these areas may improve students' understanding of morphology and enable them to produce more accurate and complex academic language.

Looking forward: eye-gaze methods in vocabulary development research.

This article presents BanglaVerb, a systematically curated and linguistically validated sentence-level dataset designed to support transitivity classification in Bangla. The dataset contains 3001 Bangla sentences, each centered on a single verb instance annotated as either transitive (1634) or intransitive (1367). It was developed to address the lack of verb-focused linguistic resources for Bangla, a morphologically rich but under-resourced language in the NLP domain. Sentences were collected from diverse public sources, standardized, and carefully cleaned to ensure textual integrity. Annotation combined rule-based pre-labeling with expert linguistic verification, resulting in a 92% majority-voting agreement among annotators, which reflects high labeling consistency and reliability. Beyond its annotation framework, the dataset provides detailed lexical and structural statistics, including vocabulary size, character length distributions, n-gram patterns, and frequency distributions that follow Zipf's law, confirming its linguistic representativeness. Baseline experiments using multiple machine learning models demonstrate strong classification performance, indicating the dataset's clarity and robustness. By bridging sentence-level structure with verb semantics, BanglaVerb offers a high-quality, openly accessible resource that can support a wide range of downstream applications, including lemmatization, morphological analysis, syntactic parsing, semantic role labeling, and the development of verb-aware language models for Bangla.

Vocabulary is a crucial aspect in all English language teaching. The greater the vocabulary mastery a person has, the better their language skills and communication abilities. Various strategies, methods, and techniques have been employed by educators to teach vocabulary to students. The use of game-based learning through the Duolingo application offers an alternative for teachers to develop students’ vocabulary, particularly in mastering new words, spelling, pronunciation, and correction. This game is suitable for teachers to incorporate vocabulary teaching into the learning process. This research was conducted to enhance students' mastery of passive vocabulary by implementing game-based learning through the Duolingo application. The study used a quantitative method with a pre-experimental approach, specifically employing a one-group pre-test and post-test design. The population consisted of all tenth-grade students, totaling 287 learners across eight classes. The sample was chosen through a cluster random sampling technique, resulting in 36 students from class E 2. Students completed an initial pre-test, participated in the treatment sessions, and then took a post-test. The results show a significant increase in learners’ passive vocabulary following the use of Duolingo as a game-based learning tool. This improvement is reflected in the rise of the mean score from 39.30 on the pre-test to 80 on the post-test. Additionally, at the 5% significance level, the obtained t-value of 4.43 is higher than the t-table value of 2.030 (4.43 ≥ 2.030) with 35 degrees of freedom. These findings confirm that the alternative hypothesis (H1) is accepted and the null hypothesis (H0) is rejected. Keywords: Game-Based Learning (GBL), Duolingo Application, Passive Vocabulary.

The purpose of this study is to explore the attitudes of Saudi EFL university learners regarding the use of AI-generated games in the process of learning English vocabulary. This mixed-methods study includes 150 responses to the questionnaire and 10 responses to a structured interview. The results of the mixed research show that most students use AI tools in learning vocabulary or playing games such as Duolingo, Quizlet, Kahoot, and WordUp. Although most respondents claim to be satisfied with AI-based games, which are interactive, amusing, and effective for retaining vocabulary, the questionnaire analysis indicates that AI games enhance vocabulary size, contextual understanding, and language abilities. At the same time, the students face multiple problems, such as technical difficulties, lack of access to the internet, lack of assistance, and an inappropriate level of gaming content. As far as gaming is concerned, the proposed solutions are increased teacher involvement, training workshops, a better selection of feedback tools, and Arabic translation to help comprehend the message. The interview data partially supports this, with results showing that AI use generally has a positive effect in stimulating interest and autonomy and reducing anxiety. In total, the findings seem to support the creation of authentic and engaging contextual vocabulary learning environments with the help of AI games. However, the positive outcomes will only be instrumental once the instructional, technical, and contextual problems are addressed by means of high-quality training and adjustments in the level of difficulty and integration.

Parent Well-Being and Language Input Predict Child Face-Voice Matching and Expressive Language Outcomes.

When do infants first begin grasping the meaning of verbs? To learn verbs - words that describe actions and events - theorists suggest that infants must employ word segmentation, event processing, and verb-to-action mapping skills. Prior research suggests that many of these skills emerge by approximately 10 months. In the current study, we examined whether 10-month-old infants understand several early verbs. In a novel action-verb pairing paradigm, infants saw videos of everyday actions while hearing matching or mismatching verbs. We tested adults on the same paradigm to verify that action-verb pairs reliably evoked an N400 mismatch effect. Adults showed an N400-like effect over frontal and centroparietal regions. Infants also showed ERP differences between mismatched and matched action-verb pairs, although the pattern differed from adults, with variation in topography and directionality. Infants' ERP response was not related to their receptive or productive vocabulary size. These findings indicate that infants were sensitive to co-occurrences between actions and verbs, reflecting emerging verb understanding and suggesting nascent semantic knowledge. We further consider alternative explanations, including the possibility that the observed ERP differences reflect early action-verb associations that may serve as building blocks for later semantic verb knowledge. These results expand our understanding of infant language acquisition by demonstrating that, by 10 months, infants are sensitive to mismatches between everyday actions and verbs.