Hasil untuk "Language acquisition"

J. Bruner

Dongqi Zheng

Large Reasoning Language Models (LRLMs or LRMs) demonstrate remarkable capabilities in complex reasoning tasks, but suffer from significant computational inefficiencies due to overthinking phenomena. Existing efficient reasoning methods face the challenge of balancing reasoning quality with inference cost reduction. We propose \textbf{Adaptive Reasoning Suppression (ARS)}, a novel training-free approach that dynamically suppresses redundant reasoning steps while preserving accuracy through adaptive certainty monitoring. ARS introduces a multi-checkpoint certainty estimation mechanism with progressive suppression thresholds, achieving superior efficiency compared to static suppression methods. Our extensive evaluation across mathematical reasoning benchmarks using multiple model architectures demonstrates that ARS achieves up to 53%, 46.1%, and 57.9% in token, latency and energy reduction, while maintaining or improving accuracy.

Jie Zhao, Wanting Ning, Yuxiao Fei et al.

In Natural Language Processing(NLP), Event Temporal Relation Extraction (ETRE) is to recognize the temporal relations of two events. Prior studies have noted the importance of language models for ETRE. However, the restricted pre-trained knowledge of Small Language Models(SLMs) limits their capability to handle minority class relations in imbalanced classification datasets. For Large Language Models(LLMs), researchers adopt manually designed prompts or instructions, which may introduce extra noise, leading to interference with the model's judgment of the long-distance dependencies between events. To address these issues, we propose GDLLM, a Global Distance-aware modeling approach based on LLMs. We first present a distance-aware graph structure utilizing Graph Attention Network(GAT) to assist the LLMs in capturing long-distance dependency features. Additionally, we design a temporal feature learning paradigm based on soft inference to augment the identification of relations with a short-distance proximity band, which supplements the probabilistic information generated by LLMs into the multi-head attention mechanism. Since the global feature can be captured effectively, our framework substantially enhances the performance of minority relation classes and improves the overall learning ability. Experiments on two publicly available datasets, TB-Dense and MATRES, demonstrate that our approach achieves state-of-the-art (SOTA) performance.

Masato Kikuchi, Masatsugu Ono, Toshioki Soga et al.

Although WordNet is a valuable resource because of its structured semantic networks and extensive vocabulary, its fine-grained sense distinctions can be challenging for second-language learners. To address this issue, we developed a version of WordNet annotated with the Common European Framework of Reference for Languages (CEFR), integrating its semantic networks with language-proficiency levels. We automated this process using a large language model to measure the semantic similarity between sense definitions in WordNet and entries in the English Vocabulary Profile Online. To validate our approach, we constructed a large-scale corpus containing both sense and CEFR-level information from the annotated WordNet and used it to develop contextual lexical classifiers. Our experiments demonstrate that models fine-tuned on this corpus perform comparably to those fine-tuned on gold-standard annotations. Furthermore, by combining this corpus with the gold-standard data, we developed a practical classifier that achieves a Macro-F1 score of 0.81. This result provides indirect evidence that the transferred labels are largely consistent with the gold-standard levels. The annotated WordNet, corpus, and classifiers are publicly available to help bridge the gap between natural language processing and language education, thereby facilitating more effective and efficient language learning.

Dinh Nam Pham, Eleftherios Avramidis

Non-manual facial features play a crucial role in sign language communication, yet their importance in automatic sign language recognition (ASLR) remains underexplored. While prior studies have shown that incorporating facial features can improve recognition, related work often relies on hand-crafted feature extraction and fails to go beyond the comparison of manual features versus the combination of manual and facial features. In this work, we systematically investigate the contribution of distinct facial regionseyes, mouth, and full faceusing two different deep learning models (a CNN-based model and a transformer-based model) trained on an SLR dataset of isolated signs with randomly selected classes. Through quantitative performance and qualitative saliency map evaluation, we reveal that the mouth is the most important non-manual facial feature, significantly improving accuracy. Our findings highlight the necessity of incorporating facial features in ASLR.

Carl Edwards, Chi Han, Gawon Lee et al.

Despite their ability to understand chemical knowledge, large language models (LLMs) remain limited in their capacity to propose novel molecules with desired functions (e.g., drug-like properties). In addition, the molecules that LLMs propose can often be challenging to make, and are almost never compatible with automated synthesis approaches. To better enable the discovery of functional small molecules, LLMs need to learn a new molecular language that is more effective in predicting properties and inherently synced with automated synthesis technology. Current molecule LLMs are limited by representing molecules based on atoms. In this paper, we argue that just like tokenizing texts into meaning-bearing (sub-)word tokens instead of characters, molecules should be tokenized at the level of functional building blocks, i.e., parts of molecules that bring unique functions and serve as effective building blocks for real-world automated laboratory synthesis. This motivates us to propose mCLM, a modular Chemical-Language Model that comprises a bilingual language model that understands both natural language descriptions of functions and molecular blocks. mCLM front-loads synthesizability considerations while improving the predicted functions of molecules in a principled manner. Experiments on FDA-approved drugs showed that mCLM is capable of significantly improving chemical functions. mCLM, with only 3B parameters, also achieves improvements in synthetic accessibility relative to 7 other leading generative AI methods including GPT-5. When tested on 122 out-of-distribution medicines using only building blocks/tokens that are compatible with automated modular synthesis, mCLM outperforms all baselines in property scores and synthetic accessibility. mCLM can also reason on multiple functions and iteratively self-improve to rescue drug candidates that failed late in clinical trials ("fallen angels").

S. Hema, S. N. S. Gandhimathi

In this era, as the education landscape continues to evolve, mobile learning has become crucial in promoting language proficiency and cultural understanding in real-life situations. As a result, the demand for cost-effective solutions has increased the importance of Mobile-Assisted Language Learning (MALL), which provides accessible tools like language learning apps, audio and video resources for language acquisition. The present research aims to analyse research trends in MALL, including the identification of leading countries, international collaboration and the establishment of conceptual frameworks. To achieve this, the study conducted a bibliometric analysis of research articles published between 2014 and 2023, sourced from the Scopus and Web of Science databases. Using key phrases and reference search methods, following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines, the study identified 398 relevant research papers. The bibliometric analysis encompasses citation patterns, co-authorship structures, bibliographic coupling, and co-occurrence analysis of keywords using ‘R programming’. The findings of the research illustrate the evolving publication trends, highlighting influential authors, prominent journals, and leading countries contributing to research on MALL. The study also includes a critical analysis of highly cited articles, providing insights from research that has significantly influenced the discourse. The findings unfold the vision into the future scope of mobile learning, envisioning its continued impact on educational paradigms and the potential for further advancements in language acquisition.

Mustafa M. AL-BAIRAQDAR, Serap ÖNEN

This study explores the potential of films as audiovisual resources for English language learners, particularly in the context of legal discourse. By offering exposure to authentic language use, films serve as a valuable tool for vocabulary acquisition. This research employs a corpus-based approach to examine films related to the field of law, with the objective of identifying the frequency of technical vocabulary encountered by students within a contextual framework, as well as the common occurrences and their alignment with specialized law dictionaries. A corpus consisting of fifteen films was compiled and analyzed using corpus analysis software. The word list generated from the film transcripts was compared against a specialized law dictionary to determine the vocabulary included in the analysis. The results highlighted the number of technical vocabulary terms present in each film, the overall quantity within the corpus, the frequency distribution of technical vocabulary, and the most frequently occurring technical terms. These findings suggest that integrating films into language learning curricula can provide learners with valuable exposure to authentic technical vocabulary and emphasize the need for educators to select films with consideration to vocabulary levels.   Cite this article as: Al-bairaqdar, M. M., & Önen, S. (2025). Analysis of disciplinerelated films to promote contextual vocabulary learning. HAYEF: Journal of Education, 22, 0031, doi: 10.5152/ hayef.2025.24031

Editorial Team of Babylonia

This issue of Babylonia centers on the multifaceted role of writing in language education, addressing pedagogical, linguistic, and technological dimensions. It examines how writing can be taught, assessed, and supported in different educational contexts, particularly in multilingual and multicultural environments. A major theme is the need to shift writing instruction beyond simple transcription or product-oriented practices. Several contributions emphasize writing as a process, involving prewriting, drafting, revising, and reflecting. This process-oriented approach is framed as essential for both first and second language acquisition. Another key focus is the integration of writing into broader language competencies. Writing is not treated in isolation but as interconnected with speaking, reading, and listening. For instance, articles demonstrate how writing activities can support vocabulary development, grammatical awareness, and critical thinking. The issue also explores digital media and writing, including collaborative writing environments, blogs, and multimodal literacies. These tools are seen as opportunities to engage learners, diversify tasks, and facilitate peer feedback and revision. However, authors caution that these approaches require thoughtful implementation and teacher training. Attention is also given to the assessment of writing, with several contributions discussing formative vs. summative methods, rubrics, and self-evaluation. One article presents an innovative assessment system that uses learner portfolios and peer review. Multilingualism is a recurring thread: writing in plurilingual classrooms is treated as a space for identity negotiation, intercultural dialogue, and cross-linguistic transfer. Some contributions highlight the pedagogical value of allowing students to draw on their entire linguistic repertoires, rather than restricting them to the language of schooling. Key contributions include: A reflection on the cultural positioning of writing across languages. A project involving cooperative writing among plurilingual students in Switzerland. An article on the use of digital storytelling to develop writing and oral expression. A detailed case study on scaffolding academic writing in French as a second language. A report on multilingual writing workshops aimed at building linguistic awareness and agency. In sum, this issue promotes a holistic, inclusive, and dynamic view of writing instruction, aiming to foster not only linguistic proficiency but also learner autonomy, intercultural competence, and academic success.

Trapoom Ukarapol, Zhicheng Lee, Amy Xin

While Large Language Models show remarkable performance in natural language understanding, their resource-intensive nature makes them less accessible. In contrast, smaller language models such as MiniCPM offer more sustainable scalability, but often underperform without specialized optimization. In this paper, we explore the enhancement of smaller language models through the improvement of their text embeddings. We select three language models, MiniCPM, Phi-2, and Gemma, to conduct contrastive fine-tuning on the NLI dataset. Our results demonstrate that this fine-tuning method enhances the quality of text embeddings for all three models across various benchmarks, with MiniCPM showing the most significant improvements of an average 56.33% performance gain. The contrastive fine-tuning code is publicly available at https://github.com/trapoom555/Language-Model-STS-CFT.

Jayanta Sadhu, Maneesha Rani Saha, Rifat Shahriyar

The influence of Large Language Models (LLMs) is rapidly growing, automating more jobs over time. Assessing the fairness of LLMs is crucial due to their expanding impact. Studies reveal the reflection of societal norms and biases in LLMs, which creates a risk of propagating societal stereotypes in downstream tasks. Many studies on bias in LLMs focus on gender bias in various NLP applications. However, there's a gap in research on bias in emotional attributes, despite the close societal link between emotion and gender. This gap is even larger for low-resource languages like Bangla. Historically, women are associated with emotions like empathy, fear, and guilt, while men are linked to anger, bravado, and authority. This pattern reflects societal norms in Bangla-speaking regions. We offer the first thorough investigation of gendered emotion attribution in Bangla for both closed and open source LLMs in this work. Our aim is to elucidate the intricate societal relationship between gender and emotion specifically within the context of Bangla. We have been successful in showing the existence of gender bias in the context of emotions in Bangla through analytical methods and also show how emotion attribution changes on the basis of gendered role selection in LLMs. All of our resources including code and data are made publicly available to support future research on Bangla NLP. Warning: This paper contains explicit stereotypical statements that many may find offensive.

Harshvivek Kashid, Pushpak Bhattacharyya

Optical Character Recognition (OCR) technology has revolutionized the digitization of printed text, enabling efficient data extraction and analysis across various domains. Just like Machine Translation systems, OCR systems are prone to errors. In this work, we address the challenge of data generation and post-OCR error correction, specifically for low-resource languages. We propose an approach for synthetic data generation for Devanagari languages, RoundTripOCR, that tackles the scarcity of the post-OCR Error Correction datasets for low-resource languages. We release post-OCR text correction datasets for Hindi, Marathi, Bodo, Nepali, Konkani and Sanskrit. We also present a novel approach for OCR error correction by leveraging techniques from machine translation. Our method involves translating erroneous OCR output into a corrected form by treating the OCR errors as mistranslations in a parallel text corpus, employing pre-trained transformer models to learn the mapping from erroneous to correct text pairs, effectively correcting OCR errors.

Shiva Ebrahimi, Xuan Guo

Tandem mass spectrometry (MS/MS) stands as the predominant high-throughput technique for comprehensively analyzing protein content within biological samples. This methodology is a cornerstone driving the advancement of proteomics. In recent years, substantial strides have been made in Data-Independent Acquisition (DIA) strategies, facilitating impartial and non-targeted fragmentation of precursor ions. The DIA-generated MS/MS spectra present a formidable obstacle due to their inherent high multiplexing nature. Each spectrum encapsulates fragmented product ions originating from multiple precursor peptides. This intricacy poses a particularly acute challenge in de novo peptide/protein sequencing, where current methods are ill-equipped to address the multiplexing conundrum. In this paper, we introduce DiaTrans, a deep-learning model based on transformer architecture. It deciphers peptide sequences from DIA mass spectrometry data. Our results show significant improvements over existing STOA methods, including DeepNovo-DIA and PepNet. Casanovo-DIA enhances precision by 15.14% to 34.8%, recall by 11.62% to 31.94% at the amino acid level, and boosts precision by 59% to 81.36% at the peptide level. Integrating DIA data and our DiaTrans model holds considerable promise to uncover novel peptides and more comprehensive profiling of biological samples. Casanovo-DIA is freely available under the GNU GPL license at https://github.com/Biocomputing-Research-Group/DiaTrans.

Marcos Zampieri, Damith Premasiri, Tharindu Ranasinghe

The spread of various forms of offensive speech online is an important concern in social media. While platforms have been investing heavily in ways of coping with this problem, the question of privacy remains largely unaddressed. Models trained to detect offensive language on social media are trained and/or fine-tuned using large amounts of data often stored in centralized servers. Since most social media data originates from end users, we propose a privacy preserving decentralized architecture for identifying offensive language online by introducing Federated Learning (FL) in the context of offensive language identification. FL is a decentralized architecture that allows multiple models to be trained locally without the need for data sharing hence preserving users' privacy. We propose a model fusion approach to perform FL. We trained multiple deep learning models on four publicly available English benchmark datasets (AHSD, HASOC, HateXplain, OLID) and evaluated their performance in detail. We also present initial cross-lingual experiments in English and Spanish. We show that the proposed model fusion approach outperforms baselines in all the datasets while preserving privacy.

Eren Dogan, M. Egemen Uzun, Atahan Uz et al.

The developments that language models have provided in fulfilling almost all kinds of tasks have attracted the attention of not only researchers but also the society and have enabled them to become products. There are commercially successful language models available. However, users may prefer open-source language models due to cost, data privacy, or regulations. Yet, despite the increasing number of these models, there is no comprehensive comparison of their performance for Turkish. This study aims to fill this gap in the literature. A comparison is made among seven selected language models based on their contextual learning and question-answering abilities. Turkish datasets for contextual learning and question-answering were prepared, and both automatic and human evaluations were conducted. The results show that for question-answering, continuing pretraining before fine-tuning with instructional datasets is more successful in adapting multilingual models to Turkish and that in-context learning performances do not much related to question-answering performances.

Sarah Annunziato

This article will discuss the design and implementation of an intermediate level Italian for HealthcareLanguage class. Since 2016, the language of Dante has experienced a marked decline in enrollment in many higher education institutions in the United States. Some recent studies related to this issue suggest that providing students with a wide variety of course options, including ones related to professional fields, is key to reversing this trend. Nevertheless, Italian presents somewhat of a challenge in this regard as it is the norm rather than the exception for learners to first encounter the language only when they begin their studies at college or university. However, in many instances, such as in healthcare, language for specific purposes courses target more advanced students. Therefore, introducing these types of classes earlier in the learning experience might prove to be instrumental in encouraging more students to continue studying Italian at advanced-level and perhaps even beyond. The present course was offered to learners who had previously completed three semesters of college-level Italian from beginner level to intermediate. People enrolled in the class had already attained an upper-intermediate level of skill in the target language. Since the course focused on healthcare, it emphasized the acquisition of new terms and communicative modes to help learners better interact with patients or clinicians in a medical setting. It was also designed around the Five C’s (Communication, Cultures, Connections, Comparisons, and Communities) made popular by ACTFL’s World Readiness Standards for Learning Languages. Overall, Italian for Healthcare Professionals relied on both an experiential and project-based methodology through which students completed tasks that reflected challenges that they might realistically encounter while working in the allied health professions in Italy or with Italian-speaking diasporas in other parts of the world. Learners reported benefiting from both the variety and realistic nature of the activities. This article will explore the need for such a course in Italian Studies, its structure, as well as examples of projects and activities that it might include. Ultimately, student response to the course suggests that such offerings can be made available to intermediate-level learners of Italian with promising results.

Mengyu He, Qin Xie

Abstract This study explores the effectiveness of Data-Driven Learning (DDL) in teaching noun collocations to pre-tertiary learners in Hubei Province, China, using the online corpus tool ‘Corpusmate’. Acknowledging the importance and challenge of mastering collocations in learning a foreign language, this study focuses on the effects of DDL on pre-tertiary learners, an area less examined previously due to the complexities associated with using corpus tools. Conducted over two months, the research employed pre-tests, post-tests, and delayed post-tests to measure learners’ comprehension and retention of noun collocations. Additionally, a questionnaire was distributed to gather feedback on learners’ experiences with the DDL approach and ‘Corpusmate’. Results indicated that the experimental group, which received DDL training, showed significant improvements in test scores compared to the control group, which used traditional resources. The experimental group’s scores remained high in the delayed post-test, suggesting that the DDL approach facilitated long-term retention of collocational knowledge, although a notable proportion of learners expressed neutral or negative perceptions of the DDL activities. These results highlight the need for further investigation into the attitudes of the participants. Overall, most participants provided positive feedback on the use of ‘Corpusmate’ in learning noun collocations. These results advocate for the incorporation of corpus consultation into language teaching practices. The study underscores that with appropriate training and tools like ‘Corpusmate’, the DDL approach can potently aid in the sustained learning of complex language elements, such as collocations, even for younger learners.

Peiyi Wang, Lei Li, Liang Chen et al.

Reasoning is a cognitive process of using evidence to reach a sound conclusion. The reasoning capability is essential for large language models (LLMs) to serve as the brain of the artificial general intelligence agent. Recent studies reveal that fine-tuning LLMs on data with the chain of thought (COT) reasoning process can significantly enhance their reasoning capabilities. However, we find that the fine-tuned LLMs suffer from an \textit{Assessment Misalignment} problem, i.e., they frequently assign higher scores to subpar COTs, leading to potential limitations in their reasoning abilities. To address this problem, we introduce an \textit{Alignment Fine-Tuning (AFT)} paradigm, which involves three steps: 1) fine-tuning LLMs with COT training data; 2) generating multiple COT responses for each question, and categorizing them into positive and negative ones based on whether they achieve the correct answer; 3) calibrating the scores of positive and negative responses given by LLMs with a novel constraint alignment loss. Specifically, the constraint alignment loss has two objectives: a) Alignment, which guarantees that positive scores surpass negative scores to encourage answers with high-quality COTs; b) Constraint, which keeps the negative scores confined to a reasonable range to prevent the model degradation. Beyond just the binary positive and negative feedback, the constraint alignment loss can be seamlessly adapted to the ranking situations when ranking feedback is accessible. Furthermore, we also delve deeply into recent ranking-based alignment methods, such as DPO, RRHF, and PRO, and discover that the constraint, which has been overlooked by these approaches, is also crucial for their performance. Extensive experiments on four reasoning benchmarks with both binary and ranking feedback demonstrate the effectiveness of AFT.

Fatma Elsafoury

In this paper, we propose a new metric to measure the SOS bias in language models (LMs). Then, we validate the SOS bias and investigate the effectiveness of removing it. Finally, we investigate the impact of the SOS bias in LMs on their performance and fairness on hate speech detection. Our results suggest that all the inspected LMs are SOS biased. And that the SOS bias is reflective of the online hate experienced by marginalized identities. The results indicate that using debias methods from the literature worsens the SOS bias in LMs for some sensitive attributes and improves it for others. Finally, Our results suggest that the SOS bias in the inspected LMs has an impact on their fairness of hate speech detection. However, there is no strong evidence that the SOS bias has an impact on the performance of hate speech detection.

Isabelle Augenstein, Timothy Baldwin, Meeyoung Cha et al.

The emergence of tools based on Large Language Models (LLMs), such as OpenAI's ChatGPT, Microsoft's Bing Chat, and Google's Bard, has garnered immense public attention. These incredibly useful, natural-sounding tools mark significant advances in natural language generation, yet they exhibit a propensity to generate false, erroneous, or misleading content -- commonly referred to as "hallucinations." Moreover, LLMs can be exploited for malicious applications, such as generating false but credible-sounding content and profiles at scale. This poses a significant challenge to society in terms of the potential deception of users and the increasing dissemination of inaccurate information. In light of these risks, we explore the kinds of technological innovations, regulatory reforms, and AI literacy initiatives needed from fact-checkers, news organizations, and the broader research and policy communities. By identifying the risks, the imminent threats, and some viable solutions, we seek to shed light on navigating various aspects of veracity in the era of generative AI.