Hasil untuk "Language acquisition"

Bakr Bagash Mansour Ahmed Al-Sofi

Linguistic landscape (LL) can play a crucial role in language learning, including vocabulary acquisition. The present mixed-methods study addresses three key objectives: (a) investigating the extent to which the LL provides opportunities for incidental vocabulary acquisition among Saudi Science and Engineering Track (SSET) students, (b) identifying the students’ perceptions regarding the LL’s in developing vocabulary acquisition and vocabulary-based skills, and (c) determining the best strategies for retaining LL vocabulary. Data were collected in a two-phase process: initial photographic documentation by students, who gathered 476 signs from public and commercial spheres in the Saudi LL from various cities in the Saudi regions, including Bisha, Abha, Jazan, Jeddah, Riyadh, Al-Baha, Taif, Dammam, Al-Ula, Hafr Al-Batin, and Tabuk. Data were also collected through an online questionnaire administered to the same students to assess their perceptions regarding the LL’s role in enhancing their vocabulary acquisition. Content analysis and preliminary descriptive statistics were employed to analyze the collected data. The results revealed that the LL played a crucial role in enhancing various aspects of students’ vocabulary acquisition, and the participants held positive perceptions of its potential. Furthermore, several helpful strategies were identified for vocabulary retention. This foundational study highlights LL’s importance in the Saudi context, serving as a critical resource that stakeholders should integrate alongside traditional classroom materials to foster students’ language skills.

Linda Zeng, Steven Y. Feng, Michael C. Frank

Multilingualism is incredibly common around the world, leading to many important theoretical and practical questions about how children learn multiple languages at once. For example, does multilingual acquisition lead to delays in learning? Are there better and worse ways to structure multilingual input? Many correlational studies address these questions, but it is surprisingly difficult to get definitive answers because children cannot be randomly assigned to be multilingual and data are typically not matched between languages. We use language model training as a method for simulating a variety of highly controlled exposure conditions, and create matched 100M-word mono- and bilingual datasets using synthetic data and machine translation. We train GPT-2 models on monolingual and bilingual data organized to reflect a range of exposure regimes, and evaluate their performance on perplexity, grammaticality, and semantic knowledge. Across model scales and measures, bilingual models perform similarly to monolingual models in one language, but show strong performance in the second language as well. These results suggest that there are no strong differences between different bilingual exposure regimes, and that bilingual input poses no in-principle challenges for agnostic statistical learners.

Giuseppe Trovato

En este artículo nos proponemos investigar el potencial de la traducción como herramienta pedagógica en el ámbito de la enseñanza del español, con especial referencia al español de la economía y los negocios (ENE) impartido a estudiantes universitarios de habla italiana. Durante muchas décadas, la traducción como técnica de enseñanza se ha visto obstaculizada por el hecho de que no estaba en consonancia con el enfoque comunicativo que ha caracterizado durante mucho tiempo la enseñanza de idiomas. Tras la publicación del Volumen Complementario (2020) del Marco Común Europeo de Referencia para las Lenguas extranjeras, el concepto de traducción como herramienta pedagógica ha vuelto a cobrar protagonismo. En concreto, nuestro objetivo es examinar en qué medida la traducción puede contribuir al proceso de aprendizaje del español como lengua extranjera abordando una parcela especializada del español, a saber, la relacionada con la economía y los negocios. Presentaremos los resultados de una experimentación llevada a cabo con estudiantes de la titulación universitaria en “Commercio estero e Turismo” de la Universidad Ca' Foscari de Venecia, con el fin de observar qué dificultades han encontrado a la hora de realizar tareas de traducción con fines pedagógicos entre dos lenguas afines (español-italiano). De este modo, podremos especular sobre la viabilidad de la traducción como herramienta pedagógica eficaz en el ámbito del español de los negocios.

Samuel Nko'o Amvene

AIMS AND SCOPE OF HEALTH RESEARCH IN AFRICA Health Research in Africa (HRA) is a peer reviewed scientific that is partnered to Health Sciences and Disease. HRA covers all aspects of medicine, pharmacy, biomedical and health sciences, including public health and societal issues. It is an “online first” publication, which means that all the publications articles appear on the website before being included in the print journal. The papers are published in full on the website, with open access. Our mission is to inform and educate all the health professionals and to promote constructive debate on health issues that matter in the management not only of diseases but of health as a whole. Acceptance of manuscripts is based on the originality, the quality of the work and validity of the evidence, the clarity of presentation, and the relevance to our readership. Publications are expected to be concise, well organized and clearly written. Authors submit a manuscript with the understanding that the manuscript (or its essential substance) has not been published other than as an abstract in any language or format and is not currently submitted elsewhere for print or electronic publication. Manuscripts must be submitted by one of the authors of the manuscript. The submitting author takes responsibility for the article during submission and peer review. The HRA editorial team is based in Yaounde (Cameroon). EDITORIAL POLICIES Ethics HRA’s Publications Policy Committee follows the recommendations of the International Committee of Medical Journal Editors (ICMJE), the World Association of Medical Editors (WAME), and the Committee on Publication Ethics (COPE) for guidance on policies and procedures related to publication ethics. The policies for HRA have been adapted from those three advisory bodies and, where necessary, modified and tailored to meet the specific content, audiences, and aims of  HRA. Peer review process Research manuscripts are initially checked by the editor in chief or section editor for identification of gross deficiencies. At this stage, the proposal may be rejected. After this initial screening, articles are sent to one or two-reviewers whose names are hidden from the author and whose review is guided by a checklist (single anonymized review). The review summary is signed by the reviewer and is not posted with article. The review process may take days to weeks to reach a final decision that is the responsibility of the editor in chief. The duration from submission to publication may take one to six months (average: 6 weeks). So, the authors should avoid contacting the editorial office less than 6 weeks after the initial submission. Plagiarism, Scientific Misconduct Manuscripts are randomly checked for plagiarism with available free tools. Those proven of plagiarism are returned to the authors without peer review. The editors reserve the right to request that the authors provide additional data collected during their investigations. The editors also reserve the right to send a copy of the manuscript and data in question to the author’s dean, university, or supervisor or, in the case of an investigation being funded by an agency, to that funding agency for appreciation. Conflict of Interest At the time of submission, authors are asked to disclose whether they have any financial interests or connections, direct or indirect, or other situations that may influence directly or indirectly the work submitted for consideration. Human and Animal Studies Manuscripts reporting results of prospective or retrospective studies involving human subjects must document that appropriate institutional review board (IRB) approval and informed consent were obtained (or waived by the IRB) after the nature of the procedure(s) had been fully explained. In any case, medical research involving human subjects should comply with the Declaration of Helsinki (2013). Authorship To be listed as an author, an individual must have made substantial contributions to all three categories established by the ICMJE (http://www.icmje.org): (a) “conception and design, or acquisition of data, or analysis and interpretation of data,” (b) “drafting the article or revising it critically for important intellectual content,” and (c) “final approval of the version to be published.” Individuals who have not made substantial contributions in all three categories but who have made substantial contributions either to some of them or in other areas should be listed in acknowledgments. Please limit the number of authors to ten when this is feasible. Content licensing - Open access compliance  Articles published in HRA are Open Access and distributed under the terms of the Creative Commons Non-Commercial No-Derivatives License (CC BY-NC-ND 4.0). Copyright The authors publishing under this license with HRA retain all rights which means that the authors can read, print, and download, redistribute or republish (e.g display in a repository), translate the article (for private use only, not for distribution), download for text and data mining, reuse portions or extracts in other works, but they are not allowed to sell or re-use for commercial purposes or re-use for non-commercial purposes; without asking prior permission from the publisher, provided the original work is properly cited. Language HRA is bilingual and accepts publications in French and English. All the publications should have an abstract in both languages. Whenever possible, picture captions and table titles should be in both languages. All accepted manuscripts are copy-edited. Particularly if English is not your first language, before submitting your manuscript, HRA advises the work to have it edited for language. This is to ensure that the academic content is well understood by editors, reviewers and readers. There are many providers that offer this service; however, the authors are liable for all costs associated with such services. Artificial Intelligence (AI)–Assisted Technology At submission, the authors should disclose whether they used artificial intelligence (AI)–assisted technologies in the production of the publication and how AI was used. However, authors should not list AI and AI-assisted technologies as an author or co-author, nor cite AI as an author. ARTICLE PROCESSING CHARGES (APC) Article submission is free of charges, but if the paper is accepted for publication, the author will be asked to pay article processing charges to cover publications costs (220-250 $), depending on the type, complexity and length of the work, and on the number of authors. To guarantee HRA's independence, APC cover publication charges such as electronic archiving, plagiarism checking, editing, peer review process, site maintenance and web-hosting, proofreading, quality check, PDF designing and article maintenance.

Kirsten H. Beier-Marchesi

Die in diesem Beitrag vorgestellte interdisziplinäre Studie wurde in einem bayrischen Kindergarten mit sieben fünf-siebenjährigen Kindern unterschiedlicher Erstsprachen durchgeführt. Sie untersucht, welche kindlichen Entwicklungen sich in leiborientierten, sprachlichen Lernszenarien ergeben und wie diese anhand von Bewegungs- und Sprachprofilen sichtbar gemacht werden können. Der Beitrag begründet aus phänomenologischer, bewegungs- und entwicklungstheoretischer sowie sprachdidaktischer Perspektive die Bedeutung einer leiborientierten Sprachbildung im Elementarbereich. Um der Komplexität der multimodalen Prozesse der Sprachaneignung gerecht zu werden, wurde ein eigenes forschungsmethodologisches Setting mit interdisziplinär angepassten Instrumenten entwickelt. Dieser innovative methodische Ansatz ermöglicht eine mehrperspektivische Analyse der sprachlichen Entwicklung, indem sprachliche, kognitive und leibliche Prozesse systematisch über Datentriangulation verknüpft werden. Die Integration der Verfahren erlaubt nicht nur eine differenzierte Erfassung individueller Entwicklungsverläufe, sondern auch eine präzise Untersuchung der Zusammenhänge zwischen Leiblichkeit und (fremd)sprachlichem Handeln in unterschiedlichen Lernkontexten.Hierfür wurden zwei subjektorientierte Verfahren eingesetzt: das Kestenberg Movement Profile (KMP) (Kestenberg-Amighi et al., 1999, 2018) und HAVAS 5 – das Hamburger Verfahren zur Analyse des Sprachstandes bei Fünfjährigen (Reich & Roth, 2007). Zusätzlich vertiefen Kinderzeichnungen und Interviews zu diesen Zeichnungen, die nach den HAVAS 5-Kriterien ausgewertet werden, die sprachliche Analyse.Um die institutionellen und familiären Rahmenbedingungen als sprachbildendes Umfeld der Kinder zu erfassen, werden leitfadengestützte Expert*inneninterviews (Kuckartz, 2007) durchgeführt. Darüber hinaus wird eine leiblich-affektive Forschungshaltung reflektiert, die sich durch einen phänomenologischen Zugang auszeichnet. Diese Forschungshaltung ermöglicht es, die subjektiven, vor-reflexiven Erfahrungen der Kinder in den Mittelpunkt zu stellen und ihre körperlichen sowie sprachlichen Entwicklungen in einem ganzheitlichen Rahmen zu verstehen.

Cheng Zhang, Guangyuan Yao

The belief that language learning abilities are malleable (L2 Growth Mindset) is crucial for sustained engagement in second language (L2) acquisition. However, the mechanisms linking L2 Growth Mindset (L2GM) to L2 engagement, particularly the role of specific positive emotions, require further exploration. This study investigates the mediating role of Foreign Language Peace of Mind (FLPOM)—a low-arousal, culturally-nuanced positive emotion characterized by inner peace and harmony—in the relationship between L2GM and the multifaceted construct of L2 engagement (behavioral, emotional, cognitive, and agentic) among 300 Chinese EFL learners. Structural equation modeling revealed that L2GM positively predicted FLPOM, and FLPOM, in turn, positively predicted all four dimensions of L2 engagement. L2GM also directly predicted behavioral and cognitive engagement. Crucially, FLPOM significantly mediated the relationship between L2GM and all engagement dimensions, acting as a full mediator for emotional and agentic engagement and a partial mediator for behavioral and cognitive engagement. These findings highlight that fostering a growth mindset can cultivate FLPOM, which then empowers learners to engage more deeply across all facets of L2 learning. The study underscores the importance of FLPOM as a key psychological resource and suggests pedagogical implications for nurturing both growth mindsets and peace of mind to enhance L2 engagement.

Abdul Rehman, Jian-Jun Zhang, Xiaosong Yang

Universal phoneme recognition typically requires analyzing long speech segments and language-specific patterns. Many speech processing tasks require pure phoneme representations free from contextual influence, which motivated our development of CUPE - a lightweight model that captures key phoneme features in just 120 milliseconds, about one phoneme's length. CUPE processes short, fixed-width windows independently and, despite fewer parameters than current approaches, achieves competitive cross-lingual performance by learning fundamental acoustic patterns common to all languages. Our extensive evaluation through supervised and self-supervised training on diverse languages, including zero-shot tests on the UCLA Phonetic Corpus, demonstrates strong cross-lingual generalization and reveals that effective universal speech processing is possible through modeling basic acoustic patterns within phoneme-length windows.

Ranjan Sapkota, Manoj Karkee

The fusion of language and vision in large vision-language models (LVLMs) has revolutionized deep learning-based object detection by enhancing adaptability, contextual reasoning, and generalization beyond traditional architectures. This in-depth review presents a structured exploration of the state-of-the-art in LVLMs, systematically organized through a three-step research review process. First, we discuss the functioning of vision language models (VLMs) for object detection, describing how these models harness natural language processing (NLP) and computer vision (CV) techniques to revolutionize object detection and localization. We then explain the architectural innovations, training paradigms, and output flexibility of recent LVLMs for object detection, highlighting how they achieve advanced contextual understanding for object detection. The review thoroughly examines the approaches used in integration of visual and textual information, demonstrating the progress made in object detection using VLMs that facilitate more sophisticated object detection and localization strategies. This review presents comprehensive visualizations demonstrating LVLMs' effectiveness in diverse scenarios including localization and segmentation, and then compares their real-time performance, adaptability, and complexity to traditional deep learning systems. Based on the review, its is expected that LVLMs will soon meet or surpass the performance of conventional methods in object detection. The review also identifies a few major limitations of the current LVLM modes, proposes solutions to address those challenges, and presents a clear roadmap for the future advancement in this field. We conclude, based on this study, that the recent advancement in LVLMs have made and will continue to make a transformative impact on object detection and robotic applications in the future.

Luyao Xu, Xiaohua Liu, Yangyu Xiao

Abstract Drawing upon Frenzel’s (2014) framework of appraisals, the current study explored language teachers’ emotional experiences and their antecedents in the online teaching context. Moreover, the interrelations between teachers’ emotions and their instructional practices were also investigated. Data were collected through semi-structured interviews with eleven language teachers from a top-tier international university in China. Our findings revealed that positive appraisals of online classroom events, including goal attainment, the capacity in and responsibility of effectively delivering lessons online, and the importance teachers attached to online teaching, tended to result in positive emotional experiences, which consequently led to motivational stimulations in more creative and productive instructional practices. By contrast, teachers who had negative appraisals of online teaching events tended to experience more unpleasant emotions and emotional vulnerability, which would possibly impede effective instructional practices and steer their teaching towards more traditional teaching methods. Our study contributes to the understanding of the relations among teachers’ appraisals of classroom events, emotional experiences, and instructional practices in online language classrooms. Implications for how to prepare teachers emotionally for online teaching are also discussed.

Masanori Hirano

With the recent development of large language models (LLMs), models that focus on certain domains and languages have been discussed for their necessity. There is also a growing need for benchmarks to evaluate the performance of current LLMs in each domain. Therefore, in this study, we constructed a benchmark comprising multiple tasks specific to the Japanese and financial domains and performed benchmark measurements on some models. Consequently, we confirmed that GPT-4 is currently outstanding, and that the constructed benchmarks function effectively. According to our analysis, our benchmark can differentiate benchmark scores among models in all performance ranges by combining tasks with different difficulties.

Chenxi Fu, Xiaoyi Wang, Zaijiang Man et al.

As recursion has been underlying any linguistic work for the last 60 years, the acquisition of recursive structures by children during language learning has become a focal point of inquiry. This study delves into the developmental trajectory of Mandarin-speaking children's acquisition of recursive possessives and locatives, assessing the impact of structural diversity on language acquisition. The research contrasts the comprehension of two-level recursive structures among children aged 3 to 7 years, employing answering question while seeing a picture task to elicit responses. The findings indicate that children do not attain adult-like proficiency in two-level recursion until the age of 6, and there exists a notable asymmetry in the acquisition of recursive possessives versus locatives. These results underscore the primacy of structural complexity and cognitive factors in the acquisition process, enhancing our comprehension of the cognitive foundations of language development and the pivotal role of recursion in child language acquisition.

Wengang Zhou, Weichao Zhao, Hezhen Hu et al.

Sign language serves as the primary meaning of communication for the deaf-mute community. Different from spoken language, it commonly conveys information by the collaboration of manual features, i.e., hand gestures and body movements, and non-manual features, i.e., facial expressions and mouth cues. To facilitate communication between the deaf-mute and hearing people, a series of sign language understanding (SLU) tasks have been studied in recent years, including isolated/continuous sign language recognition (ISLR/CSLR), gloss-free sign language translation (GF-SLT) and sign language retrieval (SL-RT). Sign language recognition and translation aims to understand the semantic meaning conveyed by sign languages from gloss-level and sentence-level, respectively. In contrast, SL-RT focuses on retrieving sign videos or corresponding texts from a closed-set under the query-by-example search paradigm. These tasks investigate sign language topics from diverse perspectives and raise challenges in learning effective representation of sign language videos. To advance the development of sign language understanding, exploring a generalized model that is applicable across various SLU tasks is a profound research direction.

Zhifan Sun, Antonio Valerio Miceli-Barone

Large Language Models (LLMs) are increasingly becoming the preferred foundation platforms for many Natural Language Processing tasks such as Machine Translation, owing to their quality often comparable to or better than task-specific models, and the simplicity of specifying the task through natural language instructions or in-context examples. Their generality, however, opens them up to subversion by end users who may embed into their requests instructions that cause the model to behave in unauthorized and possibly unsafe ways. In this work we study these Prompt Injection Attacks (PIAs) on multiple families of LLMs on a Machine Translation task, focusing on the effects of model size on the attack success rates. We introduce a new benchmark data set and we discover that on multiple language pairs and injected prompts written in English, larger models under certain conditions may become more susceptible to successful attacks, an instance of the Inverse Scaling phenomenon (McKenzie et al., 2023). To our knowledge, this is the first work to study non-trivial LLM scaling behaviour in a multi-lingual setting.

Jiaying Lu, Jinmeng Rao, Kezhen Chen et al.

Large Vision-Language Models (LVLMs) offer remarkable benefits for a variety of vision-language tasks. However, a challenge hindering their application in real-world scenarios, particularly regarding safety, robustness, and reliability, is their constrained semantic grounding ability, which pertains to connecting language to the physical-world entities or concepts referenced in images. Therefore, a crucial need arises for a comprehensive study to assess the semantic grounding ability of widely used LVLMs. Despite the significance, sufficient investigation in this direction is currently lacking. Our work bridges this gap by designing a pipeline for generating large-scale evaluation datasets covering fine-grained semantic information, such as color, number, material, etc., along with a thorough assessment of seven popular LVLMs' semantic grounding ability. Results highlight prevalent misgrounding across various aspects and degrees. To address this issue, we propose a data-centric enhancement method that aims to improve LVLMs' semantic grounding ability through multimodal instruction tuning on fine-grained conversations. Experiments on enhanced LVLMs demonstrate notable improvements in addressing misgrounding issues.

Thanmay Jayakumar, Fauzan Farooqui, Luqman Farooqui

Realizing the recent advances in Natural Language Processing (NLP) to the legal sector poses challenging problems such as extremely long sequence lengths, specialized vocabulary that is usually only understood by legal professionals, and high amounts of data imbalance. The recent surge of Large Language Models (LLMs) has begun to provide new opportunities to apply NLP in the legal domain due to their ability to handle lengthy, complex sequences. Moreover, the emergence of domain-specific LLMs has displayed extremely promising results on various tasks. In this study, we aim to quantify how general LLMs perform in comparison to legal-domain models (be it an LLM or otherwise). Specifically, we compare the zero-shot performance of three general-purpose LLMs (ChatGPT-20b, LLaMA-2-70b, and Falcon-180b) on the LEDGAR subset of the LexGLUE benchmark for contract provision classification. Although the LLMs were not explicitly trained on legal data, we observe that they are still able to classify the theme correctly in most cases. However, we find that their mic-F1/mac-F1 performance is up to 19.2/26.8\% lesser than smaller models fine-tuned on the legal domain, thus underscoring the need for more powerful legal-domain LLMs.

Renxi Wang, Haonan Li, Minghao Wu et al.

Instruction tuning significantly enhances the performance of large language models (LLMs) across various tasks. However, the procedure to optimizing the mixing of instruction datasets for LLM fine-tuning is still poorly understood. This study categorizes instructions into three primary types: NLP downstream tasks, coding, and general chat. We explore the effects of instruction tuning on different combinations of datasets on LLM performance, and find that certain instruction types are more advantageous for specific applications but can negatively impact other areas. This work provides insights into instruction mixtures, laying the foundations for future research.

Prabin Bhandari, Hannah Marie Brennan

Large Language Models (LLMs) have shown a tremendous capacity for generating literary text. However, their effectiveness in generating children's stories has yet to be thoroughly examined. In this study, we evaluate the trustworthiness of children's stories generated by LLMs using various measures, and we compare and contrast our results with both old and new children's stories to better assess their significance. Our findings suggest that LLMs still struggle to generate children's stories at the level of quality and nuance found in actual stories

Navid Rezaei, Marek Z. Reformat

Larger language models, such as GPT-3, have shown to be excellent in many tasks. However, we demonstrate that out-of-ordinary questions can throw the model off guard. This work focuses on finding answers to negated complementary questions in commonsense scenarios. We illustrate how such questions adversely affect the model responses. We propose a model-agnostic methodology to improve the performance in negated complementary scenarios. Our method outperforms few-shot generation from GPT-3 (by more than 11 points) and, more importantly, highlights the significance of studying the response of large language models in negated complementary questions. The code, data, and experiments are available under: https://github.com/navidre/negated_complementary_commonsense.

Qizhi Pei, Wei Zhang, Jinhua Zhu et al.

Recent advancements in biological research leverage the integration of molecules, proteins, and natural language to enhance drug discovery. However, current models exhibit several limitations, such as the generation of invalid molecular SMILES, underutilization of contextual information, and equal treatment of structured and unstructured knowledge. To address these issues, we propose $\mathbf{BioT5}$, a comprehensive pre-training framework that enriches cross-modal integration in biology with chemical knowledge and natural language associations. $\mathbf{BioT5}$ utilizes SELFIES for $100%$ robust molecular representations and extracts knowledge from the surrounding context of bio-entities in unstructured biological literature. Furthermore, $\mathbf{BioT5}$ distinguishes between structured and unstructured knowledge, leading to more effective utilization of information. After fine-tuning, BioT5 shows superior performance across a wide range of tasks, demonstrating its strong capability of capturing underlying relations and properties of bio-entities. Our code is available at $\href{https://github.com/QizhiPei/BioT5}{Github}$.

Margaryta Bondarenko, Brianna Butera , Rajiv Rao

This study provides an acoustic analysis of voice onset time (VOT) of voiceless stops /ptk/ in Spanish, produced by heritage speakers (HSs) of Ukrainian and of Polish who are English-dominant and beginner or intermediate learners of Spanish as a third language (L3). Given that both Ukrainian and Polish, like Spanish and unlike English, are characterized by short-lag VOT, data were collected from six Ukrainian HSs and 11 Polish HSs in their heritage language (HL), in English, and in Spanish to compare potential effects of the HL on L3 VOT production. VOT was analyzed in three task types. The goals were: 1) to determine whether VOT values produced in Spanish by Ukrainian and Polish HSs are more reflective of VOTs in the HL or in English, and 2) to determine the effect of task type on VOT. Data show that Ukrainian and Polish HSs’ VOTs in Spanish are shorter than those of L2 Spanish learners whose L1 is English, indicating a HL rather than dominant language influence on L3 VOT. Results suggest that the most crucial factors in L3 phonological acquisition are: 1) structural similarity between HL and L3, and 2) L3 proficiency (not language dominance). VOT was also affected by task type: like L1 Spanish speakers, VOT of Ukrainian HSs increases as task formality increases. This paper fills research gaps in HL and L3 phonetics/phonology as to the effects of a HL on the acquisition of subsequent sound systems in adulthood.