Dávid Horváth's review of Kazumi Uchikawa's 卒業論文と日本語教育: TBLL（Thesis Based Language Learning）の理念と実践. Thesis Writing and Japanese Language Teaching: Philosophy and Practice of TBLL (Thesis Based Language Learning). Tokyo: Coco Shuppan ココ出版, 2024.
Sourav Pal, Arghya Panda, Biprojit Bhowmick
et al.
Introduction: Crithmum maritimum(sea fennel, Chinese: 海茴香 [hǎi huí xiāng]) is an edible halophyte traditionally valued in both Mediterranean diets and Traditional Chinese Medicine (TCM), where it is prescribed for “dissolving masses,” regulating qi, and resolving damp-heat/toxins—concepts aligned with anti-inflammatory, antimicrobial, and anticancer effects. Rich in phenolic acids, flavonoids, essential oils, fatty acids, vitamins, and minerals, it shows potential against gastrointestinal tract (GIT) cancers, a growing global health burden. Methods: A systematic literature search was conducted in PubMed, Scopus, Web of Science, and Google Scholar for studies published between 2010 and 2025, limited to English language studies, using keywords “Crithmum maritimum,” “sea fennel,” “gastrointestinal cancer,” “anticancer,” “polyphenols,” “essential oils,” and “bioactive compounds.” of 167 retrieved records, 138 met the inclusion criteria: original in vitro, in vivo, or clinical research evaluating C. maritimum or its constituents for anticancer activity in GIT malignancies. Exclusion criteria removed reviews without new data, unrelated species, and studies lacking mechanistic outcomes. Results: C. maritimum exhibits preclinical efficacy against GIT cancers via modulation of p53, NRF2, and Wnt/β-catenin pathways, aligning with its traditional TCM uses. Studies identified chlorogenic acid (30–50 % of total polyphenol), gallic acid (15–20 % of terpene), limonene, sabinene, α-pinene, γ-terpinene, fatty acids, and vitamin C as key anticancer agents that promote apoptosis, inhibit angiogenesis, and attenuate oxidative and inflammatory signaling in GIT cancer models. Conclusion: C. maritimum exhibits a broad phytochemical spectrum with multitargeted preclinical efficacy against GIT cancers, consistent with its historical TCM applications. Standardization, mechanistic validation, and clinical trials are required to advance its therapeutic integration. Significance Statement: This review unites TCM tradition with modern biomedical evidence, highlighting C. maritimum as a promising natural agent for GIT cancer prevention and management, with mechanistic breadth well-suited to the multifactorial nature of these malignancies.
Other systems of medicine, Therapeutics. Pharmacology
While ethnic Chinese-Indonesians are often stereotyped as ‘economic figures’, many are also actively engaged in literature. Among their literary contributions is The Green Island Prose Compilation (2015), which is a post-Reformasi anthology of Mandarin language prose selected from the monthly Lüdao column in the International Daily News. These writings reflect personal experiences and preserve the memory of ethnic Chinese life in Indonesia. This study aims to examine how individual life narratives in the Green Island Prose Collection, shaped by social and historical context, contribute to the collective memory of Chinese-Indonesians. It explores how literature serves as a medium for documenting everyday life, preserving cultural identity, and bridging generational gaps. This research employs a descriptive qualitative method, utilizes purposive sampling to identify relevant themes of prose, and applies content analysis based on Maurice Halbwachs’ theory of collective memory. The data includes prose works that reveal recurring themes such as childhood, family, social life, and historical events. The findings indicate that the collective memory among Chinese-Indonesians is constructed from individual memories, incorporating elements of childhood, adulthood, and historical recollections. This becomes a significant shared experience and memory repository. This study not only addresses a gap in the study of the Chinese-Indonesian literature but also provides insights into the role of literature in preserving culture, shaping identity, and facilitating intergenerational communication. In this way, it enhances the understanding of the Chinese-Indonesian community within the broader context of contemporary Indonesian society.
Large language models (LLMs) are trained on vast amounts of text from the Internet, but do they truly understand the viral content that rapidly spreads online -- commonly known as memes? In this paper, we introduce CHIME, a dataset for CHinese Internet Meme Explanation. The dataset comprises popular phrase-based memes from the Chinese Internet, annotated with detailed information on their meaning, origin, example sentences, types, etc. To evaluate whether LLMs understand these memes, we designed two tasks. In the first task, we assessed the models' ability to explain a given meme, identify its origin, and generate appropriate example sentences. The results show that while LLMs can explain the meanings of some memes, their performance declines significantly for culturally and linguistically nuanced meme types. Additionally, they consistently struggle to provide accurate origins for the memes. In the second task, we created a set of multiple-choice questions (MCQs) requiring LLMs to select the most appropriate meme to fill in a blank within a contextual sentence. While the evaluated models were able to provide correct answers, their performance remains noticeably below human levels. We have made CHIME public and hope it will facilitate future research on computational meme understanding.
Large Language Models (LLMs) have achieved remarkable success across a wide range of natural language processing tasks. However, Chinese LLMs face unique challenges, primarily due to the dominance of unstructured free text and the lack of structured representations in Chinese corpora. While existing benchmarks for LLMs partially assess Chinese LLMs, they are still predominantly English-centric and fail to address the unique linguistic characteristics of Chinese, lacking structured datasets essential for robust evaluation. To address these challenges, we present a Comprehensive Benchmark for Evaluating Chinese Large Language Models (CB-ECLLM) based on the newly constructed Chinese Data-Text Pair (CDTP) dataset. Specifically, CDTP comprises over 7 million aligned text pairs, each consisting of unstructured text coupled with one or more corresponding triples, alongside a total of 15 million triples spanning four critical domains. The core contributions of CDTP are threefold: (i) enriching Chinese corpora with high-quality structured information; (ii) enabling fine-grained evaluation tailored to knowledge-driven tasks; and (iii) supporting multi-task fine-tuning to assess generalization and robustness across scenarios, including Knowledge Graph Completion, Triple-to-Text generation, and Question Answering. Furthermore, we conduct rigorous evaluations through extensive experiments and ablation studies to assess the effectiveness, Supervised Fine-Tuning (SFT), and robustness of the benchmark. To support reproducible research, we offer an open-source codebase and outline potential directions for future investigations based on our insights.
Self-supervised learning (SSL) is used in deep learning to train on large datasets without the need for expensive labelling of the data. Recently, large Automatic Speech Recognition (ASR) models such as XLS-R have utilised SSL to train on over one hundred different languages simultaneously. However, deeper investigation shows that the bulk of the training data for XLS-R comes from a small number of languages. Biases learned through SSL have been shown to exist in multiple domains, but language bias in multilingual SSL ASR has not been thoroughly examined. In this paper, we utilise the Lottery Ticket Hypothesis (LTH) to identify language-specific subnetworks within XLS-R and test the performance of these subnetworks on a variety of different languages. We are able to show that when fine-tuning, XLS-R bypasses traditional linguistic knowledge and builds only on weights learned from the languages with the largest data contribution to the pretraining data.
Gender bias in pretrained language models (PLMs) poses significant social and ethical challenges. Despite growing awareness, there is a lack of comprehensive investigation into how different models internally represent and propagate such biases. This study adopts an information-theoretic approach to analyze how gender biases are encoded within various encoder-based architectures. We focus on three key aspects: identifying how models encode gender information and biases, examining the impact of bias mitigation techniques and fine-tuning on the encoded biases and their effectiveness, and exploring how model design differences influence the encoding of biases. Through rigorous and systematic investigation, our findings reveal a consistent pattern of gender encoding across diverse models. Surprisingly, debiasing techniques often exhibit limited efficacy, sometimes inadvertently increasing the encoded bias in internal representations while reducing bias in model output distributions. This highlights a disconnect between mitigating bias in output distributions and addressing its internal representations. This work provides valuable guidance for advancing bias mitigation strategies and fostering the development of more equitable language models.
We introduce CSAR, an algorithm for inducing morphemes from emergent language corpora of parallel utterances and meanings. It is a greedy algorithm that (1) weights morphemes based on mutual information between forms and meanings, (2) selects the highest-weighted pair, (3) removes it from the corpus, and (4) repeats the process to induce further morphemes (i.e., Count, Select, Ablate, Repeat). The effectiveness of CSAR is first validated on procedurally generated datasets and compared against baselines for related tasks. Second, we validate CSAR's performance on human language data to show that the algorithm makes reasonable predictions in adjacent domains. Finally, we analyze a handful of emergent languages, quantifying linguistic characteristics like degree of synonymy and polysemy.
This paper critically reviews the conceptualisations of identity in the field of Second Language Acquisition (SLA), with particular attention to both psychological and poststructuralist sociocultural perspectives. Drawing on foundational and contemporary literature, the evolution of identity theories and their relevance to language learning, motivation, and self-construction have been reviewed. The analysis highlights how identity is shaped and reshaped through interaction with social, cultural, and linguistic environments across varied learner contexts, including migrant experiences, foreign language classrooms, and study-abroad programmes. A special focus is placed on learners of Chinese as a second or heritage language, revealing both shared and unique identity negotiation processes. While SLA identity research has expanded in scope, it remains limited by narrow theoretical approaches and insufficient attention to context-specific dynamics. This paper calls for more nuanced, longitudinal, and interdisciplinary investigations to capture the complexity of identity transformation in language learners.
China and South Korea are neighboring countries with the closest geographical and cultural proximity. From ancient times to the present, social exchanges between China and South Korea have been highly active and extensive across a wide range of fields, resulting in significant mutual influences in various aspects of society. This includes areas such as philosophy, culture, education, and technology. The educational systems of both countries have their origins in Confucianism, hence there are similarities in their current educational states. However, there are also differences in aspects such as educational systems and environments. This paper aims to compare the current educational status in China and South Korea.
It is generally agreed that first language (L1) morphological awareness, the ability to reflect upon, analyze and manipulate morphemes and morphological structure of words, can transfer and facilitate second language (L2) reading subskill acquisition. However, the facilitative role of L1 morphological awareness is unclear in the literature investigating third language (L3) reading. This study explored if and how L1 morphological awareness and L2 reading subcomponent skills contribute jointly to L3 lexical inferencing in syllabic L1 Japanese-alphabetic L2 English-morphosyllabic L3 Chinese university learners. Sixty-seven students were recruited from novice-level, first-year Chinese classes from a Japanese university. Only 56 students completed 7 computerized or paper-and-pencil tasks, including L1 Japanese morphological awareness, L2 English morphological awareness, L2 English vocabulary knowledge, L2 English lexical inferencing, L3 Chinese morphological awareness, L3 Chinese vocabulary knowledge, and L3 Chinese lexical inferencing. They also completed a self-reported proficiency questionnaire survey. Correlational and regression analyses were conducted. The results suggested that there was only a significant association between L1 Japanese morphological awareness and L3 Chinese lexical inferencing, and there were no significant correlations between L2 English reading subcomponent skills and L3 Chinese lexical inferencing. Discussion is provided regarding the crosslinguistic influence of L1 morphological awareness in L3 reading development and the implications for L3 reading instruction.
History of scholarship and learning. The humanities, Social Sciences
Mapping speech tokens to the same feature space as text tokens has become the paradigm for the integration of speech modality into decoder-only large language models (LLMs). An alternative approach is to use an encoder-decoder architecture that incorporates speech features through cross-attention. This approach, however, has received less attention in the literature. In this work, we connect the Whisper encoder with ChatGLM3 and provide in-depth comparisons of these two approaches using Chinese automatic speech recognition (ASR) and name entity recognition (NER) tasks. We evaluate them not only by conventional metrics like the F1 score but also by a novel fine-grained taxonomy of ASR-NER errors. Our experiments reveal that encoder-decoder architecture outperforms decoder-only architecture with a short context, while decoder-only architecture benefits from a long context as it fully exploits all layers of the LLM. By using LLM, we significantly reduced the entity omission errors and improved the entity ASR accuracy compared to the Conformer baseline. Additionally, we obtained a state-of-the-art (SOTA) F1 score of 0.805 on the AISHELL-NER test set by using chain-of-thought (CoT) NER which first infers long-form ASR transcriptions and then predicts NER labels.
Finding and facilitating commonalities between the linguistic behaviors of large language models and humans could lead to major breakthroughs in our understanding of the acquisition, processing, and evolution of language. However, most findings on human-LLM similarity can be attributed to training on human data. The field of emergent machine-to-machine communication provides an ideal testbed for discovering which pressures are neural agents naturally exposed to when learning to communicate in isolation, without any human language to start with. Here, we review three cases where mismatches between the emergent linguistic behavior of neural agents and humans were resolved thanks to introducing theoretically-motivated inductive biases. By contrasting humans, large language models, and emergent communication agents, we then identify key pressures at play for language learning and emergence: communicative success, production effort, learnability, and other psycho-/sociolinguistic factors. We discuss their implications and relevance to the field of language evolution and acquisition. By mapping out the necessary inductive biases that make agents' emergent languages more human-like, we not only shed light on the underlying principles of human cognition and communication, but also inform and improve the very use of these models as valuable scientific tools for studying language learning, processing, use, and representation more broadly.
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.
Large language models (LLMs) are possessed of numerous beneficial capabilities, yet their potential inclination harbors unpredictable risks that may materialize in the future. We hence propose CRiskEval, a Chinese dataset meticulously designed for gauging the risk proclivities inherent in LLMs such as resource acquisition and malicious coordination, as part of efforts for proactive preparedness. To curate CRiskEval, we define a new risk taxonomy with 7 types of frontier risks and 4 safety levels, including extremely hazardous,moderately hazardous, neutral and safe. We follow the philosophy of tendency evaluation to empirically measure the stated desire of LLMs via fine-grained multiple-choice question answering. The dataset consists of 14,888 questions that simulate scenarios related to predefined 7 types of frontier risks. Each question is accompanied with 4 answer choices that state opinions or behavioral tendencies corresponding to the question. All answer choices are manually annotated with one of the defined risk levels so that we can easily build a fine-grained frontier risk profile for each assessed LLM. Extensive evaluation with CRiskEval on a spectrum of prevalent Chinese LLMs has unveiled a striking revelation: most models exhibit risk tendencies of more than 40% (weighted tendency to the four risk levels). Furthermore, a subtle increase in the model's inclination toward urgent self-sustainability, power seeking and other dangerous goals becomes evident as the size of models increase. To promote further research on the frontier risk evaluation of LLMs, we publicly release our dataset at https://github.com/lingshi6565/Risk_eval.
The GPT (Generative Pre-trained Transformer) language models are an artificial intelligence and natural language processing technology that enables automatic text generation. There is a growing interest in applying GPT language models to university teaching in various dimensions. From the perspective of innovation in student and teacher activities, they can provide support in understanding and generating content, problem-solving, as well as personalization and test correction, among others. From the dimension of internationalization, the misuse of these models represents a global problem that requires taking a series of common measures in universities from different geographical areas. In several countries, there has been a review of assessment tools to ensure that work is done by students and not by AI. To this end, we have conducted a detailed experiment in a representative subject of Computer Science such as Software Engineering, which has focused on evaluating the use of ChatGPT as an assistant in theory activities, exercises, and laboratory practices, assessing its potential use as a support tool for both students and teachers.
Deletion in the Xp22.31 region is increasingly suggested to be involved in the etiology of epilepsy. Little is known regarding the genomic and clinical delineations of X-linked epilepsy in the Chinese population or the sex-stratified difference in epilepsy characteristics associated with deletions in the Xp22.31 region. In this study, we reported two siblings with a 1.69 Mb maternally inherited microdeletion at Xp22.31 involving the genes VCX3A, HDHD1, STS, VCX, VCX2, and PNPLA4 presenting with easily controlled focal epilepsy and language delay with mild ichthyosis in a Chinese family with a traceable 4-generation history of skin ichthyosis. Both brain magnetic resonance imaging results were normal, while EEG revealed epileptic abnormalities. We further performed an exhaustive literature search, documenting 25 patients with epilepsy with gene defects in Xp22.31, and summarized the epilepsy heterogeneities between sexes. Males harboring the Xp22.31 deletion mainly manifested with child-onset, easily controlled focal epilepsy accompanied by X-linked ichthyosis; the deletions were mostly X-linked recessive, with copy number variants (CNVs) in the classic region of deletion (863.38 kb–2 Mb). In contrast, epilepsy in females tended to be earlier-onset, and relatively refractory, with pathogenic CNV sizes varying over a larger range (859 kb–56.36 Mb); the alterations were infrequently inherited and almost combined with additional CNVs. A candidate region encompassing STS, HDHD1, and MIR4767 was the likely pathogenic epilepsy-associated region. This study filled in the knowledge gap regarding the genomic and clinical delineations of X-linked recessive epilepsy in the Chinese population and extends the understanding of the sex-specific characteristics of Xp22.31 deletion in regard to epilepsy.
Large Language Models (LLMs), such as ChatGPT and GPT-4, have dramatically transformed natural language processing research and shown promising strides towards Artificial General Intelligence (AGI). Nonetheless, the high costs associated with training and deploying LLMs present substantial obstacles to transparent, accessible academic research. While several large language models, such as LLaMA, have been open-sourced by the community, these predominantly focus on English corpora, limiting their usefulness for other languages. In this paper, we propose a method to augment LLaMA with capabilities for understanding and generating Chinese text and its ability to follow instructions. We achieve this by extending LLaMA's existing vocabulary with an additional 20,000 Chinese tokens, thereby improving its encoding efficiency and semantic understanding of Chinese. We further incorporate secondary pre-training using Chinese data and fine-tune the model with Chinese instruction datasets, significantly enhancing the model's ability to comprehend and execute instructions. Our experimental results indicate that the newly proposed model markedly enhances the original LLaMA's proficiency in understanding and generating Chinese content. Additionally, the results on the C-Eval dataset yield competitive performance among the models with several times the size of ours. We have made our pre-trained models, training scripts, and other resources available through GitHub, fostering open research for our community. Chinese LLaMA series: \url{https://github.com/ymcui/Chinese-LLaMA-Alpaca} and Chinese Llama-2 series: \url{https://github.com/ymcui/Chinese-LLaMA-Alpaca-2}