Sign language translation systems typically require English as an intermediary language, creating barriers for non-English speakers in the global deaf community. We present Canonical Semantic Form (CSF), a language-agnostic semantic representation framework that enables direct translation from any source language to sign language without English mediation. CSF decomposes utterances into nine universal semantic slots: event, intent, time, condition, agent, object, location, purpose, and modifier. A key contribution is our comprehensive condition taxonomy comprising 35 condition types across eight semantic categories, enabling nuanced representation of conditional expressions common in everyday communication. We train a lightweight transformer-based extractor (0.74 MB) that achieves 99.03% average slot extraction accuracy across four typologically diverse languages: English, Vietnamese, Japanese, and French. The model demonstrates particularly strong performance on condition classification (99.4% accuracy) despite the 35-class complexity. With inference latency of 3.02ms on CPU, our approach enables real-time sign language generation in browser-based applications. We release our code, trained models, and multilingual dataset to support further research in accessible sign language technology.
The US-Japan Alliance in the Post-COVID Era – Intensifying US-China Rivalry and the Wavering Deterrence of US Forces
ポストコロナ時代の日米同盟: 先鋭化する米中対立、揺らぐ米軍の抑止力松岡美里 国際関係理論、安全保障研究
Written by Matsuoka Misato
Translation by Rabia Devici and Katja
The article analyzes the foreign policy of Shinzō Abe, one of the most notable and unusual political figures of contemporary Japan, who was the head of Japanese government twice for a total of almost nine years. It traces how his political philosophy formed under the influence of ideological views of his relatives, prime ministers N. Kishi and E. Satō, as well as his father, Shintarō Abe. It also considers Abe’s approaches to building Japan’s relations with the countries which are most important for its interests.Abe was perceived as a devoted ally of the U.S. in Washington. He established relations of confidence with presidents Obama and Trump. For this purpose, he strengthened Japanese-American military-political cooperation, took steps to support American strategy in the Asia-Pacific region. He implemented, even if without substantial results, steps to stabilize relations with China, trying to combine policy of containing Beijing with efforts to develop bilateral ties. Abe paid substantial attention to relations with India, including for the purpose of the idea, which was put forward by him and supported by the U.S., to establish quadrilateral cooperation of “democracies” in the Indo-Pacific region – the U.S., Japan, India, and Australia.His policy in the Korean direction was not successful. The relations with Pyongyang remained in deadlock, and, with Seoul, the most acute bilateral problems were not finally solved. Abe also paid great attention to policy aiming to conclude a peace treaty with Russia on the basis of a radical improvement of Japanese-Russian ties in all spheres. The reasons for his failure in these directions are discussed in this article.The article evaluates Abe’s efforts aimed at developing governmental documents and making the Diet adopt laws determining the basic directions of the foreign and military policy of the state. The author characterizes the results of the activity of S. Abe in the sphere of foreign policy and assesses its influence on the formation of the course of the Japanese government after his resignation.
At the end of 20th century, Japan successfully overcame an environmental crisis and improved the citizens’ quality of life. The crisis, referred to as kogai in Japanese, primarily comprised physical and chemical pollution issues. Nowadays, priorities on both global and national environmental agenda usually include the climate change issue (such as creating a low-carbon economy), biodiversity concerns (and problems of land cover changes), and waste and resources management (such as implementing a circular economy, or Sound Material-Cycle Society in Japan). However, some pollution-related challenges have not yet been fully resolved. The aim of this research is to assess the quality of environment in Japan based on national statistics and monitoring data. The problems considered are as follows: air pollution, water pollution, soil pollution, vibration, urban noise, land subsidence, and offensive odors. It has been found out that, despite Japan having made significant progress in solving these problems, there still remain such issues as pollution of air with fine particles and photo-oxidants, pollution of lakes and ponds with organic pollutants and of all types of surface waters with dioxins, cadmium soil pollution. The problems of noise and offensive odors in cities have not been solved. The article also features the achievements on the way towards solving the stated problems and instruments of environmental policy (an established monitoring system, open data access, active participation in international environmental policies, active engagement of local government, and responsibility of businesses). These instruments, when used in conjunction with others, allow Japan to successfully move towards the goals of sustainable development.
Increasingly, more and more people are turning to large language models (LLMs) for healthcare advice and consultation, making it important to gauge the efficacy and accuracy of the responses of LLMs to such queries. While there are pre-existing medical benchmarks literature which seeks to accomplish this very task, these benchmarks are almost universally in English, which has led to a notable gap in existing literature pertaining to multilingual LLM evaluation. Within this work, we seek to aid in addressing this gap with Cancer-Myth-Indic, an Indic language benchmark built by translating a 500-item subset of Cancer-Myth, sampled evenly across its original categories, into five under-served but widely used languages from the subcontinent (500 per language; 2,500 translated items total). Native-speaker translators followed a style guide for preserving implicit presuppositions in translation; items feature false presuppositions relating to cancer. We evaluate several popular LLMs under this presupposition stress.
This paper presents a comprehensive analysis of InnerSource adoption processes and their evolution within enterprises. First, a comparative analysis of Japanese and global enterprises highlights differences in the state of software sharing, perceptions of its importance, and barriers to implementation. Next, this study demonstrates that InnerSource adoption involves multi-layered, topological evolution beyond conventional staged models of program evolution. The research proposes three theoretical frameworks: InnerSource Topologies, which conceptualizes collaborative structures and categorizes internal collaboration levels; the Multi-layered Incentive Model, which combines monetary and non-monetary rewards at individual and project levels; and the InnerSource Circumplex Model, which helps organizations define InnerSource forms based on their specific needs. By mapping InnerSource evolution as a circumplex rather than simple staged progression, leaders can better adjust their focus during implementation. These frameworks help refine the previously ambiguous concept of InnerSource from the perspectives of sharing scope and community growth. These findings reaffirm that successful InnerSource adoption requires the parallel pursuit of top-down program structuring and bottom-up voluntary collaboration. They also contribute to fostering a sustainable innovation culture and enhancing software-sharing practices within enterprises. Furthermore, the newly proposed frameworks, particularly the Circumplex Model, offer versatile guidelines for organizations of varying cultural backgrounds and scales, enabling them to flexibly redefine and introduce InnerSource. This research is thus expected to advance corporate software sharing and spur innovation in diverse industrial contexts.
Acknowledging that large language models have learned to use language can open doors to breakthrough language science. Achieving these breakthroughs may require abandoning some long-held ideas about how language knowledge is evaluated and reckoning with the difficult fact that we have entered a post-Turing test era.
Cross-cultural research in perception and cognition has shown that individuals from different cultural backgrounds process visual information in distinct ways. East Asians, for example, tend to adopt a holistic perspective, attending to contextual relationships, whereas Westerners often employ an analytical approach, focusing on individual objects and their attributes. In this study, we investigate whether Vision-Language Models (VLMs) trained predominantly on different languages, specifically Japanese and English, exhibit similar culturally grounded attentional patterns. Using comparative analysis of image descriptions, we examine whether these models reflect differences in holistic versus analytic tendencies. Our findings suggest that VLMs not only internalize the structural properties of language but also reproduce cultural behaviors embedded in the training data, indicating that cultural cognition may implicitly shape model outputs.
Natural Language Inference (NLI) involving comparatives is challenging because it requires understanding quantities and comparative relations expressed by sentences. While some approaches leverage Large Language Models (LLMs), we focus on logic-based approaches grounded in compositional semantics, which are promising for robust handling of numerical and logical expressions. Previous studies along these lines have proposed logical inference systems for English comparatives. However, it has been pointed out that there are several morphological and semantic differences between Japanese and English comparatives. These differences make it difficult to apply such systems directly to Japanese comparatives. To address this gap, this study proposes ccg-jcomp, a logical inference system for Japanese comparatives based on compositional semantics. We evaluate the proposed system on a Japanese NLI dataset containing comparative expressions. We demonstrate the effectiveness of our system by comparing its accuracy with that of existing LLMs.
This article provides a brief overview of the field of Natural Language Generation. The term Natural Language Generation (NLG), in its broadest definition, refers to the study of systems that verbalize some form of information through natural language. That information could be stored in a large database or knowledge graph (in data-to-text applications), but NLG researchers may also study summarisation (text-to-text) or image captioning (image-to-text), for example. As a subfield of Natural Language Processing, NLG is closely related to other sub-disciplines such as Machine Translation (MT) and Dialog Systems. Some NLG researchers exclude MT from their definition of the field, since there is no content selection involved where the system has to determine what to say. Conversely, dialog systems do not typically fall under the header of Natural Language Generation since NLG is just one component of dialog systems (the others being Natural Language Understanding and Dialog Management). However, with the rise of Large Language Models (LLMs), different subfields of Natural Language Processing have converged on similar methodologies for the production of natural language and the evaluation of automatically generated text.
The article is devoted to the activities of the Japanese intelligence community in 1874– 1945. The main work, first, against the Russian Empire, and later against the Soviet Union, was carried out by the army and navy intelligence agencies of the empire. In some cases, diplomatic missions and military gendarmerie were also engaged in secret intelligence against our country following the instructions of the supreme command of the Japanese army and navy. The beginning of Japan’s intelligence activities in Russia dates back to 1874–1875, when residents from the army and navy were sent to Saint Petersburg and Vladivostok. Before the start of the 1904–1905 campaign, Japanese intelligence managed to organize channels for obtaining reliable information about the state of the Tsarist navy and army, which largely ensured the success of its armed forces in this war. After the “Course of the National Defense of the Empire” was adopted in 1907, Japanese army intelligence focused on intelligence work against Russia, while naval intelligence began to collect information mainly focusing on the United States and Great Britain. In the 1920s, the interest of the army intelligence agencies also switched to the armed forces and military industry of Western Europe and the United States, but, after the occupation of Manchuria in 1931–1932, army intelligence focused on the Soviet direction again. Due to the tightening of the administrative, counterintelligence, and border protection regimes in the USSR in the second half of the 1930s, the collection of information by means of radio intelligence and cryptoanalysis received significant development in the activities of the Japanese intelligence community. Largely due to information from army and naval intelligence, the Japanese leadership abandoned its plans to attack the USSR in the autumn of 1941 and was informed in the spring and summer of 1945 about the upcoming war with the Soviet Union. Despite Japan’s defeat in World War II, in the mid-1950s, it resumed the activities of its intelligence community.
Benchmarks play a significant role in the current evaluation of Large Language Models (LLMs), yet they often overlook the models' abilities to capture the nuances of human language, primarily focusing on evaluating embedded knowledge and technical skills. To address this gap, our study evaluates how well LLMs understand context-dependent expressions from a pragmatic standpoint, specifically in Korean. We use both Multiple-Choice Questions (MCQs) for automatic evaluation and Open-Ended Questions (OEQs) assessed by human experts. Our results show that GPT-4 leads with scores of 81.11 in MCQs and 85.69 in OEQs, closely followed by HyperCLOVA X. Additionally, while few-shot learning generally improves performance, Chain-of-Thought (CoT) prompting tends to encourage literal interpretations, which may limit effective pragmatic inference. Our findings highlight the need for LLMs to better understand and generate language that reflects human communicative norms.
Dhrubajyoti Pathak, Sanjib Narzary, Sukumar Nandi
et al.
Language Processing systems such as Part-of-speech tagging, Named entity recognition, Machine translation, Speech recognition, and Language modeling (LM) are well-studied in high-resource languages. Nevertheless, research on these systems for several low-resource languages, including Bodo, Mizo, Nagamese, and others, is either yet to commence or is in its nascent stages. Language model plays a vital role in the downstream tasks of modern NLP. Extensive studies are carried out on LMs for high-resource languages. Nevertheless, languages such as Bodo, Rabha, and Mising continue to lack coverage. In this study, we first present BodoBERT, a language model for the Bodo language. To the best of our knowledge, this work is the first such effort to develop a language model for Bodo. Secondly, we present an ensemble DL-based POS tagging model for Bodo. The POS tagging model is based on combinations of BiLSTM with CRF and stacked embedding of BodoBERT with BytePairEmbeddings. We cover several language models in the experiment to see how well they work in POS tagging tasks. The best-performing model achieves an F1 score of 0.8041. A comparative experiment was also conducted on Assamese POS taggers, considering that the language is spoken in the same region as Bodo.
Research on food image understanding using recipe data has been a long-standing focus due to the diversity and complexity of the data. Moreover, food is inextricably linked to people's lives, making it a vital research area for practical applications such as dietary management. Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities, not only in their vast knowledge but also in their ability to handle languages naturally. While English is predominantly used, they can also support multiple languages including Japanese. This suggests that MLLMs are expected to significantly improve performance in food image understanding tasks. We fine-tuned open MLLMs LLaVA-1.5 and Phi-3 Vision on a Japanese recipe dataset and benchmarked their performance against the closed model GPT-4o. We then evaluated the content of generated recipes, including ingredients and cooking procedures, using 5,000 evaluation samples that comprehensively cover Japanese food culture. Our evaluation demonstrates that the open models trained on recipe data outperform GPT-4o, the current state-of-the-art model, in ingredient generation. Our model achieved F1 score of 0.531, surpassing GPT-4o's F1 score of 0.481, indicating a higher level of accuracy. Furthermore, our model exhibited comparable performance to GPT-4o in generating cooking procedure text.
This special issue examines representations and constructions of pregnancy, childbirth, and breastfeeding in contemporary Japanese fiction in a selection of literary texts from the 2010s to the 2020s. It thus joins ongoing conversations and existing studies concerned with the representation of reproduction and motherhood in modern and contemporary Japanese culture (Saito, 1994; Seaman, 2016; Castellini, 2017; Harada, 2021). However, the essays in this section focus on depictions of pregnancy, childbirth, and breastfeeding in terms of narrating bodies as a way to articulate women’s experiences of physical and psychological oppression within Japanese society and redefine new forms of mothering, fathering, and parenting. This research investigates the ambivalence and complexity around motherhood and embodiment in contemporary women’s fiction. At the same time, it explores the connections between literary studies and contemporary sociocultural dynamics of gender and family.
Joseph P. Vitta, Paul Leeming, Stuart Mclean
et al.
Self-efficacy has emerged as a popular construct in second language research, especially in the frontline and practitioner-researcher spaces. A troubling trend in the relevant literature is that self-efficacy is often measured in a general or global manner. Such research ignores the fact that self-efficacy is a smaller context-driven construct that should be measured within a specific task or activity where time, place, and purpose domains are considered in the creation of the measurement. Task-based language teaching researchers have also largely neglected the affective factors that may influence task participation, including self-efficacy, despite its potential application to understanding task performance. In this report, we present an instrument specifically developed to measure English as a foreign language students’ self-efficacy beliefs when performing a dialogic, synchronous, quasi-formal group discussion task. The instrument's underlying psychometric properties were assessed (N = 130; multisite sample from Japanese universities) and evidence suggested that it could measure a unidimensional construct with high reliability. The aggregate scale constructed from the instrument's items also displayed a central tendency and normal unimodal distribution. This was a positive finding and suggested that the instrument could be useful in producing a self-efficacy measurement for use in the testing designs preferred by second language researchers. The potential applications of this instrument are discussed while highlighting how this report acts as an illustration for investigators to use when researching self-efficacy.
In building interpersonal relationships through omotenashi no kokoro (Excellent Service), Keigo is a crucial aspect to consider when communicating in Japanese. Keigo, both in terms of grammar and its usage concept, is quite complicated. Therefore, the role of instructors in delivering Keigo learning materials is expected to motivate students to improve their communication skills while mastering the art of creating interpersonal connections. Communicative Language Teaching (CLT) is chosen as the teaching method. This research aims to provide an alternative teaching experience focusing on omotenashi no kokoro. The research method used is descriptive with a qualitative approach. The research subjects are students of the Japanese Language Program at Universitas Teknologi Yogyakarta (UTY) taking the hospitality course. The research is conducted through observation by analyzing CLT-based learning activities using three steps: mechanical practice, meaningful practice, and communicative practice. The study results indicate that with the CLT method, students can use Keigo through word or phrase repetition, understand the meaning of words and expressions in sentences, and freely actualize themselves through improvisation and exploration of conversational contexts. A new finding from this research is that the CLT teaching method is effective when applied to the hospitality course, emphasizing building interpersonal relationships through omotenashi no kokoro.
Zoltan Csaki, Pian Pawakapan, Urmish Thakker
et al.
Recent large language models (LLM) exhibit sub-optimal performance on low-resource languages, as the training data of these models is usually dominated by English and other high-resource languages. Furthermore, it is challenging to train models for low-resource languages, especially from scratch, due to a lack of high quality training data. Adapting pretrained LLMs reduces the need for data in the new language while also providing cross lingual transfer capabilities. However, naively adapting to new languages leads to catastrophic forgetting and poor tokenizer efficiency. In this work, we study how to efficiently adapt any existing pretrained LLM to a new language without running into these issues. In particular, we improve the encoding efficiency of the tokenizer by adding new tokens from the target language and study the data mixing recipe to mitigate forgetting. Our experiments on adapting an English LLM to Hungarian and Thai show that our recipe can reach better performance than open source models on the target language, with minimal regressions on English.
Takuro Fujii, Koki Shibata, Atsuki Yamaguchi
et al.
This paper investigates the effect of tokenizers on the downstream performance of pretrained language models (PLMs) in scriptio continua languages where no explicit spaces exist between words, using Japanese as a case study. The tokenizer for such languages often consists of a morphological analyzer and a subword tokenizer, requiring us to conduct a comprehensive study of all possible pairs. However, previous studies lack this comprehensiveness. We therefore train extensive sets of tokenizers, build a PLM using each, and measure the downstream performance on a wide range of tasks. Our results demonstrate that each downstream task has a different optimal morphological analyzer, and that it is better to use Byte-Pair-Encoding or Unigram rather than WordPiece as a subword tokenizer, regardless of the type of task.