Anak ADHD merupakan anak berkebutuhan khusus yang tidak bisa duduk diam, susah fokus dan suka mencari perhatian. Karenanya, mereka menemukan banyak kesulitan ketika belajar bahasa, terutama bahasa Mandarin. Bahasa Mandarin merupakan bahasa yang memiliki aksara dan nada yang berbeda dengan bahasa lainnya, sehingga memerlukan fokus dan daya ingat yang kuat untuk mempelajarinya. Penelitian ini bertujuan untuk menganalisis bagaimana strategi guru bahasa Mandarin mengajar siswa ADHD yang mempelajari bahasa Mandarin. Penulis menggunakan metode penelitian kualitatif dengan pendekatan studi kasus untuk meneliti proses kegiatan belajar mengajar antara guru dan siswa. Subjek penelitian ini adalah seorang guru bahasa Mandarin di kursus bahasa Mandarin Lianhe dan seorang siswa ADHD berusia 9 tahun yang telah menjalani terapi selama 5 tahun. Hasil analisis menunjukkan bahwa guru tersebut menggunakan pengalamannya dalam strategi mengajar bahasa Mandarin yang dipadukan dengan strategi mengajar siswa ADHD. Untuk meningkatkan keterampilan berbicara, guru mendorong siswa agar lebih banyak berbicara dengan menggunakan bahasa Mandarin dan memperbaiki kesalahan yang muncul. Ketika kelas, guru menjelaskan materi dengan selalu menjaga kontak mata dengan siswa. Guru juga menggunakan gambar, video, dan barang nyata untuk mengajar siswa ADHD. Ketika siswa ADHD tidak fokus mendengarkan, guru menggunakan permainan untuk menarik perhatiannya. Kata kunci: ADHD, Belajar bahasa Mandarin, Guru bahasa Mandarin, Strategi Mengajar
PT. Semen Imasco Asiatic, yang telah berdiri sejak 2020, merupakan sebuah perusahaan manufaktur semen yang termasuk bagian dari Hongshi Group asal China dan merupakan salah satu brand semen China yang dapat bersaing dengan brand lainnya, setiap dokumen tertulis dari PT. Semen Imasco Asiatic menggunakan dua Bahasa, yaitu Bahasa Mandarin dan Bahasa Indonesia. Penelitian ini bertujuan untuk mencari tahu pergeseran penerjemahan yang terjadi dalam dokumen tertulis PT. Semen Imasco Asiatic dan mencari tahu strategi penerjemahan dari dokumen tertulis PT. Semen Imasco Asiatic. Penelitian ini menggunakan metode kualitatif. Data penelitian untuk penelitian ini adalah tiga dokumen tertulis yang diberikan PT.Semen Imasco Asiatic kepada penulis. Analisis pergeseran penerjemahan dan strategi penerjemahan dilakukan dengan cara mengelompokkan kalimat sesuai dengan kategori pergeseran dan strategi penerjemahan. Berdasarkan hasil penelitian, PT. Semen Imasco Asiatic paling banyak terjadi pergeseran intra sistem dalam penerjemahan dokumen tertulis. Strategi penerjemahan yang paling sering digunakan oleh PT. Semen Imasco Asiatic adalah penerjemahan dengan kata netral.
Small Language Models (SLMs) enable cost-effective, on-device and latency-sensitive AI applications, yet their deployment in Traditional Chinese (TC) remains hindered by token-level instability - models unpredictably emit non-TC characters or code-switch into other languages. We address this practical reliability gap by creating PureTC-1B, a three-stage stabilization pipeline for Llama-3.2-1B-Instruct (an open-weight, instruction-tuned model released by Meta) using parameter-efficient LoRA adapters. Our method combines Continual Pre-Training (CPT) on TC-centric corpora, Supervised Fine-Tuning (SFT) with instruction data, and Direct Preference Optimization (DPO) using TC-adherence preferences to improve monolingual robustness without full-model retraining. On a benchmark designed to simulate real-world usage, PureTC-1B achieves a 51.3% relative reduction (micro-average) in non-TC output tokens versus the base model. On a Named Entity Translation (NET) task, PureTC-1B further reduces incorrect-language tokens by 77.2% relative to Llama-3B and 57.2% relative to Qwen-1.5B, indicating that robust TC adherence is attainable even at the 1B scale. The pipeline is reproducible, adapter-only, and hardware-friendly, offering practitioners a practical recipe to enhance language stability for TC and potentially other non-English languages.
Audio-aware large language models (ALLMs) have recently made great strides in understanding and processing audio inputs. These models are typically adapted from text-based large language models (LLMs) through additional training on audio-related tasks. This adaptation process presents two major limitations. First, ALLMs often suffer from catastrophic forgetting, where crucial textual capabilities like instruction-following are lost after training on audio data. In some cases, models may even hallucinate sounds that are not present in the input audio, raising concerns about reliability. Second, achieving cross-modal alignment between audio and language typically relies on large collections of task-specific question-answer pairs for instruction tuning, making it resource-intensive. To address these issues, previous works have leveraged the backbone LLMs to synthesize general-purpose, caption-style alignment data. In this paper, we propose a data generation framework that produces contrastive-like training data, designed to enhance ALLMs' ability to differentiate between present and absent sounds. We further extend our approach to multi-audio scenarios, enabling the model to either explain differences between audio inputs or produce unified captions that describe all inputs, thereby enhancing audio-language alignment. We refer to the entire ALLM training framework as bootstrapping audio-language alignment via synthetic data generation from backbone LLMs (BALSa). Experimental results indicate that our method effectively mitigates audio hallucinations while reliably maintaining strong performance on audio understanding and reasoning benchmarks, as well as instruction-following skills. Moreover, incorporating multi-audio training further enhances the model's comprehension and reasoning capabilities. Overall, BALSa offers an efficient and scalable approach to developing ALLMs.
Large Language Models (LLMs) possess extensive knowledge and commonsense reasoning capabilities, making them valuable for creating powerful agents. However, existing LLM agent frameworks have not fully utilized past experiences for improvement. This work introduces a new LLM-based agent framework called Retrospex, which addresses this challenge by analyzing past experiences in depth. Unlike previous approaches, Retrospex does not directly integrate experiences into the LLM's context. Instead, it combines the LLM's action likelihood with action values estimated by a Reinforcement Learning (RL) Critic, which is trained on past experiences through an offline ''retrospection'' process. Additionally, Retrospex employs a dynamic action rescoring mechanism that increases the importance of experience-based values for tasks that require more interaction with the environment. We evaluate Retrospex in ScienceWorld, ALFWorld and Webshop environments, demonstrating its advantages over strong, contemporary baselines.
Humor plays a significant role in daily language communication. With the rapid development of large language models (LLMs), natural language processing has made significant strides in understanding and generating various genres of texts. However, most LLMs exhibit poor performance in generating and processing Chinese humor. In this study, we introduce a comprehensive Chinese humor-related dataset, the Chinese Fun Set (CFunSet). This dataset aggregates existing Chinese humor datasets and includes over 20,000 jokes collected from Tieba-JokeBar, a Chinese online platform known for joke sharing. The resulting corpus comprises more than 160,000 entries. Leveraging CFunSet, we developed the Chinese Fun Model (CFunModel), the first large language model designed to handle various Chinese humor-related tasks including Crosstalk Response Selection, Humor Recognition, Joke Generation, etc. Experimental results demonstrate that CFunModel outperforms popular large language models in these tasks. Our CFunSet is available at https://huggingface.co/datasets/ZhenghanYU/CFunSet and CFunModel is available at https://huggingface.co/ZhenghanYU/CFunModel. A demostration video of our work is available at https://youtu.be/MOsISOJ66Ms.
Chinese higher education institutions have adopted a US-style tenure track system since the 1990s. This is an important reform aimed at modernizing China’s higher education system. In response, authors have begun to carry out close examination of the career system and analyse its implications in a national context (Republic of China). This study aims to present the key research themes, identify research gaps and offer recommendations from the increasing pool of Chinese-language literature on the tenure track system. A scoping review of Chinese language papers was conducted using the China National Knowledge Infrastructure (including the China Academic Journals Full-text Database, China Core Newspapers Full-text Database, China Doctoral Dissertations Full-text Database, China Masters’ Thesis Full-text Database, and China Yearbooks Full-text Database) (CNKI) database. Four major research themes were identified in Chinese discourse: (1) examining the tenure track system, (2) providing suggestions for better adaptation of the tenure track system in the Chinese context, (3) analysing the negative effects of the tenure track system, and (4) analysing the positive effects of the tenure track system. Generally, authors were concerned with the adaptation and cultivation of the US-originated tenure track system in the Chinese context and emphasized the importance of acknowledging its perceived negative influences on early-career scholars who have not received adequate attention. Overall, the authors demonstrate increasing interest in the tenure track system in China, and the literature is of variable quality. Further empirical studies are needed to analyse, evaluate and guide future improvement of the career system in the Chinese context in practice.
This study builds a Chinese Language and literature majors teaching system, by combining artificial intelligence (AI) technology and performance, with improving the teaching effect. This paper implements a classifier-based fast calculation method for successive breaking of active power flow depend on existing coordination algorithms. This method provides high-quality active power information for the static safety check and control of successive breaking. Furthermore, to improve the reliability of the power flow results, this work accelerates computation, increases the operation effect of the teaching system, and utilizes the NR method for algorithm-sensitive errors. To assess the efficiency of the implemented teaching system, a simulated classroom setting was used for testing. The implemented method was compared with Random Forest (RF) and Gradient Boosting (GB) methods with time of 15 $s$ in AI-based vocabulary check, and 35 $s$ in AI-based grammar check, power of 0.25 W in AI-based vocabulary check, and 0.59 W in AI-based on grammar check.
Abstract The development of Internet technology injects intelligent elements into Chinese language and literature education in Chinese universities. This paper will rely on the algorithms and models of Internet technology to create a new smart mode for Chinese language and literature education in Chinese universities. In terms of intelligent teaching, this paper adopts data mining and Web network technology to build two functional modules of online and offline processing and designs a user interest model based on multiple factors in offline analysis to further analyze the user behavior pattern so as to provide more accurate teaching support for teachers and students studying Chinese language and literature. In terms of teaching resource library design, this paper describes the process of teaching resource association based on knowledge points and, at the same time, solves the problem of manual entry in the traditional association method based on BERT text semantic encoding technology. The model can also recommend personalized teaching resources to users through the ITR model based on feature fusion. Comparative analysis indicates that the ITS model in this paper has a 6.24% increase in AUC value over the SKAT model, which is the next most effective model. The MAE values of the ITR teaching resource recommendation algorithm used in this paper on the data set with a training set ratio of 75% are 0.36, 0.40, 0.37, and 0.37, respectively, and the recommendation effect is significantly better than that of other algorithms in comparison. The inclusion of intelligent elements will promote the further development of Chinese language and literature education in Chinese universities.
: This paper aims to explore the importance of cultivating language skills and cultural literacy in Chinese language education for international students, and proposes corresponding teaching strategies. The paper first analyzes the current situation of Chinese language education for international students, pointing out that some students currently focus only on grammar and vocabulary in the language learning process, while neglecting the importance of cultural literacy. Subsequently, the paper elaborates on the interrelationship between language skills and cultural literacy in language learning, emphasizing their complementarity. Finally, the paper proposes some effective teaching methods to promote students' improvement in both language skills and understanding of Chinese culture.
Konflik budaya merupakan hal yang lumrah dalam kehidupan manusia, karena setiap orang dilahirkan dengan latar belakang yang berbeda-beda, dan tentunya norma, etika, moralitas dan nilai-nilai budaya tidak terlepas dari interaksi interpersonal. Konflik budaya hampir selalu terjadi dalam kehidupan nyata, namun juga dapat terwujud melalui pertunjukan seperti film. Penulis memilih film “The Farewell” (2019) dikarenakan terdapat konflik budaya antara Tiongkok dan barat yang terjadi dalam film tersebut dan film ini mendapat banyak komentar dari netizen di Douban, yaitu komunitas daring budaya Tiongkok yang berfokus pada rekomendasi dan komentar film, buku, dan musik. Dalam penelitian ini penulis ingin menganalisis bagaimana komentar netizen Douban terhadap konflik budaya dalam film “The Farewell” melalui teori perbedaan antara nilai budaya kolektivisme dan individualisme yang dikemukakan oleh Kim (1994). Data penelitian diperoleh dari komentar-komentar netizen Douban selama kurun waktu tahun 2019, yaitu tahun ketika film “The Farewell” ditayangkan, yang membahas tentang konflik budaya yang terdapat dalam film “The Farewell” dengan pendekatan kualitatif deskriptif. Hasil analisis menunjukkan bahwa dalam komentar singkat kategori “bagus”, netizen Douban setuju dan menghargai perbedaan antara nilai budaya Tiongkok dan barat yang terdapat dalam film tersebut. Namun, dalam komentar singkat kategori “umum” ada netizen Douban yang setuju dan ada yang tidak setuju terhadap perbedaan nilai budaya, ada yang lebih memihak pada nilai budaya kolektivisme dan ada juga yang lebih memihak pada nilai budaya individualisme. Dalam komentar singkat kategori “buruk”, netizen Douban tidak setuju dengan nilai-nilai budaya Tionghoa dalam film tersebut, mereka percaya bahwa konflik budaya dalam film tersebut hanya akan menciptakan stereotip orang Tionghoa di barat.
Of the four basic skills, namely, reading, is a fundamental tool that supports the development of other skills. In Chinese tertiary education, reading is a compulsory core subject for students. Although the benefits of teaching English reading through literature circles have been recognized in other countries for decades, there is still an insufficient practice of this activity among English-major students in the Chinese university context. This study aims to investigate the attitudes of Chinese English-major students toward literature circles as a method of teaching reading, and to determine the benefits of reading literature in language learning. The study employs a one-shot case study involving only one group exposed to the treatment, followed by a measure. Quantitative and qualitative data were collected to determine the participants' attitudes toward literature circles and their perception of the benefits of reading literature. An intact class of 41 students participated in this extracurricular reading activity through the convenience sampling method. The result obtained from the eight closed-ended questions indicated that the participants generally held favorable views toward literature circles. The results from the focus group interviews confirmed the benefits of reading literature in terms of cultural, linguistic, and personal enrichment. The findings suggest that more longitudinal studies involving comparison groups or qualitative studies are necessary to better understand the benefits of literature circles.
ABSTRACT Although the emergence of soft masculinity has incited much controversy in China, little research attention has been given to the discussion about the effeminate Chinese masculinity in the news media featuring public opinions at the national level. To fill this gap, this study employed critical discourse analysis to investigate the dominant discourses surrounding the “Prevention of Feminisation of Male Teenagers” debate and its ideological implications in the Chinese-language news media. The findings show that the media constructed anti-feminine discourse, nationalistic discourse, and anti-gender stereotype discourse with ideological implications for patriarchy, nationalism, and egalitarianism via various discursive strategies, specifically abstraction, authorisation, categorisation, morality, and rationalisation. This study sheds light on the diverse voices and concerns of the emergence of soft masculinity that have not been given much attention in the literature.
Against the backdrop of globalization, international Chinese language education has gradually emerged as an important bridge for disseminating Chinese culture and deepening mutual understanding among nations. Among this trend, ancient Chinese language resources stand out like a bright pearl, with their unique value receiving attention. Utilizing both literature analysis and case studies, this research delves deeply into the significant role of ancient Chinese language resources in international Chinese language education, and explores effective methods for tapping into and utilizing these resources. The ancient Chinese language resources possess multiple values in international Chinese language education. To effectively tap into and utilize these resources, this study proposes a series of strategies. Firstly, it is necessary to strengthen resource integration, systematically integrating scattered ancient Chinese language resources to form a convenient resource library, facilitating easy access and use by educators and learners. Secondly, it is essential to promote innovative teaching methods, exploring teaching techniques suitable for international Chinese language education in light of the characteristics of ancient Chinese, thereby enhancing teaching effectiveness. Simultaneously, personalized textbooks should be developed based on learners' needs and characteristics, making ancient Chinese language resources more attuned to learners' practical requirements. Furthermore, it is crucial to enhance teacher training, improving teachers' understanding and teaching abilities of ancient Chinese language resources, providing a solid talent guarantee for international Chinese language education. Lastly, modern technological means, such as artificial intelligence and virtual reality, should be fully utilized to assist in the teaching of ancient Chinese language resources, providing learners with a more vivid and visual learning experience.
Collaborative innovation systems comprise certain functions created by integrating a number of interconnected items in a certain order. These systems essentially create a connection between different elements for the achievement of a certain goal. To properly develop or transform a system, the relationships among the elements of the system must be well understood. Numerous structural models have been designed to be applied to collaborative innovation systems in higher education. Thus, the current paper deals with this gap by comprehensively analyzing the challenges that may arise for collaborative innovation systems in public higher education (PHE) in the era of industry 4.0, specifically in the context of developing countries. This study developed an integrated framework to identify and evaluate the main challenges of the collaborative innovation system in public higher education. This framework is applied to determine the subjective and objective weights of the main challenges of the collaborative innovation system in PHE in the era of industry 4.0. In addition, the framework is used to assess the preferences of PHE organizations over different main challenges of the collaborative innovation system in the era of industry 4.0. Finally, an empirical case study is taken to evaluate the main challenges of the collaborative innovation system in PHE in the era of industry 4.0. The results of this study found that; the holistic acceptance of the innovation with a weight value of 0.0614 has come out to be the most important challenge of the collaborative innovation system in PHE; in addition, the lack of technical infrastructure with a weight value of 0.0594 is the second most important challenge of the collaborative innovation system in the PHE, and educational policy has third with significance value 0.0588.
History of scholarship and learning. The humanities, Social sciences (General)
Emily Mak, Natasha Nichiporuk Vanni, Xintong Yang
et al.
Dual language learners (DLLs), especially those from immigrant families in the United States, risk losing their home language as they gradually shift to speaking English as they grow up. Given the potential benefits of bilingualism on children’s cognitive, linguistic, and social–emotional development, it is crucial to maintain children’s home language to foster bilingual development. The current literature suggests that parental beliefs toward bilingualism and the language and literacy environment are linked to children’s language development. With the growing number of DLLs living in the United States, little is known about what parental beliefs about bilingualism of their children are integrated into these bilingual households and parents’ role in home language maintenance. The present study addresses the gap in the literature by investigating low-income immigrant families, specifically Chinese American and Mexican American families, and exploring the parental perceptions of children’s bilingual language learning. Further, the present study examines the relations among parental perceptions of bilingualism, home language and literacy practices, and home language oral proficiency. Data were collected from a total of 41 Mexican American and 91 Chinese American low-income immigrant families with DLLs ages 50–88 months who had been recruited from Head Start programs and state-funded preschools in Northern California when the children were 3–4 years old. Information about shared reading frequency, home language exposure and usage, and parental perceptions of bilingualism was collected through parental interviews, and DLLs’ home language oral proficiency was individually assessed. No significant difference in home language oral proficiency was observed between the two groups. Principal Components Analysis on the parental perceptions of bilingualism measure revealed two components, “Importance of Being Bilingual” and “English over Bilingualism.” Stepwise regression analysis results show that “Importance of Being Bilingual” was associated with children’s home language oral proficiency after controlling for culture, child age, the frequency of home language shared book reading, and child home language exposure and use. The results show that parents’ positive beliefs toward bilingualism are related to the children’s use of that language and their children’s language outcomes. Implications and suggestions for home language and literacy support for DLLs are discussed.
With the advancements in large language model technology, it has showcased capabilities that come close to those of human beings across various tasks. This achievement has garnered significant interest from companies and scientific research institutions, leading to substantial investments in the research and development of these models. While numerous large models have emerged during this period, the majority of them have been trained primarily on English data. Although they exhibit decent performance in other languages, such as Chinese, their potential remains limited due to factors like vocabulary design and training corpus. Consequently, their ability to fully express their capabilities in Chinese falls short. To address this issue, we introduce the model named JIANG (Chinese pinyin of ginger) specifically designed for the Chinese language. We have gathered a substantial amount of Chinese corpus to train the model and have also optimized its structure. The extensive experimental results demonstrate the excellent performance of our model.
Recent studies have revealed that NLP predictive models are vulnerable to adversarial attacks. Most existing studies focused on designing attacks to evaluate the robustness of NLP models in the English language alone. Literature has seen an increasing need for NLP solutions for other languages. We, therefore, ask one natural question: whether state-of-the-art (SOTA) attack methods generalize to other languages. This paper investigates how to adapt SOTA adversarial attack algorithms in English to the Chinese language. Our experiments show that attack methods previously applied to English NLP can generate high-quality adversarial examples in Chinese when combined with proper text segmentation and linguistic constraints. In addition, we demonstrate that the generated adversarial examples can achieve high fluency and semantic consistency by focusing on the Chinese language's morphology and phonology, which in turn can be used to improve the adversarial robustness of Chinese NLP models.
We evaluate four state-of-the-art instruction-tuned large language models (LLMs) -- ChatGPT, Flan-T5 UL2, Tk-Instruct, and Alpaca -- on a set of 13 real-world clinical and biomedical natural language processing (NLP) tasks in English, such as named-entity recognition (NER), question-answering (QA), relation extraction (RE), etc. Our overall results demonstrate that the evaluated LLMs begin to approach performance of state-of-the-art models in zero- and few-shot scenarios for most tasks, and particularly well for the QA task, even though they have never seen examples from these tasks before. However, we observed that the classification and RE tasks perform below what can be achieved with a specifically trained model for the medical field, such as PubMedBERT. Finally, we noted that no LLM outperforms all the others on all the studied tasks, with some models being better suited for certain tasks than others.