Study-Chinese

Analysis of teacher-ChatGPT interaction in designing conflict-based communicative tasks for university-level Chinese oral expression courses, examining AI's role and impact.

Explores the feasibility of Chinese semantic parsing to Discourse Representation Structures without labeled data, proposing a data collection pipeline and a fine-grained test suite.

Introducing CPG-EVAL, the first systematic benchmark to evaluate Large Language Models' knowledge of pedagogical grammar in Chinese language teaching contexts.

Analysis of BERT-based Chinese Spelling Correction models, highlighting overfitting to error patterns and proposing a simple random masking strategy for better generalization.

A novel approach to Chinese Spelling Correction (CSC) that rephrases entire sentences instead of character tagging, achieving state-of-the-art results and improved generalizability.

An in-depth analysis of applying Deep Factorization Machines to the Duolingo Second Language Acquisition Modeling (SLAM) task for knowledge tracing at the word level.

Explores the application of the Unix DIFF utility for NLP tasks like difference detection, rule extraction, data merging, and optimal matching, highlighting its practicality and versatility.

Analysis of the winning ensemble model (GBDT+RNN) for predicting student knowledge gaps in second language learning, based on the 2018 Shared Task on SLAM using Duolingo data.

Analysis of fairness in predictive models for second-language learning, evaluating bias across platforms and countries using Duolingo data.

HSK study materials for Study-Chinese

Overview of Project MOSLA, a unique longitudinal, multimodal, and multilingual dataset capturing the complete second language acquisition process over two years.

Analysis of the KOSHIK architecture for scalable Natural Language Processing using Hadoop, covering its components, implementation steps, performance evaluation, and future directions.

This paper explores the KOSHIK architecture for scalable Natural Language Processing using Hadoop, detailing its implementation, performance evaluation, and future directions.

Analysis of using specific prompts with Large Language Models to target CEFR and EBCL levels (A1, A1+, A2) for personalized Chinese language learning.

A research paper introducing SLABERT, a novel framework using BERT to model positive and negative cross-linguistic transfer in second language acquisition, based on Child-Directed Speech data.

An analysis of how neural language models acquire a second language, exploring cross-lingual transfer, linguistic generalization, and comparisons to human L2 acquisition.

An empirical study comparing verb-noun usage in Chinese and English newspapers, and its impact on the writing of English-speaking Chinese learners.

Analysis of a research paper investigating the impact of Virtual Reality simulations on student motivation in foreign language learning, including methodology, results, and future implications.