About me
I am the Director of the Institute of AI and Language Science and a Full Tenured Professor of Computational Linguistics in the School of Foreign Languages at Tongji University. My work explores the intersection of computational linguistics, cognitive computation, speech and language technologies, and artificial intelligence. My research investigates how humans and machines process language by combining statistical modeling, deep learning, large language models, speech self-supervised learning, and cognitive-neuroscience methods such as eye-tracking, EEG, and fMRI. I am especially interested in building interpretable computational measures, such as contextual fit, that connect machine representations with human behavioral and neural dynamics. I have developed specialized large language models, speech/language analysis methods, major databases, and corpora that are used in both academia and industry. As global efforts in AI + Education continue to grow, I am committed to using data-driven and cognitively grounded AI approaches to advance language research toward big language science.
π¬ Research Focus
My work sits at the cutting edge of computational linguistics and cognitive AI, where I develop novel computational methods to understand human language processing and machine intelligence.
Cognitive Computation & Brain-inspired AI
Developing computational models that explain how humans process language, using eye-tracking, EEG, and fMRI data to understand the neural mechanisms of discourse comprehension.
Large Language Models & Reasoning
Evaluating and enhancing LLM reasoning capabilities, fine-tuning transformer-based models, and developing attention-aware computational metrics for multi-modal language processing.
Affective Computing
Understanding emotion and sentiment in human language through computational approaches, bridging cognitive and affective dimensions of language.
Advanced Statistical Analysis
Applying sophisticated statistical methods including GAMM, Bayesian modeling, and time-series analysis to linguistic and cognitive phenomena, as well as neuro data (EEG and fMRI).
Digital Humanities & Computational Text Analysis
Formal and computational models of discourse structure; computational measurement of coherence, cohesion, and information flow; discourse dependency and cross-framework conversion (RST, PDTB, dependency); multilingual discourse parsing; distant reading.
AI Methods, Ethics & Didactics
Efficient training and inference in LLMs; evaluations and benchmarking of AI models; cognitive-inspired reasoning in LLMs; critical assessment of AI biases, cultural tendencies, and societal impacts; development of DH-related curricula integrating AI literacy and ethical reflection.
π Academic Positions
π Journal Editorial Roles
π Impact & Recognition
My research has led to 30+ peer-reviewed journal publications in top-tier international venues including Cognition, Cognitive Science, Linguistics, Neural Networks, and PNAS, and 10 confernce papers in ACL and other AI top conferences, as well as 10+ publications in leading Chinese CSSCI journals such as δΈε½θ―ζ and ε½δ»£θ―θ¨ε¦. Many of my works have been reprinted in 人倧ε€ε°θ΅ζ and δΈε½η€ΎδΌη§ε¦ζζ. My research has been featured in MIT Technology Review.
π» Technical Expertise
Experimental Methods: Eye-tracking Β· EEG Β· fMRI Β· Online experiments
Languages: Chinese (native) Β· English (fluent) Β· German (intermediate) Β· Japanese (intermediate)
π Current Projects
Iβm currently working on projects that combine cognitive neuroscience, artificial intelligence, and speech/language modeling to understand how humans and machines process language:
- Large Language Model reasoning and evaluation frameworks, including methods that separate prediction from contextual fit in model readouts
- Cognitive computation models for brain-inspired AI, with a focus on how contextual compatibility shapes language processing
- Sparse localization of Mandarin lexical tone representations in speech self-supervised learning models
- Advanced statistical approaches to linguistic, prosodic, and cognitive data using GAMM, Bayesian methods, and cross-domain model comparison
- Cross-linguistic and multi-modal studies of affective computing, including how affect representation and expression decouple across alignment stages
I'm always excited to discuss research collaborations, student supervision opportunities, or innovative applications of computational linguistics, AI and cognitive computation. Feel free to reach out at sharpksun at hotmail.com!