Hongyu Nan, Haiying Zhang

Hongyu Nan

June 30, 2026

Emphatic “是”, Event structure, Machine learning, Semantic role labeling, Chinese syntax generation.

This paper proposes a deep learning-based approach for automatically generating Chinese sentences containing the emphatic “是” (shì), grounded in the relationship between its syntactic behavior and the extraction of semantic roles from event structures. We develop a two-stage neural model: a BERT-based semantic role labeler first identifies key semantic components, which then serve as constraints for a Transformer-based conditional text generator. The proposed method produces grammatically correct emphatic “是” sentences and achieves strong performance in both automatic benchmarks and human evaluations. This study offers a novel framework for computationally modeling Chinese modality.

© 2026, the Authors. Published by ISCCAC

This is an open access article distributed under the CC BY-NC license