Against the backdrop of generative AIs deep integration into news production, the linguistic characteristics of AI-generated news texts have become a key research focus. This study employs parallel corpora to conduct comparative analysis between AI-generated news and professional journalism from mainstream media outlets, utilizing corpus comparison, quantitative statistics, and discourse analysis methods to examine stylistic features and syntactic structures. Findings reveal that AI-generated news exhibits strong linguistic standardization but demonstrates stylistic homogeneity, weakened emotional expression, and pronounced lexical generalization tendencies at the stylistic level. At the syntactic level, it shows monotonous sentence patterns, uniform sentence length distribution, reliance on explicit conjunctions for logical coherence, and relatively low syntactic complexity. These characteristics reflect practical limitations in AI-generated news regarding information hierarchy construction, discursive expressiveness, and communication adaptability. The research contributes to deeper understanding of AI news language mechanisms and provides empirical references for optimizing intelligent news generation systems.
Hu Wenfang (Thu,) studied this question.