Exploration of language complexity assisted by large language models
HE Wei & HU Ziwei
Foreign Language Learning Theory and Practice ›› 2026, Vol. 200 ›› Issue (2) : 25.
Exploration of language complexity assisted by large language models
Using Large Language Models (LLMs), this study conducts a systematic analysis of the distribution characteristics of common English verbs in transitivity processes and their corresponding usages, aiming to demonstrate the complex nature of language as an organic semiotic system. Research findings reveal a distinct macro-level asymmetry in the distribution of common verbs across material, mental, and relational processes. Notably, verbs that realize material processes are predominant. In the realization of processes, English common verbs exhibit a core periphery distribution: verbs that realize material processes occupy the core, whereas those that realize relational processes are predominantly peripheral. Non-core meanings of verbs primarily originate from material processes, followed by mental processes, with relational processes contributing the least. Furthermore, approximately 80% of verbs predominantly realize a single process type, whereas around 20% demonstrate combined usage. The Pareto-style imbalance exhibited by English common verbs in their transitivity processes serves as a microcosm of language as an organic symbolic system. It specifically embodies multiple traits of the system, such as wholeness, dynamics, nonlinearity (emergence and uncertainty), asymmetry, adaptability, and evolvability. This precisely indicates that language systems are simultaneously shaped by both universal biological laws and social-cultural contexts.
Large Language Model / language as an organic semiotic system / transitivity / usage of verbs / complexity