1 Introduction

Polysymus words are widely present in language, reflecting the flexibility and creativity of language. The radiation network model is an important theory for explaining ambiguity, but it suffers from subjectivity and a lack of quantitative analysis. The corpus based behavioral feature analysis method (BP analysis method) provides a quantitative approach for the study of polysemous words, which helps to construct semantic networks of polysemous words and analyze the relationships between semantic terms. The visual verb 'see' plays an important role in Chinese, and scholars at home and abroad have conducted extensive research on its polysemy. However, domestic research mostly focuses on qualitative analysis and lacks quantitative methods. The aim of this study is to use the BP analysis method to quantitatively analyze the "look" in the Peking University CCL corpus, construct a complete polysemous semantic network, reveal its internal details, and provide a new perspective for the study of Chinese word class diversity.

2 research steps

Introduced the corpus based behavior analysis method for studying the polysemy of the visual verb 'see'. The research is divided into four steps: firstly, retrieve the use cases of "look" from the Modern Chinese Corpus of Peking University, and select 2177 corpus that meet the criteria; Secondly, manually annotate these corpora and analyze their morphological, syntactic, and semantic features; Then, create absolute frequency co-occurrence tables and relative frequency co-occurrence tables to prepare for statistical analysis; Finally, the hierarchical clustering analysis method is used to quantitatively analyze the data, revealing the internal structure and trends of "seeing". The research focuses on analyzing the usage characteristics of the four tone "look", while the one tone "look" was excluded from the study due to insufficient corpus and limited distribution.

3 Results Analysis

The study on the polysemy of the visual verb "see" shows that its semantic distribution is most commonly "observe and analyze; think", accounting for about 28%, followed by "perceive external things with the eyes" and "observe, judge and draw conclusions", accounting for 26% and 23% respectively. The frequency of using terms such as "visit, visit", "diagnosis and treatment", "arrangement, preparation (old)", etc. is relatively low, among which the term "arrangement, preparation (old)" only appears twice in 2177 corpora, accounting for 0.09%. The clustering tree diagram reveals that the semantics of "seeing" are mainly divided into two categories: "social activity meaning" and "combination of cognitive and visual meaning", among which cognitive meaning dominates and has different hierarchical structures. Through behavioral characteristic analysis, it is confirmed that "the eyes perceive external things" is the prototype meaning of "seeing", as it has the highest number of horizontal identification codes and is the most flexible to use. The constructed semantic network shows that, starting from the prototype meaning, the semantic relationships and internal connections are complex. Among them, the meanings of "visit, visit", "arrange, prepare (old)", "diagnose and treat" reflect the expansion from visual meaning to social activity meaning, while the meanings of "depend on", "observe and draw conclusions", "be seen and treated", "perceive external things with the eyes", "observe and analyze" reflect the polysemous complexity of the visual verb "see", especially the close relationship between the two meanings of "eyes perceive external things" and "be seen and treated", indicating the development of visual meaning towards cognitive meaning.

4 Conclusion

This study conducted a quantitative analysis of the polysemy of the visual verb "see" through behavioral feature analysis, revealing its eight senses, among which the visual and cognitive senses are closely related and frequently used. A study was conducted on 2177 valid corpora based on the CCL corpus, and a polysemous semantic network of "look" was constructed to explain the associations between different semantic terms. This method has advantages in the study of Chinese polysemous words, making up for the shortcomings of quantitative research in China and expanding the application scope of behavioral feature analysis.