该文尽可能依据西方原典，追溯古希腊语法到普遍语法关于“言辞成分”或词类研究的轨迹。经过三个学派250年的前赴后继，希腊语词类从三分、四分、五分、六分，最终形成八分体系。虽然引入希腊语词类模板，但拉丁语词类从四分到八分也经历了300多年。此后思辨语法总体上接受了八分。普遍语法则有九分、五分（提出名形动层包关系）、四分、八分（提出形容词独立）等。文献追溯和梳理结果表明：意义或意指才是词类的认知本质，而形式和功能等都是基于类比方法的识别手段。对词类划分要秉持开放态度，不同体系可以并存。词类划分与句子成分分析可以并行不悖，不排除相对映射。解决词类与句子成分之间的协调手段是多样化的。

This paper attempts to trace the trajectory of research on “parts of speech” (λ?γου μ?ρη) from ancient Greek grammar to universal grammar, based as much as possible on Western original texts. After 250 years of successive efforts by three schools of thought, the Greek word classes have evolved from three divisions, four divisions, and five divisions to six divisions, ultimately forming a system of eight divisions. Although Greek word class templates were introduced, Latin word classes went through more than 300 years from four to eight. Afterwards, speculative grammar adopted an overall eight divisions. Universal grammar has nine divisions, five divisions (indicating the hierarchical relationship between nouns, adjectives, and verbs), four divisions, eight divisions (indicating the independence of adjectives), and so on. The results of literature review and sorting indicate that meaning or signifying is the cognitive essence of word classes, while form and function are all recognition methods based on analogy. We should maintain an open attitude towards the classification of word classes, as different systems can coexist. The division of word classes and the analysis of sentence components can coexist without excluding relative mapping. The coordination methods between word classes and sentence components are diverse.