本研究以語料庫語言學的角度，嘗試透過大量語料庫語料，以CART決策樹分類(Breiman et al. 1984)配合VARBRUL變異數規則分析(Rand and Sankoff 1990; Robinson, Lawrence, and Tagliamonte 2001; Sankoff, Tagliamonte, and Smith 2005)等方法，證實了漢語結構助詞「的」、「得」、「地」的使用規則與實際上的語言使用是大致符合的。在使用結構助詞的認知上，因前後環境的交互作用，產生了選用結構，最後以「的」、「得」、「地」作為標記（徐陽春2004）。量化資料清楚說明「的」/「地」，「的」/「得」較不常混用，較常混用的是「得」/「地」。對於此三個漢語結構助詞的分用與合用的建議，應以整體社會語言使用者使用情形來決定。本研究建議，分用比較符合現階段整體的使用方式。本研究亦提出認知概念模型來描述三種不同的漢語結構助詞與前後環境的關聯性。本研究結果預期對於漢語辭典、華語教學、語言政策、自然語言處理做出參考與貢獻。

This study, with a perspective of Corpus Linguistics, tries to utilize a CART analysis (Breiman et al., 1984) and a VARBRUL analysis (Rand and Sankoff, 1990; Robinson, Lawrence, and Tagliamonte, 2001; Sankoff, Tagliamonte, and Smith, 2005) with a large number of corpus data to prove that the grammar rules of the three structural particles ＂de＂ in Chinese are consonant with its real use by language users, and to prove that language users are aware of the structure before selecting an appropriate particle（徐陽春，2004）. Moreover, the quantitative analyses show the lower degree of misuse between 「的」／「地」 or 「的」／「得」, but higher degree of misuse between 「得」／「地」. This study suggests that independent usage for each structural particle is required, since the quantitative analyses of corpus data show the consonance with the real use of language users. Additionally, three conceptualized models for the three structural particles have been proposed. This study expects to contribute to the composition of Chinese dictionary, Chinese language teaching, language policy, and natural language processing.