本文從複合化觀點論述漢語複合詞的興起緣由和發展方式，並提出「定型複合」為複合詞衍生的途徑。古漢語複合詞資料，以往研究已經做過相當的整理，是本文論述的重要依據。不過本文主張與以往研究未盡一致。本文全面觀照複合化早期歷史，但行文偏重定名複合詞與並列複合詞，因為它們在起源階段地位特殊。對於定名複合詞，本文大致是根據以往資料論述定名式的定型複合。至於並列複合詞，本文與以往研究分歧較大。本文主張並列複合詞最初至少有兩種來源，一種是聯合詞組內部成分發生語義溶合變化成詞，另一種是同義成分並列合成，是修辭/語用條件形成的複合詞。並列複合詞到春秋戰國轉向定型複合，自此定型複合即為漢語複合詞衍生的基本模式。

The origin of Chinese compounds can be traced far back to Classical Chinese at the ages of the Oracle Bones and the Bronze new cInscriptions. It has been widely assumed that compounds back then are all lexicalized phrases. This paper, however, shows that rules for nominal modifier-head compounds and for coordinative compounds are introduced no later than the age of the Bronze Inscriptions. The former rules apply a method called “patterned-compounding” in this paper to generate new nominal modifier-head compounds. This method, in brief, is to take a high-frequency component (modifier or head) as base form and synthesize the base form with certain component in the pattern that the high-frequency component normally occurs. In other words, it is create ompounds by analogy to existed ones. Rules for coordinative compounds, on the other hand, do not apply the patterned-compounding method originally. While the coordinative compounds accumulate in the language and high-frequency base forms come into being, rules for coordinative compounds also convert to the patterned-compounding method. The patterned-compounding becomes the most predominant way of generating lexical forms in Chinese.