佛典中存在著無數的複合詞，這些複合詞是以何種方式轉譯成漢語？這些詞是否進入 漢語？本文藉由研究《金剛經》中的275 個複合詞，將之分成傳統所謂「六合釋」的六大 類，分別是：依主釋、持業釋、相違釋、帶數釋、鄰近釋和有財釋。總括而言，梵語複合 詞的種類其多，但是在翻譯成漢語時最常使用的是偏正式複合詞，或動賓式短語。其次， 筆者比較羅什和玄奘翻譯梵語複合詞的方式，發現羅什共有155 次沒有翻譯出梵文複合詞， 而玄奘只有兩次。不僅如此，玄奘會很忠實地翻譯出複合詞前後兩個成分，而羅什有時會 省略前詞或後詞。玄奘主張忠實地翻譯原文的結果，應用在複合詞的翻譯上，就是盡可能 地使用逐詞仿譯來翻譯複合詞，不僅是逐詞翻譯，連否定前綴和一些前置詞也會用漢語一 一對應。然而，玄奘因為忠於梵文原典，而不避繁複所仿譯出的詞彙，卻不如羅什的譯語 較令人容易接受。羅什的譯詞比玄奘漢化的程度更深，順應漢語的語言結構，讓人嗅不出 翻譯的味道，也因此廣為佛教徒所接受。反之，玄奘的一些譯詞雖然較為嚴謹精確，但卻 繁複拗口，因此其譯本不受歡迎，無法在民間流行。

This paper examines 275 Sanskrit compound words found in the Vajracchedikā Prajñāpāramitāhṛdayasutra (Diamond Sutra). According to the six Chinese traditional interpretations of Sanskrit terms, Liu He Shi 六合釋, these compound words are classified into six categories: tatpuruṣa (determinative compounds), karmadhāraya (descriptive compounds), dvandva (co-ordinative compounds), dvigu (compounds with a numeral), avyayībhāva (adverbial compounds), and bahuvrīhi (possessive compounds). In Chinese there are five main types of compounds; however, most of the Sanskrit compounds are translated into modified-head structure in Chinese. Besides modified-head compounds, many Sanskrit compounds are translated into Chinese as verb-object phrases. We also study the translation techniques employed by Kumārajīva and Xuanzang based on the translation of the Sanskrit compound words. It is found that Kumārajīva did not translate the compound words for 150 times, while Xuanzang, for only two times. The former sometimes did not translate the first or the second components in compound words, whereas the latter would faithfully translate every component, including negative prefixes and prepositions. Xuanzang also used calques (loan translations), a kind of literal morpheme-to-morpheme translation, to translate Sanskrit compounds. His translation of compound words mostly sticked to the Sanskrit word structures, and was supposed to be much more correct. However, nowadays Buddhists study only Kumārajīva’s translation of the Diamond Sutra, despite his less faithful translation skills. Kumārajīva’s translation of Sanskrit compounds words uses more familiar word structures and terms in Chinese, and thus one cannot tell that they were translated from Sanskrit. On the contrary, Xuanzang though faithfully follows Sanskrit word structures, he created some terms which were unfamiliar to the public and thus difficult to understand. That is why Kumārajīva’s translation is widely accepted in the Buddhist circles.