本文重新探討「被」的句法地位以及長短被字句之間的句法衍生關係，提出三個假 設：（a)「被」爲動詞，這一觀點得到來自反身代詞「自己」、漢語片段語結構以及被字 句的控制-提升變換特性三方面新證據的支持；（b)長被字句有近距離長被字句和遠距離 長被字句之分。前者與其相應的短被字句可以具有同一底層結構，而後者沒有短被字句 形式；（c)通過重新分析一些否定測試，文章假設「被」選擇非限定性子句爲補足語， 其中不含AspP或CrsP。

This paper revisits the categorial status of the Mandarin passive marker bei and the derivational relationship between long and short bei passives. In regard to its categorial status, there have been three major analyses: bei is a preposition, or is a verb, or it has the dual status of being both a verb and a preposition. This paper reviews these views, especially the latter two, pointing out their strengths and weaknesses. On this basis, it makes three claims. (a) Bei is a verb, which is supported by the new evidence from reflexive ziji tests, Chinese fragmentary constructions, and the chameleonic properties of the bei passive alternating between control and raising structures. (b) As for the structures of the long and short passives, by reviewing the uniform and non-uniform approaches, we provide the third approach, arguing that long passives can be further divided into local long and long-distance passives. The local long passive and its corresponding short form may be derivationally related, while the long-distance passive does not have a short form. (c) By reinterpreting the facts from some negation tests, we argue that the semi-lexical verb bei subcategorizes for a non-finite clause that does not involve Aspect phrase (AspP) or Currently Relevant State phrase (CrsP).