摘要
本文認為「底」的來源是「之」，不是「者」，於是要解釋為什麼介詞「之」字只能用在語中，變成「底」後卻既能用在語中（如「飲水喫草底漢」），又能用在語末（如「問底」）。
音韻部份說明「之」字文讀保存-J-介音仍舊是「之」，白讀失落-j-而變成「底」；同時也說明「者」字不能變作「底」。
語法部份說明產生「底」字語末用法的兩條途徑。(1)由於漢代〔名＋數＋量〕＞〔數＋量＋名〕這個語序變化，先秦的〔S＋VO者〕就變成〔VO者＋S〕（如「定殷者將吏」）。
此後又產生了〔V者O〕、〔N者N〕N這兩種結構。在這三種結構中，「之」、「者」可以互易。「之」變成「底」以後，「底」、「者」繼續在語中的位置互易。受了「者」字語末用法的沾染，「底」字也逐漸能用在語末。(2)從南北朝末年開始，出現了一種新興的結構〔動詞組＋之者〕。本文假設〔之者＞底者＞底底＞底〕，結果使〔動詞組＋之者〕變成〔動詞組＋底〕。
When the particle ti 底 appeared in Late Middle Chinese, around the 8th or 9th century, it was primarily used as the marker of nominal subor-dination in the middle of a nominal phrase, hence similar to the classical chih 之. It was also used as a nominalizer at the end of a nominal phrase, hence similar to the classical che 者. Because of the dual similarity, previous scholars have proposed that ti was derived from chih, or from che, or from both. The present paper agrues that ti was exclusively derived from chih and tries to account for the occurrence of ti in phrase-final position.
The phonological section (1) shows that chih split into a doublet because the literary form preserved the -j- medial whereas the colloquial
之 OC*tj□g＞tj□i＞tj□i＞t□i＞底 MC tei＞ti
之 *tj□g＞tj□i＞ 之 MC tsi＞chih
(2) argues that all known instances of similar splits involve only the loss of -j-, never the development of -j- into -i-, and this fact supports Li Rong's view that IVth Division words, including ti, do not have medial -i- in MC, and (3) after examining the phonological patterns of interchange of characters in Tun-huang texts, due to merger, and Tibetan transcriptions of Late Middle Chinese, concludes that 者 tja＞tsja＞che could not develop into 底 tei＞ti, nor could chih and che merge into a single form.
The sections on historical grammar propose two routes through which ti, derived from chih, must have developed into a nominalizer. (1) Middle Chinese has three nominal constructions in which chih may substitute for che: (A) 〔V O che〕＋S, (B) 〔(S) V che〕＋O, and (C) N che N. (A) came about because the Han dynasty witnessed a word-order change N＋〔Num＋M〕＞〔Num＋M〕＋N, which triggered a similar word-order change S＋〔V O che〕＞〔V O che〕＋S. (B) joined (A) when in Middle Chinese the agent／patient distinction was neutralized. After chih became ti, ti continued to interchange with che in (A), (B), and (C), and this interchange-ability eventually spread to the phrase-final position previously occupied by che. (2) In the 6th and 7th centuries, before the rise of ti as a particle, there emerged a new nominalizing construction VP chih che "the one who／which…", which remained current up to the 11th century. It was proposed that chih che became ti by assimilation and haplology:
之者 tj□i tja ＞ tei ta ＞ tei tei ＝ 底底＞底 ti
〔VP chih che〕 thus became 〔VP ti〕 "the one who／which…"
The replacement of 底 ti by 的 te first occurred in 1238, in a stele recording a royal license granted by the Mongol emperor to a Taoist temple.