摘要
本文檢討西夏語具有「空間意義」的語詞。指出西夏語只有四個指示空間方位的後置詞：
“□”：標示有定的處所，用以回答“□□／何方”。
“□”：標示動作的目標，有強調正在進行中且指向明確的作用。有時也泛指物體上方，或某平面之上。
“□”：標示行動、狀況的源點，或終點。若一句之中同時有終點和源點，通常與“□、□”同現。
“□”：標示領域或範圍內存在某一行為或狀況，偏重抽象的認知領域。
這四個指示空間方位的後置詞，是由名詞性的位置詞虛化而來的。其中反映了語言之間互相影響的一個層面──詞形是本語言既有的詞彙虛化的結果，結構形式則借自異語言（阿爾泰語）。
The Russian scholar Kepping (1979) placed the following Tangut words under the category of substantative nouns:
S 1454 □ ‧u2 inside; S 0074 □ tshiaw1 above.
S 2821 □ khiu1 below; S 5232 □ ngu2 in the middle.
S 1894 □□ ‧w□2 rie2 ahead; S 1427 □ nIuo1 behind.
S 0065 □ ndr□2 from outside;
Kepping referred to these morphemes as "locational morphemes". She felt that they were a small category that had some special characteristics.
However, she also mentioned that Tangut had spatial postpositions, including: S2114 □ ndo1 to, toward; S1763 □ ‧a2 in, into; S1783. □ kha1 inside. The difference between the locational morphemes and the spatial postpositions was that the locational morphemes also had adverbial functions. They attached a semantic component of location or direction onto nouns. In Kepping (1985), she still held on to her original theory with only minor adjustments concerning the categorization of words.
A language that has such a complex system of locational morphemes in worth investigating for its own sake. Specifically, we can ask the following three questions: 1) what differences exist between the two categories? 2) What are the semantic and etymological differences between the two categories? 3) Is there a simpler method of classifying these locational morphemes? Moreover, since Tangut texts are among the oldest written records in the Tibeto-Burman language, dating all the way back to the eleventh century, any findings concerning its grammatical structure will have important ramifications for the reconstruction for the Proto-Tibeto-Burman grammatical system. Thus, it is important to examine this system carefully and see how the results compare with Kepping's findings.
My results can be summarized as follows:
Because Tangut uses a character script, it is impossible to tell from its orthography which characters are empty or full words. Each character must be examined as a minimal morpheme, and then from the vernacular texts available to us at this point, we must strive to examine each minimal morpheme's grammatical function based on its features of syntactic construction.
Following the above line of reasoning, I have found that in Tangut the locational morpheme can be divided into full words (in this case position words) and empty words (in this case postpositions) as shown below.
1) A position word is a substantative (full word) which expresses the relative positions of things. They form a listable and very versatile subclass of place words, which share the same general properties of substantatives in occurring as subjects, or as objects (the directional complement in front of the verb). When the relative position is a bound form, it is very often modified by other substantatives, such that we have the construction "Determiner-Noun" Compounds. Moreover, sometimes the compound occurs in combination with the position word as the head of the construction, which is the chief reason that they are regarded as full words. In general, the position words are as follows:
east □ vi□2 west □ lie
north □ lIa1 south □ zi□1
front, ahead □ ‧iu2, □ rie2 behind, after □ nIuo1, □ ku1
left □ ldi□1 right □ tsiwe1
above □ phu2 in the middle □ ngu2, □ ka1
below, under □ mIe2, □ khiu2, □ tsie1 inside □ khu2
inside, within □ u2 outside □ ndI□2
side, beside □ pha1, □ mbiu2 between □ zIa1
within the boundary □ ‧won1 here, there □ twu1, □ tIei2
direction, side □ rie2……
Position words may also be compounded together and combined with other full or empty morphemes to form lexically compounded position words.
2) In addition, there are four spatial postpositions, each with its own syntactic meaning. They serve only to mark the grammatical functions that make a noun or a noun phrase, the place where an event or state occurs, as defined below:
a) □ ndo1 is used to label something as a definite location. It is often used when answering the question □□ ldIon2 rie2 "where".
b) □ tshia□1 is used to express the goal of an action, which symbolizes the direction of the action in progress. Simultaneously, it may be a broad indication of the place above something, or on the surface of something.
c) □ ‧a2 is used to signal the place from which something comes or ends. If there is both a source and a goal in the same sentence, either □ ndo1 or □ tshia□1 will occur together with □ ‧a2.
d) □ kha1 is used to express the inner side of some district or category, which is partial to the abstract area of cognition.
It is easy to see that these postpositions, namely, □ ndo1 and □ tshia□1, which indicate spatial direction, have retained their substantative meaning. Moreover, I have found that some of these substantatives, namely □ (below), □ (inside, within), and □ (direction, side), have undergone the process of grammaticalization. If we put these two facts together, we can surmise that postpositions were derived from substantative position words through the process of grammaticalization. This is why Kepping (1985) had such a difficult time explaining the meanings of these characters.
Grammaticalization may occur as an internal change in a language, but may also occur when languages come in contact with one another. Proffessor Gong, Hwang-cherng (1989b) has proposed that the postposition case particles in Tibeto-Burman are a result of the influence of Altaic languages. Thus, I propose that the postpositions are a grammaticalization, of the (full) position words, with their construction pattern borrowed from the Altaic languages.