¹ This research was conducted as part of the Project on Linguistic Analysis, sponsored by the National Science Foundation. It is a continuation of work first reported in my ‘Some syntactic rules in Mandarin’, Proceedings of the I X International Congress of Linguists 191–202. Other works dealing with problems of Mandarin grammar are listed in my ‘Bibliography of Chinese linguistics’, to appear in Current trends in linguistics, vol. 2. I am indebted to Charles J. Fillmore, Anne Y. Hashimoto, Hsiao-Tung Lu, and Samuel E. Martin for constructive discussions.

² A recent exposition of transformational analysis can be found in Noam Chomsky's Aspects of the theory of syntax (Cambridge, Mass., 1965).

³ There are of course many other methods of diagramming the grammatical information contained in a P-marker, all equivalent to the tree graph. For a survey of these methods as well as a formal discussion of the tree graph, cf. Meyers and Wang, ‘Tree representations in linguistics’. Project on linguistic analysis (POLA) 56–63, 71–9 (Report no. 3r, Ohio State University Research Foundation, May 1963).

⁴ The lower-case spellings of Chinese expressions in this paper follow the official Pinyin rules. Expressions spelled in capital letters refer to classes of morphemes or classes of phonetic shapes. Thus, BU is bú, bu, b-, etc., depending on the phonological rules which will eventually operate on it.

⁵ For example, see Y. R. Chao's Mandarin primer 58 (Cambridge, Mass., 1948).

⁶ Positing the aspect markers before the verb is easily justified. One obvious reason is their interaction with BU regardless of intervening structures, as illustrated in 3e. Another is that they are mutually exclusive with emphatic markers and certain auxiliary verbs and verbal complements. These conditions of mutual exclusion are most economically stated as selectional choices before the verb. Of equal importance is the consideration that many syntactic rules need to refer to the verb and its following object; all these rules are simplified if aspect markers do not intrude between.

⁷ For Cantonese, this information was provided by Anne Y. Hashimoto. Although the situation has not yet been examined in detail, she believes that a solution similar to the one proposed here would be useful in Cantonese also. For the Min dialects, see Yuán, Jiā-huá et al., Hànyŭ Fāngyán Gàiyào 283, 312 (Peking, 1960).

It is well known that in many Germanic and Romance languages there is an aspect marker that is homophonous with the verb have ‘possess’. According to the present analysis, it is interesting to note that Mandarin also has this feature: -y⊖u (aspect) is homophonous with the verb y⊖u ‘possess’.

BU changes to méi before -y⊖u (aspect), y⊖u ‘possess’, y⊖u ‘exist’, the determiner y⊖u, and the comparative y⊖u (as in 8b). Although these five morphemes are all written with the same logograph, they have different grammatical properties. I distinguish the first of these by a preposed hyphen.

⁸ Some general features of disjunctive questions are discussed in the Mandarin primer 58 f. I follow Chao in the use of the term ‘A-not-A question’.

⁹ Cf. e.g. Dīng Shēng-shù et al., Xiàndài Hànyŭ Yŭfă Jiănghuà (Peking, 1961): whereas sentences like 6a are ‘relatively rare’, sentences like 6b–c are ‘commonly seen’ (205).

¹⁰ Cf. ‘Some syntactic rules in Mandarin’ (see fn. 1).

¹¹ Certain predicates can follow diverse and complicated subjects and function in the capacity of ‘tag questions’, in the manner of the French n'est-ce pas?. These include shì-BU-shă ‘yes-not-yes’, xíng-BU-xíng ‘feasible-not-feasible’, dùi-BU-dùi ‘correct-not-correct’, and hăo-BU-hăo ‘good-not-good’. The grammatical mechanism for deriving these constructions is essentially the same as for the other A-not-A questions. Among these constructions, shì-BU-shì has to be derived more indirectly than the others because shì is a transitive verb; i. e. through deleting the object or shifting it to precede the verb.

¹² Note that while forms like 9c have long been common in Southern Mandarin, their use in Northern Mandarin began only recently. Cf. Xiàndai Hànyŭ Yŭfă Jiănghùa 206, where the examples given include Tiān-y⊖u méiy⊖u liàng? ‘Has it become dawn?’ and Tā -y⊖u méiy⊖u qĭlai? ‘Has he gotten up?'.

It is generally thought that this usage is borrowed into Northern Mandarin as a ‘southernism’. It seems just as convincing to me, however, to explain the syntactic change on the basis of regularization. Formerly, in Northern Mandarin, deletion could take place from either the affirmative or the negative predicate in the plain A-not-A question; but if the A-not-A question contained aspect markers, the deletion had to be made from the negative predicate. In the current dialect, this is no longer true. If the restriction had persisted, we would have to add another condition to the rule T 2 in Section II, namely condition (iv): If 2 = ASP, only Change (A) is possible.

¹³ In this discussion, the solution is based purely on the relations among affirmatives, negatives, and a special type of disjunctive questions. After this paper was completed, T. L. Mei called to my attention another convergent bit of evidence. In disjunctive statements of the form Nom₁ Pred₁ kěshì Nom₂ Pred₂, where Pred₂ is the negative partner of Pred₁, the deletion extends to the leftmost member of the right verb phrase. Thus from Tā xĭhuan kàn shū ‘He likes to read books’ and Tā qùguo Shànghăi, we get Tā xĭhuan kàn shū kěshì w⊖ bù xĭhuan ‘He likes to read books but I don't’ and Tā qùguo Shànghăi kěshì w⊖ méiy⊖u ‘He has been to Shanghai but I haven't’.