This paper attempts to present the Mandarin phonological system after the generative fashion.1 We find it convenient to treat this part of the grammar in two components: namely, a syllable grammar and morphophonemics. The former attempts to designate the structure of the basic syllables independently of the syntactic component of the grammar. It consists of a set of P-rules to generate strings of phonemes for the basic syllables. The latter operates on sequences of these syllables with intrasyllabic information designated by the syllable grammar, and with category and intersyllabic information which can be given by the syntactic component. It consists of T-rules and gives a phonetic representation of sentences as its output.

It is well known that an English sentence may have its accent, or point of greatest prominence, on any of several syllables. This phenomenon has been called SHIFT.1 It is the feature of English that permits us to render such a sentence as The man can see the boy in a variety of ways, as follows:

The man can see the Boy.

The man can SEE the boy.

The man CAN see the boy.

The MAN can see the boy.

1. The present paper is a critique of Halliday's ‘Syntax an;’, the consumer' (1964).2 As avid consumers of linguistic theory (our interest stemming from; a very general concern with the psychology of language) we are naturally delighted that Halliday has our welfare at heart. He states that a theory of competence should not be regarded as the terminal point of syntactic study, but that ‘the explanation of linguistic performance’ and ‘the description of language for the purpose of various specific applications’ are important and reasonable goals. With this attitude we enthusiastically concur. Halliday outlines his basic point of departure with the following question (1964: 11–12): ‘Language may be described for a wide range of purposes, or, if that is begging the question I want to ask, there is a wide range of purposes for which a description of language may be used. The question is: do these various aims presuppose different ways of using the same description, or are they best served by descriptions of different kinds? Is there one single “best description” of a language, or are there various possible “best descriptions” according to the purpose in view?’

In his illuminating article for the Max Vasmer Festschrift, Jakobson (1956) gave a concise description of the implementation of the voicing feature in Russian obstruents. In addition to stating well-known facts about the pronunciation of contemporary standard Russian, Jakobson brought to the attention of students of Russian phonetics a fact which had been completely overlooked in previous investigations, namely the determinative role of the segment FOLLOWING the morphonemes {v} and {v,} in the distribution of voicing, rather than these sounds themselves. With respect to the feature of voicing, obstruent clusters containing medial {v} or {v,} were demonstrated to be entirely composed of voiced segments or unvoiced segments, depending on the value of the last obstruent in the cluster. Thus it was pointed out that voiceless obstruents were assimilatively voiced not only in Πρосьба [-z, b-] ‘request’, сдача [zd-] ‘submission’, ĸ жене [gž-] ‘to one's wife’, трехДнеВньІЙ [-γd-] ‘3-day’, but also in ОТ ВДоВьІ [-dvd-] ‘from the widow’, с вДовоЙ [zvd-] ‘with the widow’, к вДове [gvd-] ‘to the widow’, оΤ взгляДов [-dvzg-] ‘from the glances’, к взДохам [gvzd-] ‘to the sighs’; not only in зтот гороД [-d | g-] ‘this city, купец бьІЛ [-3|b-] ‘the merchant was (there)’ здесь же [ž-|ž-] ‘but here’, Лечь бьІ [-3|b-] ‘go lie down’, хоть дома [-d,|d-] ‘at least at home’, так зЛо [-g|z-] ‘so maliciously’, запах дьІма [-γ|d-] ‘smell of smoke’, бЛиз берега [-z,|b,-] ‘near the shore’, сквозь доску [-z,|d-] ‘through the board’ but also in как вдова [-g|vd-] [like the widow’, хоть вздохнуЛ [-d, | vzd-] ‘at least he rested’, вот в детстве [-d|vd,-] ‘but in childhood’, чтоб взять, [-b|vz,-] ‘in order to take’, поев вдовоЛь, [v |:d-] ‘having eaten one's fill’, против вдовьІ [-v|:d-] ‘against the widow’, уж в вдовах [-ž|v:d-] ‘already among the widows’, воЛхв же [-γv|ž-] ‘but the sorcerer’, без жертв бьІ [-dv|b-] ‘without sacrifices’, от моЛитв-де [-dv|d,-] ‘supposedly from prayers’, ветвь даже [-d, v,|d-] ‘even the branch’. This treatment of the voicing feature provided the basis of Halle's discussion (1959: 63–5). For some inexplicable reason, the facts brought to light by Jakobson apparently went completely unacknowledged in the Soviet Union (and elsewhere, for that matter, outside the United States) and were not reflected in the works on Russian phonetics which appeared after 1956. Thus, for instance, one notes no revision of the appropriate section in Avanesov (1958); nor in Avanesov & Ožegov (1959).

The relationship between segment and sound is a central concern of phonology.1 A framework for the definition of this relationship is that of distinctive features - proposed by Roman Jakobson and developed by Morris Halle. This framework may be looked at as having two parts: the substantive and the formal. The substantive is made up of the small number of features which, it is claimed (Halle, 1957: 67 ff.), are all that are necessary for describing all the languages of the world - fifteen by the latest count (Halle, 1964a: 329). The formal consists of a number of concepts which appear to have contributed to the definition of the relationship between segment and sound in transformational theory (Chomsky, 1964: 86). Only incidentally will I be concerned with the substantive part of the framework.