This paper is concerned with two topics, the status of the syllable and the scope of redundancy rules, in generative phonology. We begin the discussion by examining material from Lugisu, a Bantu language of eastern Uganda, which will be used as the main language of exemplification. The simple noun in Lugisu is formed by the sequence: determiner, classifier, stem. The determiner is deleted before consonant-initial stems in some syntactic environments (and in isolation; examples 1 and 3, for example, show the sequence: classifier, stem). The first examples are from Meinhof's noun class 3 (Meinhof, 1932) where the determiner is /gu/ 3 and the classifier /mu/.

The problem that provides the subject of the present study has, curiously enough, received little more than a passing glance from most grammarians. Many of them mention that adjectives can be formed from substantives by means of a dental suffix, but none seem to be intrigued by the four aspects of the problem to be examined here. The fact that these aspects of the problem have received little or no attention up to now would in itself justify this study. There is, however, a further and even weightier justification. The problem brings into focus four parts of speech-verb (through the past participle), substantive, adjective and adverb–and so provides an excellent opportunity for observing the relations between these categories. Such a chance should not be missed, especially at a time when not merely the system but the very existence of the parts of speech is being called into question.

In the existing grammars and pedagogical works on Hindi-Urdu, the category of possessive expressions has mainly been described on the analogy of the possessive in English (Guru, 1952: 434–438; Kellogg, 1938: 101–102; Sharma, 1958: 34–35; Vajpeyi, 1959: 127–133, 159–170, 302–305; Pořizka, 77–78, 149–150, 277).1 In these works, it has been suggested that the genitive postposition ka, the invariable postposition ke, the compound locational postposition ke pas, and the directional postposition ko predicate possession. In Hindi-Urdu, the following types of sentences in English are generally translated by constructions containing one of the above postpositions: John has a son, John has a book, John has a headache. Such an analysis of possessive expressions in Hindi-Urdu is unsatisfactory from the point of view of an overall syntactic description of the language, since it would imply that there are gaps in the system of possessive expressions in Hindi-Urdu, and hence that in a transformational grammar of these languages some possessive phrases at least could not be derived from underlying simple sentences.

The purpose of this paper is to give an account of some hitherto unexplained phenomena connected with the use of the Eskimo ‘relative case’. A deep structure is proposed that not only will account for the occurrence of certain surface morphemes, but also, and in a natural way, for the semantic affinities that exist between certain types of noun phrases and the transitive sentence in Eskimo. In particular, it is shown that in phrases of the type ‘the man's house’, there exists evidence for considering ‘man’ as the subject noun phrase, ‘house’ as the object noun phrase of some deep structure. The transformational machinery involved in deriving the correct surface structures is shown in some detail. A discussion of older terminology in Eskimo grammatical tradition provides additional, indirect confirmation of the naturalness of the proposed explanation.

Representation of syntactic relations by dependency trees with nodes labelled by categories of lexical morphemes and with branches effectively labelled by relational morphemes reveals the relationships of elements to which selectional restrictions apply more directly than do current phrase structure characterizations and avoids certain defects of a configurational definition of relationships. The relations are those of CASE as defined by Fillmore and more generally those of HEAD-OF and COMPLEMENT-OF. Adoption of the view that SUBJECT-OF is a surface rather than a deep structure relation eliminates the need to distinguish strict subcategorization rules from selectional rules.

This paper attempts to present the abstract phonological system which underlies the sounds produced when a Malayan variety of Engchun Hokkien is spoken.2 It is set within a generative framework and some attempt is made at a formalization of phonological rules.

It has been repeatedly suggested in the literature on generative phonology that the number of features which appear in phonological rules be used in an evaluation measure for alternative descriptions of the phonology of a language; thus Halle states (1962: 55): ‘Given two alternative descriptions of a particular body of data, the description containing fewer… symbols will be regarded as simpler and will, therefore, be preferred over the other.’ The purpose of the present paper is to examine critically the application of the proposed feature-counting evaluation measure to alternative analyses of a very restricted body of data in the phonology of Turkish. The data considered are, as far as I can see, self-contained in that the analyses to be discussed do not have any implications elsewhere in the system of phonological rules; the adequacy of the decision based on the evaluation measure in question can therefore be fully assessed in connexion with the minor problem we shall be examining.