The click sounds of two Khoisan languages, Nama and !Xóō, are described in considerable phonetic detail in order to demonstrate some of the complex, language-specific phenomena that must be taken into account in linguistic descriptions. Phonological features that can be used to classify these sounds are given; but it is shown that the detail of the phonetic phenomena can be recovered from the phonological specifications only by a considerable number of ad-hoc statements.

This paper is an attempt to incorporate a description of some segmental rules of English into the framework of cyclic phonology which has grown out of the classical SPE type of phonology, on the one hand, and Kiparsky's works, on the other. The discussion is organized around the problems of palatalization in American English. The rules arrived at are divided into cyclic and postcyclic types, and some predictions which follow from the proposed division are discussed. The analysis considers a number of modifications in the statement of rules originally introduced in SPE.

This paper criticizes the view of Hawkins (1979, 1980, 1982) that the theory of Universal Grammar must include implicational and distributional universals of word order. It shows that universals based on frequency counts of surface word order—as worked out, e.g., by Greenberg 1966—need to be reformulated in order to be relevant to the problem of language acquisition. More specifically, it discusses H's claim that the principle of Cross-Category Harmony can shed light on the proper formulations of the X' schemas of generative grammar. Finally, it is shown how survey studies of word order can be relevant to forming initial hypotheses about certain properties of grammar, and how the study of language typology might proceed within the generative framework.

The Latin vowels ě ⊖, pronounced in Late Antiquity as /ɛ / respectively, yielded rising diphthongs, usually /je we/, in Old Spanish stressed syllables, whether checked or free: FěRRU > fierro ‘iron’, C⊖RNU > euerno ‘horn’. Before certain palatal and prepalatal consonants (e.g. /j ž/), the monophthongs failed to diphthongize: SěDEAT > /sɛja/ > sea ‘let it be’, H⊖DIE > /je/ > (h)oy ‘today’, FOLIA > /fλa/ > foja ‘leaf’. However, the same environment long believed to have blocked diphthongization in Old (or Proto-) Castilian has been credited with stimulating that process to the north of the Pyrenees.

A new attack on the old problem is here undertaken, starting with the apophonically conditioned pairs RěGERE/RēGULA, TěGěRE/TěGULA, and SěCāRE/SēCULA. In a backward section of the Republic and the Empire, speakers could easily have been misled into transmuting SPěCULU ‘mirror’ into *SPēCULU etc., with /e: o:/—not subject to diphthongization—substituted for /ɛ / in words of the indicated syllabico-accentual structure.