‘Apical displacement’ is the name given to a set of sound changes in the Dravidian family, whereby alveolar and retroflex consonants which occur as Cj in Proto-Dravidian stems of the type *(Ci)VCj-V ... shift their position to produce structures of the type *(Ci)CjV- ... in a well-defined subgroup of languages, viz. Telugu, Gondi, Koṇḍa, Kui, Kuvi, Pengo, and Manḍa. With the help of a computer analysis of relevant etymologies, it has been shown that apical displacement affected less than a dozen items at a common stage of the above subgroup. The rules gradually spread to the rest of the lexicon over the following two millennia, and the change is still in progress in some languages of this subgroup. It has been demonstrated that, of the items which fulfill the structural conditions of the change, 72% are covered by it in Kui; about 63% in Kuvi, Pengo, and Manḍa; but only about 20% in Gondi and Koṇḍa. A chronological layering of lexical items is established in terms of particular combinations of languages which share the cognates-with-change. Areal diffusion is characterized as the gradual lexical spread of an inherited rule—as opposed to the traditional notion of shared innovation, which is marked by shared rules in shared innovative cognates. A large class of exceptions to sound change at any given point of time can arise when the change has not yet affected the entire eligible lexicon.

This paper maintains that some of the historical rules of consonant deletion, vowel deletion, nasalization, and initial h-deletion—all recapitulated synchronically within the transformational generative accounts of French linking—have undergone morphologization in the form of inversion, and that h-aspiré words have been assigned the feature [– vowel]. This revision is supported by substantive evidence from colloquial trends, which can only be explained by the loss of consonantal epenthesis, vocalic epenthesis, denasalization, and the integration of h-words into the vowel-initial category. It is demonstrated that the extant analysis fails in dealing with the dynamics of change for two reasons: (1) all rules are phonological, with no expectation of loss; and (2) even if rules are lost, incorrect predictions would be made.

Evidence from vowel alternations in Spanish is presented which bears on two theories of regular but non-automatic morphophonological alternations. On the ‘disjunction’ theory, alternating substrings of lexical entries are represented as disjunctions of surface alternants. The appropriate alternants are selected by rules triggered by the lexical disjunctions. On the ‘rule-feature’ theory, rule features are attached lexically to segments that have surface alternants. These features trigger rules that apply only to forms bearing the corresponding rule feature. Though similar, the two theories are empirically distinguishable. A particular analysis of the Spanish data provided by the rule-feature theory is descriptively superior to one provided by the rival theory. The inadequacies of the disjunction analysis are direct consequences of properties of that descriptive apparatus. The Spanish evidence thus supports the rule-feature theory.

Ambiguous questions like Why do they believe (that) she killed him?, where the interrogative word seems to refer to either clause, appear to offer counter-examples to the NP Ecology Constraint proposed in Cattell 1976. But sentences beginning with why are frequently not ambiguous in this way; and when they are ambiguous, they cannot always be explained by appeal to the theory that why has separate origins in the different clauses. Consequently, it is argued here that the derivation of why and other interrogative adverbs does not involve the movement of NP's, and therefore does not present counter-examples to the NP Ecology Constraint. An alternative explanation of why-questions is offered, which involves appeal to semantic and deictic criteria.

Mirror images in constituent order are found in a wide range of parallel clause types in Japanese and English. I propose three explanations for linear orderings, related to the following hypotheses: (a) all constituents may be classified as either nominal or verbal; (b) relative orderings of constituents derive from language-independent criteria for Verb-to-Complement constraints; and (c) constituents which bear higher degrees of contextually determined and semantic entailment with the verb will be closer to the verb. The evidence presented suggests that all three principles are operative in Japanese and English. Although not all orderings conform to these principles, it is argued that the three principles of ordering will form a necessary part of a theory of syntax which accounts for word order in a non-arbitrary fashion.

A complex morphological system for marking respect and social distance exists in the Nahuatl spoken in the Malinche Volcano area of Puebla and Tlaxcala, in Mexico. Principles for usage of the system are described on the basis of observation of 85 speakers in eleven communities. Usage in direct address is relatively stable in all communities. However, usage in reference, which apparently became highly elaborated during the colonial period, is being reduced in some of the modern communities. This seems to reflect the shift in the functional range of Nahuatl toward a usage manifesting indigenist solidarity, with prestige-bearing roles being increasingly encoded in Spanish.