This paper explains the Greenbergian universals of modifier and adposition ordering as accidental side-effects of diachronic derivation, arguing that disparate diachronic processes can conspire to give the effect of synchronic universals. The ordering of modifiers, for example, may result from their generation by means of a binding anaphor strategy. This strategy produces similar patterning for genitive-noun and relative-noun constructions; and these, in turn, are potential diachronic sources for adjectivals. Constrained by the normative ordering of the language, this strategy will result only in preposed modifiers in verb-final languages and only in postposed modifiers in verb-initial languages. In SVO languages, it results in either preposed or postposed modifiers.

Semantic change is carried forward by specific individuals who think in particular ways. although it can be constrained by physiology and guided by social values. The measurement of color categories provides a close look at the various propelling incentives in a domain that is probably universal. Newly refined descriptive methods enhance such observations, because they permit verification of cognitive differences on the basis of correspondence between independently elicited orders of quantifiable data. Measurable change progresses through diverse trajectories in closely related Mayan languages whose social milieux are radically distinct. In Tzeltal, variation can be explained in terms of individual cognitive shifts within universal physiological parameters. But in Tzotzil a socially enforced conservatism creates tension between the preservation of traditional categories and the addition of new ones. The tension produces a taxonomy that is deeper than any others encountered by the major world color surveys. While a model of individual cognition explains how color categories change at the basic level, a social model accounts for differences between communities.

This paper presents an integrated theory of the syntax and semantics of complement control that treats unexpressed subjects in verbal and nominal constructions uniformly, in terms of semantic control constraints that are shown to apply across discourse as well. Our account includes an explanation of the long-standing problem of be allowed to … exceptions to Visser's Generalization (that subject-control verbs do not passivize). Moreover, semantic control constraints are shown to interact with a principle governing the binding of overt and covert anaphors in such a way as to deduce the correct predictions of both Visser's Generalization and Manzini's Generalization (that VP complements are controlled by an argument of the governing verb).