In certain Eastern Algonquian languages a nasal vowel developed from the long low vowel /a:/, regardless of consonantal context. A series of experiments showed that longer vowels (regardless of height) were perceived as more nasalized than shorter ones, but only when some nasalization was actually present. Further experiments showed no evidence of an increase in nasalization for long vowels in oral contexts. If some nasalization was nonetheless introduced (either randomly or by a general increase in nasalization) into these languages, the vowels most likely to be perceived as nasalized were the long ones. This perceptual process may have been responsible for this unusual historical development.

The present study uses a multidimensional approach to trace the historical evolution of written genres of English. We briefly present a model of stylistic variation developed in our previous work, focusing on three empirically defined dimensions of linguistic variation that are associated with differences among ‘literate’ and ‘oral’ varieties: ‘Informational versus Involved Production’; ‘Elaborated versus Situation-Dependent Reference’; and ‘Abstract versus Nonabstract Style’. We then show how fiction, essays, and letters have evolved over the last four centuries with respect to these dimensions. Although they have evolved at different rates and to different extents, we show that all three genres have undergone a general pattern of ‘drift’ towards more oral styles—more involved, less elaborated, and less abstract. We discuss several possible functional and attitudinal influences on the observed patterns of drift; these include the rise of popular literacy and mass schooling, the demands of scientific and expository purposes, and conscious aesthetic preferences. In so doing, we extend Sapir's notion of drift to include the evolution of genres and the influence of nonstructural underlying motivations.

Certain structures containing comitatives appear to be counterexamples to the claim that only superficial constituents control agreement. Earlier treatments have given up the requirement that such controllers be superficial constituents in order to preserve the generalization that controllers are always constituents. In these analyses, the superficial nonconstituent controller corresponds to a deep constituent. In Tzotzil there is a better analysis for structures of this type, crucially involving the Plural Pronoun Construction, which requires neither nonconstituent nor nonfinal controllers. If this analysis can be maintained in other languages, it will make possible a more restrictive theory of agreement controllers.

On one view of syntactic structure, the expression of the grammatical relation subject is associated with an NP position. On another, this grammatical relation is expressed by a bound pronominal form. Discussion of the choice of one or the other of these has focused exclusively on the reflex of what is traditionally termed ‘subject/verb agreement’ and an ‘agreeing’ NP. This paper offers a theory, exemplified through an analysis of Luiseño, which accommodates a range of ‘subject agreement’ forms. The analysis is based on the idea that subject information can flow from a number of distinct sources. Languages might differ in regard to how many of these sources are used. One consequence of this idea is to deny a simple bifurcation into subject-pro-drop / non-subject-pro-drop or pronominal-subject / nominal-subject languages. The Luiseño analysis, therefore, forces a reconsideration of the roles of subject/verb agreement and the agreeing NP in grammatical composition.