Susan Steele

A Law of Order: Word Order Change in Classical Aztec

1 Research for this article was supported by the American Council of Learned Societies. Thanks are due to Kenneth Hale for comments on an early stage of the research; any errors are of course my own. 2 The examples are written in the (somewhat regularized) orthography of the original source. The sources cited are listed here, preceded by the code which identifies them in the article. Following the letter code is the page number where the sentence is found. The works cited are: (An)Arthur J. O. Anderson, Grammatical Examples, Exercises, and Review (Provo: University of Utah, 1973); (A)-A.J.O. Anderson, trans., Rules of the Aztec Language (Provo: University of Utah, 1973) [this is a modified version of Francis

A Typology of Functors and Categories

Recent work on visual cognition by Hoffman (e.g. 1983) challenges the traditional notion that the basic units in visual analysis, the cognitive primitives, are geometric figures.2 He argues, rather, that the cognitive primitives for vision are properties of boundaries. For example, a region of a planar image will be interpreted in one way if it is analyzed with a convex edge and in another way if it is analyzed with a concave edge. Consider Figure 1 (from Hoffman’s paper), where the rings can be taken to trace the trough or the crest of the waves, and the lines defining the waves are interpreted accordingly.

The Utility of the Proposition

The analysis of the Proposition proposed in Chapter Four has precisely the right results for the problems raised in Chapter One. Given the analysis of Chapter Four, the Sentence in (1a) is comprised of two parts — the aux nil and the Proposition hengeemali ’ariqu0073.

Luiseño Features: Background

Given its focus on agreement and other such conditions, this work is primarily concerned with phrasal units, i.e. with expressions potentially composed of more than a single word. However, the values available to such expressions are not independent of the values available to words. This is not of course an idea idiosyncratic to this work. X-bar theories claim a relationship between lexical and phrasal categories, as do unification-based theories. Thus, before we turn to the analysis of Chapters Three through Six, we must consider the values available to Luiseno words.

The Issues — Analytical and Theoretical

Competing analyses of Warlpiri have made explicit what is, I think, a relatively common assumption in regard to agreement, to wit: Agreement plays no role in syntactic composition. In Warlpiri, two (or more) elements which share certain formal properties and which are (intuitively, at least) members of a single expression need not be adjacent to one another.

Agreement, Anti-Agreement, and Order

Chapter Three argued that agreement and anti-agreement are functors with distinct consequences — agreement requires that the members of its argument share properties and yields a result with a single value for these properties; anti-agreement requires that the members of its argument have different properties and yields a result with an added value, drawn collectively from the members of the argument. The effect of agreement was demonstrated on the Constituent and the effect of anti-agreement on the Argument Structure. Chapter Four applied both to the analysis of the Luiseno Proposition. Given a display of the person and number values in the Propositional Radical (what I have called the agreement grid), agreement yields a temporal value for the Proposition; anti-agreement yields a person/number value, the Subject. Chapter Five showed how the result of agreement and anti-agreement for the Proposition solve the aux compatibility problem introduced in Chapter One. The Luiseno aux is sensitive to just those properties represented in the Proposition’s formal value — its temporal value, its Subject person/number value, and the presence and properties of the non-Subject chain.

The Utility of the Classification

Because the Constituent, the Argument Structure, and the Proposition each result from the application of functors of the same type (i.e. obligatory, listable, and non-localizable), by the classification developed in Chapter One they must be assigned the same category type (i.e. syntactically inaccessible, but phonologically accessible). We expect, then, the sort of parallel we have witnessed: All three categories — and none of a different type — are potential arguments in the rule schema in (1).