abstract

What did primitive languages look like? One way to answer this question is to examine patterns of cross-linguistic diversity exhibited by modern languages of the world.

There is now general agreement that today's languages are all equally advanced on the evolutionary scale; there are no more primitive languages left. There is also consensus that, when viewed holistically, all languages are of roughly equal complexity. Nevertheless, wihin particular grammatical domains, it is indeed possible to identify different levels of complexity across languages. To cite one well-known example: by definition, the morphology of isolating languages is less complex than that of inflectional languages, which in turn is less complex than that of polysynthetic languages. A common assumption is that languages that appear to be simple in some domains compensate for this by being more complex in other domains.

In this paper, it is suggested that cross-linguistic variation in complexity within particular grammatical domains may provide insights into possible paths of language evolution. Specifically, it is argued that within any given grammatical domain, if one finds a situation of greater simplicity in one language than in another, then that simpler situation may, under certain conditions, be considered to be characteristic of a prior stage in the evolution of language.

In this paper, the above approach is applied to the domain of syntactic categories, or parts of speech. In particular, it is claimed that languages differ greatly in their inventory of syntactic categories. More specifically, some languages exhibit an extremely simple, or impoverished syntactic category inventory. Such inventories, it is argued, may be taken to reflect the situation that prevailed in an earlier stage of language evolution.

The first part of this paper sketches the outline of a theory of syntactic categories designed to facilitate the formulation of cross-linguistic generalizations governing possible syntactic category inventories. As in categorial grammar, syntactic categories are derived from other syntactic categories by means of category-formation operators. Unlike categorial grammar, however, the present theory posits the existence of a single primitive syntactic category S0, from which all other syntactic categories are derived, by recursive application of two category-formation operators: a unary operator "kernel", forming category Xn+1 (the "kernel of X") from category Xn, plus the familiar binary operator "slash", forming category X/Y ("X slash Y") from categories X and Y.

The above framework permits the formulation of typological generalizations governing possible inventories of syntactic categories in different languages. Specifically, for any language L:

1. If X is a category in L, then all the categories in the derivational history of X are categories in L. Similarly, if X is an open category in L, then all the categories in the derivational history of X are open categories in L.
2. If X and Y are categories in L, then L possesses constructions formed with X and Y.
3. If X/Y is an open category in L, then Y is the kernel category of X.

The second part of this paper surveys the evidence to the effect that some languages possess almost maximally impoverished inventories of syntactic categories. For example, Tagalog has just one open syntactic category S0 and one closed syntactic category S0/S0, while Vietnamese has just one open syntactic category S0 and two closed syntactic categories S1 and S0/S1.

The results of this paper thus show that some contemporary languages can get by with a very simple inventory of syntactic categories. In doing so, they suggest what the inventory of syntactic categories might have looked like at a primitive stage in the evolution of language. More generally, the categorial-grammar based theory of syntactic categories presented in this paper provides for a family tree of syntactic categories of increasing complexity: a scheme which may be considered as a recapitulation of the process by which these categories evolved in human language. Thus, this paper shows how, by examining the diversity of languages spoken today, it is possible to infer how language evolved from simple beginnings to contemporary complexity.

 Conference site: http://www.infres.enst.fr/confs/evolang/