| One of the major
discoveries of the 20th century in the language sciences is that syntax
matters, both for artificial languages (Carnap) and natural languages
(Chomsky). But different traditions understand the nature of syntax
differently. In this talk we focus on two quite broad ways of
conceiving of syntax: one is the set of rules that enable speakers and
listeners to combine the meaning of expressions (i.e. compositional
syntax), and the other is the set of formal constraints on the
combinations of expressions (i.e. formal syntax). The question that
occupies us is whether all languages have a robust set of formal
constraints or whether there are languages in which most syntactic
rules are exclusively compositional. Our claims are (1) that Oneida
(Northern Iroquoian) has almost no formal syntax (as we have defined it
above) and is very close to a language that includes only a
compositional syntax and (2) that the little formal syntax Oneida does
have does not require any reference to syntactic features. Our analysis
of Oneida suggests that what is often taken as essential properties of
human languages (e.g. syntactic selection/argument structure, syntactic
binding, syntactic unbounded dependencies, syntactic parts of speech)
are simply overwhelmingly frequent.* Our research also suggests that
the function of syntax is more than syntactic feature management under
agreement, syntax is also about the management of semantic indices (in
contrast to semantic content), a function anticipated by Quine’s work
on the nature of (semantic) variables. As an illustration of the
important role of semantic indices, we show that the morphology of
Oneida provides interesting reflexes of the distinction between
semantic indices and semantic content. |