Semantic Relations vs. Abstract Syntactic Relations: Evidence from Halkomelem

Abstract

<p>Based on a partial reanalysis of a corpus of data from Halkomelem (Salish) analyzed from the perspective of classical Relational Grammar in a series of works by Donna Gerdts, I argue for a relational theory with both revaluations of syntactic relations and alternative initial alignments of semantic relations with syntactic relations. First, I look at four kinds of constructions for which Gerdts appeals to revaluation analyses: passive, causative, antipassive, and applicative. I argue that a subset of these - namely the latter three - are better analyzed otherwise. Based on these results, I examine certain grammatical phenomena which might have been understood in terms of initial syntactic relations, if multistratal analyses of the constructions in question were available. Since they are not, however, it is necessary to formulate the relevant conditions in terms of semantic relations.</p>