ABSTRACT
Anderson 1971). Within Relational Grammar (Perlmutter 1978; Davies 1984) and generative grammar (van den Besten 1985; Burzio 1986; Grewendorf 1989) these facts were treated syntactically by analyzing the surface subjects of unaccusative (or ergative) verbs as underlying objects. Certain syntactic constructions are supposed to be sensitive to this distinction in that they either apply only to unaccusatives (e.g. ne-cliticization in Italian, perfect auxiliary selection in Italian, German, and Dutch, attributive use of past participles, topicalization of subject+past participle in German), or only to standard unergative verbs (e.g. impersonal passives, creation of -er agent nouns). Cf. Das Kind hat gelacht The child has laughed’ vs Das Kind ist weggegangen ‘The child has gone away,” Hier wurde gelacht ‘Somebody laughed here’ vs *Hier wurde weggegangen ‘Somebody went away.’ Linguists working with ergative languages have criticized the use of the term ergative for the phenomena mentioned above, since they are different from the morphological and syntactic ergativity found in ergative languages (cf. Comrie 1978; Dixon 1987; Primus 1994). Every genuine ergative language is morphologically ergative, i.e. uses the zeromarked case, the absolutive, for den Stock/der Stock in the examples above. Furthermore, in a genuine ergative language den Stock/der Stock are expected to behave syntactically like subjects in anominative language. Contrary to what is expected. these noun phrases behave like surface or underlying objects in nominative languages.
