An analysis of be-passives

Abstract

The presence of the equivalents of the verb ‘to be’ in the passive constructions of various languages has prompted several analyses in which this verb was explicitly or implicitly considered a characteristic feature of passive in general. This is particularly true for transformational and some post-transformational grammars. This in turn has led to certain conclusions concerning the meaning of passive. The present paper shows that such analyses were in error and that ‘be’ is not an attribute of the passive. This is demonstrated by analyses of the meaning and form of passives in some thirty languages. The paper also presents a hypothesis as to why ‘be’ is present in the passive constructions of so many languages, namely because ‘be-passives’ are a subset of a set of sentences with nominal predicates. This hypothesis is supported by data from synchronic analyses of over thirty languages and by historical developments in several unrelated languages.