Do cognate and circumstantial complements of intransitive verbs form one ‘Range’?
In this article we examine the question whether cognate complements (as in <i>dance the tango</i>) and circumstantial complements (as in <i>climb stairs</i>) can be regarded as expressing the same semantic role. Halliday has proposed that they do: in his view they both delimit the ‘extent’, or ‘Range’, of the process. Traditionally, however, they have been regarded as distinct grammatical categories. Assuming, like Halliday, that grammatical categories are form-meaning couplings, we investigate in corpus data the two types of formal evidence proposed by him for the unified Range category: alternations and selection restrictions on determiners and modifiers. By quantifying the relative frequencies of the alternate constructions, we have found that clauses with cognate complements form the marked option – totalling on average 3.5% – in comparison with intransitives. Clauses with circumstantial complements, by contrast, alternate with intransitive clauses as well as with clauses with prepositional phrase in varying proportions. This shows that the notion of ‘location involved in process’ can be more strongly or weakly present in the semantics of verbs taking circumstantial complements. The determiners and modifiers of cognate and circumstantial complements also reflect different semantic relations of the complement to the process expressed by the verb. Cognate complements are predominantly indefinite and circumstantial complements more often definite, because the former typically construe a ‘new’ instance of the process, whereas the latter often express pre-existing locations. Attributive modifiers of the two complement types differ both quantitatively and qualitatively. Cognate complements take more qualitative adjectives, which tend to express the manner in which the process takes place. Circumstantial complements have much less qualitative modification and often express the resistance or facilitation offered by the location to the action being carried out on it. We conclude that the two types of complements express different sorts of entities with different relations to the process.