The Conceptual Inactiveness of Implicit Arguments: Evidence from Particle Verbs and Object Categorization

The linguistic system generalizes over rich conceptual structures and, at the same time, has to guarantee that the correct inferences can be drawn as to what the message is intended to convey. In this context the question needs to be answered whether entities not realized in grammar, i.e. implicit arguments, are a necessary part of the interpretation of the expression. In this paper it will be shown that this is not the case with particles (as in 'stick s.th. on'), which do not realize a reference object as their prepositional pendants do ('stick s.th. on s.th.'). The findings lead to the conclusion that any truth-conditional description of implicit arguments should discard them as being conceptually inactive and information-structurally irrelevant. As will be shown, it is only under specific contextual conditions that a construal of the corresponding argument is coerced. Only then the semantic representation needs to be adjusted, which can be characterized as an instance of lexical re-analysis. Departing from theoretical evidence it will be shown that with particles no reference object is conceptually active on the basis of two reaction time studies with an object categorization task: It is only full prepositions that prime a potential reference object which is revealed by the reaction to conceptually congruous target line drawings presented just after both prepositions and particles. In this context, two additional aspects will be dealt with. First, the type of priming triggered by inherent sortal information (as in stick the label on the ...) needs to be discussed. It will be shown that a semantic fit between the sortal restrictions activated by the preposition and a depicted object results in inhibitory effects. Second, there is evidence that certain grammatical information (such as gender) about an object expression does not enter the conceptual processing of a corresponding target object drawing, which is revealed by the fact that modifying grammatical information in the linguistic prime material proves to be irrelevant for the procedural effort the categorization task requires.