On the interpretation of null subjects of infinitival complements

Abstract

This work addresses two issues in the online processing of null subjects in infinitive complements: first, we examine whether antecedent assignment is carried out immediately, or is postponed until a later stage; second, we test whether this process is driven by verb control information or based on a distance criterion. These issues have fuelled a lively debate between the proponents of two alternative processing accounts (structural vs. lexical). We report twoERP experiments with sentences with an infinitive complement controlled by the subject or by the object of the main clause. Half of the sentences had agender mismatch between the null subject and an adjective (and its antecedent). The detection of this anomaly would indicate both the rapid assignment of anantecedent and the use of verb control information in this process. The results show that the antecedent of the null subject is assigned quickly and that this process is driven by the control properties of the main verb. These results are discussed in terms of the processing accounts mentioned above.