ABSTRACT
This chapter describes how the analysis of eye-tracking data obtained during written sentence recall can contribute to understanding the nature of processing involved in the production of subject–verb agreement. Studies in this area have primarily analyzed the errors produced in response to sentences designed to trigger subject–verb agreement errors, also known as attraction errors (Francis, 1986). These errors consist of establishing agreement between the verb, conjugated in the present indicative, and the noun that immediately precedes it, such as “le chien des voisins mange(nt)” (the dog of the neighbors eats). This paradigm has led to an understanding of the influence of semantic, phonological, and syntactic processes in grammatical processing. It has provided arguments in favor of a two-level grammatical processing model in writing production. At the first level, verb-ending number agreement is automatically activated by an immediately preceding noun (e.g., “voisins” activates the plural of “mangent” in the sentence “le chien des voisins mangent”). If the preceding noun is not the subject of the verb, an attraction error is possible. The second processing level consists of pre-graphic control, which detects such an error before it is written, by reiterating the appropriate agreement rule. Various studies have measured pregraphic control (see Largy, Cousin, & Dédéyan, 2005, for a summary), but neither the temporal course of the process nor the exact nature of the mechanism involved has been clearly determined.
