Subject-verb agreement errors in written productions: A study of French children and adults

This experiment was designed to simulate the conditions for subject-verb agreement errors, which are rarely but regularly observed in highly educated adults. Twenty-four adults and 24 children (12 years old) were orally presented with sentences to write. The sentences were in the French past indicative (the ''imperfect'' tense) and were of two types, as follows: Noun 1 [subject of the verb] + Verb 1 + Noun 2 [object of the verb] and (Adverbial phrase) + Pronoun 1 + Pronoun 2 + Verb 2 + (adverbial phrase). The adverbial phrase appeared either at the beginning or the end of the sentence. The conditions were the following: Noun 1 (and Pronoun 1) and Noun 2 (and Pronoun 2) were either matched or mismatched in number, and the sentences were either followed or not by a series of five words to be memorized. Most adults made errors when the two pronouns differed in number. But, in contrast to the results of studies using the present indicative, the extra cognitive load (the word series) did not lead to more errors. The children also made errors when the two pronouns differed, and did so whatever the cognitive load. The position of the adverbial phrase did not influence the error ratio. With the imperfect tense, it seems that making the verb number agree with its subject cannot be considered as a cognitively automatic and effortless activity, even for adults.