This paper focuses on grammar teaching with beginning writers. The study aimed to determine whether the grammatical encoding in French should be explicitly taught to improve beginner writers’ performances. Third and fifth graders were evaluated on their ability to correctly make a verb agree with its subject before (pre-test), during and after (post-test) training sessions. The training was either explicit or implicit. The explicitly trained group was requested to select the subject and the verb in erroneous sentences and to correct agreement errors. The implicitly trained group corrected only the errors they found. A global analysis showed a significant decrease in agreement errors between pre-test and post-test regardless of the training group, suggesting that agreement rules may be explicitly or implicitly taught. However, a more fine-grained analysis showed that younger writers took extra advantage of the explicit training during the training sessions, they did not do so during the post-test. Finally, explicit training benefited all participants during the post-test with sentences demonstrating syntactic ambiguity.