To locate the underlying cause of biological gender errors of oral English pronouns by proficient Chinese-English learners, two self-paced reading experiments were conducted to explore whether the reading time for each "he" or "she" that matched its antecedent was shorter than that in the corresponding mismatch situation, as with native speakers of English. The critical manipulation was to see whether highlighting the gender information of an antecedent with a human picture would make a difference. The results indicate that such manipulation did make a difference. Since oral Chinese does not distinguish "he" and "she", the findings suggest that Chinese speakers probably do not usually process biological gender for linguistic purposes and the mixed use of "he" and "she" is probably a result of deficient processing of gender information in the conceptualizer. Theoretical and pedagogical implications are discussed.