This study examines gender agreement between a complex sentential subject (containing two nouns) and a predicate adjective in second language Spanish. The data were collected using a computerized sentence completion task that measured gender agreement accuracy (correct or incorrect). Seven binary linguistic variables were analyzed: noun class of the head and attractor nouns (semantic or non-semantic), head noun morphology (overt or non-overt), gender of the head and attractor nouns (feminine or masculine), and noun class and gender congruencies (matched or mismatched). All possible combinations of the variables were considered.

To date, no study has examined all of these variables in a single experimental design assessing the second language acquisition of Spanish gender agreement. Participants were learners at three different levels of proficiency, and Spanish native speakers. Grammar and vocabulary knowledge were also examined as independent variables. Findings reveal that noun class does not affect accuracy of gender agreement. In contrast, both learners and native speakers are sensitive to the gender and morphology of the head noun, and gender congruency: participants were more accurate when the head noun was masculine and overtly marked for gender, and when the two subject nouns were of the same gender.