We focused on the classifier as semantic-syntactic language feature and examined the semantic classifier congruency effect in language production. In three experiments, participants first saw or heard a classifier (e.g., a pair of) and then had to name a picture showing a person or an object (e.g., shoes). In these classifier + noun phrases, the classifier was either semantically congruent to the noun (e.g., a pair of shoes) or incongruent (e.g., a bowl of shoes). In Experiment 1 (n = 26), English monolinguals were tested in English and results showed better performance in the semantically congruent condition than in semantically incongruent condition (i.e., a semantic classifier congruency effect). In Experiments 2 (n = 40) and 3 (n = 40), Chinese-English bilinguals were tested with both Chinese (L1) and English (L2) classifier + noun phrases intermixed in the experiments. The semantic classifier congruency effect was replicated but was significantly larger for L1 than for L2. Further the effect also differed as a function of classifier modality, indicating a larger effect with an auditory than with a visual classifier presentation. Additionally, language switch costs were found in both semantically congruent and incongruent conditions but were smaller in semantically congruent as compared to incongruent conditions. Together, these findings provide evidence for a semantic classifier effect in both L1 and L2. Furthermore, they suggest that semantic classifier congruency interacts with language switching, suggesting an influence of semantic-syntactic language features in bilingual language control.