The present paper investigates whether the processing of emotion language in the context of a second language (L2) entails motor simulations and whether simulation models extend to negation also for L2. Participants were exposed to sentences in L2 describing emotional expressions while facial muscle activity was continuously measured. Sentences mapped either directly upon the zygomatic muscle (e.g., “I am smiling”) or did not (e.g., “I am frowning”), and were presented in the affirmative and negative form. Similarly to studies involving first language (L1), the zygomatic muscle was activated when reading affirmative sentences relevant to the muscle. In contrast, and differently from what previously observed in L1, reading sentences in the negative form (“I am not smiling”) did not lead to relaxation/inhibition of the zygomatic muscle. These results extend the simulation models to the comprehension of L2 but they also provide important constraints and contribute to the debate about grounding of the abstract and concrete concepts.