It is well established that emotion is extensively influenced by language. However, previous studies mainly focus on the first language (L1) rather than the second language (L2). The current study explored whether L2 shapes emotion perception in masked priming paradigm by distinguishing emotion-label words (e.g. fear, pride) and emotion-laden words (e.g. breakup, mother). Twenty-three Chinese-English bilinguals were instructed to judge the valence of the emotional pictures that were followed by masked and briefly presented L2 emotion words. Behavioral results showed that emotional pictures primed by negative emotion-label words were processed faster and more accurate than primed by emotion-laden words. Event-related potential results further revealed that emotional pictures elicited smaller N300 primed by negative emotion-label words than by emotion-laden words. The converging evidence suggests that second language can also modulate emotion perception fast and automatically, and such modulation is dependent on emotion word type.