Monolinguals and bilinguals differ along a number of dimensions, including the way they label existing object categories (Pavlenko & Malt, 2011). In the present study, we asked whether English monolinguals, Spanish-English bilinguals, and English-Spanish bilinguals also differ in the way they use language when forming novel categories. Previous research with monolinguals shows that a shared label encourages children (e.g., Waxman & Markow, 1995) and adults (e.g., Lupyan, Rakison, & McClelland, 2007) to place objects together. Extending this work, we demonstrate that, when two objects shared a Licit Word label like “zeg,” monolinguals and bilinguals alike were encouraged to group them together. Illicit Words like “gxz,”, on the other hand, only influenced the categorization decisions of bilinguals. Thus bilinguals appear to be more flexible in their use of linguistic information in categorization. Neither group made use of non-linguistic cues (patterned frames), suggesting a unique role for language in category formation.