Language provides a finite set of labels (words) for an infinite set of possible objects that a speaker may encounter. Previous picture naming studies of language production have focused on highly familiar object stimuli that elicit uniform responses among native-speaker participants. By contrast, this dissertation explores picture naming responses to a broad range of object stimuli, from typical prototypes to unusual or unfamiliar examples of the same name. Drawing on research in lexical categorization, I document variation in native-speaker picture naming behavior and examine how this naming behavior and the underlying neural responses change as a result of language interaction in bilinguals. Behavioral studies (Chapter 2) measure differences in lexical categorization patterns (picture naming responses over many different examples of an object name) among monolingual and bilingual speakers of Chinese and English. Further, this inter-personal variation is explained in terms of speakers' unique language histories and norms of each linguistic community (native speakers of English and Chinese). I propose a statistical model which describes the role of these variables in each language in predicting categorization patterns in both the native language (L1) and second language (L2), identifying significant effects of cross-language interaction for bilinguals in both languages. Next, I introduce a new stimulus set of 407 objects sampled from several semantic domains (e.g., clothing and vehicles) and normed by native, monolingual speakers of English and Chinese (Chapter 3). This stimulus set demonstrates the extensive variation in picture naming responses among native speakers of each language and between languages. I use these norms to examine a few specific variables relating to native speaker norms identified in the previous chapter. In the final set of experiments (Chapters 4 and 5) I select a subset of 183 objects from the new stimulus set to test functional neural correlates of these categorization variables in native, monolingual speakers of each language and in Chinese-English bilinguals. Each categorization variable is associated with brain regions that uniquely respond to its variation, and activity in these regions confirms functional involvement of both L1 and L2 variables in bilinguals' L1 picture naming behavior. These findings are situated in the broader context of neurocognitive models of language production and offer a refined view of lexical semantic retrieval and selection, accounting for the variety of potential objects that speakers may encounter and variation among native speakers of a language.