Evidence from both alphabetic and nonalphabetic languages has suggested the role of orthography in the processing of spoken words in individuals’ native language (L1). Less evidence has existed for such effects in nonnative (L2) spoken-word processing. Whereas in L1 orthographic representations are learned only after phonological representations have long been established, in L2 the sound and spelling of words are often learned in conjunction; this might predict stronger orthographic effects in L2 than in L1 spoken processing. On the other hand, lexical codes are typically less integrated and stable in L2 than in L1, which might entail less pronounced orthographic effects. To explore this issue, Tibetan Chinese bilinguals judged whether Chinese spoken words presented in pairs were related in meaning. Some of the unrelated word pairs were orthographically related, and critically, this orthographic overlap induced a significant increase in response latencies. Compared to previous results from L1 listeners with the identical procedure, the orthographic effect for L2 listeners was more pronounced. These findings indicate that orthographic information is involuntarily accessed in native and nonnative spoken-word recognition alike and that it may play a more important role in the latter compared to the former.