Dialects and languages are socially meaningful signals that provide indexical and linguistic information to listeners. Are the indexical categories that are shared across languages used in cross-linguistic processing? To answer this question English (L1)-Māori (L2) bilingual New Zealanders participated in a priming experiment which included English-to-Māori and Māori-to-English translation equivalents, and within-language repetition priming for Māori and English. Half of the English words were produced by standard New Zealand English (Pākehā English) speakers and half by Māori English speakers. We find robust evidence for within-language repetition priming for both Māori-only and English-only trials. Across languages, there is L1–L2 priming: both Pākehā English and Māori English successfully prime Māori. The effect size, however, is larger for Māori English–Māori trials than Pākehā English–Māori trials. In the L2–L1 direction Māori only primes Māori English, not Pākehā English. These results support the hypothesis that indexical categories – e.g., ethnic identity – facilitate word recognition across languages, particularly in the L2–L1 direction, where translation priming has not always been obtained in the literature. Lexical items and pronunciation variants are activated through conceptual links and social links during bilingual speech processing.