This chapter introduces chief postulates common to usage-based (UB) approaches to language. The UB approach maintains that speakers’ experiences with language shape how language is stored. Experiences with specific words and word combinations in particular linguistic, discursive, and social contexts accrue in memory and subsequently contribute to patterns of variability evident in speech productions. Usage-based approaches regularly consider independent effects on lexical representations of decontextualized prior probabilities (e.g. phone/word/bigram frequencies, type frequencies), and, increasingly, contextually informed counts (e.g. lexical items’ cumulative exposure to conditioning effects of the production contexts, phone/word probabilities) are considered. This chapter offers an overview of studies exploring the connection between usage patterns and bilingual sound systems as well as studies exploring evidence of interlingual influence arising from bilingual lexical storage (schematic ties in memory). The chapter suggests potential avenues for future UB research into bilingual phonetics and phonology.

Stress placement in English loanwords into Mirpur Pahari (MP) is used to explore whether a usage-based approach can inform the incorporation of external factors (e.g., exposure to a donor language; here, English) into formal phonological analysis of loanword adaptation alongside internal factors (e.g., phonology of the recipient language; here, MP). Stress placement in English loanwords into MP shows across- and within-speaker variation between conformity to MP stress rules (formalized in classical Optimality Theory; OT) and retention of stress on the syllable that is stressed in English; this is a challenge for most theories of loanword adaptation. In our hybrid approach, variable adaptation patterns in individual speakers’ loanword realizations in production data from twelve MP speakers in the UK are correlated with degree of exposure to English, operationalized as vocabulary size, and a unified formal account is sketched through usage-based weighting of constraints in Stochastic OT.