219 Heritage Language Journal, 13(2) August, 2016 Heritage Speakers Follow All the Rules: Language Contact and Convergence in Polish Devoicing Paulina Łyskawa University of Maryland Ruth Maddeaux, Emilia Melara, and Naomi Nagy University of Toronto ABSTRACT We use a comparative variationist framework to compare variable word-final obstruent devoicing patterns in heritage Polish, English and homeland Polish in conversational speech. Phonological and lexical factors are shown to condition this variation differently in the three varieties. We have a particular interest in one other factor relevant to heritage speakers: the amount of code-switching between Polish and English by each speaker. We show that, for second generation heritage speakers, individuals’ code-switching rates are positively correlated with their rates of devoicing. Based on the qualitatively and quantitatively different devoicing patterns of heritage Polish speakers, compared to both homeland Polish and Toronto English, we argue that the phonological grammar of this group of speakers constitutes a convergence of the heritage language and the dominant language’s grammars and suggest that frequent code- switching provides the context in which these speakers’ knowledge of Polish and English patterns converge. Keywords: Polish, language contact, variationist sociolinguistics, phonological variation, devoicing INTRODUCTION Heritage languages present a case in which the dynamics of two languages in contact are unbalanced. The ubiquitous dominant language of the community may affect the heritage language, even if the latter is acquired first (L1). The tug-of-war between the two languages makes the outcome of heritage language grammars dependent on multiple factors. Some parts of these grammars may be more readily influenced by the dominant language. Furthermore, the extent of influence relies on social factors (Matras, 2009; Siemund & Kintana, 2008). We investigate the effect of contact with English on word-final obstruent devoicing in Polish conversational speech of heritage Polish speakers in the Greater Toronto Area (hereafter, Toronto), Canada, where the dominant language of the community is Canadian English. Polish is known for its robust system of obstruent-final devoicing (see Gussmann, 2007 for a general overview; Lombardi, 1999 for an Optimality Theoretic analysis; Rubach, 1984 for a cyclic and lexical phonology analysis; Slowiaczek & Dinnsen, 1985 for an experimental approach). While final devoicing exists in Canadian English as well, it has different conditioning factors and rates of occurrence. If the dominant language of Canadian English is influencing heritage Polish in Toronto, it should be possible, and even likely, to find a system of obstruent-final voicing in heritage Polish that is more similar to the English system than to the homeland Polish system.