Emergent Rankings in Foreign Word Adaptations Ellen Broselow, Marie Huffman, Jiwon Hwang, Sophia Kao, and Yu-An Lu Stony Brook University (SUNY at Stony Brook) 1. Introduction * In this paper we address two problems in loanword phonology: (i) why some foreign structures are more likely to be preserved than other equally novel foreign structures and (ii) why particular repairs for foreign structures are favored, even when the native language offers a choice of possible mappings to legal native structures. We will focus on the adaptation of the structures [ti] and [si] in words borrowed from English into Japanese. Although neither [ti] nor [si] is attested in native Japanese vocabulary, [ti] is far more likely than [si] to be preserved in established loanwords, as illustrated by the pronunciation of ‘Citibank’ as [itibaõku] (Itô & Mester 1995, 1999, 2001), where [s] is palatalized but [t] is retained. This asymmetry exemplifies 1 a problem in loanword adaptation that has been called differential difficulty (Broselow 2009) or differential importation (Kang 2011), in which some novel foreign structures are more readily accepted than others. Such asymmetries can generally be described in terms of relative rankings among markedness (M) constraints, and the Japanese asymmetric importation pattern has been analyzed as reflecting a native language ranking *[SI] >> *[TI ] (Itô & Mester 1995, 1997, 2001). However, because neither the native nor the foreign language data appear to motivate this ranking, the source of the ranking demands explanation. We argue that the *[SI] >> *[TI ] ranking reflects inherent universal differences in the perceptibility of the contrast between the foreign structures and the corresponding legal native structures. We provide experimental evidence that even speakers of English are better able to discriminate [ti/±i] than [si/i], suggesting that the ranking *[SI] >> *[TI ] emerged naturally as Japanese speakers began to accurately perceive foreign [ti] while still failing to perceive foreign [si] as distinct from the legal native structure [i]. The second puzzle posed by the Japanese adaptation patterns concerns the choice of repair for [ti] and [si] sequences, designated the too-many-solutions problem (Kang 2011, following Steriade 2001) or the problem of differential faithfulness (Broselow 2009). Although a few loanwords exhibit change to a non- palatalizing vowel (spaghetti > [supagettee]; Itô & Mester 1995, 1997, Crawford 2007, 2008, 2009), the far more common pattern involves palatalization of the consonant (team > [±iimu]). Japanese native vocabulary, where dental stops and fricatives are regularly palatalized before [i], motivates a native language ranking IDENT-V >> IDENT-C, a ranking that will account for the adaptation pattern as well. However, Kenstowicz (2007) has suggested that this ranking reflects a universal preference, based in perceptual asymmetry. On this view, the P-Map (Steriade 2001, 2009) projects universal rankings of This research was supported by funding from the National Science Foundation BCS-07460227 to Ellen Broselow, * Marie Huffman, and Nancy Squires. We gratefully acknowledge the invaluable contributions of Yukiko Asano, Sara Catlin, Jessica Fareri, Jennifer Park, Eriko Sato-Zhu, and Hisako Takahashi in preparing materials and running experiments. We are indebted to Clifford Crawford and to Larry Hyman for searching their sources for additional information on Japanese and Nupe loans (respectively), and to the audiences at the Annual Meeting of the LSA in Pittsburgh, at National Cheng-Chi University in Taiwan, and at WCCFL in Santa Cruz, particularly Junko Itô and Armin Mester, for valuable comments. We follow Crawford 2009 in using [±] and [] to represent the postalveolar sounds (represented by Itô & Mester as 1 [è] and [š]), although the Japanese postalveolar fricative, which differs in some respects from the English postalveolar fricative (Li et al. 2007), is more accurately represented as [^]. We also follow Crawford in using the term ‘palatals’ rather than ‘postalveolars’ or ‘alveopalatals’ and follow common practice in representing [] as [u]. © 2012 Ellen Broselow, Marie Huffman, Jiwon Hwang, Sophia Kao, and Yu-An Lu. Proceedings of the 30th West Coast Conference on Formal Linguistics, ed. Nathan Arnett and Ryan Bennett, 98-108. Somerville, MA: Cascadilla Proceedings Project.