Nonce-loan judgments and impossible-nativization effects in Japanese

Abstract

The core-periphery structure of the Japanese lexicon is well documented (e.g., Ito & Mester 1995ab, 1999), but there is some controversy as to whether it is synchronically productive. Productive core-periphery structure should show synchronic evidence for a HIERARCHY OF FOREIGNNESS (Kiparsky 1973) among non-native phonological structures, and should give rise to IMPOSSIBLE-NATIVIZATION EFFECTS (Ito & Mester 1999, 2001)--speakers should reject a nonce loanword that nativizes a 'less-foreign' structure while preserving a 'more-foreign' one. We carried out an experiment to collect nonce-loan nativization judgments from speakers of Japanese in order to test these claims. We found, first, that participants have a hierarchy of foreignness that is approximately like the predicted one, but differs in the relative markedness of singleton [p] and sequences of nasal+voiceless obstruent; we also found some interspeaker variation in the hierarchy. Second, while most participants showed nativization preferences that look like impossible- nativization effects, not all participants had a consistent hierarchy of preferences across all constraint pairs. These results have implications both for the phonological analysis of existing stratum-specific alternations in Japanese and for theoretical approaches to loanword phonology.

Interaction of lexical strata in hybrid compound words through gradient phonotactics

“…There is good evidence (Smith & Tashiro, 2019) that lexical strata (Itô & Mester, 1995) are a productive part of the synchronic knowledge of Japanese speakers. In Here, we focus on differences between the foreign stratum and the other three and collapse the Yamato, Sino-Japanese and mimetic strata into a single 'native' group that contrasts with the foreign stratum.…”

From experiment results to a constraint hierarchy with the ‘Rank Centrality’ algorithm

“…The forced-choice experiment results to be used for the comparison between the Rank Centrality and GLA-MaxEnt ranking algorithms are those from the study described in Smith & Tashiro (2019). This experiment investigated whether Japanese speakers have a synchronically productive markedness hierarchy for loan nativizations, such that certain 'foreign' characteristics are seen as more important to nativize than others.…”

“…This experiment investigated whether Japanese speakers have a synchronically productive markedness hierarchy for loan nativizations, such that certain 'foreign' characteristics are seen as more important to nativize than others. For discussion of the theoretical context of this study and its implications, see Smith & Tashiro (2019); the focus here is specifically on methods of determining a ranking from each participant's raw results.…”

“…This is analogous to the output of the GLA, as described in §3 above. The following section now compares results from RC and the GLA applied to the same data set-the pairwise comparison results from the experiment described in Smith & Tashiro (2019). Are the rank orders and the distances between constraints derived by the two algorithms equivalent?…”

From experiment results to a constraint hierarchy with the ‘Rank Centrality’ algorithm

“…Test-case experiment. The forced-choice experiment results to be used for the comparison between the Rank Centrality and GLA-MaxEnt ranking algorithms are those from the study described in Smith & Tashiro (2019). This experiment investigated whether Japanese speakers have a synchronically productive markedness hierarchy for loan nativizations, such that certain 'foreign' characteristics are seen as more important to nativize than others.…”