Loanword accentuation in Japanese: Corpus study, modeling, and experiments
Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Loanword accentuation in Japanese: Corpus study, modeling, and experiments

Abstract

This dissertation presents a probabilistic model of loanword accentuation in Japanese, based on large-scale corpus data, and a pair of on-line adaptation experiments, in order to gain deeper understanding of the mechanism of assigning loanword accent in Japanese. Contrary to previous work, which assigns loanword accent solely with markedness effects (i.e., language-internal principles), my corpus study (Chapter 2) reveals that faithfulness effects to English source words exist in established loanwords. Specifically, the stress pattern of English source words, as well as the epenthetic status of loanword syllables, play an important role in assigning loanword accent. These faithfulness effects are not random but systematically interact with markedness effects in a probabilistic way. My probabilistic modeling (Chapter 3), employing the Maximum Entropy Harmonic Grammar framework, shows that integrated models with both markedness and faithfulness effects outperform markedness-only models, obtaining a description that achieves a compromise between the two. My modeling also incorporates a novel proposal on the architecture of phonological grammar for loanword adaptation, motivated by the existence of accent patterns that cannot be accounted for by the interaction of faithfulness and markedness. I argue that Japanese speakers implicitly create a model of the English stress system, which I call the “Japanese Theory of English”, and exhibit faithfulness to its outputs, even if they differ from the actual source words. Incorporating this module into the model captures the accent patterns that can be characterized as hyperforeignization, in the sense of Janda et al. (1994). A pair of on-line loanword adaptation experiments (Chapter 4) were also conducted to test the faithfulness effects to stress and the interaction with markedness in experimental settings. The results confirmed the existence of faithfulness effects and the interaction with markedness in on-line adaptations, providing converging evidence that faithfulness and the interaction with markedness are part of the phonological grammar of Japanese speakers and form the basis of loanword accentuation in Japanese. Overall, this dissertation supports the view that loanword accentuation in Japanese is determined by the competition among three factors: Japanese-internal markedness principles, faithfulness to source inputs, and faithfulness to Japanese speakers’ theory of the English stress.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View