Applying Classification Trees to Stress Patterns of Japanese Loanwords in English

Abstract

This paper provides an analysis of the stress patterns of loanwords of Japanese origin in English. The ctree function of the party package in R is used to create classification trees. The stress pattern is not immediately obvious, so the algorithm is used to discover the significant predictor variables and create a classification tree. This illustration displays the stress pattern in a straightforward manner. The determining factor for the stress assignment of Japanese loanwords in English is syllable weight and vowel sonority. The presence of a final coda is the significant variable for the stress assignment in two-syllable loanwords, whereas vowel sonority is the significant variable for stress assignment in three and four-syllable loanwords. This provides support for Kenstowicz’s [1997] and deLacy’s [2002] universal sonority hierarchy.

Keywords

• loanword phonology
• classification/conditional inference tree
• suprasegmental