TY - JOUR
T1 - Japanese final-accented and unaccented phrases
AU - Warner, Natasha
N1 - Funding Information:
I am grateful to John Ohala, Yoko Hasegawa, Charles Fillmore, Nick Campbell, and Norio Higuchi for discussions of this paper, and to Patrice Speeter Beddor, Timothy Vance, and an anonymous reviewer for their helpful comments. I would also like to thank Yoko Hasegawa for help with the materials for the experiment and Mike Ward and Shawn Ying for technical assistance. Part of this work was carried out while at ATR and supported by the NSF Summer Institute in Japan.
PY - 1997/1
Y1 - 1997/1
N2 - The most common theory of Japanese pitch accent predicts that words with final-accent and those with no accent have the same f0 contour within the word, and diverge only in the mora following the word. Pierrehumbert & Beckman's (1988) theory, however, predicts a difference even within the word for the two types. Previous instrumental studies of final-accented and unaccented words, the results of which have not been conclusive, used only one- or two-mora words, which introduce a confound for Pierrehumbert & Beckman's theory. In this experiment, longer words are used, and reiterant speech is used to lower variance. All four speakers produced the words with a difference between the two accent types in the direction predicted by Pierrehumbert & Beckman's theory, and this difference is statistically highly significant. However, the speakers produced an additional difference between the two types which is not predicted by either theory. The results have implications for other theories of Japanese pitch accent. The experiment also tests, and confirms, the validity of reiterant speech as a method.
AB - The most common theory of Japanese pitch accent predicts that words with final-accent and those with no accent have the same f0 contour within the word, and diverge only in the mora following the word. Pierrehumbert & Beckman's (1988) theory, however, predicts a difference even within the word for the two types. Previous instrumental studies of final-accented and unaccented words, the results of which have not been conclusive, used only one- or two-mora words, which introduce a confound for Pierrehumbert & Beckman's theory. In this experiment, longer words are used, and reiterant speech is used to lower variance. All four speakers produced the words with a difference between the two accent types in the direction predicted by Pierrehumbert & Beckman's theory, and this difference is statistically highly significant. However, the speakers produced an additional difference between the two types which is not predicted by either theory. The results have implications for other theories of Japanese pitch accent. The experiment also tests, and confirms, the validity of reiterant speech as a method.
UR - http://www.scopus.com/inward/record.url?scp=0030642446&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0030642446&partnerID=8YFLogxK
U2 - 10.1006/jpho.1996.0033
DO - 10.1006/jpho.1996.0033
M3 - Article
AN - SCOPUS:0030642446
SN - 0095-4470
VL - 25
SP - 43
EP - 60
JO - Journal of Phonetics
JF - Journal of Phonetics
IS - 1
ER -