A Reader in Sociophonetics

196 Midori Yonezawa Morris

3. Tokyo respondents would judge Kinki pitch accent and nondevoiced
   tokens as “not from the same region.”


4. Kinki respondents judge Kinki pitch accent and nondevoiced tokens as
   “from the same region.”
   The opposite response for each token is counted as “unexpected.” “Neu-
   tral” tokens were always expected to be judged as like the respondent’s own.

Table 7.3 Tabulation (Tokyo Respondents) PA = Pitch Accent

Type Expected Unexpected

Devoiced From the same region Not from the same region

Tokyo PA From the same region Not from the same region

Nondevoiced Not from the same region From the same region

Kinki PA Not from the same region From the same region

Neutral From the same region Not from the same region

Table 7.4 Tabulation (Kinki Respondents) PA = Pitch Accent

Type Expected Unexpected

Devoiced Not from the same region From the same region

Tokyo PA Not from the same region From the same region

Nondevoiced From the same region Not from the same region

Kinki PA From the same region Not from the same region

Neutral From the same region Not from the same region

I analyzed the responses, determining the signi¿ cant phonological and
social factors for the judgments, using a multivariate logistic regression pro-
gram (Goldvarb).

5. Results and discussion

5.1 Token types

Tables 7.5 and 7.6 show the results by token types for both Tokyo and Kinki
respondents. Accent patterns are obvious clues, and the expected judg-
ments dominated. Both Tokyo and Kinki people also tended to judge tokens
with voicing variation as expected; that is, devoiced tokens were judged as
“Tokyo” or “non-Kinki,” and nondevoiced tokens as “non-Tokyo” or “Kinki.”