The present study analyzed 72 Japanese dyadic conversations between new acquaintances with regard to the choice of honorifics at the sentence level and the shifts between the use and the non-use of polite forms within the same conversation as a discourse-level phenomenon. I define "Discourse Politeness" as "a dynamic whole of functions of any element in both linguistic forms and discourse-level phenomena such as speech-level shifts and backchanneling that play a part in the pragmatic politeness of a discourse."
Focusing on these sentence- and discourse-level phenomena, I investigated situational variations in their use among dyads of speakers who vary in age and gender, important sociolinguistic variables in Japanese culture. I also examined the relationships between the sentence-level choice of honorifics and the speech-level shift as a discourse-level phenomenon, and analyzed the dynamic functions of these elements in the overall Discourse Politeness in this group of conversations.
The results revealed five points about the present use of Japanese honorifics and the manipulation of speech level shifts. 1) The use of honorifics functioned more as a stylistic choice for the speaker than as a reflection of the actual relationship with the interlocutor as traditionally explained. 2) Female interlocutors used significantly more honorifics than did males; this was the only gender-related result. 3) Only the use of the non-polite form, a minor deviation from the normative language use and dominant speech level in the setting of this study, clearly reflected the age/power relationships between speaker and hearer. 4) The percentages of types "downshifts,"--that is, "downshift from self," "downshift from interlocutor" and total downshifts, and of "upshift from interlocutor" as discourse-level phenomena clearly reflected the age/power relationships between speaker and hearer. In other words, speakers tended to downshift more frequently when talking with younger interlocutors, and to upshift from their interlocutor more frequently when they were talking with older interlocutors. These findings were statistically significant, as shown by an ANOVA. 5) Non-marked utterances that did not include any linguistic politeness markers accounted for about 25-30% of the total number of utterances in each conversation. This finding may indicate that speakers used non-marked utterances to avoid acknowledging the speakers' asymmetric relationships by choosing linguistic politeness markers that already embedded the vertical human relationships. Thus, they may have functioned as a discourse politeness strategy.
Based on these findings, I return to the universal theory of politeness, proposed by Brown and Levinson (1978, 1987), and claim that it is necessary to incorporate the concept of "Discourse Politeness" into studies of politeness. Doing so will enable researchers to contrast politeness behavior in languages with honorifics and without honorifics within the same framework, and to construct a more comprehensive "Discourse Theory of Politeness."