Two Reasons for Not Reducing Inferences of Social Deixis to Conventional Implicature

다국어 초록 (Multilingual Abstract)

1. In "Pragmatics and Social Deixis" (1979) and Praomatics (1983), Levinson argues that inferences arising from social deixis are non-cancellable as well as detachable, non-truth functional, and non-caculable like conventional implicatures. Therefore, he claims, social deictic items such as French tu, vous, Tamil referent honorific nka, etc, carry conventional implicatures.
Contrary to what Levinson claims, however, I will show in this paper that social deictic item do not carry conventional implicatures by pressenting two sorts of empirical evidence; 1) while conventional implicatures are a part of the semantic content of the sentence with which they are associated, inferences of honorifics are not; 2) whereas conventional implicature is non-calculabel, inferences of honorifics are calculable.
Before presenting in detail the above mentioned evidence, I will go over briefly with examples what led Levinson to conclude that social deictic items carry conventional implicature. I hope that the following summary of Levinson's argument makes this paper accessible to those readers who are not familiar with notions like conventional implicature, detachable, etc, explicated in Grice (1975).
2. In order to argue that inferences of honorifics can not be reduced to either presupposition or conversational implicature, one of whose main features is a cancellability, Levinson tries to show that inferences of honorifics are not cancelled by conjoined sentence. For example, given that the presupposition of the utterance "John didn't manage to stop in time," is "John tried to stop in time." (1) below shows that the presupposition of the first sentence, i.e. "John tried to stop in time." is cancelled by the conjoined sentence underlined:
(1) John didn't manage to stop in time. But in fact, he didn't even try.