Rises and Falls. Studies in the Semantics and Pragmatics of Intonation Marie Nilsenova Abstract: This dissertation explores the meaning and use of rising intonation in English and French. The central theme was provided by the generally accepted claim that there exists a typical question melody in American English. The term `question' (as opposed to `interrogative'), however, was found to be problematic because speakers tend to interpret it in various ways: at the two extremes, as 'response seeking utterances' including imperatives and requests for acknowledgment or, for other speakers, only as sentences with interrogative markers. This finding undermines existing conclusions regarding the interpretation of certain contours as 'questions'. Here, a pragmatic definition of questioning utterances was employed instead, based on the typical function of an interrogative, namely seeking an evaluative response. Some criteria for identifying evaluative response-seeking utterances in a conversation were introduced. In an experiment, it was found that evaluative response-seeking declaratives can be identified by subjects even outside of context. Furthermore, there is a set of contours that facilitates the interpretation (though their presence is neither sufficient nor necessary). These contours can best be captured with Gunlogson's definition of 'final rise' and were here described with broad ToBI alphabet as containing the nuclear tunes L*H-H%, H*H-H% and L*L-H%. It was found that these three contours are a stronger predictor of the evaluative-response seeking interpretation than other properties of the utterance (e.g., the presence of a hetero-cognitive predicate, which is otherwise linked to this particular interpretation as well). At least in this sense, the meaning of intonation is thus in no sense "weaker" than the meaning of lexical items. Next, rising and falling polar interrogatives in American English were explored in their context of use. According to the existing studies, falling interrogatives appear to be biased towards a positive response in some contexts and towards a negative one in others. This discrepancy can be captured by the decision-theoretic approach to the use of polar interrogatives, which assumes that there are two types of bias: one, where the speaker wishes for the proposition of the same polarity to be true, and another, where she expects the proposition with the opposite polarity to hold (in both cases, receiving an affirmative answer results in a higher utility than receiving a negative one). The decision-theoretic description, however, cannot account for all the data; this observation was further confirmed in a perceptual categorization task designed to test the association between nuclear tunes and types of bias. In the task, it was also found that final rises are frequently associated with speaker's wish for the affirmative answer to hold (desired state bias), while the low fall is linked to speaker's informational bias towards the negative answer. The empirical observations regarding the use of rising intonation on declaratives and interrogatives were tied to an update semantic formalization, exploiting the supposition that the primary linguistic interpretation of final rises is uncertainty (rather than questioning). In terms of the universal biological codes for pitch interpretation, their linguistic and paralinguistic adaptation was thus treated uniformly and the frequent questioning effect of rising declaratives in context was derived from the maxims of rational conversation. The proposal does away with the hybrid category of question/ rising declaratives and simply treats them as declarative sentences containing an operator of epistemic uncertainty (the final rise). It can also account for the use of rising utterances, both declarative and interrogative, in contexts which were problematic for existing theories, and for the association of rising utterances in general with politeness. Finally, some preliminary findings regarding the use of final rises in French were reported. In two corpus studies, a significant link was found between final rises (expressed with the help of the INTSINT alphabet) and polar questions, as well as final rises and discourse topic openings. While the association of rises with topic openings was stronger than the association with questions (suggesting the operation of the Effort Code), there appeared to be no noticeable tendency to use final rises with topic-opening questions. It was concluded that a more fine-grained system for intonation transcription is needed to explore the issue in detail. Keywords: