The article puts forward a discourse-pragmatic approach to the notoriously evasive phenomena of contrastivity and emphasis. It is argued that occurrences of focus that are treated in terms of ‘contrastive focus’, ‘kontrast’ (Vallduví & Vilkuna 1998) or ‘identificational focus’ (É. Kiss 1998) in the literature should not be analyzed in familiar semantic terms like introduction of alternatives or exhaustivity. Rather, an adequate analysis must take into account discourse-pragmatic notions like hearer expectation or discourse expectability of the focused content in a given discourse situation. The less expected a given content is judged to be for the hearer, relative to the Common Ground, the more likely a speaker is to mark this content by means of special grammatical devices, giving rise to emphasis.

<p>This dissertation presents a new theoretical solution to the sociological <i>problem of observability</i>: the question of the extent to which and by what means individuals "observe" or infer mental states of other individuals, thereby sharing knowledge with them. The answer offered here states that the social situation of <i>dialogue</i> permits a speaker to use utterances to compel a hearer to generate specific and expectable assumptions about some of the speaker's intentions and beliefs.</p><p>In order to show precisely why and how dialogue possesses this capacity, the dissertation proceeds deductively. Dialogue is defined as a situation where interlocutors (1) are <i>compelled to overhear</i> what the respective other is saying, (2) apply <i>socially shared semantic rules</i> to decode utterances into private cognitive representations, and (3) act <i>as if </i>they expect that any utterance they make will be met with a <i>reply of acceptance </i>rather than a reply of rejection. It is demonstrated that the bilateral operation and anticipation of these constraints allows the hearer of an utterance to make a systematic guess at the intentions and beliefs that led its speaker to produce it.</p><p>Drawing on the works of H. Paul Grice, the dissertation shows that the hearer's guess becomes systematic by focusing on an <i>underlying informative intention</i>. It corresponds to the intention the speaker could anticipate<i> </i>the hearer would ascribe to him. By means of this expectable imputation, the hearer arrives at an <i>adequate </i>explanation of what social goal the speaker's utterance was meant to achieve.</p><p>The treatise concludes by analyzing the specific conditions under which a minimum sequence of three turns leads to <i>mutually ratified shared knowledge</i>. Whereas the status of merely shared knowledge is fundamentally precarious, mutually ratified shared knowledge is mutually recognized to be mutually known and, therefore, constitutes a societal solution to the problem of observability.</p>

This dissertation presents a new theoretical solution to the sociological problem of observability: the question of the extent to which and by what means individuals "observe" or infer mental states of other individuals, thereby sharing knowledge with them. The answer offered here states that the social situation of dialogue permits a speaker to use utterances to compel a hearer to generate specific and expectable assumptions about some of the speaker's intentions and beliefs. In order to show precisely why and how dialogue possesses this capacity, the dissertation proceeds deductively. Dialogue is defined as a situation where interlocutors (1) are compelled to overhear what the respective other is saying, (2) apply socially shared semantic rules to decode utterances into private cognitive representations, and (3) act as if they expect that any utterance they make will be met with a reply of acceptance rather than a reply of rejection. It is demonstrated that the bilateral operation and anticipation of these constraints allows the hearer of an utterance to make a systematic guess at the intentions and beliefs that led its speaker to produce it. Drawing on the works of H. Paul Grice, the dissertation shows that the hearer's guess becomes systematic by focusing on an underlying informative intention. It corresponds to the intention the speaker could anticipate the hearer would ascribe to him. By means of this expectable imputation, the hearer arrives at an adequate explanation of what social goal the speaker's utterance was meant to achieve. The treatise concludes by analyzing the specific conditions under which a minimum sequence of three turns leads to mutually ratified shared knowledge. Whereas the status of merely shared knowledge is fundamentally precarious, mutually ratified shared knowledge is mutually recognized to be mutually known and, therefore, constitutes a societal solution to the problem of observability.