Open Access

This paper argues that conversation analysis has largely neglected the fact that meaning in interaction relies on inferences to a high degree. Participants treat each other as cognitive agents, who imply and infer meanings, which are often consequential for interactional progression. Based on the study of audio- and video-recordings from German talk-in-interaction, the paper argues that inferences matter to social interaction in at least three ways. They can be explicitly formulated; they can be (conventionally) indexed, but not formulated; or they may be neither indexed nor formulated yet would be needed for the correct understanding of a turn. The last variety of inferences usually remain tacit, but are needed for smooth interactional progression. Inferences in this case become an observable discursive phenomenon if misunderstandings are treated by the explication of correct (accepted) and wrong (unaccepted) inferences. The understanding of referential terms, analepsis, and ellipsis regularly rely on inferences. Formulations, third-position repairs, and fourth-position explications of erroneous inferences are practices of explicating inferences. There are conventional linguistic means like discourse markers, connectives, and response particles that index specific kinds of inferences. These practices belong to a larger class of inferential practices, which play an important role for indexing and accomplishing intersubjectivity in talk in interaction.

Building on а corpus of 70 hours of German driving lessons, this paper studies the use of declaratives vs. imperatives for instruction. It shows how these linguistic resources are adapted to different praxeological, temporal and participant-related environments. Declaratives are used for first instructions, task-setting and post- trial discussions. They exhibit complex syntax and do not call for immediate compliance. Their high degree of explicitness conveys how the action is to be carried out. Imperative instructions overwhelmingly correct ongoing actions of students or respond to their failure to produce expected actions. They exhibit minimal argument structure. They are reminders which presuppose that the student monitors the scene and can perform the action unproblematically. They index that requests have to be complied with immediately or even urgently.

Dieser Artikel gibt einen Überblick darüber, wie grundlegend Wissen als Voraussetzung, Gegenstand und Produkt von Verständigungsprozessen für die Organisation von Gesprächen ist. Zunächst wird ein kognitivistischer Zugang zu Wissen mit einem sozialkonstruktivistischen kontrastiert. Es werden zum einen kommunikative Gattungen, die auf die Kommunikation von Wissen spezialisiert sind, dargestellt; zum anderen wird gezeigt, wie Wissen auch dann die Gestaltung der Interaktion bestimmt, wenn der primäre Gesprächszweck nicht in Wissensvermittlung besteht. Vier Dimensionen werden angesprochen: a) Das mit dem Adressaten geteilte Wissen (common ground) ist Grundlage des Adressatenzuschnitts von Äußerungen (recipient design); b) geteiltes Wissen wird in Verständigungsprozessen konstituiert; c) der relative epistemische Status der Gesprächspartner zueinander wird durch Praktiken des epistemic stance-taking verdeutlicht und bestimmt selbst die Interpretation von Äußerungsformaten; d) epistemischer Status, soziale Identität und Beziehungskonstitution sind eng miteinander durch moralische Anspruchs- und Erwartungsstrukturen verknüpft.