The central theme of this dissertation is multimodal viewpoint – that is, the way that bodily actions work together with spoken language to express the perspective of characters within a narrative. Drawing on research in cognitive linguistics, conversation analysis, gesture studies, and signed language studies, this dissertation argues that communication is inherently multimodal. Analyses of quotations in narratives elicited from dyads of American English speakers show that speakers mark viewpoint changes by whatever expressive means are available to them – bodily articulators such as posture, facial expression and gaze, manual gestures, and prosody in addition to language.

To investigate factors affecting the multimodal expression of viewpoint in quoted speech, we distinguish quoted monologues and dialogues and between ‘ordinary’ direct speech quotation and fictive interaction, i.e., creative extensions of direct speech quotation allowing speakers to quote objects, pets, attitudes, and so forth. We find that the multimodal expression of viewpoint is sensitive to the type of quote and the number of quoted entities. Speakers’ bodily actions are found to differentiate between quoted characters. We also find that various bodily actions are often used conjointly in order to create a full representation of the quoted character.

On the basis of these findings it is argued that human cognition – including language and communication – is rooted in the body, and is inherently multimodal.