Abstract
<title>Abstract</title> <p>This study contributes to ethnomethodological research on gestalt contexture by extending it from relatively stable and symmetrically available perceptual objects to dynamic settings marked by practical urgency and asymmetric perceptual access. Using an artificial agent’s situated conduct as a point of comparison, we explore how an experienced sighted guide’s actions methodically segmented observable and hearable properties of participants’ surroundings, and indexed various temporal and normative orders—such as those of the ongoing traffic—as constituent details of the practical activity of “crossing the street”. Our empirical observations help specify the set of practices glossed by the terms “real-time visual assistance” and “real-time descriptions” in recent research on human and human–agent interaction. We document how a human guide’s grammatically declarative statements do not merely describe a pre-existing scene (immediately experienced as coherent) but, rather, progressively make selected features of the surroundings into constituent details of a shared here-and-now.</p>