= Visual world paradigm
The <visual world paradigm> investigates spoken <language> processing by recording <eye tracking> while a listener views a scene and hears an utterance. The timing and destinations of fixations provide evidence about which visible referents are being considered during <sentence comprehension>. The scene can contain real objects or displayed pictures.
In <Michael K. Tanenhaus> and colleagues' 1995 experiment, a head-mounted tracker recorded eye movements while listeners followed instructions to manipulate real objects. An instruction temporarily allowed a towel phrase to describe either an apple's location or a destination. Comparing scenes with one or two apples changed the referential usefulness of that phrase and affected interpretation.<Integration of Visual and Linguistic Information in Spoken Language Comprehension by Michael K. Tanenhaus and others>[pp. 1632–1634, apparatus, one/two-referent conditions and Figs. 1–3]{ref}
The crucial comparison holds much of the linguistic input constant while manipulating relevant visual context. Fixations require time to plan and execute, so they must be aligned with the utterance and interpreted through the task's alternatives. This makes the paradigm a controlled link between a theory of comprehension and a timed behavioral record.
Back to article page
