Script activation in lipreading

In three experiments the roles of script, basic and low-level contexts in lipreading (Abbott et al., 1985) were studied. Three scripts were used: a restaurant, a railway station, and a clothing store. Presence or absence of script headers constituted a between-subjects factor. Basic and low-level context, as well as test-sentences (referring to basic and low-level of abstraction), were within-subjects factors. Also, different kinds of working memory tests were employed to assess potentially important information-processing skills for lipreading. The results show that (a) lipreaders benefit from a compatibility between context level and sentence type, (b) the directions of inferences were asymmetric such that the basic context prevented inferences at the low-level of abstraction, but not vice versa, (c) low-level context in general was most useful for skilled lipreaders in the script group, (d) skilled lipreaders in the non-script group took most advantage of basic level sentences, and (e) working memory capacity did not prove to be important for lipreading skill. Guessing and pure lipreading control experiments established that each of these potentially confounding factors only accounted for about 5% of the lipreading performance, and thus, can not explain the obtained interactive effects between visual information conveyed by lip movements and context. Theorizing departing from prototypicality vs. script accounts of inference-making was critically evaluated, and a predictive script activation mechanism, relevant to lipreading skill, was suggested.