Using brain imaging to gauge difficulties in processing ambiguous text by non-native speakers

Processing ambiguous text is an ever challenging problem for humans. In this study, we investigate how native-Arabic speakers face problems in processing their non-native English language text which involves ambiguity. As a case study, we focus on prepositional-phrase (PP) attachment ambiguity whereby a PP can be attached to the preceding noun (aka low attachment) or the preceding verb (aka high attachment). We setup an experiment in which human participants read text on a computer screen and their brain activity is monitored using near infrared spectroscopy. Participants read two types of text: one involving PP-attachment ambiguity and the other unambiguous text which is used as a control for comparison purposes. The brain activity data for ambiguous and control text are clustered using hierarchical-clustering technique available in Weka. The data reveal that Arabic speakers face more difficulty in processing ambiguous text as compared to unambiguous text.