McGurk fusion effects in Arabic words

Our earlier work, on incongruent audiovisual segments embedded in English wordand phrasestimuli, showed that McGurk fusion is sensitive to the phonological context of a segment. In this paper we describe the preparation and early testing of Arabic word stimuli with an incongruent phonetic segment. We confirm that in the context of Arabic words, fusion similar to that in English words still occurs, but is affected by differences in phonological contrasts between Arabic and English speech. In previous work on English, we have contrasted fusion in onset and coda sites at wordmedial and word-edge positions. In Arabic this taxonomy of sites is rendered questionable by disagreements about syllabification and by the widespread occurrence of geminate consonants, which possibly behave as both coda and onset. We develop an improved site taxonomy, design a small lexicon of Arabic words that covers place and manner contrasts at all types of site, and outline some of the problems encountered when creating incongruent data for this lexicon. Finally, we report on fusion experiments using 10 participants that suggest there are no coda-onset distinctions in the mental models that mediate word perception amongst native Arabic speakers.