In Prunet, Béland, and Idrissi 2000, we presented evidence from an aphasic subject that argued for the morphemic status of Arabic consonantal roots. We predicted that inaudible glides in weak roots should resurface in metathesis and template selection errors, but at the time the relevant data were unattested. Here, we present such data, obtained from a new series of experiments with the same aphasic subject. Arabic hypocoristic formation offers another case of glide resurfacing. Both sources of data confirm that Arabic consonantal roots are abstract morphemic units rather than surface phonetic units.
Glide Resurfacing in Aphasic ErrorsThe morphemic status of the consonantal root in Semitic is of special interest to morphologists because it provides a test for competing theories of word formation, most current versions of which reflect Hockett's (1954) distinction between morpheme-based item-and-arrangement and word-based item-and-process and word-and-paradigm models. Morpheme-based models, such as Halle's (1973), operate in an input-to-output fashion by appealing solely or partly to affixation processes. By contrast, word-based models function in an output-to-output fashion and posit morphological rules turning words into other words (e.g., Aronoff 1976, S. R. Anderson 1992) or redundancy rules (e.g., Jackendoff 1975) and functions (e.g., Stump 2001, Blevins 2003 relating fully formed words to one another.Morpheme-based analyses of Semitic recognizing the consonantal root have represented the majority position in Western linguistics for most of the last two centuries. Word-based analyses of Semitic, on the other hand, deny the existence of the consonantal root, often seeing it as This article is dedicated to the memory of three great Semiticists with whose friendship and advice we were graced: