The appearance of Glides in Classical Arabic defective verbs Abdellah Chekayri Tobias Scheer 1. Introduction 1 In Classical Arabic, so-called weak verbs subdivide into two categories: deaf verbs with identical C 2 and C 3 (e.g. madad "to extend") and those showing a glide in either C 1 , C 2 or C 3 . It has been shown in Chekayri & Scheer (1996) that the distribution of [y] and [w] is predictable in the latter class. That is, the glide appearing in some forms of a given verb is the output of a derivation originating in a vowel, that is V 2 for defective (glide in C 3 ) and hollow (glide in C 2 ), V 1 for assimilated (glide in C 1 ) verbs. 2 The nature of this derivation is apophonic in the sense of Guerssel & Lowenstamm (1996). According to this view, weak verbs are underlying biliterals that resort to apophonic glide-creation iff a situation arises where a consonant is needed in order to fill in a vacant position of the template. Chekayri & Scheer (1996) predict which glide will appear in which verb. In the present paper, we investigate the circumstances under which this glide surfaces throughout conjugation. Its presence vs. absence is considered to be unpredictable in traditional work such as, among others, Siibawayhi (1988) and Fleisch (1979). In these sources, the reader is simply given an amorphous list of glide-triggering and glide-preventing contexts that are not claimed to share any property. We intend to show that 1) the appearance of the glide throughout conjugation of defective verbs IS predictable, 2) its prediction can only be achieved when assuming Chekayri & Scheer (1996) and 3) the same insertion- strategy Classical Arabic recurs to elsewhere in the grammar is at work here. That is, a glide is apophonically created iff a phonologically "illegal" 3 situation arises. We start in section 2 by recalling the relevant parts of Chekayri & Scheer (1996). Sections 3 and 4 present the excessively complex system woven by the conjugation of defective verbs, stating relevant surface generalisations and putting forth an analysis thereof. Finally, it is shown in sections 5 and 6 that two misbehaving personal forms in fact reveal the more general opposition of derived vs. underived verbs.