Arabic dialects are characterized by the occurrence of geminate consonants in word-medial and word-nal position. is article relates the patterning of Arabic geminates to the ongoing controversy in phonological theory concerning the representation of geminate consonants. Two views are contrasted: the prosodic length analysis of geminates whereby a geminate is underlyingly a single consonant phoneme linked to two C-slots, and the moraic weight representation where a geminate is underlyingly a single consonant linked to a mora. We speci cally argue that the patterning of geminate consonants in Arabic dialects largely supports the moraic weight representation. Evidence comes from phenomena such as the patterning of word-nal geminates, the behavior of geminates with respect to stress, geminates in loanwords, and geminates in rst language acquisition. We show that each of these phenomena supports the moraic weight representation of geminates.

We present a strategy for dependency annotation of corpora of second language learners, dividing the annotation into different layers and separating linguistic constraints from realizations. Specifically, subcategorization information is required to compare to the annotation of realized dependencies. Building from this, we outline dependency annotation for coordinate structures, detailing a number of constructions such as right node raising and the coordination of unlikes. We conclude that branching structures are preferable to treating the conjunct as the head, as this avoids duplicating annotation.

We examine some non-canonical annotation categories that license missing material (ellipses and enumerations). In extending these categories to learner data, the distinctions seem to require an annotator to determine whether a sentence is grammatical or not when deciding between particular analyses. We unpack the assumptions surrounding the annotation of learner language and how these particular phenomena compare to competing analyses, pointing out the implications for annotation practice and second language analysis.

This study presents data on the first language development of final consonant cluster acquisition in Cairene Arabic. We compare the production of final consonant clusters of two siblings (an older brother and a younger sister) acquiring Cairene Arabic in a monolingual setting when both were 2 years, 8 months (2;8). Since one child had more target-appropriate clusters than the other at that age, we get a glimpse of the developmental path of final consonant cluster acquisition in Cairene Arabic. Our findings include that pharyngeal-initial final clusters are acquired early and that gemination is the common “repair” strategy for clusters not yet acquired. We conclude by relating our findings to theories regarding the nature of first language phonological acquisition.