The Emergence of the Second Hand in Sign Language Phonology: From Underlying to Surface Representations

On the basis of corpus data of Sign Language of the Netherlands (NGT), we examine a number of issues related to the use of the non-dominant hand in signed language discourse. The prolongation of the final state of one hand of a two-handed sign while the other hand continues to produce other linguistic material (a ' hold') has been analysed in the literature as a prime piece of evidence for the presence of prosodic domains such as the phonological phrase in the organisation of sign languages. It thus provided evidence for a prosodic level of organisation more generally. It was not clear in those studies, however, how these prosodic phenomena can be represented and how they relate to the underlying lexical form of signs. We propose a representation with three levels: lexical, surface level spell-out and post-lexical. An underlying lexical representation itself (such as provided by the Dependency Model of signs that we use here) cannot handle phenomena such as spreading across multiple lexical items. For these phenomena, an intermediate surface level is needed, to connect the underlying representation to the linear surface form (syllables) holding the spell-out of underlying features. The surface level spell-out in turn is integrated into the prosodic structure of the sequences of signs, where postlexical phenomena such as (meaningful or purely prosodic) spreading across signs and phrases occur. With regards to the second hand, the underlying lexical representation does not distinguish two independent articulators. A spell out process is needed to determine the dominant hand, as well as the position of the selected fingers, the orientation, and other features of the weak hand. Our overall conclusion is that: a) the prolonged presence of the non- dominant hand can indeed be seen as a phonological process; b) this process can be analysed in terms of spreading of features and feature clusters of manual signs; and c) other surface phenomena (such as cross-articulator assimilation or 'echoing') can also be analysed as feature spreading, and in that sense are not fundamentally different from full holds. Thus, by creating an explicit multi-layered formal representation, heterogeneous post-lexical phenomena such as buoys, prosodic spreading, weak prop and manual 'echoes' can be analysed in a unified manner. Our analysis strengthens the view of signed and spoken languages being similar even in the domain of post-lexical phonology, where at first sight the modality-specific phonetic differences in the surface form appear to be substantial.