Are We DONE Yet? A corpus-based analysis of DONE as a sentence-boundary diagnostic in DGS

Abstract

This paper looks at the lexemes DONEα and DONEβ in the Public DGS Corpus to find out whether they have a tendency to occur more frequently in sentence-final position than elsewhere and are therefore good cues for identifying sentence boundaries in German Sign Language (DGS). Since the Public DGS Corpus does not contain annotations of sentence boundaries or continuous syntactic tagging, we used time-aligned German translations as provisional sentential units. DONEα and DONEβ were selected since the literature suggests that they are best analyzed as right-peripheral heads. Additionally, we compared the distribution of DONEα to that of ALREADYα, which served as a baseline. The results suggest that DONEα and DONEβ indeed occur more often at the end of a translation tag than elsewhere in DGS. A more detailed analysis of a randomly selected subset of tokens of DONEα and DONEβ revealed that some of their functions are associated more strongly with sentential boundaries than others. Restrictive focus and discourse marker uses occur more consistently in the final position than, e.g., aspectual uses.