In this paper we study the representation of negation in UD treebanks. We show that the existing annotations are often inconsistent with the guidelines and that there are ill-motivated differences in annotation of constructions across and even within languages. Moreover, we argue that even if the annotation of the two negation-related features (Polarity=Neg and PronType=Neg) were consistent, these two features would be inadequate for straightforwardly expressing the semantics of negation because they relate to the word level only and hence to form rather than meaning. We therefore propose to add two features, Negated=+ and DoubleNegated=+, which directly encode when a predicate is semantically under negation, and thereby allow a straightforward semantic interpretation of a UD parse in terms of negation.

This paper presents a new dataset with Discourse Representation Structures (DRSs) annotated over naturally-occurring sentences. Importantly, these sentences are more varied in length and on average longer than those in the existing gold-standard DRS dataset, the Parallel Meaning Bank, and we show that they are therefore much harder for parsers. We argue, though, that this provides a more realistic assessment of the difficulties of DRS parsing.