AnCora-UPF: A Multi-Level Annotation of Spanish

Abstract

There is an increasing need for the annotation of multiple types of linguistic information that are rather different in their nature, e.g., word order, morphological features, syntactic and semantic relations, etc. Quite frequently, their annotation is combined in a single structure, which not only results in inadequate annotations of treebanks and consequent low-quality applications trained on them, but also is deficient from a theoretical (linguistic) perspective. We present a new corpus of Spanish annotated on four independent levels, morphology, surface-syntax, deep-syntax and semantics, as well as the methodology that allows for obtaining it with fewer cost while maintaining a high inter-annotator agreement. 1