Annotation of person marking constructions in the Corpus of Hungarian Lyrical Poetry

Principles and practices

Abstract

This paper presents the annotation scheme for the manual annotation of the Corpus of Hungarian Lyrical Poetry. The corpus will consist of 400-600 annotated texts, grouped into four sub-corpora: 20th century lyrical texts from the canon of Hungarian public education, contemporary lyrical texts, slam poetry texts, and song lyrics. The manual annotation is based on automatically generated and manually checked annotations of lemmas, parts of speech and morphosyntactic features. The manual annotation of syntactic properties proposed in the annotation scheme follows a dependency analysis approach and allows us to obtain quantitative data on person marking constructions in Hungarian lyrical texts. Besides the annotation of verb-dependent relations, the paper also presents the annotation of specific phenomena such as auxiliary verbs, vocatives, elliptical structures, and nominal predicates. The annotation scheme was tested using a test corpus of 16 texts. We also provide some examples of the types of quantitative data that can be extracted from the annotated corpus.