Metrical Annotation of a Large Corpus of Spanish Sonnets: Representation, Scansion and Evaluation

Metrical Annotation of a Large Corpus of Spanish Sonnets: Representation, Scansion and Evaluation. Borja Navarro-Colorado1, Mar´ı Ribes Lafoz2, Noelia S´anchez 1Department of Software and Computing System, 2UA Language Centre University of Alicante, Alicante (Spain) borja@dlsi.ua.es, chitty@csidiomas.ua.es, noeliasanchezlopez@gmail.com Abstract In order to analyze metrical and semantics aspects of poetry in Spanish with computational techniques, we have developed a large corpus annotated with metrical information. In this paper we will present and discuss the development of this corpus: the formal representation of metrical patterns, the semi-automatic annotation process based on a new automatic scansion system, the main an- notation problems, and the evaluation, in which an inter-annotator agreement of 96% has been obtained. The corpus is open and available. Keywords: metre, scansion, corpus annotation, poetry, Spanish 1. Introduction Even though the application of Natural Language Process- ing techniques to literary texts is not a new task, nowadays it is still undergoing a new stage of development.1 For in- stance, the macroanalysis approach (Jockers, 2013), based on the study of large periods of the history of literature through the computational analysis of large corpora, needs the development of large and richly annotated corpora of literary texts. Our interest is focused on the computational analysis of Spanish Literature, and specifically on the metrical and semantic aspects of Spanish Golden-Age Sonnets (Navarro Colorado, 2015). 16th- and 17th-Centuries Span- ish poetry is considered one of the best periods of the His- tory of Spanish Literature (Rico, 1980 2000; Terry, 1993; Mainer, 2010). It was the time of great, famous and “canon- ical” Spanish poets such as Miguel de Cervantes, Lope de Vega, Garcilaso de la Vega or Calder´on de la Barca, among others. Due to the importance given to this period, it has been deeply studied by scholars since the 19th century. We are persuaded that new approaches based on the macro- analysis framework (Jockers, 2013) (or “distant reading” (Moretti, 2007; Moretti, 2013)) could shed new light on this period. We have compiled a large and representative corpus of Spanish sonnets written during the 16th and 17th centuries, in which the metrical pattern of each verse has been an- notated following a semi-automatic process. In this paper we will present and discuss the three main aspects of this corpus: 1. a formal representation of metrical information; 2. the semi-automatic annotation process based on a new automatic scansion system; 3. the evaluation of the annotation. The paper is organized as follows: after a brief State of the Art, section 3. shows how the corpus of Spanish Golden- 1See, for example, the Computational Linguistics for Litera- ture Workshop series. Age sonnets has been compiled and and how the metri- cal patterns have been formally represented; section 4. de- scribes the scansion system that has been specifically de- veloped to annotate these metrical patterns, the manual revision and the main problems detected; and section 5. presents the evaluation of the annotation. The paper fin- ishes with some conclusions. 2. State of the Art Nowadays there are several projects devoted …

Content truncated to 500 words