Parallel corpora

Lefer, Marie-Aude [UCL]

This chapter gives an overview of parallel corpora, i.e. corpora containing source texts in a given language, aligned with their translations in another language. More specifically, it focuses on directional corpora, i.e. parallel corpora where the source and target languages are clearly identified. These types of corpora are widely used in contrastive linguistics and translation studies. The chapter first outlines the key features of parallel corpora (they typically contain written texts translated by expert translators working into their native language) and describes the main methods of parallel corpus analysis, including the combined use of parallel and comparable corpora. It then examines the major challenges that are linked with the design and analysis of parallel corpora, such as text availability, metadata collection, bitext alignment, and multilingual linguistic annotation, on the one hand, and data scarcity, interpretation of the results and infelicitous translations, on the other. Finally, the chapter shows how these challenges can be overcome, most notably by compiling balanced, richly-documented parallel corpora and by cross-fertilizing insights from cross-linguistic research and natural language processing.