Annotating factuality in the TAGFACT corpus

The author of a text states, either implicitly or explicitly, his/her commitment with respect to the certainty of the situation being narrated (i.e. its factual status according to the author’s stance). In the area of corpus linguistics and natural language processing (NLP), the annotation of factuality has experimented a growing interest since it simplifies complex processes such as information retrieval or fact checking. This article describes the labels used for the annotation scheme of the corpus TAGFACT. It presents some innovative aspects such as the labelling of situations according to their dynamic or non-dynamic nature. Non-dynamic situations are further classified as states, absolute truths and properties of an eventive nature. Even though this scheme has been created for the annotation of a Spanish corpus, it is applicable to many other languages.