To this day, more than twenty major media outlets in several countries use bots to expand, streamline and diversify their coverage of sporting events (Rojas Torrijos, 2019).Some apply them for continuous and last-minute tracking of results on Twitter (The Washington Post), others for the generation of graphics to enrich their live events (The Telegraph), while an increasing number of media organizations decide to use them to improve their local coverage (the MittMedia group in Sweden, or the Press Association agency for local press groups in the UK).

The spread of this technology in the field of sports is directly related to the nature of the competitions themselves. These have an important statistical base that favours the management of organised data and allows for the programming of informative routines given the cyclical and repetitive nature of matches and competitions and their monitoring in the media.

This increase in the use of automated information-writing bots has generated a debate within the profession in recent years. The debate has to do with setting boundaries between the functions of machines and people, as well as the consequences of their use in a journalistic world that were more agile, diverse, and of particularly higher quality.

So far, many of the experiments developed by benchmark international media organizations have shown that bots can be useful to the industry in covering more topics, expanding their audience, and enriching coverage with data that is more abundant and faster, and if possible, in real time through new digital and mobile platforms.

3. Case study: AnaFut, the new editor of El Confidencial

Bearing this context in mind, the present article addresses the case study of the first technology of this type developed by a Spanish media company for the automatic writing of sports reports. The case involves a bot nicknamed AnaFut, developed by the digital native El Confidencial through its journalistic innovation laboratory (El Confidencial Lab) in September 2017, and this media began using it in early 2018 to produce football match reports from the Second Division B and Third Division leagues on a national level.

3.1. Objectives and hypothesis

On one hand, this study analyses the structure and content of these automatically produced texts in order to extract quantitative and qualitative measurements on the type of language that is naturally-produced, and on the other hand, its consequences on product quality within the context of the Sports section of this media company.

The research is based on the hypothesis that the routine and statistical nature of football narratives favours the application of artificial intelligence in this area of information over others, and that the automation of sports reports is the result of the refinement in the programming of a series of pre-established editorial linguistic models that are generated from data and repeated without the need for human intervention once the texts have been published.

Therefore, the nature of the reports automatically generated by this bot causes the resulting text to have the tendency to reflect a series of linguistic repetitions, especially in the use of verbs and expressions when it comes to situations of victory, defeat or a tie in the course of a match, but which are amendable and avoidable from the moment the human editor prepares the machine.