doxa.comunicación | 29, pp. 235-254 | 245

July-December of 2019

José Luis Rojas Torrijos and Carlos Toural Bran

ISSN: 1696-019X / e-ISSN: 2386-3978

As far as the headlines of the reports are concerned, they are informative and restrained, in many cases nearly always pointing out the contending teams, and in most cases the result. Inclusion of the result in the headline of these reports is justified because in this case, no data sheet is included, so all elements that usually appear in the data sheet must appear in the text.

Image 4

Screenshot of the headline of a report published on 03-03-2019.

As in the headline, there are hardly any subjective elements in the subheading (Table 1). The only two cases in which a certain subjectivity is apparent are those referring to a 1-0 result as a “worthy triumph”, and a 1-1 final as “a poor result” for the home team. As such, in both cases the subjective weight resides in using a descriptive adjective.

Regarding the subheadings and how they appear in combination with the headline of the reports, analysis of the texts produced by the bot reveals two models that appear most frequently: a subheading that merely repeats the headline data (the result if mentioned, team names, and even certain verbs and nouns) (Image 4); or one that points to an obvious fact (e.g., a tie is “a draw”, or “two points were ceded”, or phrases such as “they will fight for a win in the next match”).

Table 1

Combination headline-subheading

Classification

Number of reports

% of total

Result in headline

51

63,75%

Subheading repeats result

13

16,25%

Subheading does not repeat/obviousness

31

38,75%

Subheading does not repeat/statistics

5

6,25%

Subheading does not repeat/subjectivity

2

2,5%

No result in headline

29

36,25%

Subheading with result

17

21,25%

Subheading without result/repeats headline

8

10%

Subheading without result/statistics

4

5%

Source: created by the authors

However, in 11% of the cases analysed, the subheading enhances and enriches the headline by introducing some statistical data in context (Image 5). This would therefore be an improved subheading model, which on other occasions would