Looking into international research groups' digital discursive practices: Criteria and methodological steps in the compilation of the EUROPRO digital corpus

Research in Corpus Linguistics 8/2: 87–102 (2020). ISSN 2243-4712. <https://ricl.aelinco.es> Asociación Española de Lingüística de Corpus (AELINCO) DOI 10.32714/ricl.08.02.05 Looking into international research groups’ digital discursive practices: Criteria and methodological steps in the compilation of the EUROPRO digital corpus Daniel Pascual – Pilar Mur-Dueñas – Rosa Lorés University of Zaragoza / Spain Abstract –The EUROPRO digital corpus was designed by the InterGedi research group, based at the University of Zaragoza (Spain). The main focus of InterGedi is the analysis of the textual resources used by international research groups as part of their dissemination and visibility strategies. The corpus comprises a collection of 30 international research project websites funded by the European Horizon2020 Programme (EUROPROwebs corpus). By looking into their websites, 20 projects were observed to maintain a Twitter account and the tweets from these accounts were the basis for the compilation of the EUROPROtweets corpus. This paper delves into the criteria used for the selection of the research project websites and the methodological steps taken to classify, label and tag the verbal component in these websites and tweets. The paper discusses the challenges in the compilation of the corpus because of the dynamic, hypermodal, and hypermedial nature of the digital texts it contains. The paper closes by underlining the potential uses and applications of EUROPRO in order to gain insights into the digital discursive and professional practices used by international research groups to foster their visibility online. Keywords – corpus design; digital discourse; research project websites; Twitter; e-visibility; Computer-Mediated Communication 1. INTRODUCTION 1 Professional practices are increasingly influenced by digital communication. This is no exception for scholars who need to deploy digital discursive practices, especially when it comes to disseminating the results of their research. It is not only necessary for 1 The compilation of the corpus has been carried out within the framework of the project Towards Greater Visibility and Dissemination of Scientific Research: A Linguistic, Rhetorical and Pragmatic Study of Digital Genres in English as a Language of International Communication (grant number FFI2017-84205-P). We would like to thank Dr Enrique Lafuente-Millán for coordinating different stages during the process of compilation. We would also like to express our gratitude to our research colleagues in the InterGedi (Interpersonality in Digital Genres) research group for their collaboration in this endeavour. Finally, we are also grateful to Dr Rosana Villares Maldonado, a research fellow from our research group CIRES (Gobierno de Aragón – H16_17R), for providing technical support in the compilation of the corpus. 88 academics to produce primary output, which certifies and legitimises new knowledge (Puschmann 2015: 31), but also to disseminate it broadly, which is frequently done online and in English. With the aim of undertaking lexico-grammatical, pragmatic, discursive and genre analyses of digital texts in the international research project websites, the InterGedi research group 2 compiled a database of 100 research project websites funded by the European Horizon2020 Programme, henceforth H2020. Out of these websites, a corpus of 30 webs fulfilling the criteria described below was compiled (see Section 2.1), and the texts were downloaded and tagged. This digital corpus was named EUROPRO and consists of two collections: EUROPROwebs which includes the texts downloaded from the 30 research project websites that were selected, and EUROPROtweets which includes the tweets from the 20 projects which had a Twitter account. In order to compile the EUROPRO digital corpus, the World Wide Web was used ‘for’ a corpus rather than ‘as’ a corpus (Fletcher 2013), as a careful selection of webs and their texts was made to compile our own corpus. This is an effective alternative to foster fine-grained analyses but requires solving the challenges of the digital environment before the texts can be processed and worked on, as will be discussed presently (see Section 2.2). EUROPRO can also be described as a specialised and static corpus (Gries and Newman 2013: 259), since it delves into the context of international research communication by compiling, at a given point, digital instances of researchers’ discursive practices for the dissemination of their projects. The analysis of this type of discourse through specialised and ready-made corpora, as is the case in the EUROPRO digital corpus, is advantageous because such corpora tend to be of a manageable size. Compiling a specialised ready-made corpus allows for more qualitative analyses and may help overcome the de-contextualisation of texts and of the particular discursive and linguistic features that are analysed. Such a de-contextualisation of texts and their features is a frequently criticised methodological aspect in corpus-based studies. In this sense, the present paper intends to provide metadata which may contribute to contextualising the digital material included in the corpus, thus facilitating its use and its application. 2 For information on the research group, see http://intergedi.unizar.es/ 89 The EUROPRO digital corpus is also static in that the texts were compiled at a specific moment and not modified or extended, regardless of their evolution in the digital sources where they are hosted. As a result, and to deal with the organic nature of websites and Twitter accounts, our corpus captures texts as published final products which constitute the basis for our analyses. The compilation of the corpus over time would have been complex and so would have been its use, since multiple, on-going versions of the texts would have been available. Instead, we decided EUROPRO to remain static and to retrieve enough contextual information, as to cater for the process of text crafting and publication. While it is true that texts online can be easily accessed and saved, important challenges and decisions need to be made when compiling corpora emerging from ‘Computer-Mediated Communication’, since these are distinct from those conformed by off-line texts or speech genres (Collins 2019). Some of these challenges are related to the selection of the specific sites and the texts in the corpora to ensure representativeness and to the coding of contextual information, which is of great relevance, especially in the website. The EUROPRO digital corpus contains textual documents in a reduced form consisting of character strings (Beißwenger and Storrer 2008: 297), but including a prominent layout and structure as well as multimodal elements (see Section 2), in the understanding that the combination of different modes —verbal, visual and audiovisual— makes meaning as a multimodal ensemble (Kress and van Leeuwen 2001; Jewitt 2016). The aim of this paper is to describe the selection and nature of the texts compiled in the EUROPRO digital corpus and to justify the criteria followed in its compilation as regards size, balance, representativeness and topic (Sinclair 2005). Section 2 provides the description of the corpus. Section 2.1 discusses the criteria followed for its compilation while Section 2.2 outlines the methodological aspects considered when compiling, downloading and storing such texts in the belief that texts need to be gathered according to explicit design criteria (Tognini-Bonelli 2001: 2). Section 2.3 offers information on the process of labelling and tagging the EUROPRO digital corpus and Section 2.4 describes some important contextual factors of the corpus. Finally, Section 3 deals with the uses and applications of the corpus and Section 4 provides some concluding remarks. 90 2. DESCRIPTION OF THE CORPUS The websites and Twitter accounts selected for the EUROPRO digital corpus emerge as part of the communication and dissemination plans included in the European Horizon2020 Framework Programme for Research and Innovation. Thus, they are taken to be instances of current digital scientific writing practices which, among others, serve the purpose of accounting for the adequate investment of public expenditure. From a technical viewpoint, research project websites, as all websites, contain fluid texts which are featured by hypermediality and divided into webpages. These webpages play the role of dynamic nodes and host verbal and audiovisual content that users can navigate through hyperlinks (Djonov 2007: 145). In the case of the research project websites, they render visible a hybrid nature interweaving traditional and digital sources in their design and content organisation. Consequently, the overall function of the genres and texts housed in the research project websites sits “uneasily somewhere between a commercial, technical description of the product and a more formal report on facts” (Stein 2006: 5). These websites serve as repositories of the activities and productivity of the project, as transmitters of the current values of scientific research and as venues to strategically engage with interested users and make the research available to a broad audience (Lorés 2020: 1). The tweets gathered in the EUROPROtweets collection are intended to illustrate the use that research groups make of social networks for the dissemination of their projects, assuming that “interactions in social media contexts may enable selfpromotion strategies that result in social or economic gain” (Page 2012: 182). The dynamicity, immediacy and addressivity of social media such as Twitter enable research groups to develop a distinct kind of communication about the projects. In Twitter, they report on scientific progress and also devote space to daily issues, related topics and social and professional bonds. Hence, tweets are maximised to mediate everyday routines of professional research work and connect users, collaborators and beneficiaries at scholarly events (Kuteeva 2016: 440). In all, Twitter users may enact an ‘ambient identity’ to address the mass online audience and to construe an experience of semiotic belonging to different groups (Zappavigna 2014: 2–3). Such an identity originates in users’ discourse choices, in the values conveyed through them and in their exploitation of the affordances of this medium. 91 2.1. Criteria for the compilation of the EUROPRO digital corpus The EUROPRO digital corpus emerges from a database of 100 websites of H2020 projects. It comprises two collections of texts: EUROPROwebs and EUROPROtweets. The former contains 30 research project websites funded under the H2020 programme with a word count of 394,072 and an average of 13,136 words per website. We here followed Biber et al. (1998: 243), who point out that “representativeness refers to the extent to which a sample includes the full range of variability in a population.” Therefore, we attempted to collect a specialised corpus which included samples of websites, within the context of the H2020 programme, and would allow to draw reliable insights on aspects of genre theory, metadiscourse, pragmatic strategies and multimodality. Several criteria were established on the selection of the 30 websites for the corpus to be “a good sampling” (Koester 2010: 69). First, the research projects whose websites were selected had to aim at knowledge creation and dissemination in their respective disciplinary fields and not at training PhD students or professionals. Second, we followed a convenience sampling method which entailed choosing research projects with at least one member from the University of Zaragoza (Spain) or a research institution based in Zaragoza. This allowed to complement our text-based analysis with valuable contextual evidence from potential informants, as discussed in Section 3. Third, at some point, the date of the projects should coincide with the development of our own research project (2018–2021), so that the most recent digital academic practices could be studied. Given the importance of social media in general —and Twitter in particular— for dissemination purposes, EUROPROtweets was compiled as an extension of EUROPROwebs. This collection of Twitter accounts of 20 research projects consists of 4,219 tweets containing 88,970 words, with an average of 211 tweets and 4,449 words per account. The use of social media is indeed highlighted by the communication plans endorsed by Horizon2020 and is generally adopted by research groups. The choice of Twitter as the object of study was also based on an observational analysis of the range of social networks maintained by the H2020 research projects within the representative sample, as described in Table 1 below. Out of the 30 research projects, only eight did not make use of any social networks to disseminate their research results. 20 research projects made use of Twitter, which was the most frequent social network, 13 made use 92 of LinkedIn, ten made use of Facebook and five made use of YouTube. Interestingly, it was also observed that over 50% of the webs were linked to, at least, two social networks, as shown in Table 1. NUMBER OF SOCIAL NETWORKS NUMBER OF H2020 PROJECTS PERCENTAGE OF H2020 PROJECTS 0 8 26.6% 1 6 20% 2 8 26.6% 3 6 20% 4 2 6.6 % Table 1: Range of social networks maintained in EUROPROtweets by the research projects of H2020 2.2. Methodological decisions in the compilation of the EUROPRO digital corpus Once the selection of websites was determined, several methodological decisions were made for downloading and storing the texts. All texts from the websites were downloaded and labelled using different codes which referred to the pages or sections of the website where such texts were housed, as illustrated in Table 2. We excluded those texts housed in the websites as external downloadable documents, mainly deliverables in PDF format, as they were not considered to share the same digital nature and purpose as those texts generated for the website. DIFFERENT LABELS USED FOR THE SECTIONS CODE FOR MENU SECTIONS Home; Homepage HOME About; Objectives; Project; Summary ABOUT Partners; Researchers; Consortium; Related projects PARTNERS Work packages; Actions; Demos WORK News; Events; News and events; Blog NEWS Outreach; Publications; Reports; Deliverables; Repository OUTPUT Table 2: Codification for salient menu sections and range of headings of sections in EUROPROwebs 93 After that, information related to the extent to which the text could be directly accessed from the website menu was also recorded. In cases where the menu sections were included in the codes provided above, no additional information was required. However, if the section showed up in an unfolding menu of options, the code SUB (subordinate) was noted. Likewise, if the section was included in the website but had no label in the menu, the code EMB (embedded) was added. In this way, a representation of the options of the menu section could give us insights into the preferred position and relevance of some sections or pages throughout the websites. All texts were downloaded in May 2019 because of the dynamic nature of the research project websites. At that time, the projects in the sample had been developed to different extents just like their websites. For this reason, it was key to record information about the start and end date of the projects, as well as information about their degree of development when the texts were compiled. This information may be of great importance when discussing and interpreting the data retrieved from the analysis. Compiling a ‘Monitor Corpus’ (McEnery and Wilson 2001), which would need to be updated regularly, was disregarded as a feasible objective in our own research project. Such a compilation would entail constantly comparing updated versions of the websites and tracing them at different points in time throughout the duration of the project which, adding to being extremely time-consuming, would not have been relevant for the purposes of our research. The ‘hypermodal’ and ‘hypermedial’ nature (Petroni 2014) of the websites led to the tagging of (external, internal and peripheral) hyperlinks, 3 of visuals, such as tables, figures, pictures or logos, and of videos and audios. Thus, although the focus of the analyses, as well as the corpus, is mainly grounded on the verbal component, these multimodal and multimedial elements were not overlooked in websites, since they are affordances that combine with the verbal component as meaning-making devices (Kress and van Leeuwen 2001; Jewitt 2016). Similarly, because of the importance of layout and web design aspects, we stored screenshots for every page in the website as part of 3 Internal hyperlinks specifically link to other sites within the project website while external hyperlinks lead to external sources of information. Peripheral hyperlinks refer to project-related pages and downloadable documents which are located outside the project website. 94 the corpus, since this would allow us to go back to them when analysing verbal features in the research project websites. The methodological decisions to compile EUROPROtweets were similar to those taken in the compilation of EUROPROwebs. Here, given the dynamic nature of Twitter accounts, all tweets and retweets were downloaded in June 2019. Since these platforms are also hypermodal and hypermedial (Petroni 2014), tweets were coded and tagged for hyperlinks and multimodal elements such as pictures, videos or GIFs. One key feature of Twitter accounts is their potential interactivity. As a result, we retrieved and saved information about (1) the number of likes in each tweet at the specific date, (2) the number of retweets by other users and (3) the number and types of hashtags (#) used in the tweets by the research group and their mentions (@) to other Twitter accounts. This information should be taken into account when carrying out textual analyses based on this corpus, as there may be a correlation between discursive choices and their likely dynamic nature. 2.3. Labelling and tagging of the EUROPRO digital corpus The verbal component of the 30 websites of H2020 which conform EUROPROwebs was downloaded and saved into TXT format documents, both as a document corresponding to the whole content of the website (labelled ‘NAMEOFTHEPROJECT’) and as documents corresponding to common web sections or pages which were labelled with the codes pointed out in Table 2. The tagging of EUROPROwebs was performed manually. First, a number of general tags was determined in the light of the texts downloaded and saved, namely <hyperlink>, <image>, <video>, <table>, <graph>, <map>, <presentation>, <questionnaire> and <language>. The whole corpus was then annotated using these tags, as illustrated in Figure 1 below. Specific codes — metadiscursive, pragmatic, ethnographic or multimodal— were further applied to the different analyses undertaken by the members of the InterGedi research group. 95 Figure 1: Example of manual tagging of EUROPROwebs Manual annotation was leveraged over XML language in the belief that interpretation and evaluation play a significant role in this process at a syntactic, semantic, discursive and pragmatic level (Collins 2019) in the specialised texts that make up the EUROPRO digital corpus. Since ‘big data’ tends to be advocated for in the use of XML —and since the corpus does not contain large collections of dissimilar texts— a consistent coding and tagging carried out by the InterGedi members was preferred. This system facilitated keeping in mind the design and layout of the sites, pages and texts under analysis at all times and in a clear visual way. A similar procedure was followed in the compilation of EUROPROtweets. The verbal component of the tweets was downloaded and saved into a TXT document corresponding to each account and labelled ‘NAMEOFTHEPROJECT_T’. The tagging of EUROPROtweets was also carried out manually. As Figure 2 illustrates, several tags were determined in the light of the texts downloaded and saved: <link>, <hashtag>, <image>, <video> and <language>. 102 Corresponding author Daniel Pascual University of Zaragoza Department of English and German Studies Calle San Juan Bosco, 7 50009. Zaragoza Spain e-mail: [email protected] received: January 2020 accepted: July 2020