Who are the users of national open access journals? The case of the Finnish Journal.fi platform

In this paper we study the diversity of users of open access articles on the Finnish Journal.fi platform. This platform hosts around hundred open access journals from Finland publishing in different fields and mainly Finnish and English languages. The study is based on an online survey, conducted on 48 journals during Spring 2020, in which visitors were asked to indicate their background and allow their location and download behaviour be tracked. Among 668 survey participants, the two largest groups were students (40%) and researchers (36%), followed by private citizens (8%), other experts (7%) and teachers (5%). Other identified user categories include journalists, civil servants, entrepreneurs and politicians. While new publications attract a considerable share of the views, there is still a relatively large interest, especially among students, in older materials. Our findings indicate that Finnish language publications are particularly important for reaching students, citizens, experts and politicians. Thus, open access to publications in national languages is vital for the local relevance and outreach of research.


INTRODUCTION
During the past few decades, Responsible Research and Innovation (RRI) and Open science policies have underscored the importance of societal impact of research (Gerber et al., 2020;(Mathies et al., 2020;Pölönen et al., 2018). In 2016-2017, 8% of the Finnish universities' entire output of peer-reviewed journal articles was published in national journals .
The share of English language publications is currently increasing in many European countries, especially in the social sciences and humanities (SSH) . In non-Anglophone countries, policy-makers, evaluators and also researchers may question the value of publishing in languages other than English. Consequently, national journal publishers-in Finland mostly learned societies and research institutions-need increasingly to argue for their importance to research and society in order to motivate high-quality submissions from authors, and to legitimize funding of their publishing operations and transition to open access (Late et al., 2020). Indeed, both multilingualism and open access are claimed to facilitate the dissemination of research knowledge beyond academia to support, for example, teaching, learning, enlightenment, critical debate, professional practice, innovation, and decision-making (Elsabry, 2017;Sivertsen, 2018;Zuccala, 2009). Yet, evidence of societal impact of national journals and open access remains relatively weak.
Traditionally, the impact of research has been measured by counting articles and citations in international journals indexed in Web of Science Core Collection and Scopus databases, in which publications in languages other than English are almost invisible. According to Kulczycki et al. (2020), these commercial databases covered only 3%-8% of almost 94,000 peer-reviewed journal articles published in the local languages of seven European countries by more than 50,000 SSH researchers in 2013-2015. In any case, citation analysis provides only a partial picture of research impact because not all publications and citations from research literature are counted, researchers do not cite all research they read, and not all readers are researchers (Mohammadi et al., 2015;Nicholas et al., 2005). Similarly it is important to note that downloading, reading and use are not the same thing. A publication might be downloaded but never read, or it might be scanned, for example for locating a reference to another paper, but never actually read. Moreover, reading might connote anything from briefly scanning the abstract, to selective or more in-depth reading of the text. Thus, here we mainly study 'use' in terms of a request for a particular source, rather than actual reading (cf. Kurtz & Bollen, 2010).
A new suite of methods and data, altmetrics, are developed for capturing the wider outreach of research and the different types of users of research publications (Wouters & Costas, 2012). Nevertheless, the most promising altmetrics sources for readership or user profiles, such as Mendeley, mainly cover research published in English language journals (Hammarfelt, 2014;Mohammadi et al., 2015). Readership metrics-or rather usage metrics-have been proposed in addition to metrics that are based on interactions in social media, as an alternative method for assessing the uptake of academic publications (Duy & Vaughan, 2006;Haustein, 2014). Studies in the use of online resources have focused on a range of aspects, for example the age of sources used (Nicholas et al., 2005), disciplinary differences in uptake of online materials (Talja et al., 2007) and usage depending on age of publication. A recent study on reading practices in Finnish academia points to PhD students and professors as the most frequent users of journal articles, and the articles were almost always accessed electronically (with the exception of humanities scholars who still read printed journals). Moreover, scholars were mainly interested in recent publications, and this was especially noticeable in the case of journal articles (Late et al., 2019 In the literature we often find a division between publications directed mainly to an audience of fellow researchers, and those which aim for broader public dissemination. In a discussion on the audiences of humanities research, Nederhof (2006), divides the audience into three groups: international scholars, researchers on the national or regional level and a non-scholarly audience. This division, we argue, may be applied more broadly in order to understand the different users of open access journals in national languages. Yet, we should be careful to view these categories as mutually exclusive-researchers may be active both internationally and nationally-and a fourth, and significant group of 'scholarly' users, students, could be added. Further divisions are obviously possible, and one important audience to consider is professionals-teachers, doctors, lawyers etc.-for whom access to academic publications are of great importance in their work.

Key points
• Students (40%) and researchers (36%) are the largest user groups of articles at Journal.fi platform, while those outside academia make up 24% of 668 users participating in the study.
• While 46% of users of English language articles are researchers, their share of national language (Finnish and Swedish) articles is only 25%.
• Students made greater use of the national language articles and older articles on Journal.fi than other user groups-the opposite of researcher users.
• The vast majority, 97%, of all visits to national language articles originate from Finland, while 63% of visits to articles in foreign languages-mainly English-come from foreign locations. Visitors who wished to participate in the survey had to choose one role from the following list of choices offered in Finnish, Swedish and English language: Student, Researcher, Private citizen, Other expert, Teacher, Journalist, Civil servant, Entrepreneur and Politician. No definition of roles was provided. Teachers, for example, can include anyone identifying themselves as a teacher in different contexts, such as higher or elementary education. The tracking cookies stored data about the articles and abstracts published in the 48 journals, which the participants visited during the survey. Each visitor who indicated their role and permitted tracking cookies was identified with a unique hash key.

RESEARCH QUESTIONS, DATA AND METHODS
The analysis of survey results is conducted using the basic categories of users, articles and visits: • Users: number of participants identified with unique hash key.
• Articles: number of unique articles visited by users. Results are limited to articles published in 48 journals participating in the survey study. Article is counted as visited if at least one participant user visited the article's abstract or full-text.
• Visits: number of articles visited by all participating users counted together. We counted only one visit per user per article, so multiple visits to the same article, or to both abstract and full-text, by a single user are excluded.
The visitors' IP address was used to determine a geolocation. In the analysis, we group users to three categories: (1) Finland, (2) Nordic countries (including Denmark, Iceland, Norway and Sweden), and (3) other foreign locations. In addition, information about the year and language of publication of articles was gathered from the Journal.fi platform. To analyse the age of articles, we divided them into six groups according to the year of publication: 1881-1999, 2000-2009, 2010-2014, 2015-2016, 2017-2018 and 2019-2020. As regards languages, we distinguish two broad categories: (1) national languages (Finnish and Swedish), and (2) foreign languages (mainly English). The population of speakers, the political, administrative and cultural status, as well as the role in scholarly communication of languages differ between regions and countries. In studies concerning publication languages, a distinction is often made between English and other languages. Also languages having a formal status as official and/or national languages in certain countries can be regarded as a special group of 'local' languages . According to the Constitution of Finland, 'the national languages of Finland are Finnish and Swedish'. Finnish is the language of 87% of the population, while Swedish-the main administrative language until the late 19th century-is the first language of 5%. In the case of Journal.fi platform, the number of articles in languages other than Finnish, Swedish and English is very small. Therefore, all foreign languages are grouped as one category.
While the novel mixed-method approach allows us to scope the diverse groups of users of the research articles published in a subset of 48 journals available on the Journal.fi platform, it also has clear limitations. The unique hash key is created and stored in the user's browser, so different users coming from the same IP address are separated as long as they use their own devices. However, if one user accepts the survey and other users happen to visit Journal.fi platform from the same device, and the cookie was not deleted, more than one person is counted as one user (one hash key). If a user changes the device or deletes cookies and accepts the survey again, she will be counted as two separate users (two hash keys). Some participants may have accessed Journal.fi from more than one country. The period of time, during which the tracking cookies stored information on the visited articles, varies between participants (ranging from 54 days to 1 day). All this may result in a certain degree of inaccuracy in the counting of the actual number of users, visits and geolocations. Also the self-selection of participants to an online questionnaire targeted at platform users may lead to underrepresentation of certain groups that may be less willing than others to participate in the study. Despite these limitations, the survey provides new and valid information on the range of users as well as the language and age of articles they visited.

Role of users
Total of 668 users participated in the survey (Table 1). Survey participants were asked to indicate the role in which they searched or used the research accessible at Journal.fi. platform.

Geographical origin of users
The country of origin of the participants was identified based on the IP-address. In all, 68 countries were identified as places from which Journal.fi was accessed by the participants. Besides Finland (371 users), the 10 most frequently represented countries were United Kingdom (34), United States (29), Sweden (25), Germany (13), Estonia (12), Netherlands (11), India (10), Indonesia (10), Italy (8) and Japan (8).
As shown in Fig. 1, the majority of participants, 56%, came from Finland. Foreign users amounted to nearly a half of the survey participants, 5% coming from the Nordic countries (Sweden, Denmark, Norway and Iceland) and 39% from the rest of the world. Visitors from the Nordic countries amount to 11% of all users from the foreign geolocations, which is also far above their share of world population (less than 0.3%).

Geographical origin by user role
Participants in all user roles come from both Finland and foreign geolocations (Fig. 1). Some categories of users, such as private citizens and experts, come predominantly from Finland. Only in the case of researchers the share of participants from foreign locations is larger (56%) than the share of participants from Finland (44%).
Access to national language and foreign language articles by user role As shown in Fig. 2, in case of the national language publications students (42%) are clearly the largest group, and besides researchers (25%), also private citizens (12%) and other experts (11%) figure prominently among users. As regards the foreign language publications, researchers (46%) and students (38%) are more clearly the main user groups.

Access to old versus new articles
The survey participants visited a total of 1,546 articles from 48 journals participating in the survey study. On average, participants visited three articles but 63% visited only one article. The most active participant visited 148 articles during the period the survey was active (maximum of 54 days). The share of participants visiting two or more articles was considerably larger among participants from Finland (50%) than from the Nordic (28%) or other foreign locations (21%).
As shown in Fig. 3, the survey participants used also the old material available on the Journal.fi platform but focused strongly on the latest contents: while articles published in 2019-2020 amount to 8% of all Journal.fi articles, 25% of the articles visited by the survey participants were published in 2019-2020 (the number of articles are provided in Table 2).
A vast majority, 70.5%, of the 1,546 articles visited by participants are in Finnish, 28.4% in English, 0.6% in Swedish, and 0.6% are in German and Italian. This reflects the overall language distribution of articles available on Journal.fi platform, however the share of English language articles is larger among articles visited by the survey participants (28% compared to 19% overall).
The 1,546 articles received a total of 2018 visits by 668 participants (counting only one visit by participant per article). As shown in Fig. 4, in the case of articles published in the national languages, as well as those published in the foreign languages, the majority of visits, 35%, concerned the most recent articles published in 2019 or in the beginning of 2020 (the number of articles are provided in Table 3).

Age of articles by user role and language
There are, however, considerable differences between user categories in the age and language of visited articles. As shown in Fig. 5, while students account for 44% of all article visits, these are much less focused on the latest articles compared to other groups. This seems to suggest that older publications provide strong support for learning. Two-thirds of all the visits by the survey participants were to national language articles (1356) while one-third were to foreign language articles (662) available on Journal.fi platform. As shown in Fig. 6, the share of visits to national language articles is much larger among students and private citizens (78% and 84%, respectively) than the average among all user categories (67%). A slight majority (55%) of the article visits by researchers were to foreign language publications, however it is clear that also researchers are prominent users of national language publications. National and foreign language articles serve markedly different audiences in terms of the geographical origin of users. As shown in Fig. 7, the vast majority, 97%, of all visits to national (Finnish and Swedish) language articles are by survey participants from Finland. As expected, the foreign (mainly English) language articles available on the Journal.fi platform serve a much more international audience: users from Finland account for 37% of the article visits, while 5% of them come from the Nordic users and 58% from the rest of the world.  Share of visits by survey participants to all articles, and to articles in national and foreign languages by year of publication.
FIGURE 2 Share of survey participants using national language publications (327 users) and share of survey participants using foreign language publications (386 users) by self-reported role.

DISCUSSION AND CONCLUSIONS
This paper shows that open access publications in national languages can be central for reaching important users of research both within and beyond academia, including experts, private citizens, teachers and students. In fact, in the specific context of Finland, these groups appear much more prone to read research that is published in national languages compared to English language publications. This is a strong argument why open access publishing platforms, which provide access to publication in national languages, are an important part of the knowledge infrastructure.
Further studies would be needed to investigate to what extent our findings are generalisable across different countries. In many countries, recent years have seen a shift in the research policy from a narrow focus on international reach and excellence towards a broader agenda where notions of social impact and responsible research have been highlighted. Clearly, the question of open access, and availability of research in languages other than English is part of this broader agenda. This study, despite its limitations in terms of scope and size, does point to the importance of research published in national languages, and it strengthens recent claims regarding the need for preserving and strengthening the language diversity in scholarly communication .
The vast majority of Finnish OA journals are Diamond OA journals, meaning that they are free for both authors and readers (Late et al., 2020;Linna et al., 2020). A recent study finds that also in other countries OA journals in general, and Diamond OA journals in particular, play an important role in supporting multilingual scholarly communication (Bosman et al., 2021). This Diamond OA study also underscores the fragile economy of Diamond OA journals, often published by learned societies and research institutions, relying mainly on voluntary work of researchers. Diamond OA journals often lack editorial and technical support to meet the best technical standards of OA publishing, as exemplified by DOAJ or Plan S criteria.
One of the main recommendations of the Helsinki Initiative  Share of visits to national and foreign language articles available on Journal.fi platform by self-reported user role by survey participants. In particular, our results highlight the important role of national language publishing in a small language area: there are around five million native speakers of Finnish language, most of whom reside in Finland. This is in stark contrast, for example, to French language, which has over 76 million native speakers and hundreds of millions of more or less proficient speakers around the world. A much larger number of researchers worldwide also study French than Finnish society, culture and language. While the French-language articles on the Erudit platform have a wide international readership beyond Canada and France (Cameron-Pesant, 2019), the Finnish language content from the Journal.fi platform serves predominantly scholarly, professional and general audiences in Finland.
Our results also point out, in the specific case of Journal  Tennant et al., 2016). Another possible explanation could be that the Journal.fi platform hosts mainly SSH journals, which may not offer straightforwardly applicable information for business interest. It is also plausible that Journal.fi is not yet a well-known service among this group of users.
Publications from the most recent years attract most users, and this applies not only for researchers and students, but even more so for private citizens and 'other experts'. Hence, the rapid dissemination and accessibility to recent research is of great importance not only for researchers themselves but also to a wider audience. While new publications attract a considerable share of the views, there is still a relatively large interest, especially among students, in older materials. Thus, the use of materials, stretching as far back as 1883 highlights the need to digitalize and keep older contents accessible online.
In terms of method, we find that the approach of using an online questionnaire, which targeted active users of the platform, was an effective method. However, the self-selection of participants may lead to the underrepresentation of certain groups. For example, it could be that some groups are more prone to take part in the survey (students) compared to others (researchers). This is a question that would be interesting to study further using other methods for sampling users.
While this study focuses on the use of OA-publications in national languages, we deem that the methodology could be further developed in order to understand how academic knowledge is used in broader society. For example, users could be asked about their purpose for using a specific material, and such studies could further our understanding of the wider societal impact of research. More detailed studies looking at specific fields would also allow for more exact conclusions to be drawn regarding the use of research from different academic disciplines.
Finally, this study focused on a relatively small language context of Finland, and similar approaches for studying the use of