Introduction
Stanley Hoffmann might have been pleased to see that after nearly 45 years, his reflections on the failings of the International Relations (IR) discipline,Footnote 1 which he attributed to the convergence between 20th-century growth of America the nation and that of IR the ‘American’ discipline, remain relevant. He might even have been gratified to find that his assertion, that America’s ‘political preeminence’ was largely the reason for the discipline’s failure outside the United States,Footnote 2 has been referenced in numerous scholarly works.Footnote 3 But Hoffmann might also have experienced some unease that, despite his insights and those of the many works that have followed his, the situation about which he expressed such displeasure – that of an American ‘scientific’-oriented ‘quest for certainty’, a general tendency to ignore history, and an overall American dominance in the discipline – remains arguably unchanged.Footnote 4
One of the many works drawing inspiration from Hoffmann’s original treatise posits that the continuing disciplinary imbalance can be attributed to the persistence of dependent development.Footnote 5 The authors argue that through socialisation and other ‘institutional incentive structures’, peripheral scholars replicate ‘core-western modes of knowledge production and dissemination’,Footnote 6 thereby ensuring that the asymmetric relationship remains constant. Among their many recommendations for breaking the dependency cycle and moving towards a symmetric interdependent IR, the authors note the importance of going beyond the ‘linguistic hegemony of English’.Footnote 7
This article sets out to explore the assumption behind this recommendation, first by interrogating the theoretical understanding that linguistic hegemony contributes to a dependent core–periphery relationship, and then by measuring the veracity of the claim that a linguistic hegemony of English exists in IR scholarship. Readers should note that while we acknowledge the problematic binary distinction between terms such as ‘core and periphery’ or ‘West and non-West’,Footnote 8 let alone the increasing recognition of the diversity that exists within each of these,Footnote 9 we are choosing to use them here in a linguistic sense corresponding to an Anglo-American ‘core’ and a ‘periphery’ of other native language users, a distinction that is developed in more detail below. With this in mind, a number of works have already looked into the question of Anglo-American dominance in terms of IR publication practices, yet most have focused on journals in the core, namely, on ‘leading’ IR journals,Footnote 10 which, with a few exceptions, are predominantly published in the United States or United Kingdom,Footnote 11 where the linguistic hegemony is not only well documented but also understandable. It is less clear however whether this pattern remains true in the periphery, thereby perpetuating even further the linguistic homogenising impact. This study investigates therefore periphery-based centres of knowledge production and dissemination, to see in what ways they too may be implicit in maintaining the English-language hegemony in IR and thus serving as an impediment to a more symmetrically interdependent vision of disciplinary development.
Linguistic unilateralism
There is little question that English has become the dominant language in the production and dissemination of scholarly knowledge, both in terms of the medium of instruction in higher education,Footnote 12 as well as in the realm of academic publications.Footnote 13 The picture is especially clear in the natural and basic sciences, where it has been calculated that 98 per cent of all scholarly articles are now written in English,Footnote 14 but is largely the case as well in the social sciences and humanities – albeit at slightly lower percentages and varying considerably according to country and discipline.Footnote 15
What does this mean for scholars whose native language is something other than English? An overview of the many works written on the subject in a variety of fields suggests two broad approaches in responding to the question – the first being that it is not a problem, the second being that it is, but that it can be addressed and may in fact be diminishing.
Among the first group are those who argue that English domination in the realm of international publishing is not really a problem because publishing constitutes only a small part of what scholars do.Footnote 16 If we take into consideration the full range of their academic activities we actually see a growth of plurilingualism, e.g. in the case of Catalan, which, one author argues, gained its status as an academic language in the 1970s.Footnote 17 Another ‘positive’ take could be argued from the perspective of the various cognitive studies showing the advantages of multilingualism; for example, those showing that bilinguals outperform monolinguals on a variety of tasks involving executive controlFootnote 18 or that functioning in a second language increases creativity.Footnote 19 One could conceivably take these works as evidence that non-native English-speaking (NNES) scholars are actually in a privileged position compared with the many native English-speaking (NES) scholars who remain monolingual.
The second broad position admits that English dominance has been a reality but highlights various reasons why it need not be considered a key factor impacting knowledge construction. Some emphasise that the real ‘problem’ is not about English but is actually one of academic writing, which, they argue, is a distinct learned skill that presents challenges to everyone. In other words, there is no significant difference if you are learning to write academically in English as a native English speaker or as someone who speaks another language as a native tongue.Footnote 20 Others agree that, yes, there is an imbalance for NNES scholars, but all we need to (can?) do is help them with their English through translating and editing services, combined with promoting greater awareness and self-reflection on the part of native English speakers.Footnote 21 Still others point to the rapidly growing body of work acknowledging the different varieties of Global or World ‘Englishes’.Footnote 22 Noting the clear evidence of the increasing use of these multiple varieties of English, some might posit that any possible ‘problem’ of English dominance that existed in academic scholarship is gradually becoming irrelevant. Each of these arguments is flawed.
Suggesting that the situation is not as bad as you may think because publishing is only a part of what scholars are expected to do ignores the fact that for most scholars around the world, international publishing is not a choice if they hope to find or retain employment. Case studies abound showing rising expectations for scholars to be actively engaged with the international (and therefore English-medium) publishing world. These studies come from all corners, from South AmericaFootnote 23 to EuropeFootnote 24 to Asia,Footnote 25 and extend even to doctoral students needing to publish in English in order to graduate.Footnote 26 The studies often point to state policies encouraging English-language publishingFootnote 27 or to institutional expectations.Footnote 28 Of course, there is more to being a scholar than publishing articles in international journals, but downplaying the particular importance of that aspect of the profession paints an unrealistic portrayal of academic life. As for the possible argument that bi- or multilingualism is actually an advantage, yes, it has been shown to be so for any number of tasks and even for delaying certain degenerative diseases of the brain. However, none of those studies involved tasks testing verbal skills, let alone writing. It would be unreasonable to conflate, for example, creativity in a group design project with the creativity necessary for writing a scholarly article.
Turning to the idea that academic writing in English presents an equal challenge to everyone regardless of their native spoken language, it is perhaps unsurprising that its most prominent supporters,Footnote 29 despite being extremely well-respected linguists, are themselves native English speakers. Beyond the personal accounts that we have probably all heard asserting the falseness of this claim, there are also published rebuttals. Hanauer, Sheridan, and Englander,Footnote 30 for example, report on a survey of 384 Mexican and Taiwanese researchers, comparing their experiences of writing research articles in their native language and in English. The results showed that for these scholars, writing in English was significantly more difficult and anxiety-provoking. If you mistrust such self-reports, as Hyland does, the argument has also been rebuffed linguistically. Flowerdew powerfully outlines how, in terms of acquisition of syntax, phonology, and pragmatics, NNES writers of academic English are disadvantaged vis-à-vis their native English-speaking counterparts.Footnote 31 Finally, even if you discount both personal reports and linguistic arguments, there are emerging works providing evidence of a bias in publishing against NNES writers,Footnote 32 including a randomised control study showing that abstracts written in ‘non-standard’ academic English are more likely to be rejected than abstracts with the exact same scientific content written in ‘standard’ academic English.Footnote 33
The last two approaches seem based on the premise that English dominance in academic life is perhaps a positive factor, either as an evolving lingua franca that might ease worldwide scholarly discussion or, at worst, a reality that cannot be changed, so we should instead focus on raising awareness of and reducing the burdens it poses for NNESs. Much has been said on the lingua franca question, but we must keep in mind that the ‘English as a Lingua Franca’ modelFootnote 34 is primarily based on English as it used between speakers of other languages, in which judgements of what is appropriate or not are based on the idea of mutual intelligibility.
Simple ‘intelligibility’ is far from the accepted standard in written academic discourse and is not likely to become so in the foreseeable future. This point has been made particularly well by Fregonese, who writes that English is not currently functioning as a lingua franca, in which language is streamlined to facilitate communication, but rather is being used in the ‘vernacular sense, where eloquence is a prime determinant of getting people’s ear, and therefore as a source of ‘linguistic disenfranchisement’.Footnote 35 In other words, English is playing a gate-keeping role, where native-language skill in using the language is necessary to be heard. In terms of the proposed responses aimed at making the best of this situation, a number of works have called for more consciousness of the overt and hidden advantages that NES scholars have when publishing.Footnote 36 They have suggested creating more ‘transcultural spaces’Footnote 37 and proposed greater ‘linguistic reflexivity’ to help scholars recognise how ‘socio-linguistic units of categorization unconsciously organize our perception’ and can affect the writing process.Footnote 38 Heightened awareness of the issue is indeed an important start, as long as it does not serve only to soothe the consciences of NESs and prevent further steps from being taken.
As one of the sometimes proposed ‘further steps’, providing support in the form of editing assistance or translation is well intended but also inadequate. The editors of a major IR journal outline the practical complexities they faced even with their decision to simply offer translations of article abstracts. For example, should the same language(s) be used for every article? And, most fundamentally, which one(s) should be used, other ‘hegemonic’ languages for a wider readership or less dominant ones in order to truly shake up the current linguistic power dynamic?Footnote 39 Even when these questions are answered, there remain theoretical and philosophical complications with translation. As many have discussed,Footnote 40 English is far from being a ‘value-neutral code’, and translators may unwittingly be promoting further English dominance by either destroying the epistemological infrastructures of the original works (when translating into English), or leading to English rhetorical patterns being ‘calqued’ upon works in other languages (when translating from English).Footnote 41
If the solution to the current English dominance is reduced to translation, editing, and reflexivity, then at best we can reach what has been described as a state of ‘imposed multilingualism’ on the part of NNESs, in which they keep their languages, but the languages cease to be ‘theirs’ due to regulation and standardisation.Footnote 42 In such a scenario, nothing is achieved with respect to resolving the real crisis caused by English-language dominance, which is the loss of epistemological and methodological diversity and a gradual loss of unique, culture-specific ways of understanding reality.Footnote 43 Seeking epistemological diversity – the ultimate, long-term goal of movements to globalise the IR discipline and others – while at the same time turning a blind eye to increasing linguistic uniformity is counterproductive at best, as it fails to combat the hierarchising and homogenising of disciplinary knowledge. Current pressures in virtually all fields of research to get published in indexed, high impact, largely English-language journals therefore risk the production of increasingly formulaic texts. As a result, scholarship overall will suffer as ‘scholarly diversity becomes an oddity in an expanding sea of “normality”’,Footnote 44 in terms of questions asked, methods used to answer those questions,Footnote 45 and perspectives taken to interpret the answers.
Linguistic dependency in the social sciences
Discussions over Anglo-American dominance in the social sciences, both in general and more specifically in terms of language, are certainly not unique to the IR discipline. In fact, though they may have started later, the issue has been very actively debated in other areas of study, ranging from lawFootnote 46 to applied linguistics.Footnote 47 In the broad field of geography, the ‘Hoffmann moment’ did not arrive until 1998, when Berg and Kearns wrote of how they experienced disciplinary practices that ‘valorise American geographies as “unlimited”, and marginalise other geographies … as “limited”’.Footnote 48 Subsequent works abound, however, from early efforts to propose disruptive solutions to English-language hegemony,Footnote 49 to more recent works showing empirically how knowledge circulation continues to revolve around the anglophone worldFootnote 50 and pointing out the detrimental effects of this ongoing ‘skewed transnationalization’.Footnote 51
Anthropology has seen the emergence of a ‘World Anthropologies’ project aimed at pluralising existing disciplinary views beyond North Atlantic-centric discourse and the hegemony of the English languageFootnote 52 and has its own empirical studies exploring the structural mechanisms behind the imbalance. Mathews, for example, describes how today’s ‘audit culture’, stemming from the corporatisation of academia, interacts with the politics of language usage to disenfranchise anthropologists in non-English-language societies.Footnote 53
The debate over English hegemony in the field of management has been particularly active, with many works outlining a phenomenon not dissimilar from that of ‘dependent development’. Early empirical works provide evidence through both the perspectives of non-native English-speaking academics in peripheral locations who feel obliged to publish in EnglishFootnote 54 and through citation studies of leading Western journals that describe an ‘ideological complex’ that keeps intact a core–periphery division in the knowledge-production process: a neo-colonial intellectual domination by the West that ‘parallels economic domination maintained through control of top of the value chain’.Footnote 55 More recently, Boussebaa and Tienari write of the need to play the ‘game’ of publishing in top, indexed, inevitably English-medium journals or risk becoming irrelevant, and how this leads international management scholars into a trap of becoming ‘agents in the (re)production of core–periphery hierarchies’.Footnote 56
Within the IR discipline there is a well-established body of scholarship discussing Anglo-American or ‘Western’ domination, including calls for decolonising, decentring, or in various ways pluralising the study of world politics,Footnote 57 and the push for change coming in recent years from the ‘Global IR’ initiative.Footnote 58 Indeed, these can be seen as rooted in a long tradition of works more broadly exploring the power dynamics of knowledge production in IR. Such works have successfully revealed its gendered underpinningsFootnote 59 and the harm this has wrought on both the lives and careers of scholarsFootnote 60 and on the discipline itself by limiting and distorting the range of knowledge, ideas, and methods available for understanding international and global politics.Footnote 61 Other works have exposed the complicit relationship between IR knowledge and IR’s racist practices,Footnote 62 and how these have impacted everything from the studyFootnote 63 and teaching of IRFootnote 64 to its very historyFootnote 65 and resulting discourses.Footnote 66
Disappointingly, although these works represent a vitally important effort in not just raising awareness of but also seeking ways to reduce the harms produced by power imbalances in IR knowledge production, they all too often neglect or unreflexively accept the role that language plays in maintaining the imbalance. Loke and Owen, for example, offer a very interesting typology capturing the complexities of the various ways that IR scholars around the world negotiate their knowledge-production practices.Footnote 67 Yet despite the authors’ extreme sensitivity to methodological and epistemological transparency as well as their adherence to a critical agenda, the language issue remains unexamined, aside from an unabashed admission that all but one of the 26 interviews on which the study is based were conducted in English. One might compare this with a similar effort from another discipline, Steyaert and Janssens, which also seeks to categorise the various ways in which global scholars in management studies conduct their academic practices, but in their case, the authors explicitly question the scholars’ use of English and argue that simply accepting English dominance, even if only for pragmatic reasons, prevents more imaginative engagement with possible routes to language multiplicity.Footnote 68
Overall, the number of works focusing specifically on the linguistic aspect of Anglo-American domination of the IR discipline are few, even though the reality of an English hegemony has been clearly acknowledged in an editorial piece.Footnote 69 The rare exceptions include Akyesilmen and Ishmeal, who examine the use of language in the ongoing overemphasis on Anglo-American and Eurocentric concepts in IR,Footnote 70 and Pellerin, who compares French and English conceptualisations of a critical IR term, globalisation, to show how languages – and therefore perspectives – differ and thus demand a rethinking of IR as a universal body of knowledge.Footnote 71 Steffek and Lasshof analyse citation patterns in the leading German-language IR journal, Zeitschrift für Internationale Beziehungen, to show that even when writing in German, the majority of the works cited are in English.Footnote 72
More than four decades after Hoffmann published his concerns about an American hegemony in the IR discipline, and after dozens of works exploring the broader ‘Western’ or ‘core’ hegemony in IR theorising, publishing, and education, the role that language plays in ensuring the continuance of current patterns remains underexamined. The domination of a single language, English, arguably plays a key role in promoting what has been described as the ‘dependent development’ of the IR discipline in the periphery, maintaining the current core hegemony, and standing as an obstacle to true globalisation of IR. To genuinely explore possible ways of introducing greater multilingualism in IR scholarship, it is important to have a fuller picture of the extent and nature of the current practices. This study contributes to that effort by focusing on knowledge production and transmission practices in the linguistic periphery.
Methodology
In order to assess the degree of English dominance in academic publications in the periphery, we first put together a list of IR journals from countries outside the ‘inner circle’ of the English-speaking world. The term describes the first of Kachru’s three concentric circles of World Englishes,Footnote 73 the innermost of which refers to those countries traditionally dominated by the ‘native speaker’ varieties of the language (e.g. the United Kingdom, United States, Canada). The ‘outer’ circle refers generally to former British colonies, where English is used as a second official language (e.g. India, Pakistan, Nigeria), and the ‘expanding’ circle describes all other countries where English may be learned as a foreign language but has no official role in daily life. Although a controversial typology in modern-day sociolinguistics, Kachru’s division is useful for a simple defining of a linguistically based ‘periphery’.Footnote 74 One exception made in this study was to exclude South Africa despite it being considered an ‘outer’ circle country, as English is nonetheless broadly accepted there as the language of science.
To create our database, we first compiled a list of all journals from the 39 ‘outer’ and ‘expanding’ circle countries that were included in the Web of Science ‘International Relations’ category and in the Scopus ‘Political Science and International Relations’ category. Each of the resulting 135 journals was then entered into the database along with information about their language of publication (entirely in English; entirely in one or more other, local languages; or ‘multi-language’, defined here as English plus one or more local languages); country of origin (location of the publishing institution, association, or university); and impact factor. For the last of these we relied on the Scimago Journal Rank (SJR) categorisation of journals (Q1, Q2, Q3, and Q4), as it covers more of the non-core journals we were focusing on than does the Web of Science or Scopus, and because it is a size-independent prestige indicator that measures both the number and relative importance of citations by looking at where they come from. Total numbers of journals according to SJR ratings can be seen in Table 1 below, and a full list of journal distribution by country can be seen in Appendix A.
Table 1. Numbers of journals according to Scimago journal rankings.
We then analysed article-citation records to better understand the nature of the attention that works published in these journals are receiving. This analysis included looking at whether these articles are being cited in leading IR journals, whether there are particular linguistic patterns in the citations process (to what extent do scholars cite intra-linguistically versus inter-linguistically), and also whether there is a relation between the language of publication and the number of citations received.
For these citation analyses, we focused only on articles published in the journals categorised as ‘multi-language’, in order to be able to compare according to article language without having to compare across journals. In total, we collected 2,503 articles from 32Footnote 75 multi-language journals between the years of 2015 and 2019, in order to allow adequate time for articles to have been cited. We excluded special issues during that period, as their focused content might arguably attract irregular attention and skew the data. For each article, we noted both the number of total citations received in Google Scholar as of the last week of May 2021, as well as the languages and publication venues of the articles making those citations.
To calculate the citation averages for each language, we took the Google Scholar citation numbers, and then, for each journal, divided the total sum of citations for articles in a particular language by the total number of articles written in that language. To explore whether these articles were being cited in ‘leading core IR’ journals, we considered the top 10 journals according to the 2020 Scimago rankings,Footnote 76 Political Science and International Relations category. One of these, the Journal of Peace Research, is published by the PRIO Peace Research Institute in Oslo and is therefore itself a ‘non-core’ journal according to our linguistic categorisation. We therefore discounted it from the top 10 list and included instead the 11th-ranked journal on the list.
Finally, to further delve into a possible connection between language used in publications and signs of dependent development, we made an in-depth analysis of the works published in one multi-language journal, Uluslararsı İlişkiler [International Relations], which has since 2004 published in both English and the local language (Turkish). A single journal was examined to provide maximum comparability of the works, and this particular journal was selected as, unlike many of the other multi-language journals, it had a fairly evenly balanced number of works in each language. We conducted a content analysis of the 162 articles (93 in Turkish, 69 in English) published between 2015 and 2022. For this content analysis, we created a three-part ‘framework of dependencies’ – Resource Dependency (whose literature is being cited?); Conceptual Dependency (whose theories are being referenced?); and Problematique Dependency (whose questions/topics are being addressed?) – and explored whether differences in these areas emerged between the articles published in Turkish and in English.
Findings
Language of publication and journal rank
The first step in our analysis was simply to explore whether there was a relationship between the journals’ rankings and their language of publication. Are English-language journals in the linguistic periphery more likely to be ranked higher internationally than their non-English medium counterparts?
Table 2 below shows the breakdown of the journals according to their Scimago ranking and their language(s) of publication. A quick look reveals the fewest number of journals in the highest-ranking category, with only 12 Q1 journals. These journals come from 10 different countries, but Norway and Germany dominate (3 and 2 respectively). Of these highly ranked journals, the most striking characteristic is that they are predominantly English-medium, with 9 of the 12 publishing exclusively in English, and the remaining three being ‘multi-language’, i.e. publishing in English along with other languages.
Table 2. Journal numbers according to ranking (SJR) and language of publication.
In terms of the 29 second-ranked (Q2) journals, the picture becomes a bit more mixed, though still roughly half (15) are English only, and a majority (24) are multi-language. Of the five Q2 journals publishing exclusively in local languages, two are from France and the others are from Chile, Russia, and Spain. In the case of the two French journals, Politix and Revue Française de Science Politique, we have categorised them as ‘local language’ because they do not officially publish in English, but it should be noted that they do actually publish 2–3 articles per year translated into English. In terms of Q3 journals, the English-language dominance continues, with half (18) being English-only, a third (12) being multi-language, and 6 being non-English.
Only with the Q4 and non-ranked journals do we see a somewhat different picture. Just 3 of the 19 Q4 journals and 8 of the 38 non-ranked are in English only, while the majority (12 Q4s and 24 non-ranked) are multi-language, and the remainder (5 and 4 respectively) are in only local language(s).
While these descriptive numbers suggest a relationship between language of publication and ranking, we ran a correlation between the two factors, using this time only the 97 ranked journals and their language categorisation. Because of the lack of normal distribution among the groups, we employed a non-parametric Kruskal-Wallis test, which revealed a significant result (see Table 3). A post hoc Dunn test showed that the English-only journal group was significantly different from the other two journal groups. In other words, there is a relationship between language category and ranking, with English journals being ranked higher than their multi-language and non-English counterparts.
Table 3. Correlation between language of publication and rank.
The non-ranked journals
Although we were not able to include the 38 journals without rankings in the statistical test, we were still interested in learning what we could about why these journals were not included in the impact-factor rankings we looked at, and whether the editors saw any connection between the language of publication and ranking scores. We reached out by email in English to the 34 unranked journals with publicly available contact information and then, because they constituted the largest single-language group among the unranked journals, again in Spanish to the 15 unranked Spanish-medium journals. In total, we received responses from the editors of 17 of these publications.
In terms of their ‘unranked’ status, four journals noted that they were not yet a part of the rankings because they are indexed with the Web of Science Emerging Sources Citation Index, which does not give impact factors. A total of six journals (three of which were also part of the Web of Science Emerging Sources index) also noted that they are indexed with Red Iberoamericana (REDIB), which is a widely used platform for open-access scientific and academic content largely in Spanish, but which is not taken into consideration for the SJR.
Turning to the question of whether language of publication is perceived as having any role in whether a journal is able to be ranked by the major international indexes, we received a number of interesting responses. Only one of these explicitly drew such a direct connection, arguing that it stemmed from the indexes themselves. The editors of this Spanish-language journal wrote to us that they had been rejected twice by Scopus, and that as part of the explanation for these rejections, they were told they should ‘consider publishing more papers in English’. The editors went on to explain to us that this was clearly against the purpose of their journal, as they aim to ‘extend and disseminate the specialist academic international relations literature in the professional Spanish speaking academic setting’ and to ‘enrich the reflection on the discipline in the Spanish speaking academic community’; thus they had made the decision to continue publishing only in Spanish, even if it meant being excluded from various international indexes.
Two other responses suggested a similar feeling about pressure to publish in English coming from the indexes but did so less explicitly. The editors of a Colombian journal wrote to us that they had been trying, unsuccessfully, to be included in international indexes, but that ultimately they had become uncomfortable with what they felt were the ‘criteria to be indexed’, including English-medium production, and had therefore given up their efforts: ‘we have decided to maintain our independence from the international editorial policies in order to maintain our purpose’. Similarly, the editors of an unranked German-language journal also explained that they were ‘unsurprised’ by their journal’s failure to be included in certain international indexes, because ‘German-language journals are barely noticed on the international market’.
Interestingly, nearly half of the responses we received made reference to such a pressure for English stemming not from the indexes, but from the researchers themselves. The editors of a Belgian journal wrote to us that, even though their journal was established as a French-language journal 30 years ago, they have recently decided to accept articles in English and officially become a bilingual journal due entirely to the growing number of English-language submissions they are receiving from scholars. Similarly, the editor of a Spanish-language journal explained: ‘I have pressures from researchers that prefer to publish in English in order to get more recognition’, and the editor of a French journal told us that he receives requests to introduce English content ‘from the community of researchers who wish to publish in higher impact [English language] journals’. The editors of a Russian journal dismissed the idea that the pressure they were experiencing to move towards English medium came from ‘any explicit demands by Scopus’ but that it was rather ‘a natural phenomenon’ based on the desires of the researchers submitting works and ‘seeking wider recognition’ of those works.
Whether it stems from above (the indexes and their explicit or implicit expectations) or below (the scholars submitting articles), journals in the periphery are clearly facing increased pressure to publish in English, if not entirely than at least partially. The rationale behind the push to publish in English seems clearly to be about connecting to a larger audience, with the assumption that this will result in more citations and thus better rankings, both for the journals and for the individual researchers, in the form, for example, of higher h-index scores. In order to explore this connection between English publications and the wider dissemination of ideas, we decided to compare the citation records of works published in English and in other languages.
Citation profile of articles from non-core journals
The next step of our analysis explored the attention – in the form of citations – that articles published in these journals are receiving internationally. This analysis included looking at whether any of the articles were being cited in ‘top’ IR journals in the core, and also comparing whether English-language articles were indeed receiving more attention than their foreign-language counterparts.
Turning first to the highest ranked (Q1) journals, only three are multi-language: Swiss Political Science Review, Colombia Internacional, and the Revista de Ciencia Politica. The first of these officially publishes in four languages (English, German, French, and Italian) but in our 2015–2019 timeframe only had works in English, German, and French, with the vast majority in English. Because of the non-normal distribution of the data, which proved to be true for the majority of the multi-language journals, we were unable to conduct statistical tests. However, as can be seen in Table 4, the median citation scores of the articles in the three languages is dramatically higher for the English language articles (22.16) than either the German (3.71) or French (5.0).
Table 4. Median citation scores of Q1 multi-language journals.
We were able to run Mann Whitney U tests on the citation averages for articles in the two multi-language Q1 South American journals. For the first, Colombia Internacional, although the average citation rates were higher for the English-language articles than for those written in Spanish or Portuguese, no significant difference was found. In the second, Revista de Ciencia Politica, the Mann Whitney test result was significant (P < .05), with the Spanish-language articles receiving on average more citations than the English-language ones.
Turning to the remaining 29 of the ranked (Q2, Q3, Q4) multi-language journal articles, the majority (19) either did not have normal distribution of the numbers of articles or simply had such low numbers of citations that the averages could not be tested statistically for significance. The remaining 10 with sufficient counts and normal distribution all received statistically insignificant results on a Mann Whitney U test. If we look at the medians descriptively, they suggest a higher citation average for English-language articles, but the picture is a bit murky. For the 10 journals with normal distribution but no statistically significant correlation results, the median scores for English articles were nonetheless higher for eight of them, the two exceptions occurring in journals that published articles in German and Portuguese.
For the journals on which no statistical tests could be conducted due to very low numbers of citations or to non-normal distribution, the results were basically split: English articles had higher medians in 9 journals, and other languages had higher medians in 10. It is important to note that 6 of the 10 with higher ‘other’ language medians were Spanish-language journals, and in many of these, the number imbalance was extreme, e.g. 170 Spanish articles and two English-language articles (Politica y Sociedad). Obviously, the extreme imbalance in numbers makes any kind of comparison risky; moreover, it seems important to consider that IR publishing practices in Spanish cannot be compared with those in other ‘outer’ or ‘expanding’ circle languages as the Spanish-language community in IR is particularly well established, a point we will talk about more in the discussion section.
Given that only one of the statistical tests run on articles in 12 journals was statistically significant, it is impossible to draw a clear connection between language of publication and recognition, as measured by citations. However, if we look at the numbers in a purely descriptive manner, including in the citation averages for all of the ranked multi-language journals (regardless of distribution), there is a suggestion of higher medians for English-language articles, particularly if we remove Spanish from consideration.
Core citations of non-core articles and cross-linguistic citation patterns
We also looked one step deeper into the citations of articles in our subset of multi-language journals from 2015–2019, to see first whether they were being cited in ‘top core’ IR journals, and second to get a preliminary picture of whether any patterns emerged in terms of cross-linguistic citations.
Regarding citations in ‘top core’ IR journals, the most striking finding is that of the more than 2,500 articles collected, only 17 (0.68%) received any citations at all from works published in our list of top 10 IR journals from the linguistic ‘core’. Of these, six of the cited articles were from a single journal, the Q1 rated Swiss Political Science Review, with five published in English, and one in German. Overall, among the 17 articles receiving ‘top’ citations, 10 were written in English, 4 in German, 3 in Spanish, and 1 in Italian. What is notable here is not the dominance of English-language articles being more cited, but the overall lack of attention being paid to any works published in these ‘periphery’ journals by writers publishing in the leading disciplinary venues.
In terms of cross-linguistic citation practices overall, our preliminary and limited data suggest that English is the primary language of the articles citing the works in our database, regardless of the articles’ own original languages. In other words, English-language articles are more likely to be cited by other English-language articles, but other language articles are also more likely to be cited in English-language articles. The few exceptions to this are some degree of cross-linguistic citations between Spanish-language articles and between Spanish and Portuguese. This imbalance is possibly due to the simple fact that more works are published in English overall, so the likelihood of more citations being in English is to be expected, and suggests that these practices of intra- and inter-linguistic citations need to be explored in more detail.
Content analysis of Uluslararsı İlişkiler articles (2015–2022)
While the above correlations and editorial comments revealed a clear connection between the dominance of English and recognition in the global discipline, the final stage of our analysis directly addressed the issue of dependent development by looking at the content of published works. For this analysis, 162 articles in one multi-language journal were examined to see whether, according to language of publication, different patterns emerged in whose literature (core/periphery) was cited, whose theories were referenced, and whose topics were addressed.
Resource dependency
To examine whose literature was being cited, we looked at scholarly citations of journal articles, books, and conference papers, and discounted newspapers, websites, or policy reports such as those from the European Union, United Nations, or World Bank. Total numbers of citations were counted and then divided between those referencing articles in English and those in Turkish. An extremely small number of citations of articles in languages other than English or Turkish were not considered for this analysis.
Rather than any major differences between the Turkish and English articles, the most striking finding is the overall common practice of citing predominantly English works (see Table 5). While the Turkish articles cite on average more than twice as many local-language works than do the English articles, the overwhelming majority of references (>80% in Turkish articles, >90% in English articles) are to English-language works. While such a large discrepancy certainly reflects in part the sheer volume of English-language scholarship that exists, it also inevitably illustrates a clear linguistic dependency on English language works as resources.
Table 5. Language of cited articles.
Conceptual dependency
Turning to the question of ‘whose theories’ are being referred to, the citations were also examined specifically for references to works and scholars commonly associated with IR theoretical ideas or perspectives, and these were then divided on a linguistic basis between those rooted in the Anglo-American core (e.g. works of E. H. Carr, Hans Morgenthau, Joseph Nye, Robert Keohane, Alexander Wendt, John Ruggie, Martha Finnemore, Robert Cox, Immanuel Wallerstein, J. Ann Tickner) and those from outside the Anglo-American core (e.g. Michel Foucault, Jacques Derrida, Frantz Fanon, Fernando Cardoso). The 162 articles were then grouped into four categories: (1) those citing core theories/theorists only; (2) those citing both core and non-core theories/theorists; (3) those citing non-core theories/theorists only; and (4) those citing no particular theories/theorists.
Here again, no major differences were found between the works published in Turkish or in English; rather, a common practice of citing primarily Anglo-American core theories and theorists dominates (see Table 6). While it may not be surprising that only three articles in total refer exclusively to non-core theoretical perspectives, it is unexpected that the number of pieces with combined core/non-core perspectives is also very low. Taking these latter two categories of combined and exclusively non-core together, there is little difference between the Turkish and English articles, with a total of just 17.4% of English articles and 11.8% of Turkish articles giving at least some recognition to non-core perspectives. These low numbers reveal the obvious reliance on Anglo-American theories and are a clear sign of theoretical dependency.
Table 6. Theoretical sources of articles.
Problematique dependency
For a third angle on dependency, we looked at the questions or topics being addressed, and, if relevant, the geographical area(s) being discussed. We broke this into two parts: (1) the overall subject of the articles (e.g. foreign policy, international law, security studies); and (2) the country(ies)/region(s), if any, being focused on. For the former, the articles were categorised as a total of 12 broad ‘subjects’. For the latter, we took into consideration any geographical cases being given major consideration (i.e. mentioned in the title or abstract). These countries/regions were then divided again along a linguistic understanding of core and non-core, with the ‘core’ referring to countries/regions of the earlier described English-language ‘inner circle’, and the non-core referring to the ‘outer’ and ‘expanding’ circles (see Table 7). The articles were then grouped into three: (1) solely core-focused (e.g. an article about the future of the United Kingdom); (2) solely non-core-focused (e.g. an article about Saudi Arabian and Iranian competition); and (3) core/non-core combined (e.g. a comparative analysis of the European Union and the African Union on human protection) (see Table 8).
Table 7. Subjects of articles.
Table 8. Country/regional focus in articles.
Any differences that can be noted in terms of the ‘problematique’ of these works again seem to exist not according to the language of publication, but overall in terms of these works emerging primarily from a non-core country and being disseminated in a journal based in the periphery. While the resource and conceptual angles of analysis show distinct signs of core dependency, the questions being asked and the specific problems being discussed lean towards non-core recognition. The single most commonly discussed subject of the works was ‘foreign policy/diplomacy’ (nearly a third of English-language articles, and over 22 per cent of Turkish ones), the vast majority of which were focused on Turkey or other non-core countries/regions.
Perhaps the most interesting result linguistically was about works on IR theory, of which there were 11 written in Turkish, and 1 in English. These numbers were unexpected given recent discussions in the Turkish IR disciplinary communityFootnote 77 suggesting an internal core/periphery divide in which a Western-educated, English-speaking ‘elite’ dominates discussions over theoretical issues, while a locally educated ‘periphery of the periphery’, writing in Turkish, is thought to be ignoring such topics.Footnote 78 The works in Turkish noted here indicate that there is in fact discussion of theoretical issues taking place in Turkish. They also serve as a reminder of the tragedy of the lack of awareness of these works, as shown in the earlier discussion of core citation of non-core articles, as many are raising possibly unique critical perspectivesFootnote 79 that could help pluralise existing disciplinary views.
As a final note on the journal examined here, we learned that recently a decision was made to move from being a ‘multi-language’ journal, to publishing only in English. We reached out by telephone to learn the reasons for this change and received the following insights from someone with knowledge of the decision. His blunt response to why they had made this switch was that ‘frankly, as a magazine, we thought it was a decision we had to make in order to survive’. He then linked survival entirely to citation numbers. He explained that the journal had failed to see a rise in recent years of their citation numbers internationally, a problem the editors attributed to the works being in Turkish: ‘We’ve observed that Turkish academicians who publish internationally do not cite publications in Turkish.’ He spoke of the critical need to secure the journal’s position in particular international indexes and explained that to do so ‘we have to take references from prestigious international journals, unfortunately the citations at the local level do not mean much’.
Our findings here suggest that, unfortunately, while publishing exclusively in English does seem to correlate with higher rankings for periphery journals, overall, citations of works in periphery journals remain very low in core journals, regardless of language, making it unclear whether this journal’s policy decision will have the desired effect.
Discussion: Dependency, linguistic unilateralism, and homogenisation
This study sought to explore the role that the English language plays in IR academic knowledge production and dissemination by looking at the publication practices of journals based in the linguistic periphery (majority non-native English-speaking countries), with the aim of better understanding the possible linguistic barriers they help construct against a true globalising of the IR discipline.
The data from our broader quantitative investigation show that an English hegemony in scholarly publication practices continues outside the ‘inner circle’ of English native-speaking countries. The vast majority of IR journals from the ‘outer’ and ‘expanding’ circle countries choose to publish in English or English plus one or more local languages. Even among the latter ‘multi-language’ journals, again the majority publish most works in English rather than the other language(s). Linguistic unilateralism has clearly expanded into the periphery itself. Moreover, in doing so, it has created its own dependent structures and mechanisms, at the heart of which is the omnipresent desire and need for recognition.
Our communications with journal editors reveal an understanding that English-language journals and articles gain more recognition globally, an assumption that is backed up by our finding that English-medium journals are statistically more likely to be higher ranked internationally than either non-English or multi-language journals. Moreover, our data show that, with the possible exception of articles written in Spanish, articles written in English are more likely to be cited overall – paralleling results from a similar citation study in the natural sciences.Footnote 80 Moreover, the data from our email exchanges exposed that, while there may be some external pressures from indexes to have English as the language of publication, the primary pressure comes from within. Periphery scholars are themselves demanding that local journals allow them to publish in English, as it is seen as the only way to gain broader recognition. These scholars understand that the knowledge-production complex operates in English, so they try to comply with this standard of linguistic assimilation even if it goes against their natural instincts and long-term benefit. The system demands that they ask ‘what do you need me to do, how do you need me to act, and what do you need me to leave behind in order for you to listen to me?’.
The spread of linguistic unilateralism is thus not only a clear sign of dependency but reflects an underlying linguistic racism, ensuring that inclusion in the global discipline is possible only through the dominant language. Some may consider the ‘linguistic racism’ label harsh, but it seems warranted when we consider that on the other side of the picture is an apparent deep-seated inferiority complex leading many hard-working periphery scholars to feel that the only way to succeed is to act, think, and write in the Anglo-American core’s language. The inherent ‘racism’ ensues in the sense that the Anglo-American core sees no anomaly in expecting periphery scholars to be proficient in English if they want their academic quality to be recognised. The practice is exacerbated by a parallel phenomenon within the periphery itself. While the linguistic core dominates and dismisses periphery disciplines at the global level, locally, the English-utilising ‘core of the periphery’ dominates and dismisses the non-English-proficient ‘periphery of the periphery’.
Our data suggest that Spanish presents some degree of exception to these observations. The editors of some Spanish-medium outlets we communicated with expressed less resignation and a greater sense of resistance to the English ‘standard’. Their words reflected a self-confidence necessary to consider rejecting the requirements for fitting into the English-dominant system, a strength that stems perhaps from having viable alternatives, from an extensive Spanish-speaking audience to broadly accepted indexes such as REDIB. Granted, the native Spanish-speaking world is the second-largest language group globally (after Mandarin Chinese), so the Spanish experience may not be fully transferable to academic communities in other smaller language groups. Still, it may provide glimpses into how a multilingual global IR might successfully work, as will be discussed below.
While English articles do generally get cited more than non-English ones, the citation data point to the tragic degree to which the leading voices in IR knowledge production – those publishing in the top IR journals – fail to acknowledge their periphery colleagues overall. With less than 1 per cent of the articles examined being cited in top journals, it is obvious that vast amounts of knowledge carrying different perspectives, insights, and information are being ignored. Let us not forget that the majority of the articles examined were in English, so language accessibility is not an excuse for this oversight. The real underlying problem is one of choices being made – choices by all scholars, from both the linguistic core and periphery, to primarily engage with voices from the linguistic core.
Finally, our in-depth look at article content from one multi-language journal showed further clear evidence of dependency in terms of resources cited and concepts used. Regardless of the language of the article text (English or Turkish), the overwhelming majority of resources cited are in English, and the theories/theorists referenced are almost exclusively from the Anglo-American core IR. Significantly however, this high degree of resource and conceptual dependency is not seen when it comes to the problems and questions being addressed in these works. Indeed, the data showed the problematique diversity that is evident in the periphery and the local scholars’ strong desire to engage in discussions of issues relevant to them.
Herein lies the main problem of the dependent structure and the linguistic unilateralism which serves as the hidden hand keeping that structure intact: it leads to assimilation and ultimately to a homogenisation of the scholarship produced. Research begins with problems and questions. Our findings show that, while periphery scholars are clearly capable of identifying periphery problems and questions of relevance and importance, they remain dependent on core resources and concepts. They are presented with conceptual toolboxes, and encouraged to use them, in a core-friendly language, to construct frameworks of analysis and literature reviews that are familiar to a core audience, in order to more likely be deemed suitable for recognition in the form of publication or citation. The linguistic core, through its dominance of the resources and concepts, blocks a ‘normal’ analytical process which could potentially result in periphery issues being turned into new theories, concepts, and research agendas that could in turn get cited and become future resources. Even if those periphery problems do get studied or shown interest, linguistic racism prevents them from being studied from the perspectives of the periphery itself. Instead, their problems end up, at best, being presented and redefined by the core – essentially, being ‘core-splained’ through the lenses of core resources and concepts.
Policy recommendations for a multilingual IR
In looking at English dominance in academic scholarship, two conflicting ideologies emerge. The first argues that multilingualism is a resource that should be tapped not just to ease the burden for non-English speakers, but for healthier disciplinary development and more fruitful knowledge production. The second promotes a unilinguist approach, arguing that it is more efficient because a common lingua franca makes the sharing of knowledge easier.
Some may reluctantly join the second group for pragmatic reasons. Unilingualism appears to have already won out, and English is the clearly dominant language in knowledge production and dissemination. Perhaps this is why the issue of language does not receive that much attention in disciplinary discussions, and even linguists focused on the field of English for Research PurposesFootnote 81 seem to have given up, opting to promote greater emphasis on ‘text mediators’ (i.e. translators/editors) or advocating that journal editors universally adopt a more ‘open and flexible orientation to English in publishing’ in which NNES writers’ submissions are ‘judged only on content not language’.Footnote 82 These ‘solutions’ may help the individual scholar here and there but do nothing to break the dependency cycle. As has been noted in the natural sciences, the extent to which English dominates ‘assure[s] the dependence of less developed nations on the scientific and technological development of a few states, and largely in a single language’.Footnote 83
We firmly place ourselves therefore in the multilingual camp. A unilingual IR is an oxymoron, and a unilingual global IR is even more of one. Global IR cannot afford to be unilingual, because a common language carries with it a homogenising impact that kills diversity and creativity. Academic writing and publishing play a vital role in constructing, disseminating, and legitimising knowledge. Once we exclude NNES scholars, either through assimilating them fully or setting up impossible barriers, their knowledge is either marginalised or appropriated, while Western knowledge is legitimated and reproduced.
The IR discipline has long suffered from linguistic unilateralism, which continues to serve as a fundamental and under-debated part of the Anglo-American disciplinary hegemony that Hoffmann pointed out decades ago. Linguistic unilateralism directly shapes the nature of research inquiries and therefore the outcomes of those inquiries. It also shapes the nature of the ‘inquirers’ themselves, again with a major homogenising impact. Because of linguistic unilateralism, dependency and a fixation on IR theory importation have taken hold in the periphery and have naturally turned it into a satellite for core IR. Linguistic unilateralism can thus be seen as the mother of dependency and the fuel that keeps the knowledge-industrial complex running.
Multilingualism can be scary. Many will argue that the dissemination of knowledge in multiple different languages will lead to fragmented parochialisms and IR provincialism, and the loss of any sense of a unified production of knowledge. We propose however, that multilingualism, despite its complications, is a necessary stage for a healthy globalisation of knowledge. Moreover, although this study and the recommendations below focus on the level of knowledge dissemination, the urgency behind this call for a multilingual IR is actually based at the level of knowledge production. English will undoubtedly continue as a primary tool for communication needs in academia, but its destructive homogenising impact at the production level must be mediated. To prevent even further homogenisation of knowledge in IR, periphery scholars must be relieved of the pressure to fit into the current dependent structure, and the production of original works from different perspectives and in different languages must be incentivised. While the responsibility for this effort will have to be shared, it must start where the power lies, in the linguistic core.
First, the core must acknowledge and promote the existing alternative linguistic hubs of knowledge production and dissemination. This effort begins with leading organisations such as the International Studies Association (ISA) and the British International Studies Association (BISA) moving towards greater multilingual practices. For their annual conferences, they might allow the submission of proposals in languages other than English and offer translating services for a percentage of the presentations – both to and from English. Major organisations must provide venues for facilitating such multilingual contact if they wish to promote dialogue and help to diminish the linguistic disconnection and resulting parochialism that currently exists.
Core scholars must continue to engage in increased reflexivity, but if their desire for global IR is genuine, they must also assume a more proactive role. One concrete step is to engage in more co-authoring of works with scholars from the linguistic periphery. This recommendation comes with a caveat, however, that in these relationships, the value-added contribution of the periphery scholar must be more than just an exotic name and affiliation, but rather one of offering new concepts, insights, and perspectives. To do this, collaborating periphery scholars should not be praised for just copying core practices from different angles but should be allowed and encouraged to introduce local values and ideas.
Periphery scholars obviously also have a critical role to play, particular those ‘hybrid’ scholars who, because of educational background and linguistic ability, are able to function in both the linguistic core and periphery. These bilingual ‘hybrids’ are the obvious potential ambassadors of a multilingual IR, but they too often succumb to the dependent structure. It is practical and easier for them to cash in on the prestige of a core identity in the local environment, importing ideas and selling them locally or regurgitating them back to the core in familiar ways but with slightly different shades. Ideally, these hybrid scholars should expend at least some of their energy on producing and disseminating knowledge in their local languages, thereby contributing to the accumulation of value in local outlets. With these scholars’ added prestige and contributions, that accumulated knowledge may have a hope of eventually being acknowledged in the core. This appears to be one of the few ways that home-grown ideas, which are essential for global IR, can be both produced and subsequently introduced into the core knowledge base.
To be fair, we recognise the pressures that many of these scholars receive from their home institutions, and the paradox they face that publishing in local languages may not be recognised for promotion. Nevertheless, these hybrid scholars’ position as true agents of global IR is critical. Only they have the linguistic tools to help us through a transition period between today’s pure English dominance to a future of mutually interdependent linguistic hubs in which original concept cultivation is valued and recognised. These linguistic hubs can then export their finest and most relevant contributions to an open access ‘Babelian’ bazaar of ideas – a true global IR.
Since this study has focused on publishing practices, our final recommendation is for the journals. Journals in the linguistic core, as the centres of power in the knowledge marketplace, must take the lead and set new standards of accepted practice. Promising new practices such as including short video-recorded article summaries can be expanded to include optional videos in different languages. This option would allow all scholars a means to share their work’s basic ideas in a language of their choice – possibly a ‘native’ language for NNESs, but also a ‘foreign’ language for multilingual scholars wishing to reach out to broader audiences. Core-based journals should also purposefully select multilingual associate editors whose primary job is to go through non-English periphery journals and seek locally produced works to be brought into the wider discussion. They should put out calls for special issues or joint authored works on those concepts to encourage competitive conceptual analysis. They may also invite meta-analyses of research from different linguistic hubs of IR scholarship in order to highlight ideas and perspectives that might not otherwise get wide exposure.
By cultivating interest in periphery products, the core journals can also help pave the way for the growth and development of periphery journals. As periphery products come to be seen as more valuable, periphery scholars will have greater incentives to produce and publish locally. Freed to work on local ideas and in local languages, these scholars may actually produce better products. To borrow an expression, this rising tide will lift all boats, as better products mean better journals, which can then actually compete with the core journal outlets in a free-access, fair-trade marketplace of ideas. Only then do we have a chance for interdependent knowledge production, and truly global, multilingual IR.
Acknowledgements
The authors would like to thank Bilkent University PhD student Mr Volkan İmamoğlu for his great help in conducting the quantitative analysis for this article.
Appendix A: Journal distribution by country and impact factor (Scimago 2020)
Open access
