The semantics of diversity in higher education: differences between the Global North and Global South

Inspired by neo-institutional theory, we explore whether the semantics of diversity appears to be global and universal through computer-assisted content analysis of 2378 publications. Diversity discourses are dominant, but only in the USA and Canada, UK and Ireland and Europe, not being present in Asia, Africa, the Middle East and Latin America. Diversity is interpreted differently across regions influenced by the local socio-political settings. Academic literature on diversity first appeared in the USA and Canada in the mid-1970s in relation to race and gender. In other English-speaking countries, diversity gained momentum only in the mid-2000s, with inclusion, gender, ethnicity and cultural diversity being the dominant terminologies. Later in that decade, diversity appeared in the academic literature in Europe, often framed as inclusion and gender. We did not find any evidence that the semantics of diversity has become global or universal and, therefore, question the cultural globalisation and the worldwide standardisation of academic knowledge around the valorisation of individual and collective differences.


Introduction
In recent years, diversity-based discourses in students' recruitment and retention in higher education have gained considerable momentum. A wide range of arguments, including economic, social justice, internationalisation and equity, have been used to increase and support student diversity in higher education (Hakkola, 2019), with a noticeable surge in policies and programmes targeting student diversity (Langholz, 2014). However, diversity in itself is a broad and continuously evolving concept, with several meanings and interpretations. Over the years, the understanding of diversity has differed from an emphasis on race/ethnicity and gender to including religion, income, family characteristics, disabilities, chronic health conditions and sexual orientation. This pace of evolution of diversity and its meaning and characteristics that have gained prominence is shaped by the historical and current political conditions and the socio-demographic composition of various countries (Jiménez & Lerch, 2019).
The literature on diversity commonly assumes that the concept has uniformly percolated to higher education systems globally, acquiring similar meanings, definitions and terminologies. In particular, the neo-institutionalists assume that the spread of diversity both inside and outside higher education has been universal and in line with the westernised principles and understandings of equality, despite the apparent differences across various higher education systems (Ramirez et al., 2009). This implies that academics and scholars often attempt to standardise human variation in higher education by relying on classifications that take for granted their application to different contexts, even worldwide. Representative of this approach is the so-called Big 8 model classifying diversity in terms of fixed categories (race, gender, ethnicity/nationality, organisational role /function, age, sexual orientation, mental/physical ability and religion) (Langholz, 2014;Plummer, 2003). Hence, literature dedicated to the study or promotion of diversity and the classifications employed in diversity research also often presuppose universalistic and globalising trends, by relying on a standardised terminology or semantics to approach human diversity.
Studies on the increasing valorisation of diversity have, until now, fallen short in drawing on sources that allow for a worldwide comparison of the academic language of diversity based on regions and socio-political settings. There is a consensus that diversity is a common idea in the Global North, but it is still unclear whether it has become a part of a world culture that has also permeated to higher education in the Global South, as hypothesised by neo-institutionalists (Frank & Meyer, 2020;Schofer et al., 2021). Therefore, it is important to understand the trends in the spread of diversity and evaluate if the spread of the semantics of diversity in higher education has really been uniform across different regions representing the Global North and South as well as the regional differences in aspects of diversity that gain prominence based on political and social developments.
In this article, we examine the evolution and the flow of the semantics of diversity, i.e. the different meanings assigned to the term across regions by focusing on academic publications. Specifically, we aim to ascertain whether the spread of the semantics of diversity has been following a globalising and universalistic trend. A globalising trend would indicate a uniform spread of the semantics of diversity in higher education across different regions representing the Global North and South, whereas a universalistic trend would point towards similar conceptualisation of diversity across regions. This implies that an interest in diversity in higher education would be observed across all countries and regions with similar definitions and focus on various aspects of diversity, for example, gender, race, ethnicity, sexual orientation and mental/physical ability. Anticipating globalising and universalistic trends, we would not expect to observe any major regional differences in terms of the spread of the idea of diversity or its aspects that have gained prominence (Frank & Meyer, 2020). On the other hand, differences in the meaning of diversity across regions would indicate the role that local contexts play in the diffusion and construction of this universal concept. By examining these trends, we endeavour to produce the first study of the worldwide evolution of diversity in higher education. We venture such an analysis with bibliometrics and content analyses of semantic maps built on titles, keywords and abstracts of 2378 scholarly publications from Scopus (2020). Such an analysis based on a current review of the literature on diversity in higher education expands the theoretical debates on the possible convergence of academic topics worldwide.
Our study offers a bird's-eye view of the academic literature on diversity available to scholars using this terminology. It expands upon previous reviews that focused on separate topics, such as sexual orientation (Bradbury-Jones et al., 2019), religious diversity (Small & Bowman, 2012) or specific mental disorders/illnesses (Nuske et al., 2019) as well as those discussing standardising conceptualisations of diversity into different dimensions or characteristics based on a small number of academic publications (Bush et al., 2017;Byrd, 2019;Langholz, 2014). Our aim is to contribute further to the theoretical discussions regarding academic isomorphism in disparate societies and the underlying egalitarian agenda in higher education beyond the West. We start by reviewing the trends expected by a neo-institutional view of diversity. We then present bibliometrics and computer-assisted content analysis, subsequently describing the regional meanings of diversity in the literature. We conclude by highlighting the need for studies on diversity in higher education in other locales to broaden the discussions on academic isomorphism.

Theoretical framework: globalisation and universalisation of diversity?
The valorisation of diversity relies on an idea of progress linked to the ethical principle of equality. Both the valorisations of equality and diversity are not historically common around the globe. It was initially a western pursuit, but since the Second World War, it is said to have become increasingly global (Ramirez et al., 2009). The 'goodness' of diversity relies on western liberal discourse that expects that policies should benefit everyone and not only particular groups (Archer, 2007). This view of the society juxtaposed approaches stressing a shared sense of community instead of protecting the agenda of individuals or groups of individuals, influenced by either the doctrine of Christianity or mono-cultural nationalisms (Ramirez et al., 2009). The creation of a human rights regime after World War II exemplifies initiatives that valorise the individual among human beings (Ramírez et al., 2007). This regime has impacted upon approaches to justice and equity globally, for example, international human rights instruments and social movements as well as education systems (Tsutsui, 2017). More recently, after the fall of the Berlin Wall, the influence of a global culture that carries models of universalisation was further strengthened, generating an even greater tendency to emphasise the importance of diversity issues (Schofer et al., 2021). Jiménez and Lerch (2019) explain the most recent revival of diversity based on the diminishing influence of communist states, which had involved their emphasis on communal values as opposed to individualised rights and their denial of marginalisation (although it may have existed).
In the current scenario, the global expansion of higher education has allegedly provided the push towards the convergence of diversity and its associated terminologies to similar classifications and become increasingly isomorphic. We use the term academic isomorphism (Frank & Meyer, 2020;Schofer et al., 2021) to explore the role of academics (and their publication outputs) in the spread of common terminologies associated with diversity. The global expansion of higher education has created a community of educated professionals, with shared values, beliefs, norms and information in each country. And, thus, it could be assumed that these educated citizens with shared beliefs and opinions may play a role in developing a common understanding and diffusion of diversity discourses globally (Schofer, Ramirez, & Meyer, 2021). Thus, it can be argued that scholars around the world are major actors in translating diversity into a common knowledge that is diffused in academic conferences, classrooms and publications. Influenced by the neo-institutional approach in sociology known as the world society theory, our initial assumption is that the semantics of diversity has global and universal trends.

Globalisation of diversity
Empirical studies have provided evidence on the globalisation of diversity. Diversityrelated offices now operate in 59% of US universities (Kwak et al., 2019) and 20% of those in in Germany (Oertel, 2018). Baltaru (2018) also reveals a general increase in the number of administrative staff responsible for promoting diversity in the UK. Frank and Meyer (2020) also mention the commitment towards diversity and its different dimensions through the statements from universities in the USA, Austria, and South Africa. Jiménez and Lerch (2019) provide empirical evidence that supports our assumption on the globalisation of diversity -occurring in two waves -through their analysis of school textbooks. The first wave appeared in the 1920s and declined in the 1950s, with its renewed valorisation starting in the mid-1970s and having even greater emphasis after the end of the Cold War. These findings contradict the hypothesis of a single trend of globalisation after World War II. Whilst they do not explain the resurgence of interest in the topic, this can probably be attributed to reforms promoted by the counterculture movement of the 1960s. Jiménez and Lerch (2019) also speculate that exclusionary rationalism can challenge this trend, but recognise that they cannot explore this in countries of the Global South, because of the lack of representative textbooks from these regions.

Universalisation of diversity
We also expect a trend towards universalism in the semantics of diversity. This universalism refers to the rationalisation of knowledge in the belief that it is standard and universal, such that it is comprehensible and applicable worldwide. This second expectation is based on the theoretical assumption that universities are increasingly linked by a shared world culture that privileges knowledge that is universal (Frank & Meyer, 2020). The striving for universalism appears in this historical moment in which the interconnection of societies facilitates that universities are pressured by common cultural changes. With globalisation, universities share an increasingly interconnected institutional environment that exerts pressure for the adoption of ideas, structures and practices. Representative of isomorphic trends is the standardisation of diversity through the adoption of the so-called Big 8 (Langholz, 2014;Plummer, 2003) which operationalises eight different dimensions of classification.
It is also possible that the semantics of diversity has globalised, whilst the local context may shape its meaning. Rhetoric and everyday life can differ, and ideas can be used to achieve specific goals or even to prevent change from taking place governed by the socialpolitical climate (Meyer & Scott, 1983), with certain dimensions being prioritised over others. Hakkola (2019), for example, finds that, in practice, the criteria of race and ethnicity prevailed in the admission policies of two universities in the Midwestern USA. In Germany, the nationality dimension is conceptualised in terms of 'migration background', which, according to Neusel (2017), makes foreigners or nationals not born in Germany indistinguishable from German-born first-generation descendants of immigrants in national and higher education statistics. The lack of isomorphism outside Europe and North America would also reflect the intellectual programme of authors promoting the development of social sciences (Collyer, 2016;Connell, 2016). For example, in South Africa, owing to its long history of Apartheid, diversity criteria based on race and class are employed in the admission policies (Erasmus, 2010;Soudien, 2010). It is unclear whether these understandings merge with the broader model of diversity of the American literature.

Method
We downloaded the titles, abstracts and keywords of publications from Scopus (2020), where the word 'diversity' in higher education was explicitly mentioned. Our basic search combined 'higher education' or equivalent terms and the word 'diversity' and similar terms ('gender', 'race', 'ethnicity') in the titles, abstracts or keywords (Table A-2). In addition, we also examined articles with different words describing human variation in their titles (and not in the abstracts), with or without the word 'diversity', to validate the publication trends ( Figures A-5, A-6, A-7, A-8). We first examined the proportion of publications with the semantics of diversity by year and place of affiliation of the authors, as classified by Scopus. We analysed the selected publications through computer-assisted content analysis (see Lee & Esterhuizen, 2000) using VOSviewer. Next, we conducted content analysis of the titles and abstracts, with the aid of semantic maps. Our computer-supported analysis using VOSviewer assisted us with word frequencies and co-occurrence maps, which were further examined through content analysis.
We only included English language texts in our search. This is because all of the articles in Scopus have a title translated into English, even if the article is written in another language. We tried to compensate for a potential bias of publications in English in our bibliometric analysis by examining the proportion of publications on diversity out of the total numbers of publications on higher education in each region ( Figure A-4), as opposed to only considering the total number of publications. Looking at the proportion of publications allowed us to observe the emergence of the weight of the semantics of diversity in relation to broader academic discourses, even if some regions tend to publish less in English. We had a sufficient number of publications from Latin America and Asia focusing on higher education (over four thousand) in journals indexed in Scopus (Figure A-9) that allowed us to examine the proportion of literature dedicated to diversity. To further validate our results, we also triangulated our results with newspapers ( Figure  Two independent researchers manually coded 2378 publications by reviewing the titles and abstracts with reference to diversity in higher education in any form. Inter-rater agreement between the two researchers was high (97.4%). A consensus was then found in cases that had received a different classification. We later divided the articles into eight sociopolitical regions after downloading separate files from Scopus: the USA and Canada, UK and Ireland, Oceania and Europe, together representing the Global North, and Asia, Latin America, the Middle East and Africa, representing the Global South.
We analysed the selected publications through computer-assisted content analysis (see Lee & Esterhuizen, 2000) using VOSviewer. This approach allowed us to examine a much larger number of texts than content analysis and visualising relationships between words through semantic maps (Leydesdorff & Nerghes, 2017). The first unit of our analysis in the semantic maps was the words selected through publications or bibliometrics of texts in Scopus. In the semantic map, each word (equivalent to a node in the network literature) is assigned a weight and a score depending on its frequency and the strength of its relationship with other terms in the literature. The size of the word and its circle in the cluster represent the word's frequency in the titles, abstracts and keywords. The second analytical category was word clusters. These are built by computing horizontal and vertical coordinates for each word to locate them in each cluster of the map depending on their proximity in the sentences of the analysed texts. We then labelled the clusters built in each semantic map corresponding to the uncovered meanings about diversity that we were seeking in our analysis.
Next, we conducted content analysis of the titles and abstracts with the aid of the semantic maps. We followed the steps of a simple content analysis as identified by Flick (2018) that starts by defining units of analysis, in our case words. Then, codes are used to explore relevant passages, in our case titles and abstracts; formulate categories, which are the clusters; and build new categories, i.e. clusters across regions.
Our computer-supported analysis using VOSviewer assisted us with word frequencies and co-occurrence maps, which were further examined through content analysis. We identified the main topics in each cluster, assigned names and classified them according to two different categories: the meaning and context of diversity. We read the abstracts or the articles, identified and quoted key authors from each region and used citations to guide our choice of articles' representability. We also manually colour coded the clusters that VOSviewer had identified, giving same colours to clusters in different regions that reflected the same topics. Through this convention, we could highlight similarities and differences across regions. Regions were the third category we used to interpret how diversity is approached in the literature.
We do recognise the limitation in the selection of articles for our content analysis of publications that included only those that use the term 'diversity', with or without other terms, such as gender, race or ethnicity and not articles that could refer to human variation, but did not use this terminology. This selection is justified because our main focus was on publications that standardise their discussions in terms of diversity as a single model that becomes universal. Thus, we were especially interested in the vocabulary of diversity that subordinates new or previous conceptualisations of human differentiation. Besides, a content analysis of all existing publications without the terminology of diversity was not only beyond the scope of our study, but also, impossible, given the large quantity of unrelated literature that would be listed, if we searched for other keywords, such as 'race' or 'women' without 'diversity' in the search query for titles, abstracts and keywords. The triangulation of our main findings with trends observed in newspapers (Figure A-10) (Nexis Uni, 2021) and international organisations working on diversity (Figures A-11 and A-12) (Yearbook of International Organizations, 2020) also allowed us to verify that our analysis was consistent with information from other sources. As in the articles, our search was made in English even if the original names and titles were in other languages.

Regionalisation-not globalisation-of diversity
In order to ascertain whether the spread of diversity in higher education has followed a globalised trend, we calculated the proportion of articles on diversity in relation to the total number of all published articles on higher education in Scopus for each of the regions in our analysis. Comparing the publication output on diversity from each of the regions showed the extent of engagement with higher education diversity across regions. Contrary to our assumption, the publication numbers show that diversity in higher education is mostly a discourse discussed by the authors in the Global North, especially those from English-speaking countries (Fig. 1). Specifically, the USA and Canada lead in the number and proportion of publications on diversity, calculated based on total output of publications on higher education in each region in 2019 (176, 1.2% of publications), followed by the UK and Ireland (31, 0.60%), Oceania (198, 0.67%) and Europe (164, 0.14%). In Asia (12, 0.06%), Africa (5, 0.14%), Latin America (5, 0.13%) and the Middle East (7, 0.12%), only a small proportion of the total publications focus on diversity in higher education.
The publication trends also reveal that the earliest publications on diversity originated in the 1970s in North America, especially in the USA and much later (early 1990s) in other English-speaking countries. The relationship between these observed trends and the local social and political context is described in detail in the next section. In Asian countries, the publications on diversity in higher education are still minimal, especially in relation to the total number of publication outputs in the field of higher education from these regions. For example, China ranks third next to the USA and the UK in terms of the number of publications in the field of higher education, but with very few publications on diversity in higher education.

Regional Interpretations-not universalisation-of diversity
In order to understand the semantics of diversity and the characteristics that gain prominence in each of the region, we inductively created eight different clusters from the analysis of the literature, representing four different dimensions of diversity, i.e. gender, race/ethnicity, inclusion and cultural diversity ( Table 1). The numbers in each region represent the ordinal number of the cluster within each map, highlighting the size and importance of the cluster, based on the frequencies of words in the analysed texts. For example, in the USA, the teaching cluster is the dominant cluster followed by student, race/ethnicity, medical schools and STEM. After analysing the semantic maps and reading the literature, we further differentiated the clusters based on the meaning of diversity and the (higher education) context where diversity is prominently discussed or promoted. This implies the discussion of diversity with regard to teaching and learning in classrooms or diversity of students or specific disciplines where diversity debates were prominent.
There are four prominent aspects of diversity that emerged in our analysis of the literature: gender (green), race/ethnicity (yellow), cultural diversity (brown) and inclusion (blue). The emphasis on these dimensions varies across regions. Whilst the link between diversity and the gender dimension is clear in all of the regions, others, such as the connection of diversity to inclusion, can only be seen in the UK, Ireland and Europe. Similarly,  Fig. 1 Proportion of articles with 'diversity' and 'higher education' in Scopus by region (1972Scopus by region ( -2019 race and ethnicity are only discussed in the USA, Canada, the UK and Ireland, whilst cultural diversity is only prominent in Oceania. The low number of publications from Latin America, Asia, Africa and the Middle East did not allow for the construction of separate maps for these regions or for forming cohesive conceptualisations for analyses. At the right end of the table, we represent four contexts where the semantics of diversity is applied: teaching (red), student (orange), STEM (grey) and medicine (pink). We observe that the teaching cluster with its focus on the diversity in teaching practices is the only cluster that is common across all regions. In the USA and Canada, diversity is commonly identified in relation to teaching and to the student population. STEM and medical school are also topics related to the context where diversity is applied, but to a lesser degree. In the following section, we explain the regional use of the meaning of diversity and its applications to different contexts.

All regions: gender
The link between diversity and gender can be traced to the early debates on diversity in the mid-1970s and continues to occupy a prominent position in the diversity literature even today, although with different emphasis across regions. Even though the discussions on 'gender' are now, at least in theory, open to other gender identities, the common phrasing still refers to the gender binary, i.e. 'female' and 'male' or 'male-dominated' (e.g. Dennehy & Dasgupta, 2017). Only in the USA and Canada, do 'LGBT' and 'transgender' (e.g. Anderson et al., 2009) appear as recurrent terms, although both stand independently, without further connections to the rest of the semantic network. In this region, gender does not form an independent cluster. Rather, we observe that the discussion on gender is closely linked to the diversity literature in STEM disciplines, thereby highlighting the gender disparity in STEM.
In the UK and Ireland, 'gender', 'woman' or 'man', 'ethnicity' and 'disability' are terms represented near to each other, which means that they are often jointly used (e.g. Grace & Gravestock, 2008). The same occurs in Oceania, where despite a denser semantic map representing interconnected semantics, discussions about gender diversity are distant to those about diversity in terms of 'culture' and 'language' (e.g. Engels-Schwarzpaul, 2015). These terms are used to contrast the various aspects of Western and indigenous cultures. In Europe, diversity in terms of 'gender' and 'woman' is usually connected to the rest of the diversity literature and is discussed in relation to 'policy' and 'culture' (e.g. Mählck & Thaver, 2010). In turn, the term 'disability' is also mentioned and is located near to discussions about gender. Still, the relatively small size of this node does not show a predominant thematisation of diversity in connection with 'disability'.

USA and Canada: race, medical schools and STEM
The semantics of diversity in the USA and Canada dates back to the first discussions relating to racial discrimination in higher education (Fig. 2). The first publications emphasise racial discrimination (Johnson et al., 1975)  Absher, 1997). Race was a part of the early semantics of diversity and together with gender continues to be a major theme in the US diversity literature. 'Affirmative action' (e.g. Turner et al., 2011) and 'critical race theory' or 'CRT' (e.g. Evans-Winters & Hoff, 2011) are dominant themes in the literature focusing on race and diversity. The idea of affirmative action is linked to the policies aimed at increasing the representation of underrepresented/minority groups at educational institutions (e.g. Cohen, 1997). Specific activities in the literature evoked by this topic are those concerning 'selection' and 'admission' (e.g. Squire et al., 2018) or 'admission policy' (e.g. Glasener et al., 2019) in relation to an 'applicant' representing an 'underrepresented minority' group (e.g. Gibbs et al., 2014). This policy also thematises and strongly links to the 'gender' (e.g. Pittman, 2010) and 'race' clusters.
The race cluster also contains authors describing historical events favouring affirmative action policies (e.g. Glasener et al., 2019). These include the 'supreme' court in the 'graz', 'fisher' (in 'texas') (e.g. Hode &Meisenbach, 2017), 'gutter' vs 'bollinger' andthe 'gratz' vs 'bollinger' cases (e.g. Green, 2004), in which the Supreme Court allowed for policies favouring the admission of underrepresented minority groups at the University of Michigan, but established limits. Discussions also relate to 'California', mainly highlighting the diversity initiatives at the University of California (e.g. Koretz et al., 2002;Mohr & Lee, 2000), referred to as the 'immigrant's university' (e.g. Douglass and Thomson (2010) and California State University (e.g. Saetermoe et al., 2017), where the rulings of the Supreme Court on affirmative action are discussed (e.g. Richardson & Lancendorfer, 2004). The cluster also contains publications where the decision by university administrators to frame race in terms of cultural diversity is discussed (e.g. Berrey, 2011). Related to this, the literature also focuses on the 'racial climate' (e.g. Johnson, 2011) and 'microaggressions' in contexts such as those of a 'white institution' (Harwood et al., 2012), highlighting the exclusion and experiences of racial minority groups.
The rest of the clusters in the USA refer to contexts in which diversity is discussed. The 'student' cluster represents the emphasis put, by some authors, upon the 'college student' or the 'undergraduate student', mainly focusing on the composition and characteristics of students in higher education. Emphasis is given to the physical characteristics, prominently 'white', 'black', 'Asian', 'Asian American' or 'European American' (e.g. Halualani, 2010). Diversity in the 'medical school' (e.g. Saha et al., 2008) is prominent together with related areas, such as 'medicine' (e.g. Carr et al., 2015), 'dentistry' (e.g. Freeman et al., 2016), the work of the 'physician' (e.g. Grumbach, 2011) and the 'nursing program' (e.g. Wilson et al., 2006). The focus on medicine and other related fields stems from the need to increase the competencies of medical/health professionals to address the requirements of patients from diverse backgrounds.
In the STEM cluster, there is a clear emphasis on 'woman' in 'science' (e.g. Smith et al., 2015) that represents two large nodes with strong links to 'engineering', 'stem' and 'mathematics' at a 'research university' (e.g. Glass & Minnotte, 2010). Other contexts where gender discussions are visible relate to 'state university' and 'black colleges', 'minority serving institutions' and 'HBCU' (historically black colleges and universities) (e.g. Mader et al., 2016). References are also made to the 'underrepresentation' of women in the 'workforce' and in academic 'career' using the term ('leaky') 'pipeline' (e.g. Pender et al., 2010), a term used to represent underrepresentation of women in doctoral studies (e.g. Holley & Gardner, 2012), dentistry (e.g. Formicola et al., 2009) as well as tenure-track faculty positions or STEM careers (e.g. Pender et al., 2010).

UK and Ireland: inclusion, gender and ethnicity
Publications on diversity in higher education in the UK and Ireland appeared only after 1998, with a steep increase after 2014 (Fig. 3). The 'inclusion cluster' first appeared in the literature only in the second half of the 2000s. The inclusion cluster expands the meaning of diversity to include fair and equal participation of all individuals in higher education, regardless of their backgrounds. In the UK, the abolition of the vocational or polytechnic sector in 1992 galvanised the interest of researchers 1 teaching 2 gender 3 race/ethnicity 4 student 5 inclusion 6 cultural diversity 7 STEM 8 medical school Fig. 3 Semantic map of the UK and Ireland towards higher education access (Kettley, 2007). This agenda gained further interest with the focus of the Higher Education Funding Council of England (HEFCE) and the New Labour Party on widening participation (Archer, 2007;Storan, 2006). The inclusion cluster mainly focuses on eliminating discriminating practices and policies in higher education to enable equal participation. Themes were related to the establishment of procedures and institutional policies as well as practices to promote 'equality' and 'social justice' through inclusive higher education (Archer, 2007). This is also the only region in which 'age' (e.g. Sheard, 2009) is commonly alluded to. Diversity is also understood in connection to 'ethnicity' and 'ethnic diversity', which builds a third separate cluster. In the UK, there is no reference to 'race' and the language referring to human variation adopts less polemic terms, such as ethnic groups, community or population (e.g. Keita & Kittles, 1997). The semantic maps show that the debate about race or ethnicity is not central in other regions, at least not in relation to a general idea of diversity.

Oceania: gender and cultural diversity
In Oceania, publications, focusing on diversity, are recent, having only appeared since 2009. Cultural diversity is often discussed in conjunction with race and ethnicity (Cumber & Braithwaite, 1996). Additionally, as in the USA, the focus of this cluster also seems to be on increasing the cultural competencies of students, who are expected to work with diverse populations after completing their studies, e.g. medicine and nursing students. Cultural diversity forms an independent cluster in 'Oceania', where a notion of diversity associated with the broad idea of 'culture' plays an important but secondary role in the literature on diversity in higher education (Fig. 4). In this region, 'culture' as a main node connects to 'learning' in the teaching cluster (Bowden et al., 2015). Authors usually argue that 'teaching' and 'learning' are influenced by 'international students' (e.g. Moore & Hampton, 2015) as well as by 'teachers' (e.g. Deakins, 2009) or 'educators' (e.g. Weuffen et al., 2017). The role of 'culture' and 'language' is also emphasised in teaching and learning (e.g. Moore & Hampton, 2015), as are ethnic cultures and non-English languages (e.g. Engels-Schwarzpaul, 2015). Discussions on diversity are framed in terms of community and culture, but not as race itself, as in the USA and Canada.

Europe: inclusion and gender
The terminology of diversity started to appear in higher education literature in Europe only towards the end of 2000, where it is still now relatively rare (21 publications each year), if the total number of publications in higher education and Europe's position as the second region is considered (Fig. 5). We speculate that an event that correlates with the surge in the number of publications was the European Commission's emphasis on the creation of knowledge societies and on widening higher education participation to include previously underrepresented groups, especially after the (European debt) economic crisis (European Commission, 2011). This may have also resulted in the increasing number of publications focusing on aspects of diversity of inclusion and integration (e.g. Rossi, 2010), disability (e.g. Moriña et al., 2015), together with inclusion and integration of students (e.g. Danowitz & Tuitt, 2011). This is the main theme that creates a cluster with gender (e.g. Sullivan & Kedrowicz, 2012), both representing major themes discussed in the early years (Fig. 5).

Global South -Asia, Africa, the Middle East and Latin America: diverse meanings of diversity
In Asian countries, publications on diversity in higher education are still minimal in relation to the increasing number of scientific publications on higher education. For example, whilst Chinese academics have taken third place after the USA and the UK and Ireland based on the number of publications in general higher education literature, publications on diversity do not follow this general trend. The first and foremost place of publication is Malaysia (13 total publications), with a focus on terms such as ethnic diversity in universities promoted by 'affirmative action policies' (Yeoh, 2006). China follows with only 10 total publications focusing on, for example, student loans and access (Shen, 2009) or the experiences of Chinese students in the USA (Zhao & Biernat, 2017), a surprising figure considering the explosion of publications on higher education since the 2000s (see Fig. 2). In Africa, a vast majority of articles (42 out of 50) on diversity have been published by scholars from South Africa focusing on issues of access and equity in connection with digital education (Badat, 2005) as well as the lack of demographic representation in higher education, thus emphasising the need to expand higher education access (Winberg, 2006). Additional topics are race and gender (London et al., 2009). These issues reflect the commitment of the government to reforming higher education based on the principles of equity and access against the background of the country's long history of apartheid (Mählck & Thaver, 2010). Similarly, in Latin America, Brazil (13 out of 32) has the highest number of publications on diversity, focusing on, for example, quality evaluation, including aspects of diversity and equity (Bertolin & Leite, 2008) and disability (Guarinello et al., 2009). Lastly, most articles in the Middle East are from Israel (16), where the first articles thematise diversity as multiculturalism, mainly focusing on Israeli Arabs and Jewish students (Al-Haj, 2003;Eilam, 2002

Diversity is not globalising
Against our expectations, we observe that the focus on diversity in higher education is not global. There is a movement in the semantics of diversity across countries, but this is confined to countries in the Global North. The terminology of diversity originated in and spread to other regions from English-speaking North America, where the antidiscrimination agenda connected to the civil rights and feminist movements was strongly associated with the vocabulary of diversity. This later moved to other English-speaking countries in the mid-2000s and, finally, to the higher education literature in Europe at the end of the same decade. This moment of adoption of diversity in higher education corresponds to the 'second wave' of diversity found by Jiménez and Lerch (2019). In line with their findings, we also found that, after the fall of the Berlin Wall, the spread of the semantic of diversity further gained momentum, but this impetus was restricted to Europe and Oceania.
Scholars in Latin America, Africa (with the exception of South Africa), Asia or the Middle East appear to be less engaged in using the semantics of diversity as used by the academics in the Global North. The reasons for the lack of reception of diversity may differ from region to region. Common to countries in these three regions is the limited influence of the feminist agenda, as evidenced by the later and the reduced founding of gender studies in Asia, Latin America and Africa between the 1980s and 1990s (Ramirez & Wotipka, 2008). In Latin America, there has not been a civil rights movement equivalent to that in the USA. This can be put down to the fact that policies about social differences and accessibility are more determined by Spanish-descendent 'criollos', whose upper classes are not interested in discussions about socio-economic background or ethnic groups (Mignolo, 2011).
A similar disconnect to the social movements in the USA or Europe may explain the lack of focus on diversity in Africa, with a notable exception of South Africa. South African universities seem to connect well to the model of diversity, highlighting different human characteristics that should be valorised and included, but the connection to vindicating academic discourses seems to be exceptional for the region. This may be traced back to South Africa's cultural and linguistic linkages to Western countries resulting from its early colonisation by the Dutch and the British. Since diversity discourses are quite prominent in these Western countries, South Africa appears to be an exception even with its long history of institutionalised racial segregation (Kirby and Dempster 2018). In some Asian countries, such as the regional powers in Asia (e.g. China and India) or the Middle East (e.g. Iran), it is possible that the lack of engagement with diversity is related to a lack of the required social legitimacy and even to local regulations that do not promote diversity or even restrict it based on characteristics, such as ethnicity or sexual orientation (Dillon, 2018;Kakoti Borah, 2018;Nourmoahammadi & Seifi, 2020).
Different levels of engagement with the semantics of diversity seem to occur despite common isomorphic trends in making universities more entrepreneurial. Universities in countries with a socio-political atmosphere permeated by communal values (e.g. China) or nationalism (e.g. Russia, Turkey or contemporary India) may be committed to managerial trends in higher education linked to rationales of economic efficiency. However, at the same time, they seem to be decoupled from discussions on diversity that are viewed as representing social progress in other regions. In these places, the increasing explosion of categories to improve teaching or characterise the increasing student population and, hence, faculty may continue without the language of diversity and the ideal quest of reducing marginalisation.

Diversity is not universalising
Second, we did not find a trend towards the universalisation of diversity. The literature is segmented even in regions where the language of diversity is adopted. The only theme that unites scholars from some Western countries is their linkage to the semantics of diversity to promoting gender equality. The discussion of diversity in terms of racial underrepresentation is exclusively an approach in the USA and Canada, but adopted with the considered more acceptable term of ethnicity (or ethnic diversity) in the UK and Ireland. Inclusive education approaches to diversity tend to be a central topic in the UK and Ireland, and Europe, whilst the language of diversity linked to disability seems to occur only in Europe. Lastly, cultural diversity is more emphasised in Oceania. The different regional and disciplinary emphases given to diversity show that literature using its language is not a kind of academic knowledge that is easily transferable or standardisable. Against our expectations, we did not observe discussions about certain underrepresented constituencies formulated under the umbrella term of diversity. The Big 8 model of diversity (Plummer, 2003) is a common form of thinking about the notion in the USA, although these eight labels are not the only way diversity is evoked by academics, who engage in an egalitarian agenda inside universities in other regions. The representation of diversity in academic literature seen in toto is not that of eight big clusters. Our semantic maps illustrate a 'Big 2' composed of gender and race/ethnicity and a 'Small 2' referring to cultural diversity and disability linked to inclusion. Religious diversity, nationality, sexual orientation and age are minor, almost non-existent topics, when compared to all the ideas of diversity that currently circulate.
Other definitions of diversity emerge locally, decoupled from other narratives. For example, Moore and Hampton (2015) show that in Germany, whether a student or academic is considered to increase diversity depends on the country where they or their parents were born, their so-called migration background (Neusel, 2017). The alternative concept of 'educated nationals' (Bildungsinländer) often differentiates people according to where they obtained their undergraduate degrees (Langholz, 2014). Diversity may be related to the low incomes of students in the USA (Radey and Cheatham 2013). On the other hand, academics or practitioners emphasising certain forms of human characteristics are not willing to categorise them within the broader conceptual umbrella of diversity. For instance, feminists in Germany have actively opposed the institutionalisation of diversity within diversity management offices, because they view this institutionalisation with a view that is managerial and focuses on efficiency as a threat to their programme for gender equality (Krell, 2015). Academics around the world who engage in the language of diversity only rarely simultaneously discuss human diversity according to different dimensions, such as the Big 8. There seems to be a lack of academic isomorphism around a common semantic of diversity.

Conclusions
Our study provides an updated review of the literature on diversity in higher education that also has implications for the theoretical discussion about theoretical assumptions about worldwide academic isomorphism. The influence of diversity at the global level is more limited than portrayed in the neo-institutional (Jiménez & Lerch, 2019;Ramirez et al., 2009;Tsutsui, 2017) and diversity literature (Bush et al., 2017;Byrd, 2019;Langholz, 2014). We did not find that the semantics of diversity has become global or universal. Only countries in the Global North adopt the vocabulary of diversity, and there are strong regional differences in its conceptualisation as well as the topics covered. Even in the USA and Canada, diversity experts in fields such as medicine and STEM build their own narratives about what constitutes diversity. We conclude that, whereas growth trends and the content of publications on diversity seem to be more linked to pedagogical models and organisational practices that rationalise teaching, based on the population at universities, they do not seem to support the creation of universalistic academic knowledge. We further reflect that the lack of universalisation and standardisation of diversity as a legitimate topic of research in higher education may be related to a lower global influence of Western liberalism than is often assumed by the diversity experts in the higher education and globalisation literature (Ramirez et al., 2016;Tsutsui, 2017).
The regional and segmented spread of the semantics of diversity in higher education literature suggests that the cultural content currently being disseminated along with the expansion of schooling and higher education is not completely related to a standard knowledge system that globalises and becomes universal. Academic isomorphism may be an observable trend in higher education in other topics, for example, procedures designed in the Global North to rationalise teaching through student evaluation of teaching (Pineda & Seidenschnur, 2021a, 2021bvan Dalen, 2020) or research through bibliometrics (Pineda et al., 2019;Pineda 2015). However, academic knowledge is less universal when it comes to academic knowledge stemming from North American and European valorisation of individual differences.
Increasing rationalisation may couple better with procedures and models evoking efficiency in capitalist economies that are globalised and less to the valorisation of diversity in societies that may place a different value on egalitarian principles. Assumptions about academic isomorphism and general cultural globalisation seem to be, indeed, more prone to critiques from authors claiming to have a better understanding of local trends in the Global South (Collyer, 2016;Connell, 2016;Mignolo, 2011). Universities around the world may increasingly strive for efficiency, but not necessarily engage similarly with diversity debates. New research using novel methods that encourage the viewing of different regions concurrently and interpret results within national and regional socio-political atmospheres might contribute to a wider discussion on the semantics of diversity. Such research endeavour could well challenge the assumption that the Western liberal world view is linked with an increasing globalisation and universalisation of diversity.