Language ideologies and dangerous, celebratory, and hegemonic multilingualism

There is no neutral version of multilingualism. Following Heller (Reference Heller and Heller2007:2), we view language use as the deployment of ‘ideologically-defined resources and practices’, which make power relations central to any speech context. In an organizational setting, then, power is not separate from but rather exhibited through relational processes like communication (Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012:161). Research on multilingualism therefore involves considerations of the symbolic value of language and how it can serve as a mechanism of domination.

In our study of multilingualism at the United Nations, we focus on language ideologies, that is, beliefs about language that are socially shared and well-established, to the extent that they are assumed to be common sense (Woolard Reference Woolard, Schieffelin, Woolard and Kroskrity1998, Reference Woolard2020). There is an established body of research examining language ideologies at the national level, which tends to show that multilingualism is perceived as a problem by and for nation states (e.g. chapters in Blommaert Reference Blommaert1999; Muehlmann & Duchêne Reference Muehlmann, Duchêne and Heller2007; Pujolar Reference Pujolar and Heller2007; Blommaert, Leppänen, & Spotti Reference Blommaert, Leppänen, Spotti, Blommaert, Leppänen, Pahta and Räisänen2012). This is unsurprising, since multilingualism exists in stark contrast to national monoglot ideologies whereby, from a statalist perspective (Pohl Reference Pohl1984), linguistic and territorial borders are presumed to coincide. In such a nationalist model, multilingualism can destabilize linguistic imaginaries in which citizens are assumed to share a common language and identity (Anderson Reference Anderson1991). As such, national language policy and planning ‘has been dominated by… politicostrategies of languages…which entrench the use of a single language in public administration and education’ (Fettes Reference Fettes, Maurais and Morris2004:37). From such a perspective, multilingualism is ‘dangerous’ because it is ‘imagined as being disordered, impure and abnormal’ (Blommaert et al. Reference Blommaert, Leppänen, Spotti, Blommaert, Leppänen, Pahta and Räisänen2012:9). It is contexts where multilingualism is perceived as ‘dangerous’, that is, anathema to the nation state, that can result in policies that disfavour linguistic pluralism (see Johnson Reference Johnson and Tollefson2012), whilst also promoting a hierarchical view of particular language resources and skills—most often those possessed by the elite (Fettes Reference Fettes, Maurais and Morris2004).

Despite the dominance of monoglot ideologies within the nation-state world order, the coherence and value of these have been under scrutiny and disruption due to both external factors (e.g. rising immigration, mobility of workers and the economy, technological change) and unresolved internal dynamics (e.g. indigenous linguistic minorities) (Pujolar Reference Pujolar and Heller2007). Further challenges have arisen with the emergence of international and supranational organizations like the European Union (Van Els Reference Van Els2005; Ammon Reference Ammon2006, Reference Ammon and Spolsky2012; Fidrmuc & Ginsburgh Reference Fidrmuc and Ginsburgh2007; Wright Reference Wright2009; Cogo & Jenkins Reference Cogo and Jenkins2010; Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012; Gazzola & Grin Reference Gazzola2013; Barbier Reference Barbier, Gazzola, Torsten and Wickström2018; Kruse & Ammon Reference Kruse and Kheng2018; Leal Reference Leal2021; Ringe Reference Ringe2022), the Council of Europe (e.g. Sokolovska Reference Sokolovska2016, Reference Sokolovska2017, Reference Sokolovska and Pavlenko2023), the International Committee of the Red Cross (Garrido Reference Garrido2022), and the United Nations (e.g. McEntee-Atalianis & Vessey Reference McEntee-Atalianis and Vessey2020, Reference McEntee-Atalianis and Vessey2024, Reference McEntee-Atalianis, Bianco and Spolsky2025; McEntee-Atalianis Reference McEntee-Atalianis, Gazzola, Grin, Cardinal and Heugh2023, Reference McEntee-Atalianis, Bianco and Spolsky2025; McEntee-Atalianis & Tonkin Reference McEntee-Atalianis and Tonkin2023). These organizations have privileged (restricted) multilingualism as a mechanism for and a symbol of unity, working to achieve aims such as international cooperation, peace and security, and human rights (United Nations 2023). In some cases, supranational organizations have become recognized as alternative and legitimate sources of ideological discourses on multilingualism (Muehlmann & Duchêne Reference Muehlmann, Duchêne and Heller2007:104). Such a championing of multilingualism is arguably ‘celebratory’ in that it orients towards a predominantly and sometimes even exclusively positive version of multilingualism, in principle, if not in practice. Celebratory multilingualism often focuses on the ‘value’ of language as a source of ‘pride’ and/or ‘profit’ (Duchêne & Heller Reference Duchêne and Heller2012). In terms of the former, language can have identity value as a marker of authenticity within wider celebrations of cultural diversity; for the latter, language can be construed as a commodity with instrumental value, exploited for profit in a globalized market context (Duchêne & Heller Reference Duchêne and Heller2012). Whether for ‘pride’ or for ‘profit’, celebratory multilingualism is problematic in many ways, not least because of simplistic and reductive renditions of sociolinguistic realities. These include the assumption, for example, that ‘mere possession of selected languages and mastery of the right accent [serve] as a guarantee of individuals’ success and ultimate satisfaction’ (Martin Rojó Reference Martin Rojó, Tollefson and Pérez-Milans2017:563).

Despite the growing trend to embrace ‘celebratory’ multilingualism in some quarters, several researchers have observed remarkable similarities in the language ideologies that underpin these discourses. In other words, regardless of multilingualism being rejected or embraced, the ideological underpinnings often remain the same (Muehlmann & Duchêne Reference Muehlmann, Duchêne and Heller2007:96). The emergence of international and supranational organizations does not mean that new discourses on multilingualism are necessarily progressive or that inequalities are resolved. In fact, Pujolar (Reference Pujolar and Heller2007:77–78) contends that this has led national elites to develop new ideological frameworks, strategies, and tactics to maintain their legitimacy and dominant positions. Ultimately, this means that the same powerful languages (/varieties) may remain dominant and ‘elite’ (Fettes Reference Fettes, Maurais and Morris2004; Barakos & Selleck Reference Barakos and Selleck2019) in international and supranational regimes. Pujolar (Reference Pujolar and Heller2007:78) explains:

In this sense, we see ‘hegemonic multilingualism’ (Krzyżanowski & Wodak Reference Krzyżanowski, Wodak, Hülmbauer, Vetter and Bühringer2010) at play. This version of multilingualism assumes that only selected languages will (and should) have status in supranational organizations. Krzyżanowski & Wodak (Reference Krzyżanowski, Wodak, Hülmbauer, Vetter and Bühringer2010:117) draw on Gramsci’s (Reference Gramsci1971) notion of hegemony to explain how multilingualism becomes a strategic project, combining domination (power) and intellectual and moral leadership. If multilingualism only accounts for some (and not all) languages, then it contributes to the disenfranchisement of some speakers. Thus, hegemonic multilingualism can arise in cases where a multilingual language regime obtains ‘buy in’ from members (e.g. nation/member states in a supranational organization) even when this may ultimately result in their own disadvantage. One example of such disadvantage emerges in contexts of employment in supranational organizations when this is dependent on multilingual proficiency in elite languages; this ultimately restricts employment access for citizens of countries where elite languages are not national languages (Garrido Reference Garrido2022). As such, hegemonic multilingualism results from ideologies shaping social and political orders.

The perpetual impact of nation states/linguistic politico-strategies (and their concomitant linguistic/monoglot interests) in the international world order is perhaps unsurprising if we consider how nation states responded to the loss of sovereignty and legitimacy in a postnational era. As Pujolar (Reference Pujolar and Heller2007:86) observes, many states shed the trappings of nationalism and turned to invest in international networks and alliances that allowed them to ‘negotiate their interests and strive for a share of power in the global arena’. Through this outsourcing and recalibration, nation state interests were mobilized and masked, laterally reintroduced into supranational organizations via reference to third parties. Notably, these third parties are often founded on linguistic principles. Pujolar (Reference Pujolar and Heller2007) refers to these as ‘language power blocs’: international solidarity alliances founded on common linguistic/cultural agendas between nation states which previously shared constituencies or colonial subjects. Language power blocs are key to the mobilization of ‘new’ discourses about multilingualism within a supranational context, because these blocs promote established dominant languages. One key example here is La Francophonie, which arguably perpetuates France’s political influence over its former colonies, and especially Africa. This is argued to occur via the reframing and repositioning of French not as a national language, but rather as the language of an international alliance, a language used to articulate human rights rhetoric, advocating for peace, democracy, and cultural diversity (Vigouroux Reference Vigouroux2013; Humbert Reference Humbert2023). The role of France in La Francophonie fits Pujolar’s (Reference Pujolar and Heller2007:86) argument that such linguistic power blocs are characteristically led by former imperial states projecting cultural policies in the form of international cooperation over their former colonial subjects. ‘Thus’, he explains, ‘they allow these states to exploit cultural and economic relationships stemming from the imperial period in a way that masks the colonial dimension of these alliances’ (Pujolar Reference Pujolar and Heller2007:86).

Dangerous, celebratory, and hegemonic multilingualism provide valuable ‘multilingual ways of seeing’ (Piller Reference Piller2015) language ideologies in a supranational organization like the United Nations. Adopting a critical stance with regard to the functions and values assigned to multilingualism in a postnational era (Sokolovska Reference Sokolovska and Pavlenko2023), we examine verbalizations, metapragmatic acts, and linguistic practices in United Nations debates on the 1995 multilingualism resolution. As explained in the next section, we focus, in particular, on how members achieve this via their ‘access to the floor’ (Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012) in the context of language policy debates and via their use of multilingualism as positioning devices within these debates.

The study of language ideologies in language policy contexts

In the field of language policy, the importance of language ideologies is well-established. For example, beliefs and values about language constitute one of the three pillars in Spolsky’s (Reference Spolsky2009) model of language policy. Language ideologies are also well-established within the discursive approach to language policy (Barakos & Unger Reference Barakos and Unger2016), in that ideologies underpin discursive constructions that are part of the policy cycle (Sheyholislami & Vessey Reference Sheyholislami, Vessey, Gazzola, Grin, Cardinal and Heugh2024). In their work on the language policy of EU institutions, Krzyżanowski & Wodak (Reference Krzyżanowski, Wodak, Hülmbauer, Vetter and Bühringer2010:118) explain that ‘language ideologies are always produced and reproduced in and by means of various discourses’. However, in practice, research on discourse invariably focuses on what Woolard (Reference Woolard2020:2) calls ‘verbalizations’, that is, explicit metalinguistic discourse. A focus on metalinguistic discourse has become so mainstream that it tends to overshadow other expressions of language ideologies, including mental constructs, embodiments, and linguistic practices (Woolard Reference Woolard2020:2).

Work in critical sociolinguistics has highlighted that language ideologies are ‘inscribed into human bodies and densely connected with affect’ (Spitzmüller, Busch, & Flubacher Reference Spitzmüller, Busch and Flubacher2021:5). These local and individual experiences build into practices and interpersonal alignments, which have a bearing on language policy. As Savski (Reference Savski2025:132) explains, language policy emerges from the codification of associations between particular types of emotional responses and particular forms of language. Thus, the study of social positioning and stance (Jaffe Reference Jaffe2009) can enable us to understand how evaluative practices connect with ideologies in language policy contexts. Positioning and stance emerge via language practices that reveal ‘the micro-processes of symbolic domination’ (Heller 1995/Reference Heller and Wei2007:374). Barrault-Méthy (Reference Barrault-Méthy2012) and Johnson (Reference Johnson and Tollefson2012) have shown the importance of positioning in language policy research, since this can have implications for agency in policy making. Specifically, a speaker’s self-positioning as ‘expert’ alongside their institutionally privileged role as ‘arbiter’ of language policy entails alignments and disalignments with other participants; (dis)alignments can result in the marginalization of some participants, who become ‘mere receivers or implementers of policy’ (Johnson Reference Johnson and Tollefson2012:116) and/or can lead to morphogenesis or morphostasis in language policy planning and/or implementation (McEntee-Atalianis & Hult Reference McEntee-Atalianis and Hult2025). At the same time, positioning depends to some extent on ‘access to the floor’ (Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012), and—drawing on Goffman’s (Reference Goffman1959) dramaturgical metaphor—the public-facing ‘front stage’. In other words, the meaning of positioning depends on the ability of individuals to perform such roles and for their interlocutors to perceive such performance as meaningful.

Here, we focus on front stage performance of language ideological positioning and how this can be examined via a variety of different means (see McEntee-Atalianis & Vessey Reference McEntee-Atalianis and Vessey2025). One example of how this positioning is expressed is via metapragmatic actions, that is, communicative acts about language, such as voting on language policy. Voting is pragmatic in that it completes a social act (i.e. communicating meaning in terms of support/dissent of a motion); however, in a language policy context, voting is metapragmatic in that it is a communicative act about language. In other words, voting means adopting a position on a language issue, creating ‘conditions that make speakers’ use of language possible and effective’ (Caffi Reference Caffi and Brown2006:82).

The act of voting, however, is contextually determined. The concept of ‘deliberative voting’ addresses the nature of voting procedures within a deliberative environment where voting is not an ‘aggregative’ or ‘fixed final collective decision’, but rather a marker of ‘the degree of acceptance of a position within a deliberating group and the reasons the deliberators have for accepting or rejecting the position’ (Serota & O’Doherty Reference Serota and O’Doherty2022:1). Although voting ‘enables us to conceptualize the act of disagreement through an explicit social action’ (Serota & O’Doherty Reference Serota and O’Doherty2022:5), it is ‘a low-information signal’ (Chambers & Warren Reference Chambers and Warren2023:4). It is only by examining the act of voting within its deliberative context that it can be considered ‘a discursive phenomenon’ (Serota & O’Doherty Reference Serota and O’Doherty2022:1). As such, the examination of language policy deliberations (i.e. verbalizations about language) is essential to understanding the meaning of voting in a language policy context.

In turn, in order to examine verbalizations, language practices must be identified as ideological choices. Halliday (Reference Halliday, Fontaine, Bartlett and O’Grady2013:25–26) has explained that the notion of ‘choice’ underpins all human activity, and semiotic (meaning-making) activity involves choice: ‘Whenever something is chosen, something else is not chosen; that is the meaning of choice. What is not chosen, but could have been, is the meaning of that particular instance of choice’. While no language choice is neutral, in a language policy context it is especially pertinent to consider how a verbalization has been encoded in a particular language or variety (Sheyholislami & Vessey Reference Sheyholislami, Vessey, Ajšić and Vessey2025). In this sense, linguistic practices demonstrate another ‘front stage’ performative aspect of language ideologies, practices which underpin even seemingly innocuous constatives in language use, allowing individuals to create and act in a social world (Woolard Reference Woolard, Schieffelin, Woolard and Kroskrity1998; Lønsmann Reference Lønsmann2014). The significance of language choice is especially pertinent in the case of supranational institutions that promote multilingualism. In their research on a European Union institution in Luxembourg, Lovritis, Langinier, & Ehrhart (Reference Lønsmann2024:2) explain that the language choices of national representatives are not ‘personal’ but rather ‘of significant importance’ because they manifest metalinguistic stances, that is, expressions of affective or normative values about language. Thus, linguistic practices (and, more specifically, language choices) express alignment with other speakers of that same language. Collective alignment via shared linguistic practices can be a mechanism used in a language policy context to signal agency and promote geopolitical interests. Applying this to Pujolar’s (Reference Pujolar and Heller2007) notion of ‘language power blocs’, we can see that these blocs may employ a common language as a shared linguistic practice that demonstrates alignment and a desire to mobilize the agendas of respective national members within a supranational context.

Thus, in supranational organizations, there are many mechanisms for participants to ‘perform multilingualism’ (Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012:159), and the three foregrounded in this article are (i) metapragmatic actions (here, assessed via voting records), (ii) verbalizations, and (iii) linguistic practices. In the next two sections, we turn to how these manifest in United Nations language policy and how they can be analyzed in practice.

Context and data

The first United Nations General Assembly resolution (2 (1)) on the use of languages within the organization was established on 1 February 1946. However, much of this policy evolved over the subsequent thirty years (see McEntee-Atalianis Reference McEntee-Atalianis, Jessner and Kramsch2015, Reference McEntee-Atalianis, Gazzola, Grin, Cardinal and Heugh2023, Reference McEntee-Atalianis, Bianco and Spolsky2025; McEntee-Atalianis & Vessey Reference McEntee-Atalianis and Vessey2020, Reference McEntee-Atalianis and Vessey2024). In addition to resolution 2 (1), there existed at the time of the resolution’s formation many provisions in resolutions of UN system bodies relating to issues of multilingualism and/or languages, but these were not recorded in one document; rather, they were dispersed across different agenda items and discussed in different UN organs. Despite the lack of a single official policy, by 1973, the status of the United Nations’ six official languages had been established: Arabic, Chinese, English, French, Russian, and Spanish were all deemed to be ‘equal’ within the organization.

Over the course of the next three decades, however, it became clear that despite their equality in policy, these languages were not all equal in practice. In the mid-1990s, a group of eighteen Member States began to lobby for the inclusion of a separate agenda item on multilingualism in the General Assembly and for a resolution on the topic. A further twenty-nine Member States supported their request and the accompanying draft resolution (A/50/147/Add.1 and A/50/147/Add.2). On 22 September 1995, the General Assembly included multilingualism in its agenda and decided to consider it at regular intervals (see A/50/PV.3 and A/50/250). With a sponsorship of sixty-eight member states, the draft resolution was adopted as resolution 50/11 on 2 November 1995, with 100 member states voting in favour, thirty-five against, and twenty-nine member states abstaining. Since 1995, multilingualism has been regularly included as an agenda item in the General Assembly, with the resolution being debated, redrafted, and adopted more or less biannually, becoming more elaborate over time. The most recent resolution (at the time of writing), adopted in September 2024 (A/RES/78/330), consists of ninety-two paragraphs (following a three-page preamble, setting out the background to the resolution) separated into seven sections with ‘operative paragraphs’ (stating the opinion of the organization and actions to be taken) in relation to multilingualism and the Secretariat, the Department of Global Communications, human resources and staff language training, language services personnel, peace and security, and language rights and development. This contrasts with the first resolution (A/RES/50/11), which consisted of just three pages and (after preambulatory paragraphs) ten short paragraphs of text relating briefly to the language training and skills of the Secretariat and language services (and financial support in relation to the latter), communication requirements for the Secretariat, the provision of multilingual publications and library resources, and language service provision for internal Organizational meetings. Since the first resolution was established, changes to subsequent resolutions have been adopted via consensus, without the need to vote after member state discussions. However, the changes to the resolution have coincided with (and resulted from) the deliberations about multilingualism in the General Assembly becoming more extensive and elaborate over time. These deliberations consist of speeches made by member state representatives (referred to here by reference to the relevant member state), which occur multilingually in the United Nations’ six official languages; they are also available monolingually on the UN website as official translations.

To examine how the United Nations member states have debated multilingualism via metapragmatic actions, verbalizations, and linguistic practices, we focus on a corpus of official records of General Assembly discussions relating to multilingualism between 1995–2022. Agenda items and speeches relevant to the multilingualism resolution were identified in the documents housed in the United Nations Digital Library (UN iLibrary), dating from the original discussion and vote (22 September 1995) until 2022 (the most recent at the time of data collection). While all documents were downloaded, only the content relevant to multilingualism (which was clearly delineated in specific sections) was extracted into separate files, which were then used to compile a corpus. To account for linguistic practices and metapragmatic actions, all files were labelled for the language used by the speaker and their vote on the 1995 resolution. Accounting for these details allowed us to establish the extent to which verbalizations aligned with language choices and voting records. In total, the corpus consists of 57,500 words from fifty-five member states in all six official languages of the United Nations (Arabic, Chinese, English, French, Russian, and Spanish). While all languages are accounted for in the metadata, only English translations of the data were used for compatibility with corpus linguistic programs (on this, see below).

Findings

The 1995 Multilingualism resolution passed by majority vote, with 100 member states voting in favour, thirty-five against, and twenty-nine abstaining. Those voting in favour also tended to dominate the floor discursively, producing over 77% of the deliberative debates about the 1995 resolution and its later iterations in agenda items across the next two decades. The alignment of metapragmatic action (voting in favour) and verbalization (i.e. quantity of speech) affords some insights about agency in language policymaking: language policy decisions may depend (at least in part) on the extent to which speakers discursively dominate the floor (Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012; McEntee-Atalianis & Vessey Reference McEntee-Atalianis and Vessey2024). Comparatively, the thirty-five ‘Against’ voters produced less than a quarter of the amount of data and the twenty-nine ‘Abstainers’ produced less than one tenth of the amount of data as ‘Yes’ voters (see Table 1).

In addition to ‘Yes’ voters dominating the floor, French was the most widely used language in the debates, constituting 36% of all data, followed by English (27.6%) and Spanish (13.5%). Chinese (as noted above) was only used in one single instance by the Singaporean diplomat in 1995, where the representative code-switched between English and Chinese. Likewise, Russian and Arabic constituted very small proportions of the data (3.3% and 6.4% of the total corpus word count, respectively). The remainder of the speeches used code-switching between different languages (as in the Singaporean example). France was the individual member state that dominated the floor in these settings (interventions from the French delegate comprise 14.3% of the entire word count), with other countries contributing only a maximum of 4–5% of the total word count (e.g. the Senegalese, American, Tunisian, and Spanish representations). Of the remaining fifty member states that contributed, most only uttered a few hundred words, with Sudan contributing as few as 110 words (0.2% of the total corpus).

While French and English are the majority languages used, they are also used along metapragmatic lines: French speakers exclusively voted ‘Yes’, whereas 78% of English discourse was produced by delegates abstaining or voting against the resolution. While languages besides French were used by ‘Yes’ voters, French was used in 46.6% of the total ‘Yes’ vote. By comparison, Spanish was the language used in 17.4% of the ‘Yes’ vote data, with English only used for 7.8% of the ‘Yes’ data. In contrast, English was used by all those who voted against the resolution and English was used in 76.7% of the ‘Abstain’ vote data (see Table 2).

Table 2. Languages used according to vote.

With the knowledge that English and French were the primary languages used to advocate for or resist the resolution, we now turn to the verbalizations used to express these positions.

In our examination of verbalizations, we sought to establish alignments and disalignments between the different ideological positionings. To that end, we examined the twenty-five top-ranked statistically significant words (‘keywords’) in the interventions of ‘Yes’ voters (see Appendix A). In turn, we also examined the top twenty-five keywords (and their embeddedness within interventions) used by ‘No’ and ‘Abstain’ voters (see Appendix B).

First, we observe that both camps unsurprisingly employ a vocabulary and register particular to the UN organization. In the ‘Yes’ data, this includes keywords such as initiative and department, and also lexis formally expressing opinion or will, common to the genre of diplomatic debate at the UN. For example, ‘Yes’ voters use keywords like recall (often intertextually) as an argumentation strategy to draw attention to previous actions, agreements, or texts to support claims and persuade the audience. In example (1), the Sudanese delegate uses recall as an intertextual appeal to UN precedent, serving to underscore prior agreements and repeated calls for action via repetitive use of deontic modality, expressing obligation (“should be commensurate”; “should not be subject to”).

As typical of this genre, we also witness keywords like honour and behalf, which demonstrate how speakers strategically position and refer to themselves and their belonging within the context of a debate, that is, that they have the honour of speaking on behalf of a larger group. As evidenced in example (2), representatives of these groups tend to include lengthy predicative statements with detailed and positive evaluations of their respective network. In the case of the Peruvian intervention in example (2), details about the Group of Friends of Spanish include the size of its membership (“over 20 Spanish-speaking countries”), the geopolitical scope of its influence (“on three continents”), and its international standing (“similar groups in Geneva and Vienna”).

Further, the keyword behalf is used by speakers as part of a referential strategy, indexing the adoption of an alliance position that foregrounds international solidarity efforts founded on the basis of linguistic, cultural, or (post)colonial agendas—that is, Pujolar’s (Reference Pujolar and Heller2007) ‘language blocs’. Although all speakers in the United Nations’ plenary invariably speak on behalf of a larger constituency, in the case of ‘Yes’ voters, only two speakers use behalf to highlight their role speaking for their respective member states (the Beninese and Haitian representatives); the rest (N = 20) use behalf to highlight their role speaking for alliances. The significance of these international alliances helps to explain why Spanish is a keyword: Spanish is mentioned more than the other official languages because of the repeated reference to the alliance ‘Friends of Spanish’.

A second theme arising from the keywords, common to the ‘Yes’, ‘No’, and ‘Abstain’ voters, is evaluation. In the case of the ‘Yes’ voters, this involves keywords that convey a positive stance towards multilingualism, including both instrumental and identity-based evaluations. Multilingualism is represented as a tool to be used, enabling “effective participation” (example (3)). At the same time, multilingualism is a “pillar” of the UN (see Appendix A) and an expression of UN values. Speakers adopt economic terms, arguing that languages “enrich” the organization (example (3)), contributing to its “intellectual wealth” (example (4)).

For ‘Yes’ voters, the adverbial keyword always serves to intensify the speech of member state representatives who codeswitch. Of the twenty instances of always by ‘Yes’ voters, eight are produced in mixed language speeches, often with highly favourable self-presentation strategies. For example, both the Tunisian and Moroccan delegates use Arabic and French to positively evaluate their respective countries. The Tunisian diplomat, for example, claims that Tunisia is an “age-old civilization which has always been characterized by moderation” (1995_A_50_PV.34, Tunisia), whereas the Moroccan representative claims Morocco “always played a significant and internationally recognized role as a humanitarian, cultural, civilizational and religious focal point as well as a bridge between Africa, Europe and the Arab world” (2022_A_76_PV.81). Likewise, the Andorran diplomat uses four different languages (French, Spanish, Catalan, and English) along with the keyword always to emphasize their privileged position on multilingualism (see example (5)). The speaker articulates the inherent value of language, which is “the natural bearer of tolerance and mutual respect”. Thus, the delegates who codeswitch advocate for multilingualism through the deployment of linguistic practices (codeswitching), metapragmatic acts (i.e. voting in favour of the Resolution), and through their verbalizations, including their own positive self-presentations.

Turning to the ‘No’ and ‘Abstain’ subcorpora, we also observe keywords pertaining to evaluation, but this evaluation is not positive as in the case of the ‘Yes’ voters’ discursive construction of the multilingualism resolution. For example, ‘No’ and ‘Abstain’ voters use the keyword implication to negatively evaluate the effects of the Resolution, sometimes using adjectival intensifiers such as negative, grave, and far reaching. In example (6), we see the Thai delegate drawing on multiple keywords, alongside intensifiers, in their articulation of “[strong] reservations” about the Resolution, because of its “[grave] implications” for personnel.

The keyword discriminate is even more explicitly evaluative, and the use of this term in context is very fixed, occurring alongside another keyword to form discriminate against. In terms of the target of discrimination, speakers refer to nationals (another keyword) “whose mother tongue is not among the six official languages of the United Nations” (see example (7)). Analysis of the deliberations reveals that the feared discrimination relates to concerns about imposing competence in both English and French as a requirement for applicants to positions within the Secretariat and/or for promotion. Many note that their nations’ languages are already not supported within the organization and therefore insisting that their nationals learn both English and French would be unfair and discriminatory, even colonial. Such a policy would instead favour those individuals for whom English or French is a national language. In example (7), we see the Nepalese representative highlighting the implications of the Resolution on nationals “including those of my own country” and in example (8) we see the Malaysian diplomat positioning their nation amongst those “who are already disadvantaged”. Such positioning does not refer to a formal alliance of member states, but rather an informal assemblage of individual member states who are not otherwise represented within the organization. Identifying this assemblage in the data is possible via an examination of the verbalizations (i.e. explicit concerns about discrimination) and linguistic practices (i.e. the adoption of English).

Notably, ‘No’ and ‘Abstain’ voters do not negatively evaluate multilingualism per se. Rather, speakers express reservations about the implications of (parts of) the Resolution for those who are already at a disadvantage. As explained by the Finnish representative, a vote on the Resolution “must not be construed as being in any way against multilingualism”, since multilingualism is “of great importance to all States Members of the United Nations” (example (9)). Rather than adopting an alliance position and codeswitching (as in the case of some ‘Yes’ voters), in this case the diplomat speaks on behalf of his nation, Finland, and employs the English language to highlight that the country is “constitutionally bilingual” and “values linguistic and cultural diversity”.

The individual positioning adopted by the representative for Finland is an example of a final theme emerging from ‘No’ and ‘Abstain’ voters: keywords denoting individual member state names (e.g. Israel, Finland, Bangladesh). These keywords occur in self-referential excerpts in speeches from respective member states. Speaking with this member state perspective occurs most in the discourse of ‘No’/‘Abstain’ voters, who tend to explicitly verbalize their national status and use the English language as a medium of expression. This is a marked difference from the alliance positions adopted in the ‘Yes’ corpus, highlighted by keywords like honour and behalf, and the predominant use of French, Spanish, and code-switching (see example (2) and Table 2).

Discussion and conclusion

This article started by asking how multilingualism is discursively constructed in UN debates on multilingual language policy, and if verbalizations align with language practices and metapragmatic action (i.e. voting). The findings showed that ‘Yes’ voters tended to adopt a ‘celebratory’ approach to multilingualism, invariably representing its positive instrumental or identity value. Crucially, this version of multilingualism was promoted most by the member state representatives whose own languages were included in the definition. In other words, since the 1995 resolution ultimately served to underscore and promote the organizational status quo (wherein six languages have official status), this version of multilingualism was most strongly supported by the member states whose languages would continue to enjoy status.

The positive representations promoted by ‘Yes’ voters were predominantly articulated through the French language, since French speakers exclusively voted in favour of the 1995 draft resolution. French was used by a range of member state representatives, including France, but also various countries throughout La Francophonie. It is unsurprising that French was the language used most by those promoting the multilingualism resolution; there is a long history of the French language being used to mobilize principles of cultural and linguistic diversity, especially in the face of perceived Anglo-American imperialism and linguistic homogenization (Vigouroux Reference Vigouroux2013; Garrido Reference Garrido2022; Humbert Reference Humbert2023). Nonetheless, another language central to ‘Yes’ voters was Spanish (especially adopted by members of the “Friends of Spanish”), who mobilized a celebratory discourse on multilingualism that included the common language. This also helped to explain why the term Spanish was statistically significant within the ‘Yes’ corpus, since speakers aligned themselves with the bloc by referring to the group by name. Together, these examples indicate the existence of ‘language blocs’ (Pujolar Reference Pujolar and Heller2007), which appear to exert power and influence in supranational contexts such as the United Nations via the use and promotion of a common language.

In turn, the ‘No’ voters tended to adopt a more cautious and nuanced discourse on multilingualism, which was not ‘celebratory’, nor was multilingualism portrayed as ‘dangerous’ (Blommaert et al. Reference Blommaert, Leppänen, Spotti, Blommaert, Leppänen, Pahta and Räisänen2012). Rather, ‘No’ voters tended to highlight the ‘hegemonic’ (Krzyżanowski & Wodak Reference Krzyżanowski, Wodak, Hülmbauer, Vetter and Bühringer2010) nature of organizational multilingualism that, through the Resolution’s strengthening of the status quo, serves to further marginalize and disadvantage those whose languages were not already included in the policy. In this rendition, it is not multilingualism that is dangerous, but rather the hegemony of the language blocs that is highlighted as detrimental to some member states. Their imposition of ‘elite plurilingualism’ (Fettes Reference Fettes, Maurais and Morris2004) continues to hold sway through the ‘buy in’ of members. Here, too, verbalizations tended to align with metapragmatic actions, in that ‘No’ voters’ discursive constructions of multilingualism were coherently cautious and nuanced; also, speakers tended to position themselves as representatives of individual member states (rather than as representatives of blocs).

In the interventions from ‘No’ and ‘Abstain’ voters, English was by far the most commonly used language, suggesting that it was a language adopted by those whose other (national) languages were not recognized by the organizational language policy. In other words, although there was no reference to an English-language bloc (e.g. akin to the labels “Friends of Spanish” or “French speaking group”), it is evident that English nonetheless served as a unifying way of positioning speakers. Perhaps inadvertently, English still becomes recentred within multilingualism debates as it is the main language available to a considerable number of member states whose other languages are not included in the organizational model due to its global status. Although the dominant role of English is generally presupposed in international organizations (e.g. Wodak et al. Reference Wodak, Krzyżanowski and Forchtner2012; Leal Reference Leal2021; Garrido Reference Garrido2022; Ringe Reference Ringe2022; McEntee-Atalianis Reference McEntee-Atalianis, Gazzola, Grin, Cardinal and Heugh2023, Reference McEntee-Atalianis, Bianco and Spolsky2025), the ideological function and consequence of the role of English is not always accounted for within the constraints of the language policy model of a supranational organization like the United Nations, wherein member state representatives must choose from a limited range of different languages. The use of English is not neutral, nor is the use of any other language. However, the case of English is particular because of the multilingualism policies that are in place to prevent the dominance of English and yet result in English being used more by a wider range of member states due to the constraints of the organizational language policy model. Thus, the discourse analysis shows that there are no ‘Friends of English’, nor is there an ‘English-speaking group’, and yet these alliances are borne out through language practices that align through metapragmatic actions and these, in turn, have a bearing on language policymaking at the supranational and organizational level.

In this sense, then, we see that discursive constructions indeed align with language practices and metapragmatic actions. We also observe evidence of hegemonic multilingualism with the organization, resulting from the repackaging of old monoglot interests in supranational contexts. At the heart of this context is contention over what multilingualism is as opposed to what it should be. Multilingualism is a contested and unsettled term, disputed by those whose languages are already protected in the system and those who denote the term more broadly (e.g. to include considerations of the need for other languages to be part of the UNs work more broadly and for equity/inclusion, etc.). In this sense, there is apparent tension between the ‘roots’ and ‘routes’ of multilingualism (Garrido Reference Garrido2022), with the former pertaining to the organizational multilingualism, steeped in the post-WWII era, and the latter pertaining to the connections made between multilingualism and the international community. Thus, the imaginaries of multilingualism in the data emerge from ‘the flow of language ideologies from nation-state institutions to transnational ones’ (Muehlmann & Duchêne Reference Muehlmann, Duchêne and Heller2007:108).

Our research has shown that rather than focusing exclusively on verbalizations (often referred to simply as ‘discourse’) (e.g. Krzyżanowski & Wodak Reference Krzyżanowski, Wodak, Hülmbauer, Vetter and Bühringer2010), the examination of verbalizations alongside language practices and metapragmatic actions can provide a more complete and nuanced understanding of the nature of language policymaking in supranational organizations (Sheyholislami & Vessey Reference Sheyholislami, Vessey, Gazzola, Grin, Cardinal and Heugh2024). Although this study focussed exclusively on ‘front stage’ (as opposed to ‘back stage’, Goffman Reference Goffman1959) performances of language policy deliberations in this context, we have found that corpus linguistic tools offer valuable new dimensions to enhance our understanding of language policymaking, allowing us to account for a tripartite model encompassing language practices and metapragmatic actions alongside verbalizations. Our results suggest that organizational efforts to stem the dominance of English may inadvertently lead to its growth when policies create forced choices within a hegemonic version of multilingualism. If the role of English is rising even within the context of multilingual policies of supranational organizations, it remains to be seen how multilingualism will fare into the future. Supranational organizations like the United Nations are facing mounting pressures and decreasing support from membership within an increasingly nationalistic world order (Debre & Dijkstra Reference Debre and Dijkstra2023). The reemergence of nationalism, coupled with the astronomical rise of technology, raises questions about the future of language policy for international organizations, especially in terms of ‘how to find disappearing multilingualism’ (Kelly-Holmes Reference Kelly-Holmes2024:8).