KINYARWANDA AND KIRUNDI: ON COLONIAL DIVISIONS, DISCOURSES OF NATIONAL BELONGING, AND LANGUAGE BOUNDARIES

The development of the Bantu languages Kinyarwanda and Kirundi is entangled within the colonial histories of Rwanda and Burundi, first under German and then Belgian rule. From the turn of the twentieth century on, missionaries compiled grammars and dictionaries of the two mutually intelligible languages, contributing to the development and instrumentalisation of two prestigious varieties out of a larger dialect continuum. In this contribution, I trace the missionary and colonial activities of corpus planning and textualisation and summarise how Kinyarwanda and Kirundi turned into official languages with distinct linguistic boundaries. The central research question is how speakers of Kinyarwanda and Kirundi thereafter came to be identified as “Rwandans” or as “Burundians,” with each language indexing a specific national categorisation. Tentatively, I contrast these developments with contemporary fluid practices in multilingual neighbourhoods.


Two Bantu Languages, Two Nations: On Similarities and Differences
Overviews of Bantu language subgroups, as part of the major phylum of Niger-Congo languages, vary in their assumptions of the exact number of languages, mostly ranging between 440 and 660 (for an overview of different counting methods, see Nurse and Philippson 2003). The difficult task of counting and differentiating closely related languages and their dialects becomes evident in the continued attempts to classify specific groups and dialect continua in the Bantu area (e.g., compare Guthrie 1967-1971with Maho 2009). In the two neighbouring Central African countries Rwanda and Burundi, two strikingly similar languages serve as official languages and encompass altogether more than 20 million speakers; Kinyarwanda and Kirundi. The group of Bantu languages that includes Kinyarwanda and Kirundi was first labelled as subgroup "D" according to Guthrie; then as "West Highland" by Hinnebusch et al. (1981) and Nurse and Muzale (1999). It was then regrouped by Meeussen to group "J" and (re-)organised by Maho (2009: 58) in the most recent classification as "JD/Ruanda-Rundi Group." Kinyarwanda and Kirundi are commonly treated as two distinct languages, with their own dialects (some of which, yet not all, are listed by Simons and Fennig 2018). Generally, both belong to the JD subgroup of Bantu languages (see Maho 2009: 58); Kinyarwanda is classified as JD61 with the isocode [kin] and Kirundi as JD62 with the isocode [run]. The genetically closest language varieties are listed as Fuliiro (JD63), Vira (JD631), Subi (JD64), Hangaza (JD65), Ha (JD66) and Vinza (JD67). Interestingly, Kimenyi's (1978) assessment and Harjula's (2004) work on Giha (or [Ki]Ha) suggest that Kinyarwanda, Kirundi and Giha are remarkably closely related and mutually intelligible (see also Botne 1990). However, as no dialectological or dialectometrical studies exist and no further analysis has been carried out it is difficult to precisely define how close or distant the different varieties actually are from each other. According to speakers, there exists a strong similarity and mutual intelligibility between these three languages, however this does not extend to most of the remainder of the JD60 subgroup. 1 Apart from striking lexical and grammatical resemblances, the standardised versions of Kinyarwanda and Kirundi also reveal salient linguistic differences; for instance, in tense and aspect formatives 1 The neighboring varieties Kinyabwisha (DR Congo) and Rufumbira (Uganda) can be considered varieties of Kinyarwanda (even though speakers treat them as distinct languages, too), while Kinyamulenge (DR Congo) reveals a divergent lexicon and morphological patterns that are deviant from Kinyarwanda and Kirundi. All three varieties are not listed by Maho (2009). For an overview, see Nassenstein (2016Nassenstein ( , 2018. However, they are listed in the Glottolog  (see Botne 1990 on the future tense), conjoint-disjoint marking (see Ngoboka andZeller 2016, Nshemezimana andBostoen 2016) and, of course, in the rich divergent lexicons that range from everyday core lexemes to more specialised vocabulary. While both languages were already in use in the two precolonial kingdoms, the borders of the Rwandan and Burundian monarchies did not concur with today's national borders of the two postcolonial nation states: Migration, waves of expansion and reconquering of occupied space by local authorities had kept the boundaries variable and in a constant state of flux (see also Vansina 2004 for the history of the Rwandan Nyiginya Kingdom and its shifting boundaries). The turn from precolonial monarchies to one colonial protectorate under German rule from 1894 and Belgian rule from 1916 had a major impact, not only on the peoples' autonomy and sociopolitical situation, but also on the language ecology of Ruanda-Urundi (i.e., the colonial German Residenturen Ruanda and Urundi, which were separate residences within the protectorate German East Africa).
In this article, I aim to sketch the development of today's languages Kinyarwanda and Kirundi by critically assessing the colonial practices of "designing" languages out of a linguistic continuum. This process was a part of language/corpus planning and included linguistic and orthographic interventions by missionaries, colonial agents and Rwandans/Burundians working under colonial rule. I pursue the question of how speakers of these languages were grouped along ethno-national lines of separation (inhabitants of Rwanda vs. Urundi, or today: Rwandan citizens vs. Burundian citizens) and how they associated with either Standard Kinyarwanda, or Standard Kirundi as national languages. I then contrast this with my contemporary analysis of migrant speakers' agency (the result of large migrant waves since the 1960s) and their messy linguistic practices in neighbouring Uganda; where it seems that speakers employ fluid and creative strategies beyond the dialectal lines of national belonging. My contribution is far from exhaustive, but rather a first approach to the colonial processes of the artefactualisation and reification of languages. 2 2 I am very grateful to all interlocutors both in Rwanda and Burundi (where I conducted first interviews on language, nationality, borders and ethnopolitical categorisation between 2012 and 2016) and all interview partners in Kampala (whose identity has been anonymised). Moreover, I warmly thank the issue editors for their efforts and patience, and the three anonymous reviewers for their ideas, corrections and

The Fixation of Colonial Languages: Indexing Nation(-hood) and Identity
Rwanda and Burundi have sometimes been described as "false twins" (Chrétien 2003: 34), as two political systems that both co-existed with neighbouring borders in precolonial times, which were then lumped together to one colonial territory Ruanda-Urundi from 1894 on, before being finally reversed into two political mirror images in the postcolony (followed by comparable outbreaks of violence in more recent times). Both countries are characterised by similar cultural practices and traditional monarchic systems; with, prior to the European arrival, one king, the mwami, and a strictly organised and militarised royal court. Both countries are furthermore connected by the mutual intelligibility of their national languages, Kinyarwanda and Kirundi. In Rwanda, Kinyarwanda is an official language beneath English, French and, more recently, also Kiswahili. In Burundi, French, English and Kirundi serve as the official languages of the country since 2014.
Historically, both languages emerged from the pairing of an historical continuum bound to the precolonial kingdoms and colonial interventions, i.e., processes of language planning, textualisation and standardisation. These interventions were common endeavours throughout large parts of colonial Africa, yet the apparent resemblance between these two prescribed languages (as dealt with in early grammars) was evident to missionaries and colonial agents alike. Missionaries first began to compile grammars and dictionaries focused on the varieties that were already used at the royal courts of the two realms -the prestigious varieties spoken around the capitals Nyanza (Rwanda) and Kitangoulé (Burundi; see Chrétien 1968: 50). The specific choice to compile first grammar sketches and dictionaries of these two widespread varieties, and thus textualise them, occurred both for pragmatic reasons (due to their wider distribution) and as a result of ideological factors, as the policy of "indirect rule" led to both the clerical and colonial authorities being built on the preexisting models. The work on specific varieties of Kinyarwanda and Kirundi suggestions. It has to be noted that this first (preliminary) overview paper does not claim to offer ultimate results but intends to give a first introduction to the topic, anticipating more research output in the future. All remaining shortcomings are thus my own responsibility. and their implementation in the christianisation process of the colony was therefore closely tied to the colonial oeuvre itself. Moreover, missionaries were among the first Europeans to settle in today's Rwanda and Burundi, a late movement at a stage when Tanganyika, the Congo, Buganda, etc. had already witnessed the arrival of Europeans.
While the German Count von Götzen, the first European to invade today's Rwanda, 3 only proceeded into the interior of the realm in 1894, missionary groups had settled down in the region and were sending reports to Europe since the 1870s. On the Protestants' side this was the London Missionary Society, while on the Catholic side the Missionnaires d'Afrique (White Fathers/Les Pères Blancs) were active (see below). Among the first European explorers to set foot in Urundi was John Richard Burton in 1858 (Chrétien 2003: 29) when arriving with Speke at the eastern shores of Lake Tanganyika and moving northwards up to today's Uvira. The more detailed descriptions of the kingdom of Burundi, including notes on the monarchy and the king's much-feared warriors, actually stem from Burton's accounts (see Chrétien 1968: 50).
Several scholars have already critically looked at the practices of textualisation, orthography-making and artefactualisation of African languages, for instance Blommaert (2008) in his account of how languages were reified and developed into distinctive entities; Irvine (2008) in her seminal paper with regard to colonial ideologies; Fabian (1986) and Mortamet and Amourette (2015) in terms of the colonial making of Kiswahili; Ferrari (2012) for colonial archives of Lubumbashi Swahili in DR Congo; Pugach (2012) as a colonial overview of German East Africa; and Storch (2018) with a broader overview of the inherently colonial discipline of African Linguistics. In my own studies, I have approached cross-border variations in the tripoint of Rwanda, DR Congo and Uganda from the theoretical angle of "border thinking" (cf. Mignolo 2012). The discussions cover the similarities and differences between closely-related varieties (Kinyarwanda, Kinyabwisha, Rufumbira) and how they are perceived as meaningful by speakers on the basis of a "colonial difference" but concurrently filled with new meaning through postcolonial agency and speakers' own interpretations of geographical and epistemological borders. Also, the treatment of language, discussed critically as a "decolonial option" and speakers' "delinking" processes, can be related to the speakers' affiliation with either side of geolinguistic lines of belonging. Agyekum (2018), in a recent paper, discusses the notion of "linguistic imperialism" in regard to acts of documentation and preservation of African languages, and also addresses language shifts and the forceful prescription of one dominant variety. Abdelhay, Makoni and Makoni (2018), in their recent article, "When God is a Linguist: Missionary Orthographies as a Site of Social Differentiation and the Technology of Location," address the orthographic conventions that were often arbitrarily applied to African languages by outsiders and self-proclaimed experts. The colonial endeavours to either lump similar languages together or draw a boundary between them, through orthographic conventions or deliberate language planning, are described by Gilmour (2006) for a similar case to that of Kinyarwanda/ Kirundi, namely, dealing with isiZulu and isiXhosa in South Africa.
In comparison to other languages in the colonial system, such as Lingala which was planned and underwent processes of grammatical adaptation by Scheutist missionaries like de Boeck, the languages in Rwanda and Burundi were not fully designed and "constructed" by missionary and colonial agents. Before the arrival of the Europeans, Kinyarwanda and Kirundi were already employed in both kingdoms -Ikinyanduga 4 in southern Rwanda and Ikiruundi in central Burundi -yet with a lot less linguistic unity in the two kingdoms than in (post)colonial times. The missionary and colonial interventions, 5 therefore, rather focused on lexicon, resulting in status planning initiatives and contributing to the 4 In the present work, for reasons of clarity and in order to avoid confusion, I use language names with an augment i-in order to denote dialects of the two languages such as Ikigoyi or Ikiruundi, while I use the more westernised labels Kinyarwanda and Kirundi for the standard varieties and national languages. Colonial toponyms are usually also given alongside their contemporary equivalents for better orientation, often separated by a slash, e.g. Kisenyi/Gisenyi and Usumbura/ Bujumbura. The colonial conglomerate of Ruanda-Urundi is maintained in its original form, in all other cases Rwanda and Burundi are used to denote the two contemporary countries. 5 It has to be mentioned that missionary and colonial initiatives did not always correspond. During the time of the German protectorate, missionaries were already engaged in writing grammars of both languages and therefore in the process of artefactualising and textualising the two languages, while German officials focused compilation of dictionaries, favouring a specific dialect over others. The argumentation was often based on discourses of the "right" vs. "wrong" language and also on nationalism, i.e., what it meant to be a "Rwandan" or a "Burundian," and what implications that had on one's language use. This was also pushed by local intellectuals who worked in favour of the colonial powers. 6 The most salient and visible adaptations were a part of the primarily orthographic alignments of textualisation processes (turning languages into written artefacts).
During the colonial period, the close relation between and striking similarity of the two standardised languages was already subject to numerous discussions by linguists. In a paper, Barakana (1952) summarised the controversial idea of a potential "(re-)unification" of the two languages in the journal Civilisations. He stated that the missionary work on Kinyarwanda and Kirundi had increased the differences between both languages: [Seen from a linguistic perspective, the arrival of the Whites, particularly the missionaries, has even accentuated the gap that exists between the two languages by the fact that numerous technical terms had to be fixed and invented, referring either to religion, other branches of education and modern life, and that this work has been accomplished partly without prior understanding between the two countries] (Barakana 1952: 73; my translation) The controversial discourse on whether to demarcate the two ausbau languages, textualised by missionaries as distinct languages, or whether to "unite" them, as suggested in Barakana's paper, was a recurrent political topic towards the end of the colonial period.
rather on the promotion of Kiswahili. It was only under Belgian colonial rule that these activities from church and colonial state were more aligned. 6 Apart from these intellectuals, one should not ignore the agency and interests of a diverse array of actors, by far not limited to European colonial agents and missionaries. I am grateful to one anonymous reviewer for this important comment.
Moreover, the colonial policy of bundling the higher education institutions together in only one of the two countries (e.g. plans to establish a university in Usumbura/Bujumbura) was viewed with much suspicion by Rwandan and Burundian elites and criticized along with further standardisation tendencies that were feared to actually bring the two languages closer together (rather than make them more different). The Burundian mwami, among others, expressed his concern in regard to the standardisation of the two vernaculars in a public speech, partly cited by Weinstein (1974).
According to Henry Morton Stanley's early map of the Great Lakes region, the boundaries between the two realms were clearly defined along the Kanyaru river when the Europeans arrived in the area (Chrétien 1968: 52). Also, an early German "Völkerkarte" [ethnic map] by Karl Weule (drawn before 1916 and taken from the German colonial atlas) clearly locates the Warundi in the south of the German colony of Ruanda-Urundi and the Wanjaruanda (according to German orthography) in the northern regions. 7 This shows that encompassing the internal distinctions and colonial classifications of the inhabitants of the colony, Rwandans and Burundians were perceived as different "peoples," or, based on European nationalist thought in the age of imperialism, as sorts of "nations" (see Blommaert and Verschueren 1992).
On the contrary, the well-known yet less demarcated linguistic borders of both languages were strengthened through missionary interventions. Local dialects spoken in border areas were increasingly neglected (such as Ikinyabweeru in northern Burundi, which, according to speakers, is very close to Ikinyanduga) as the standard varieties were used in churches throughout the colony and taught by missionaries at the periphery. In this context, the realm's former dominant variety competed with the local dialects (such as Ikibo around Usumbura/Bujumbura and Ikigoyi around Kisenyi/Gisenyi).
Altogether, it can be said that the preference for a specific dialect of Kinyarwanda and Kirundi, respectively, turned into a colonial sign of nationality and nationhood. Speakers whose languages were similar or equivalent to Kinyarwanda, were categorised as "Rwandans," while speakers whose language corresponded with standardised Kirundi, were hitherto understood to be "Burundians." These indices of nationality did not correspond with the linguistic diversity found in precolonial Rwanda and Burundi, as pointed out by Vansina (2004) and visible in different dialectal maps (see Maps 2-3). Especially along the periphery of the precolonial kingdoms, the realisations of Kinyarwanda and Kirundi corresponded much more with the local dialects than with the standardised national language proclaimed by missionaries and others.
The fixation of these two languages, as will be sketched out in the following sections, primarily led to the development of "imagined communities" in Anderson's (2006Anderson's ( [1983) terms. He states that in many of the nations and nation-states that were created "only a tiny fraction of the population 'uses' the national language in conversation" (p. 46), with a particular reference to the African continent. He labels the frequent incoherence between print-languages, national consciousness and nation-states in these contexts as "discontinuity-inconnectedness" (ibid.). Blommaert and Verschueren (1992: 359) also speak of "nations" or "peoples" as "natural discontinuities," as "folk perceptions of which conceptualizes them in much the same way as species in the animal kingdom." In contrast to many other colonies, where the creation of nation-states remained a purely imperial product with little influence on people's actual language practices, the system of indirect rule in colonial Rwanda and Burundi led to a wide application of standardised Kinyarwanda and Kirundi, especially under Belgian rule. 8 In the following decades, the implication of the excessive use of these national languages for missionary purposes, colonial aims, and as unifying national emblems was that each language became indexical of its speakers' national belonging: Kinyarwanda speakers were therefore regarded as Rwandans, Kirundi speakers as Burundians. This was also transferred to adjacent borderland regions, where "Rwandophone" speakers in DR Congo and Uganda were for a long time (and arguably still are) considered to be "Rwandans" despite their Congolese or Ugandan nationality. 9 The indexical qualities of the colonially fixed and stylised languages -stylisation here standing for language planning and dialectal levelling -implied mostly perceived differences rather than actual linguistic, cultural or ethnic 10 differences, to some extent based on stereotypes and rumors (i.e., what it meant to be Rwandan/Burundian). The concepts of (imagined) nationality and nationhood are here grounded in processes of stigmatisation and human categorisation, following colonial models of drawing boundaries between territories, people, and languages.

Precolonial Empires: Nations, Languages, and Borders in a State of Flux
Throughout the two long-lasting kingdoms of the Rwandan and Burundian aristocratic courts, the borders of the realms were far more flexible than those later established on the basis of the Brussels conference in 1910, which still determine the current national borders.
When the Germans arrived in Burundi, they "found a situation bordering on chaos" (Lemarchand 1970: 49) with struggles between the king Mwami Mwezi Kisabo and numerous opposing chiefs in different parts of the realm. This caused the mwami to display a willingness to cooperate with the Germans, with the intention of bringing back under control the "breakaway" parts of his realm that the rebellious Batare had seized, especially in the north and northwest. Here, due to a long-lasting fratricidal dispute, the borders of the monarchy were not clearly marked and disputes over fragile local alliances with other chiefs occurred frequently. There was a fixation on national boundaries throughout the ongoing German construction of Standorte der Schutz-und Polizeitruppe (locations of protection and police troops). In Usumbura/Bujumbura, Gitega etc. this therefore caused an artificial image of a volatile realm, and also lead to the language policy of turning the prestigious dialect Ikiruundi into the standardised language Kirundi.
The regions at the periphery of the Rwandan realm, for instance, were in a steady state of flux, depending on treaties with adjacent border groups. During the era of King Rwabugiri from 1867 on, numerous royal expeditions were undertaken in the outlying regions (Vansina 2004: 211-212), in order to occupy adjacent areas and negotiate treaties.
Some of the shifting precolonial borders of Rwanda are shown in Map 1, revealing the expansions and losses of the realm between 1867 and 1910. In roughly thirty years, the king's power had vastly expanded up to Butembo in the northwest, towards Ankole in the northeast and encompassing Bushubi and others before decreasing again when the colonial powers occupied the area. Bufumbira, to the north, fell to the British and was hitherto ruled from Kampala, mostly with ethnic Baganda authorities from the capital city. This shows how fluid and volatile the borders were and consequently how arbitrary the markers are, which were used to categorise somebody from Bugoyi or Bufumbira as belonging to the Rwandan realm or not. Therefore, it can be argued that these practices of drawing lines of national belonging (to Rwanda-Urundi, to German East Africa vs. belonging to British East Africa, etc.), were only established through colonial powers. 11 11 Local hegemonies, especially as part of a "pacification campaign," were simply replaced (and, at first, seemingly stabilised) by colonial and church authorities. Missionaries such as Léon Classe (see below) actively supported these colonial initiatives of replacing local Hutu lineage heads with Tutsi elites from the royal Modern Africa: Politics, History and Society | 2019 | Volume 7, Issue 1

Map 1: Historical borders of the Rwandan Kingdom between 1867 and 1910
At the time of the arrival of European powers at the northeastern borders of the Rwandan realm, the borders were not clearly demarcated, and territorial claims were controversial. A considerable degree of confusion about the exact boundaries often created political tensions, especially between the Belgians (to the west) and the British (to the northeast). This confusion also had an impact on Rwandan citizens' self-identification in the border regions; such as the socalled Mfumbiro Massif region (or today's tripoint in the Virunga volcanoes), of which it was always unclear whether it was British-ruled, German-ruled or Belgian-ruled, due to the contradictory indications on colonial maps. Ngologoza (1998: 58) states, with regard to the Kigezi region, i.e., Bufumbira and adjacent parts of Rukiga-speaking territory around today's Kisoro and Kabale in Uganda, that "at this court (Carney 2014: 32). The apparent strengthening of the northern and western borders of the Rwandan realm were, however, only temporary and colonially secured politics that increased the tensions between the colonised groups. Imposed nationality, or European nationalist ideologies, throughout the African continent on the basis of arbitrary geolinguistic borders, had a heavy impact on the colonised individuals' sense of belonging; on their self-positioning and identification. The official adoption of colonial languages such as Kinyarwanda and Kirundi as a result of linguistic interventions and artefactualisation was not always necessarily rejected by speakers: In the northern periphery, for instance, the colonial practice of "imposed nationality" (Rwandan or German-ruled vs. Ugandan or British-ruled vs. Congolese, thus Belgian-ruled) also conveyed linguistic citizenship (Stroud 2001) to speakers, who then expressed geospatial belonging through performed language. This also led to Rufumbira, as a northern Kinyarwanda dialect, being perceived by speakers as an autonomous language, separate from the colonial divisions between German East Africa and British East Africa. Rufumbira speakers, whose national belonging to the Rwandan realm was not clear until 1910, were then turned into "Ugandans" overnight, when the Brussels conference clarified colonial lines of division. In contrast, the direct neighbours to the Bafumbira who spoke Ikirera, were thus turned into "Rwandans."

Kinyarwanda: Of Missionaries and Orthographies
Despite plentiful and richly varied historical analyses of precolonial and colonial Rwanda, very little has been said about the actual language planning processes. These processes fixed the specific Ikinyanduga dialect of Kinyarwanda as the variety for the Rwandan kingdom. Furthermore, it was later promoted as the language used by missionaries under German rule, despite German authorities actually favoring Kiswahili as the widespread language for their colony (this was also the case in Tanganyika). After the founding of the German Residency, missionaries were first required to use written messages from the Court, as the only form of communication with the Germans (in order to avoid unclear and ambiguous oral communication). This was supposed to be entirely held in Kiswahili "[b]ecause Kinyarwanda was so rich in nuance" and included "richer possibilities for misunderstandings." Furthermore, Europeans generally communicated in Kiswahili and the knowledge of Kinyarwanda was not widespread (Des Forges 2011: 95). The Royal Court eventually agreed and even sent one of their clients to learn Kiswahili in order to be able to communicate in written form with the Germans.
It was only later that writing in Kinyarwanda became one of the prime goals on the agenda of church representatives: Several missionaries and intellectuals took part in the textualisation of Kinyarwanda, whose decisions laid the groundwork for later orthographic changes. Among them was Eugène Hurel (1911Hurel ( , 1921, whose linguistic output was a grammar, a Kinyarwanda manual and a dictionary. While the first dictionary had already been compiled by Harry H. Johnston (1902), the first German dictionary under colonial rule was written by Felix Dufays (1913) and considerable work on Kinyarwanda was later carried out by R.P. Schumacher (1921, etc.).
Other notable linguists/intellectuals contributed to the development of Kinyarwanda as it is now written and used in schools, universities and for official purposes. They actively engaged in transforming diverse language practices into one standardised variety. Among them was the missionary Léon Classe, who came to Rwanda in 1901 (as a White Father) and was made Vicaire Délégué for Rwanda in 1907. In 1912, he eventually became the right hand man for bishop Hirth in Ruanda-Urundi (Strizek 2006: 189). Missionaries began to focus on politics and linguistics, as they had a hard time finding converts particularly due to their social stigmatisation (given the name inyangarwanda "repudiators of Rwanda"; see Carney 2014: 27). Classe pursued the idea of christianising Rwanda through elite conversion and therefore supported the foundation of the Tutsi monarchy (ibid.: 28), which, to some extent, unconsciously contributed to an aggravating schism within the colony. In terms of linguistics, he emphasised the fact that there was one language, which he (mis)took as a symbol of national unity, stating that "there are few peoples in Europe in which one finds together the three factors of national cohesion: the same language, the same religion, and the same customs" when compared with Rwanda (ibid.: 34). Classe's understanding of Rwanda as one nation and, building on this thought, that all Kinyarwanda speakers were one homogenous linguistic unit, ignored the fact that Rwanda had always been linguistically diverse, before and after the arrival of the Europeans (Des Forges 2011). His idea of a "linguistic unity" in Rwanda was undoubtedly bound to his knowledge of the royal court and their use of the preferred dialect Ikinyanduga. The reality, however, was far more complex.
The linguistic diversity in Rwanda was striking from the early days of the Nyiginya kingdom on, as summarised by Vansina. He also stresses that the feeling of belonging to "one nation" was mainly an invention of the twentieth century, and thus tied to colonialism and missionary activities: "The linguistic and cultural unity of the country today did not exist in the seventeenth century and Rwanda is not a 'natural' nation. It is the product of the expansion of the culture of the Nyiginya court that began in the eighteenth but occurred mainly during the nineteenth and early twentieth centuries. Rwanda really became a nation in the twentieth century. It is an error to imagine that all the forebears of the present-day speakers of the Rwandan language must have spoken it in the past and in the same way as now, and therefore that the language was always as unified and standardized as it now is" (Vansina 2004: 198).
This becomes evident in dialectal maps of Kinyarwanda that display the linguistic diversity of the country and the large number of Kinyarwanda dialects, out of which Ikinyanduga was chosen. It was adapted in terms of corpus planning, language planning, orthography and then standardised, due to the fact that the Rwandan court of Nyanza was located within the Ikinyanduga-speaking area (see Map 2).
The problems in reducing the many diverse Kinyarwanda varieties to one principal language, to be used throughout the colony, did not necessarily lie in the fact that missionaries "falsified" the language due to their own limited knowledge. Instead, their -at times very detailed -knowledge of the language and their advanced speaking skills led to ideologically subjective choices and decisions being made in matters of morphosyntax, lexicon etc. Often, their interventions knew no limitations, mainly due to the monopolisation of education and the privileged socioeconomic positions of individuals working for the church. In 1930, for instance, Léon Classe had convinced the colonial (Belgian) authorities to transfer the power over all colonial secondary schools to the missionaries. At times, missionaries were perceived as being extraordinarily competent in the language, leading to their decisions being easily accepted and often appearing irrevocable, as implied by Vansina: Five different orthographic changes, as listed by Niyomugabo and Uwizeyimana (2018: 309), took place in the textualisation of Kinyarwanda. The orthography was first designed by European missionaries, then further implemented by the above-mentioned Classe (1938), along with T. Bagaragaza (1974), A. Nsekalije (1985 and J. D. Mujawamariya (2004), 13 before finally being officially implemented by the Rwandan government on 13 October 2014, with the purpose of simplifying the orthographic conventions. These recent initiatives, however, did not radically change the orthography. 14 It is a commonly held view that the consolidation of nationhood was not only achieved on the basis of colonial violence and foreign missionaries but also that "the intelligentsias were central to the rise of nationalism in the colonial territories" (Anderson 2006(Anderson [1983: 116). Alexis Kagame (1912Kagame ( -1981) played a dominant role as an intellectual in colonial Rwanda, in a role analogous to what Anderson (ibid.) labels "bilingual literacy." The "homogeneity of language and literary tradition" is cited by Nzabatsinda and Mitsch (1997: 99) as one of the main factors that determine a nation, alongside religious practices, sociological behavior and people's coexistence. A homogenous language was also institutionally considered as an element that established (national) identity (ibid.). Kagame, a Catholic priest, doctor of philosophy, writer and thinker, also primarily promoted a unified national language. Standard Kinyarwanda and his "life and works were thus concurrently tied to a project to identify, cultivate, promote, protect and illustrate a literature that was conceived as national" (ibid.). Equally, his prolific writing was intended to textualise Kinyarwanda orature of all genres, and to establish a corpus of written poetic and non-fictional texts "conceived as national." Heavily criticised for a "centralizing 'Rwanda' speech" (ibid.) in his attempts to express Rwandan unity in poetic and traditional texts, he was often confronted with the accusation of not correctly respecting the dynastic tradition at the royal court and its historical details. Kagame's linguistic work included tentative attempts to standardise and "stabilise" Kinyarwanda, whereby "its autonomy with respect to the languages most like it, such as the Kirundi" (ibid.: 103) was of major importance. This is summarised by Nzabatsinda and Mitsch (ibid.) as a practice to "distinguish the particularities of Kinyarwanda from those of other languages whose forms (especially lexical) presented similarities." In more recent times, especially after 1994 when Rwanda was liberated by the RPF/FPR and reunified as one nation, the role of the single unified Kinyarwanda was further strengthened. Standardised Kinyarwanda is frequently used in public speeches by the president and other public figures as well as being taught at the University of Rwanda and in schools. Despite the common claim that "most known dialects in Rwanda […] need a core protection" through public social awareness (Turikumwe 2013), research on dialectal variation and linguistic diversity does not appear to be of primary importance among Rwandan linguists (with a few exceptions, such as Bazimaziki 2018; Nkejabahizi 2007Nkejabahizi , 2010Rwigamba et. al 1998). Among Bantuists, Standard Kinyarwanda is usually subject to specific grammatical phenomena in linguistic studies; with all the available grammars dealing with the standardised variety (Kimenyi 1980(Kimenyi , 2002. The largest lexical compilation and corpus planning initiative spanned over several decades and was eventually completed by Coupez et al. (2005). Due to its scale, it can be seen as having largely contributed to a further development of the Kinyarwanda lexicon (and its standardisation).

Van der Burgt's Works on Kirundi
The development of today's Kirundi is mainly bound to the work of missionaries, whose interventions on the language throughout precolonial Urundi turned Kirundi into the contemporary official language, mainly through the writing of grammars and compilation of dictionaries. While Kinyarwanda was fully standardised and underwent several orthographic changes during the twentieth century, Kirundi largely remained as it was described in A. E. Meeussen's (1959) early extensive grammatical study, which today is still the most comprehensive description. In the early colonial days, several other grammar sketches were published, namely François Ménard (1908), and as a more extensive study, van der Burgt's (1902) grammar.
The Dutch missionary Johannes-Michael van der Burgt arrived in Urundi in 1892 as part of the Congrégation des Missionnaires d'Afrique (Les Pères Blancs), the White Fathers. After some months in Uzige, he founded the mission of Mugera in 1899, not far from today's city of Gitega, together with Father van der Wee (Chrétien 2010: n.p.). Until 1908, he served at the newly founded mission of Kanyinya in the north of the colony, after his most comprehensive work had already been published. His dictionary (van der Burgt 1903a), shows the rich lexicon of Kirundi but also contains ideological adaptations based on van der Burgt's "racial obsession," namely "the inclusion of the Tutsi category into a Hamito-Semitic stock" (Chrétien 2010: n.p.). This work remains as debated as his ethnographic study Un grand peuple de l'Afrique équatoriale: Eléments d'une monographie sur l'Urundi et les Warundi, which appeared in the same year (van der Burgt 1903b). Both his dictionary entries and his ethnographic work reflect the spirit of their time, with racialising and pseudo-scientific judgments about the group compositions and origins of the inhabitants of colonial Ruanda-Urundi (although they were at times repeated in later ethnographies, such as Meyer [1916]).
In correlation to this controversy, van der Burgt's 630-page comprehensive dictionary reveals a clear orientalist perspective on Kirundi, its speakers and their worldview, labeled by Chrétien (2010) as "fantasmes d'un missionnaire" [fantasies of a missionary]. "[T]he Burundese culture was thus trapped by this imagination born out of the European libraries and it was described as the result of degeneration" (ibid.), as van der Burgt sought various far-reaching explanations for the diversity among Burundians, who were at this time understood to make up one "Burundian nation" as speakers of a textualised variety of Kirundi.
During the compilation, van der Burgt received help from a native interpreter thought to be a young Tutsi, Bernard Kitwe, who supported Van der Burgt's pursuit of knowledge of Kirundi, leading to the publication of a dictionary in 1903, despite van der Burgt himself actually having problems preaching in the language (Chrétien 2010). Thus, it can be assumed that his young interpreter contributed to large sections of the lexicon entries, which were then translated and annotated into French, German and Kiswahili, based on Van der Burgt's subjective explanations and assumptions. The large, longterm impact that this dictionary had on Burundian scholarship and Kirundi linguistics shows the potentially disastrous effects of colonial textualisation practices (and the apparent irreversibility of their written formations).
Similar to the Rwandan situation, Kirundi orthographic debates have also been constant matters of discussion: The colonial language was adapted, not necessarily to pursue German colonialists' aims, but in order to turn it into a widespread medium for gospel and give the missionaries a means of communication to try to achieve their principal aims. Kirundi diverges from Kinyarwanda orthography in several specific ways, such as in the differing graphemes <by> in Kinyarwanda and <vy> in Kirundi for the phoneme [ß j ] as in ababyeyi/ abavyeeyi "parents." Orthographic discussions were already subject to colonial and missionary debates in the early colonial period: After World War I, van der Burgt announced that order should be established through orthographic unification. This shows that orthographic adaptations could be used to target cross-regional unification among the different colonies ("die Engländer," "alle Franzosen"), through encouraging the subjective observations and notations of missionaries and colonial agents. In contrast to a speaker-centred perspective (whose judgements and experiences were completely omitted), the colonial orthography was intended to be pragmatic, at least for European users. This also concerns Europeans' efforts to establish unified standard varieties, as the purpose was especially to benefit the christianisation campaign and a colonial (as well as epistemological) implementation of power. Bukuru (2003: 4) describes what is nowadays perceived as Standard Kirundi as "a dominant dialect […] spoken in the Burundi central highlands" and in "the traditional regions of Mugáamba, Kirimiro, Buyeénzi and Butuutsi, which historically constituted the heart of the old Burundi monarchy." He further states that "the same dialect is currently used in all official domains, such as schools, churches, administration, and politics," yet he underlines that, despite its status as the official language, it has not yet been fully standardised.
Map 3: Kirundi dialects (adapted from Bukuru 2003, Mayugi andNdayishinguje 1985) Against what is today often taught as Standard Kirundi, the dialectal situation in Burundi presents itself as much more diverse than most colonial sources reveal (they were mostly written and published for the ideological support of missionaries in regard to their work and language planning initiatives). Material is scarce and most studies are not easily accessible, with the available dialectal analyses to be mainly found at the University of Burundi. Map 3 shows the major dialects and reveals that the prestigious central dialect is not actually traditionally present in the region of the capital Bujumbura. However, according to speakers, dialects such as Ikibo (the western dialect from Bujumbura) and Ikiragane (south) are less used today than they were before.

Overcoming Boundaries: On Fluid Language Use in Multilingual Neighbourhoods
Based on an analysis of Kinyarwanda and Kirundi speakers' contemporary fluid language use, the language planners were not too successful, as will be shown in the following Sections, despite their extensive efforts of standardisation and stabilisation.
From 1959 on, the first pogroms of the Hutu opposition against the Tutsi (and moderate Hutu) minority took place and migration from Rwanda to neighbouring Uganda steadily increased. In the following decades, many thousands of Rwandans settled in western Uganda, especially around Mbarara, as well as in different neighbourhoods of the Ugandan capital Kampala. It was only in the mid-1990s that the Rwandan Genocide was finally ended by a military liberation of Rwanda, operated by high officials of the Ugandan army whose families had formerly fled Rwanda due to persecution.
In 2015, around the outbreak of violent conflict in Burundi due to President Nkurunziza's debated third electoral term, large waves of refugees poured into neighbouring countries, predominantly Rwanda, the DR Congo, and Uganda. Uganda already held a Kinyarwandaspeaking population from the first migratory waves in the 1960s, which therefore mingled with newly arrived Burundians. In specific transit neighbourhoods of the Ugandan capital, interactions commonly take place between long-term Rwandan heritage speakers, more recent Burundian refugees, "Rwandophone" traders, students and former rebels from DR Congo as well as Rufumbira speakers from southeastern Uganda. These multilingual spaces, where Kirundi, Kinyarwanda, but also Luganda (another Bantu language from the central regions in Uganda), English and French are spoken, either have a long history as business hubs (often centred around bus terminals), or are places of quick encounters, fleeting relationships, nightlife and markets. Bakuli, located in the old city centre of Kampala, is such a neighbourhood where Kinyarwanda and Kirundi speakers meet and interact: Based on preliminary interviews conducted in Kampala, it can be observed that the fixed divisions between the two languages of refugees and migrants are perceived as an intelligible continuum rather than as two artefactualised entities: In contrast to the colonial design and schism of two closely-related languages, contemporary language use in these busy neighbourhoods shows a high degree of fluidity and strategic crossing of language boundaries. Migrants, travelers, traders and others deconstruct and "disinvent" language boundaries through linguistic mimicry, play and the reversal of identities in business interactions, which allows them to indexically present themselves as "Burundians" or "Rwandans" whenever this is suitable or appropriate to their specific communicative aim. This may, in further studies, potentially allow for a deeper understanding of the development process that turns "indices" (of nationality, belonging etc.) into fixed "markers" and eventually into "stereotypes" or "categories" (see Labov's [1972] model of indexicality; see also Hirschauer [2017]).
In order to conduct a more detailed analysis of contemporary speakers' patterns of translanguaging and linguistic accommodation in opposition to (pre-)colonial divisions, their language biographies need to be included, as is recurrent in numerous sociolinguistic studies (Busch, Jardine and Tjoutuku 2006;Franceschini 2001 etc.). Conflict-induced migration patterns and colonial and missionary categorisations along with the practices of fixing languages, speakers and nations, are key aspects in the fluid practices seen in Ugandan neighbourhoods such as Bakuli. The speakers' processes of self-positioning (in contrast to their classification as "Rwandans" vs. "Burundians") may potentially be approached with methods employed in perceptual dialectology (Preston 1999; for Kiswahili, see for instance Njoroge and Githinji 2018), allowing more fine-grained modes of differentiation. In the study of language and migration, De Fina and Georgakopoulou's (2012) focus on storytelling, and the narratives around speakers' trajectories as a discursive pool, could also yield fruitful results in the analysis of language use in a neighbourhood like Bakuli.

Conclusion: Categorisations of Speaker, Language, and Nation
In this article, I have sketched the processes of textualisation and linguistic intervention by missionaries and colonial administrators in regard to two strikingly similar linguistic varieties: Kinyarwanda and Kirundi. These are two dialects chosen out of a fluid continuum, based on varieties used in precolonial monarchies and then implemented through colonial policies. The textualisation, standardisation and orthography-making by scholars such as van der Burgt, Classe, Kagame and others have contributed to a correlation between constructed linguistic entities and national territories, demarcating linguistic boundaries between Rwandans/Kinyarwanda speakers and Burundians/Kirundi speakers, based on colonial nationalist ideologies. The dilemma of African linguistics becomes evident in the description and documentation of languages that were developed, introduced or employed in the colonial era: Languages sketched along their speakers' imagined ethnolinguistic boundaries or a language planner's geolinguistic lines of division often do not correspond with speakers' agentive or creative strategies in interaction.
In divergence from the colonial categorisation of speakers as either "Rwandans" or "Burundians," according to a stencil-like measurement technique of their language practices against two monolithic linguistic blocks (i.e., the standard varieties), speakers constantly position themselves and make use of new and fluid categories. In a way, today's discipline of (African) linguistics is repeatedly confronted with early colonial and missionary studies that are seldom undisputed or uncontroversial -while some scholars approach this dilemma by acknowledging the disastrous impact of colonialism, evident in such emergent subdisciplines as (post-)colonial linguistics, others continue to dissect and essentialise fluid language practice, dividing varieties along city boundaries, nation-states or based on other criteria. Chrétien (2003: 10) reflects upon the colonial violence of grouping and categorising people based on arbitrary icons, indices or markers by mentioning the resulting distortions that continue to affect contemporary speakers and scholars from the Great Lakes region: "One can imagine the holes, distortions, and intellectual myopia engendered by the depth of the divide born from the colonial partition and the primacy of Eurocentric frame works. What needs to be thought anew involves spaces, temporali ties, and themes." Therefore, re-examining the broader context of colonial and postcolonial lines of division that contributed to today's nation-states and that categorised or "sorted" speakers according to arbitrary boundaries, remains one of the most urgent tasks in critical studies conducted in the field of African linguistics. Forthcoming work in the field of missionary and colonial linguistics needs to address the discrepancy between colonial categorisations and the strategies of speakers to position themselves.