A Diasystematic Approach to Multilingual Ecology: The Case of Mbum Speakers in Ngaoundéré, Cameroon

Abstract

Speakers of the minority language Mbum (Cameroon) live in a multilingual ecology coping with different degrees of interferences from a wide variety of other languages and dialects. They consciously and/or unconsciously insert constructions from different languages in different communicative contexts and often apply mixed codes altogether. Sometimes, they reckon that a given linguistic unit does not belong to one specific language from their repertoire but feel that it is somehow ‘around’ in the multilingual space they navigate. Such statements highlight an epistemological dilemma that becomes more and more apparent in language contact theory: While it is evident that in multilingual ecologies assumptions of single, articulate, static, and ideally variation-free language systems do not at all reflect speakers’ realities, the concept of individual languages that influence each other to various degrees while keeping their grammars neatly apart is generally presupposed for descriptive and analytic purposes. A possible solution for the dilemma lies in the Diasystematic Construction Grammar approach which conceives of a grammar as community specific and not as language specific.

1. Introduction

Speakers of the minority language Mbum (Kebi-Benue, Adamawa, Cameroon) live in a multilingual setting on a daily basis. Whether they are part of some remote village communities where interference with other languages may be less obvious or they intermingle with a multitude of speakers of various languages, lects, and varieties in an urban hot spot like Ngaoundéré (the regional capital of the Adamawa province in Cameroon) they all cope with different degrees of interference from other languages and dialects. Speakers in such an environment may or may not be aware that they use material from different languages in different communicative contexts and often apply mixed codes altogether. Sometimes they even reckon that a given linguistic unit cannot be said to belong to one or the other specific language from their repertoire but feel that it is somehow ‘around’ in the multilingual space they navigate.

A fruitfully employed term for such complex situations is ‘multilingual ecology’. Following this metaphor from environmental studies, one conceives linguistic varieties as practices inscribed in a local system where they fill niches defined by their relations with other varieties therein and by their functions in the milieu. The milieu equals the sociolinguistic specifics of a given locality and may in turn develop stimuli that prompts the varieties to adapt, thus eventually leading to linguistic evolution and change. Such complex situations that occur mainly in contexts of small-scale multilingualism are by no means extraordinary in Cameroon, Africa or even on a global scale. Multilingualism and its impact on language use and development is and has been common all over the world and may be considered rather the rule than the exception [1, 2, 3].

On the backdrop of the ever-growing body of research on all aspects of multilingualism and language contact, its concomitant linguistic, cognitive, and social processes and mechanisms, an epistemological dilemma becomes more and more apparent: On the one hand, it is evident that in multilingual societies assumptions of single, coherent, static, and ideally variation-free language systems do not reflect speakers’ realities at all. On the other hand, presupposition of well-defined and coherent language systems that influence each other to various degrees while keeping the respective grammars neatly apart are applied for descriptive and analytic purposes.

This line of thinking dominates even the most recent and sophisticated theories of language contact as it is part of the discipline’s scientific legacy. Given that modern linguistics developed on the backdrop of European nationalist ideas mainly in the nineteenth century, the concept of named languages reflecting national identities established itself as undisputed preconception. In so far, the beginnings of contact linguistics with scholars like Weinreich [4] and Haugen [5] were no exception, and most recent theories of language contact [6] still need to keep the involved languages in a multilingual context conceptually apart.

So, there is a dilemma between observed realities and theoretical considerations. On the one hand, scholars realize the ubiquity of multilingual/-lectal societies, observe interlingual similarities on all grammatical layers, and model the cognitive interaction of systems in multilingual minds as operating in one single module [6]. On the other hand, they need defined entities like named languages with specific grammars that allow for clear boundaries between native and foreign material to describe and analyze contact-induced processes and outcomes in the involved language systems. I experienced such dilemmas on several occasions in my own analyses and writings. In a paper on the effects of limited input for language learning, I look at strands of communications where speakers exploit their multilingual repertoires. Even though my analysis of quite a few of the occurring constructions leads to inconclusive results as to their linguistic origins, in the interlinear transcription I propose nonetheless various named languages as their sources. Thereby I conform to the implicit presupposition of language contact being indeed contact of separated grammatical systems that can be traced in the multilingual output while—at least in some instances—this assumption does not really hold on closer inspection ([7], pp. 8, 9). To overcome this dilemma, it seems like one has to get rid of the classic line of thinking that puts the grammar of a given named language center stage and look closer at the ways speakers and groups use the linguistic means at their disposal.

Given that such an endeavor has an iconoclast touch to it, there are currently only a few attempts leading in such directions. For instance, Watson [8] opts for a prototype approach to study multilingual data to produce “more accurate and flexible descriptions based on both the elicited judgements and observed usage of speakers, and both allow and account for variation. Such descriptions can enable an analysis of multilingual discourse whereby individual linguistic features are interpreted within the sociolinguistic context in which they are uttered.” ([8], p. 126).

Another attempt to challenge the dilemma comes from scholars with a background in the construction grammar paradigm [9, 10, 11, 12, 13]. They develop an approach to common language structures which regards multilingualism as an integral part of a grammatical system of a group of speakers. They argue for a usage-based construction grammar where speakers in bi- or multilingual groups create a common system for all their languages. This so-called Diasystematic Construction Grammar (DCxG) [9, 10] pictures the (speech)community as the primary carrier of grammar, not the single language. It thus becomes possible to integrate multilingualism as a basic feature into a coherent grammatical description representative for whole speech communities. The term ‘diasystem’ was coined by Uriel Weinreich to capture dialectal variation in a kind of overarching supergrammar. He stated that it is mainly an analytic construct that combines two systems which have partial similarities. But he also added that this does not mean that it is always a scientist’s construction only but that diasystems are experienced in a very real way by bilingual speakers [14]. A diasystem can thus be understood as a higher-order system with its component units being abstractions from individual systems that are based on speaker’s experiences. Insofar, the term expresses exactly what the DCxG approach tries to bring forward in terms of grammar writing.

In this chapter, I like to demonstrate how such an approach may look like when applied to non-standardized and often under-described languages and what one may gain from it. To that end the paper is organized as follows: I firstly describe the multilingual environment of Ngaoundéré and northern Cameroon. I then shortly present some young urban speakers of ethnical Mbum who navigate this multilingual ecology and served as language consultants for this study. In this section, I also reflect on the term ‘multilingual ecology’ and its relation to linguistic Darwinism. In the third section, I briefly lay out the theoretical foundation of DCxG and discuss its epistemological advantages in coping with the kind of multilingual ecology at hand. The fourth section exemplifies the DCxG approach with primary fieldwork data stemming from the aforementioned groups of speakers. The last section then sums up the findings and gives an outlook to further possible steps to model speaker realities in such multilingual ecologies.

2. The multilingual environment of Ngaoundéré

The northern territory of Cameroon bordering on Nigeria, Chad, and the Central African Republic is one of the linguistically most diverse areas worldwide [15]. Ngaoundéré is the regional capital of the Adamawa Province situated on the Adamawa plateau and the economic gateway to the northern provinces. The already highly diverse linguistic ecology of the area is currently intensified by growing influxes of foreigners from other parts of Cameroon and neighboring countries due to a multitude of currently ongoing regional crises and concomitant migration.

Besides French (and English) as the official language(s) of Cameroon, a local variety of Fula (Atlantic), namely Adamawa-Fulfulde, fuctions as the lingua franca of the region and is also the language of the local noble authority. Moreover, many indigenous languages of different families (mainly Adamawa and Chadic) are spoken, and other major African languages (e.g., Hausa, Kanuri, and Shuwa Arabic) are also present in the city. Mbum, from the Adamawa family, and Adamawa-Fulfulde are considered the local languages in Ngaoundéré with Mbum claiming autochthonous status.

Apart from French and some specific high-prestige varieties of the major African languages, which are rarely encountered in Ngaoundéré, standard orthographies and generally acknowledged reference grammars are extremely scarce. Again with the exception of French, none of the multiple languages is used in a coherent way in education or the mass media. Therefore, in the multilingual ecology of Ngaoundéré, French, even in its colloquial Cameroonian version, is a high-prestige language only mastered by a few very well-educated people and rarely used in everyday conversations. In this respect, it is also not competing with the local lingua franca which is always the first choice when people are not sure of the linguistic competences any given addressee might have. This general make-up of the local multilingual ecology has contributed to what has been called ‘diffused’ linguistic systems [16] and a situation where ‘language’ is generally a more diffuse phenomenon [17] rather than a clearly defined entity.

In this environment, Mbum is no different from any other language. Notwithstanding the fact that beginning with early colonial writers, the Mbum and their language have been under scrutiny from researchers for over a century, there is no comprehensive description that reflects its far-reaching regional variability. Throughout the decades, different local varieties were described from different angles, for different purposes and with different scope which gives us some notion of Mbum’s dialectal and chronological diversity [18].

Despite the above-mentioned ‘diffuse’ language concept, the acknowledged diversity may also lead to discussions about the correctness of a lexeme, a construction, or any other linguistic unit among speakers. This happened several times during fieldwork sessions and even culminated in interviewees arguing about ‘real’ and ‘false’ Mbum. The occasional phone call to some elder in a village for a ‘correctness judgment’ would only rarely settle the issue.

2.1 The language consultants

The interviewees of the present study were language consultants who worked with the research team on several occasions between 2018 and 2023. They belong to two different social groupings that I will shortly describe in terms of their current social relations and life modes.

The first group contains five friends/colleagues who live in Ngaoundéré. They are all young men (between 20 and 27 years) and unmarried. They often pass their free time together, and two of them live in the same compound in one of the city’s suburban areas. They all were born and raised in the same rural area some hundred kilometers east of the city and thus, in principle, share the same dialectal background as their L1. I got to know two of them during the first project phase when I studied the use of the local lingua franca Adamawa-Fulfulde among motorcycle taxi drivers (MTD) in Ngaoundéré [19]. Later, when the project’s scope shifted to the varieties of Mbum, my connection with these Mbum speaking MTDs proved very fruitful as they established contacts with other friends of their small group. Two of which work in a repair garage for trucks while the third runs his own small roadside garage mainly for puncture repair and small mechanical issues of the ubiquitous motorcycles in town. All five speakers speak and understand at least three different languages. They are fluent in Mbum and Adamawa-Fulfulde and all have at least some notion of French. In addition to these shared competences, several other languages are part of their individual repertoires (e.g., Gbaya (Ubangi), Masa (Chadic), and Dii (Adamawa)) that mainly reflect individual biographies and family histories.

The second group of language consultants is made up of mostly young speakers, both women and men, that are all from the Mbum-speaking village of Nganha, some 50 km northeast of Ngaoundéré. However, during semester time, they all live in Ngaoundéré or nearby settlements, as they are all students (on different levels and subjects) at the University of Ngaoundéré. The first contact with that group was established through one single student who responded to an online search for Mbum speakers via Facebook in 2020 when the corona crisis prevented fieldwork. When in October 2022 the first post-corona field trip took place this online acquaintance helped to establish contact in real life with his peers from the Nganha-student group and also introduced us to his family and the noble authorities in his natal village. The student group is not a formally established entity but relies on the members’ solidarity as daughters and sons of the natal village. Language-wise, they again share the usual set of repertoire languages, i.e., Mbum, Adamawa-Fulfulde and French, with the latter usually on a higher competence level than in the first group. The individual biographies again add further languages to the individual repertoires.

All language consultants consented to provide us with social background information and language data that stem from different tasks. The social questionnaire comprised the usual personal questions about age, parents, education level, residencies, and known languages (including self-estimation of competence levels) alongside questions on mobility patterns, friendship relations, and concomitant contact frequencies.

The linguistic data stem from a variety of tasks. We asked each individual to retell two picture stories in Mbum without specifying the variety we are interested in. We also recorded group interviews, discussions, and work-related encounters. The language data are then transcribed and tagged for interferences from other languages and lects on all levels. This work is still in progress but yields already interesting data that will be presented in Section 4 below.

2.2 The multilingual ecology

All our consultants have the necessary means at their disposal that are needed for successful communication in the complex linguistic ecology that the Adamawa plateau and, more specifically, the city of Ngaoundéré represents. The use of the ecology metaphor in connection with languages goes back to Haugen [20] and is further elaborated by Calvet [21]. He indicates that languages as practices are inscribed within a worldwide system, being organized into constellations, within which every language has its niche, defined by its relations with other languages and by its functions in the milieu. The milieu, or in sociolinguistic terminology—the linguistic landscape of a given locality—also acts on languages by stimuli to which languages respond in the mode of regulation and change, which can lead to linguistic evolution and change [22]. When we break down the ‘worldwide system’ and look at a much smaller context where nonetheless many different languages and varieties interact accordingly, the term multilingual ecology seems a logical and appropriate choice. It is the same as linguistic ecology only highlighting the fact that the repertoires of the speakers in such an ecology are by definition multilingual.

The ecology metaphor is also closely related to evolutionary or (neo-)Darwinian linguistics. The idea that linguistic units behave like genes in that they replicate, evolve through mutation, and occupy niches they adapt to is also well in line with the usage-based approach advocated by DCxG. In so far, multilingual ecology and DCxG complement each other very well (see also section 3).

The speakers choose a construction from their multilingual repertoire that they consider appropriate for any given communicative situation. The environmental stimuli are the various communicative contexts, different people, and topics. The multilingual construction they use shows specific interference patterns and traits that exist in their ‘niche’ of use and may be described accordingly. To correctly account for the construction used in any specific niche, it is imperative to take the characteristics of that niche into account. The diasystematic approach is specifically tailored for this enterprise.

3. The Diasystematic Construction Grammar

Diasystematic Construction Grammar builds on the premises of construction grammar (CxG) that understands language as non-modular and usage-based. It is mainly based on Goldberg’s cognitive construction grammar (CCxG) [23, 24] and Croft’s radical construction grammar (RCxG) [25, 26]. There is nothing in the basic principles of these usage-based CxG’s that actually requires a language system to be restricted to just one language. There is no necessity for language boundaries in the input, in language usage, or in the general cognitive principles which are the basis of constructional networks ([9], p. 46). Based on these premises, the pivotal point of DCxG reads like this:

“The grammatical description of a language system in a multilingual environment – i.e. the socially conventionalised set of all structural elements shared by a specific speaker group as well as cognitively stored and processed by the individual speakers – must include structures of all languages or varieties involved, and the social establishment and individual acquisition of such a system must be inherently multilingual” ([8], p. 140).

DCxG sets out from two basic assumptions, namely that any manifestation of language contact happens in the context of individual and collective multilingualism and that multilingual speakers do not rely on separate modules in cognitive processing, but select the linguistic elements they apply from their multilingual repertoire according to the given communicative setting [27]. It furthermore builds on the capacity of multilingual speakers to notice formal and functional similarities of linguistic material in their different languages. The general tenets of all CxG approaches are also part of DCxG:

1. grammar is conceived as an inventory of constructions, i.e., form-meaning pairs that form a continuum involving everything from lexically filled constructions (such as words) via partially filled constructions (e.g., in inflectional paradigms) to maximally schematic ones (such as syntactic or prosodic patterns);

2. more schematic and more concrete constructions are connected through inheritance links, forming a network of interdependent elements;

3. speakers learn schematic constructions by categorizing them on the basis of the available input, resulting in an economic representation of their linguistic knowledge;

4. not all constructions are rule-based and productive, many are irregular in that they are not predictable from more abstract structures ([28], p. 247).

To these basic principles, one might add that CxG—and DCxG in extension—understand language as an ‘emergent’ process compatible with the tenets of the neo-Darwinian evolutionary synthesis. The concepts of divergence and hybridization are also relevant for language change and variation and play out in scenarios of isolation of and/or contact between speech communities. The phenomenon of flexibility in interaction with the environment reflects the capability of constructions to adapt to ever-changing communicative settings and respond to specific communicative needs ([29], pp. 113, 114). DCxG’s multilingual approach puts special emphasis on hybridization and adaptability.

Speakers build the grammar of their varieties through abstraction and generalization. That is, filled constructions that behave identically in some respect instantiate a single, more abstract construction, thus forming the overall network of constructions that is the grammar, in CxG terminology: the construction.

The same principles hold in language contact situations: Interlingual identification of constructions leads to the establishment of shared constructions that are non-language specific in the speakers’ construction, which in turn contains the sum of all language-related constructions. These shared constructions are called diaconstructions. In other words: Multilinguals are assumed to organize their entire linguistic knowledge into one single multilingual construction. This construction includes not only constructions from different languages, but beyond that also constructions that are language-unspecific. Constructions that are specific to a certain language (defined as a shared set of linguistic expressions representative of a certain speaker community) appear as idioconstructions in the multilingual construction [30]. In this model, whether and to what extent linguistic elements are assigned to a single language or shared by several languages is not a question of belonging to one or the other language system, but is determined at the level of the individual constructions that interact with each other in a cross-language construction network ([9], p. 140).

The big difference between DCxG and other attempts to describe languages in contact is the fact that most forerunners remain on the individual speaker level. For instance, Weinreich ([4], pp. 9, 10) already hinted at the possibility of a bilingual speaker identifying different (lexical) signifiers (i.e., phonological forms) from separate languages with one common signified (i.e., meaning) as integral part of one compound sign. On the level of the involved languages, however, the two systems are always kept apart, and contact-induced interference was defined as deviation from the codified norms of speech of the contacting languages. The same holds true for the various theories on code-switching, which—for obvious reasons—have to presuppose that the codes between which the switching occurs belong to separate entities [31, 32].

More recently, however, the notion of the ‘linguistic repertoire’ accounts for structural elements from different languages at the disposal of individual speakers to effectively communicate in a multilingual context ([33], p. 2). While this notion still relies on the idea of separate language systems to account for the make-up of the repertoire, on the individual level the systems need not to be necessarily separated anymore. Languages of such repertoires may have different functions in multilingual communities and are used for different purposes in different contexts. There is also the possibility of a ‘multilingual mode’ (in extension of Grosjean’s ‘bilingual mode’ [34]) in a group of speakers where the various languages of the repertoire are mixed without any clear-cut purposes and some sort of mixed code becomes the default means of communication. This leads to the apparently radical view that multilinguals do not categorically distinguish between separate language systems at all, but that their linguistic knowledge consists of a common ‘repertoire’ of elements and structures for all of their languages and varieties ([6], pp. 208, 209). From this repertoire they then choose whatever is appropriate in the current communicative context. The repertoire approach thus does away with the idea that linguistic structures belong to a particular language system a priori and that languages are pre-existing entities at all [10].

In DCxG terminology, the repertoire equals a single multilingual construction which includes constructions from different languages and language-unspecific ones. While such constructions can sometimes be described as generalizations across pre-existing language-specific ones, this is not always the case, in particular with schematic constructions [27]. For example, speakers who are part of the multilingual ecology of Ngaoundéré may have a language unspecific presentative construction at their disposal that they might fill with lexical constructions from one of the languages present in their repertoire or with unspecific lexical constructions all together (cp. 4). In DCxG, language-specificity is not a feature of the ‘language system’ or ‘grammar’ as such, but rather an optional property of individual constructions. More precisely, it is a pragmatic property and, hence, part of a construction’s meaning along with other pragmatic properties. In other words, whether and to what extent linguistic elements are only assigned to a certain language or shared by several languages is not a question of belonging to one or the other language system, but is determined at the level of the individual constructions that interact with each other in a cross-language construction network ([9], p. 140).

4. A DCxG description of Ngaoundéré-based speaker groups

To substantiate the above-made claims on adequacy in describing speech in a multilingual ecology with the DCxG approach, I introduce and discuss some data that stem from the language consultants introduced above. The first language examples come from the small group of Ngaoundéré-based friends that all work in some sort of mechanic-related business, hence: the mechanics (M). Their education levels are quite divers ranging from no schooling (apart from Koran school) through elementary school (6 years) to college (collège) level (6 + 3 years) which results in quite different levels of competences in the official language French. However, as they all work in blue-collar jobs where French is only sporadically used their main options for communication rely on the indigenous languages they combine in their repertoires.

4.1 Diaconstructions of the mechanics group

The first example for a diaconstruction came to my awareness when I was working with a consultant (M1) who works as a motorcycle taxi driver. In various recordings taken with him, a lexically filled construction signifying ‘dog’ shows up. The first appearances are in the retelling of two picture stories, where dogs feature quite prominently. I had him retell the stories in what he called his ‘normal’ version of his native language Mbum. Throughout these texts, he uses the word form góy, pl. góy-rì. In a later recording, he sits in the roadside motorcycle garage of his friend who just works on some engine problem of a customer. M1 tells his friend M2 how he recorded the picture story about some stray dogs. In this encounter M1 uses several times his ‘normal’ Mbum word but also employs the Adamawa-Fulfulde version of the construction. It is related to Mbum but adopts noun class markers as it is typical for Fulfulde: góy-rù, pl. góy-jì. Also, in this encounter, in one occasion, M1 employs the ‘standard’ Fula version ɓoosaa-ru but corrects himself immediately. Interestingly, in the same recorded encounter, the customer asks a clarifying question about the story, and M2 repeats what has been said, now using again another word for dog: váa. This construction which is from another Mbum-Dialect (Mbum-Pana) is then used alongside the Fula-influenced version góy-rù throughout the remainder of the encounter between M1, M2, and the customer. In Figure 1, the lexically filled diaconstruction for the meaning ‘dog’ in this particular setting of speakers is depicted in opposition to the usual sound meaning pairs accounted for in language-centered grammars. The diaconstruction leaves the actual phonological representation unspecified but depicts the various choices the speakers have to realize it. These choices are governed by the language-specific pragmatic usage of the construction as defined by social features within the groups of mechanics under scrutiny. In the situation described above, the two mechanics converge on their common L1 and the Fulfulde of Ngaoundéré (NgF) for the main part of the conversation which seems to be their usual in-group way of speaking. There are no discernable prompts that govern the choice between the two lexical fillings of the construction. As soon as M1 and M2 realize that the customer is also a speaker of Mbum, although from a different area, they change their in-group Mbum filling for the dialectal version of the customer alongside the ubiquitous NgF-constructions. M1’s slip to standard Fulfulde is immediately corrected as he is aware of this being not part of the appropriate diaconstruction for the situation.

Figure 1.

Schematic representation of a lexical diaconstruction for the meaning ‘dog’ in the mechanic group of speakers in opposition to representations in language-specific grammars.

The next example will show that there are more schematic diaconstructions than the lexically filled ones. It also brings out more clearly the fact that diaconstructions may be language unspecific. When I recorded the already mentioned retelling of the picture stories (about the stray dogs), all respondents frequently used a presentative construction to introduce the individual pictures of the story using a locative preposition in combination with a locative adverb. All five members of the M-group applied at least two versions of this construction that they all considered part of their competence in Mbum. Figure 2 shows that the first element háa is identical in two languages and cannot be connected exclusively to either of the languages involved, at least synchronically. The second element is diachronically related to either Mbum or the local Fula variety, but in my data from the M-group also seems to be freely interchangeable.

Figure 2.

Schematic representation of the presentative diaconstruction in the mechanic group. The scheme itself is language unspecific just like the preposition used. The speakers may choose between the lexical filling ɗo from Ngaoundéré Fulfude or záy from local Mbum.

Not only did the group members use both expressions while retelling the picture stories but there were various occasions in other recordings where both versions were used by the same respondent. Thus, the brackets signaling the language-specific pragmatic function of the local adverb are obsolete in the group of mechanics. It is just a sort of free variation in the phonological shape.

4.2 Diaconstructions in the student group

As already mentioned, the consultants subsumed under the header ‘student group’ are not a defined group with regular interactions among them but more like a loose-knit circle of people who know each other and share a common background in terms of natal village, upbringing, and education.

One of the consultants in that group constructed an example of possible linguistic variation that—according to him—anyone of the students might use in one or the other circumstances without hesitation. The examples (1a) to (1d) [35] show versions of a sentence that range from a ‘pure’ Mbum via two mixed versions to an apparently ‘pure’ Adamawa-Fula sentence. According to the consultants that were asked about this, they all claimed to understand all versions perfectly but would probably use one of the mixed examples when speaking among each other. While (1a) depicts the standard way that Mbum (M) speakers in Nganha would phrase it, examples (1b) and (1c) show versions with different proportions of Fula (F) parts, and (1d) depicts a Fula sentence in Ngaoundéré.

Although it looks as if the examples 1b, 1c are just the usual mixed utterance that we know from so many descriptions of code-switching, there is more to it than just intra-sentential change of language. Synchronically speaking, the 1sg pronoun mì is the same in both languages. Only the fact that it is marked by tone reveals its Mbum-nature (Fula not being a tone language). The tonal marking for clearly Fula-origin elements further hints to the fact that they are used in the tone language environment of Mbum. The emphatic 1sg version mìn is a reinterpretation of the original 1pl.excl pronoun of Fula [35], so far only known from the speakers of the student group. The first verb in 1b, originally from Fula, seems to be derived by the Mbum suffix marking verbal nouns. In terms of standard Mbum grammar, this would, however, result in a slightly different reading as the sentence now consists of two verbal nouns following each other, yielding something like ‘My thinking of the doing’. The same suffix would allow an interpretation as in 1c, where it is interlinearized as the continuous marker from Fula. The example 1d is a bit of a conundrum. Apart from its tonal markings, it looks like a ‘normal’ Fula sentence from Ngaoundéré, but its meaning would then be different from the one the student speakers gave to it. The personal pronoun would be interpreted as 1pl.exclusive, and the i-suffix on the finite verb would be the completive marker. The meaning of the sentence would thus be ‘we (without you) thought of doing (it)’. This interpretation, however, was refused by the consultants who all understood the sentence with an emphasis on the 1sg and in a continuous form. This leaves the question of the nature of the i-suffix in this context open.

A diasystematic interpretation of these possibilities starts with the schematic representation of the progressive aspect employing a finite verb in an imperfective mode followed by an infinite verb form. This is the unmarked choice for progressive aspect construction in both languages, therefore language unspecific. The different outcomes of example 1 are then governed again by the language-specific pragmatic usage of the different constructions as defined by the socio-pragmatic features encoded in language specificity. Thus, the ‘new’ emphatic mìn ‘1sg’ is only available when the construction appears in the context of the student peer group. The filling of the progressive scheme will likewise depend on the addressees of the phrase. In the peer group, the Fula verb tamm- in connection with a suffix -à would frequently be used. Given the overall progressive scheme, our interpretation in 1b seems less likely. Accordingly, the suffix -à probably represents the Fula continuous marker. In an even more Fula-centered context, the final lexical construction might use the Fula filling for ‘doing’ waɗugo in its infinite form.

The baseline of these interpretations is, again, that the members of the student peer group have a shared language repertoire that allows them to adapt freely to the communicative needs at hand. They use a language-unspecific scheme from their shared construction and fill it with appropriate material from said repertoire. Interestingly, they also develop new uses of former language-specific constructions that they reinterpret. For the emphatic pronoun, the cognitive motivation (1pl.excl. (F) > 1sg.emp) is relatively easy to grasp. The reinterpretation of the progressive scheme by using two verbal nouns is more difficult to fathom. These last facts also illustrate how diaconstructions may develop in such multilingual ecologies: Schematic constructions are categorized on the basis of the available input leading to an economic representation of the linguistic knowledge of the speaker group using it. It mostly relies on the cognitive process of interlingual identification of the constructions shared by the members of the specific speaker group.

However, not all outcomes in a specific speaker group-based construction may be rule-based and productive. The mapping of the Fula ‘1pl.excl.’ onto the ‘1sg.emp’ in the students’ group’s construction and even more so the filling of the presentative scheme with a verbal noun at the place of the continuous verb form or the use of a completive in this position shows that they may be irregular in that they are not predictable from more abstract structures.

5. Conclusion and outlook

Given the fact that multilingual speakers are by far outnumbering monolinguals on a worldwide scale, the need to factor in multilingualism into grammar descriptions seems obvious. This is even more pressing in contexts where multilingual ecologies are the norm and the languages involved are more diffuse concepts for the speakers. This is also true—to some extent—for the describing linguists who run into problems in such contexts as the grammatical descriptions of named languages therein need to cope with far-reaching dialectal varieties and contact-induced elements that are often difficult to disentangle as originating from separate defined grammars.

The here-presented DCxG approach looks at grammar from the viewpoint of the speakers. It helps to understand how and to what end they use linguistic elements that—in more traditional descriptions—would be attributed to grammars of different languages. The speaker-centered approach regards grammar as a set of constructions that may or may not stem from different languages but that form the entirety of the grammatical means at the disposal of a group of speakers. This group is defined by social relations and shared experience in the same multilingual ecology. In extension this also means that speech communities in such ecologies may be defined more in terms of the specific shape of their shared diaconstruction than by anything else, e.g., ethnicity, religion, or common ancestry. Examples of such hybrid communities are not only found in small-scale multilingual contexts but also in the ever-growing superdiverse megacities [36], where the DCxG approach may also fruitfully be applied.

The above-presented speaker groups may communicate among each other in a way that would not be possible in other settings and sometimes be regarded as a wild mix of different languages by an observing linguist. Nonetheless, the communication works effectively and should in principle be able to be captured by descriptive grammar. The only way to do this, then, is to describe the speaker group-specific constructions in their entirety and suppress the urge to put a language name tag on its different components. It is just the construction that speakers of group × have at their disposal and that they can adapt readily to any need at hand.