Exploring the complexity of multilingual spaces: embracing diverse perspectives of linguistic non-understanding

1.1 Navigating linguistic (non-)understandings in multilingual settings

In multilingual conversation and research, reflections on linguistic (non-)understanding often hold the key to unlocking deeper insights into the complexities of language use and the nuances of interpretations and understandings. In our article, we investigate multilingual spaces as we harness the power of multi-perspective data and self-analysis through the triangulation method, refined for researching multilingual settings (Weidl and Goodchild forthcoming) to critically reflect on linguistic non-understandings. Our investigation focuses on the sociolinguistic dynamics of diverse and manifold language environments, offering novel insights for researchers navigating their research within multilingual contexts. Situations, spaces, and potential for linguistic non- or mis-understandings are ever-present within social communicative exchange. Linguistic non-understanding is not exclusive to multilingual communication, but multilingual contexts often provide particularly rich grounds for its occurrence due to the varied linguistic repertoires at play. In our study, we apply methods like triangulation and multi-perspective analysis to explore linguistic non-understanding within multilingual settings, though these methods are adaptable across various communicative contexts. In this article, we shed light on the limits of comprehension that all repertoire users encounter, a phenomenon referred to as ‘spaces of linguistic non-understanding’ (from now on LNU) (van Hest and Jacobs 2022). Therefore, throughout this article, we will reflect on the challenges and affordances of incorporating LNU in research practice.

We illustrate the advantages of employing in-depth ethnographic data collection methods and multi-perspective analysis through concrete data examples and sharing of experiences. Additionally, the article addresses the challenges we have encountered and offers critical reflections on our performance as researchers conducting research multilingually (see e.g. Costley and Reilly 2021; Holmes et al. 2022). As qualitative researchers aligning with non-positivist approaches, we embrace the position of being a “non-knower” (Holsapple 2022: 341) as a prerequisite for future understanding.

Following a clarification of the terms and concepts used in the subsequent sections, we will describe our methodology, especially focusing on the triangulation method that has been applied and refined by the authors for nearly a decade, in Section 3. Using data from our ethnographic based studies on multilingualism in the Casamance, we will present various spaces of LNU. The data examples in Section 4 include everyday language use, backed by thick descriptions of how data were collected and analysed, the speakers’ social contexts, realities and spatial relationships, while concurrently reflecting upon our position and analysis as researchers on a meta level, including on-site interpretations and future analysis. Here, we clearly point out that a sole interpretation from a researcher’s perspective includes the risk of misinterpretation and can exclude relevant points in analysis, which, however, can be mitigated through a triangulation approach and “ethnographic understanding of local language practices” (King 2023: 731). Furthermore, we engage in “linguistic reflexivity” (Rolland et al. 2023: 647), whereby we reflect on the language(s) used (or not) by participants and researchers, with a focus on the researcher’s changing repertoires across time and space. Overall, we recognise the repertoire users as experts for their own linguistic practices, and that varied LNU are a part of the complexity of research on multilingualism. In the concluding Section 5 we will present analytical advantages of such an in-depth and multi perspective approach.

1.2 The research setting and reflections on the researchers’ positionality

The setting from which we present our data is itself highly diverse: the data originate from the Casamance area in southern Senegal (see Figure 1), where a large number of (un-)named languages are present. French, the sole official language in the country’s institutional sector, is used in only a small fraction of households, estimated at 15 %, even less so in the Casamance area and is rarely used in its standardised form. Nonetheless, exposure to French is inevitable during upbringing in Senegal due to its omnipresent use in the media, its exclusive use in formal education and its fluid use with other languages in everyday conversations. In contrast, Wolof, the most widely spoken language in the country, with over 90 % of the population using it on a daily basis, holds limited ethnic representation and identity value in our region of interest (see, for example, Swigart 2000; Johnson 2005). Besides French, both English and Portuguese are influential due to the proximity to The Gambia and Guinea-Bissau borders as well as Arabic, for religious purposes. Furthermore, people in the area speak additional languages of wider communication (LWC) such as Joola languages (Joola Fogny, Kaasa, etc.), Mande languages (Mandinka, Sose, etc.), or a Portuguese-based Creole, which can be integral to their identity. Additionally, many much smaller languages exist in the region, often spoken by only a few thousand or a few hundred speakers, (e.g. Baïnounk Gubëeher, Joola Eegimaa, etc.). These languages serve as primary identity markers for many individuals and are considered the patrimonial languages of the villages, meaning they are rather connected to a geographical space and the male founder of a village rather than reflecting the linguistic practices of all inhabitants (Goodchild 2018; Lüpke 2016; Weidl 2018, 2022]).

Figure 1:

Map of research area in the Casamance, Senegal (soascrossroads.org).

In a setting as such, many inhabitants grow up in a highly multilingual manner (cf. Sagna et al. 2022), using translanguaging practices, constantly expanding and modifying their linguistic repertoires. We understand translanguaging as being a theoretical approach rooted in people’s actual linguistic practices, who do not necessarily pay attention to the supposed boundaries of a priori defined (un)named languages (e.g. Li Wei 2018; Otheguy et al. 2015 also see Section 2). We use translanguaging then to describe both the wider theoretical approach and as a way of describing people’s practices, thus if we use the term ‘translanguaging practices’, then we mean specifically the concrete communicative practices people engage in. People engage in communicative practices such as translanguaging, using their linguistic repertoires. We understand linguistic repertoires as (linguistic) resources that the individual can use to communicate, gained throughout their lives, which may be associated with different times and spaces, and emotions (e.g. Blommaert and Backus 2013; Busch 2012, 2015], 2017]; Pennycook and Otsuji 2015). Moreover, we recognise that the linguistic repertoire is fluid, layered and changeable and in studying repertoires, one should also look at what is not used, for example a repertoire may also include (named) languages for identity rather than communicative purposes (see Ducos 1983 on the Casamance context). The person’s linguistic repertoire is inherently individual, yet is tied to wider societal and historical processes (Blommaert and Backus 2013; Goodchild forthcoming). As researchers and authors of this article, we shared only a small part of our linguistic repertoire, common knowledge and cultural understanding (King 2023) with participants in our research settings, however, our repertoires also developed in the process of investigation (see also Section 4).

Interested in various facets of multilingual language use, we, as linguists and researchers growing up outside our research areas, thus consistently found ourselves among the repertoire users most linguistically challenged during data collection. This momentary state however changed during data analysis, where we got detailed insights from various perspectives, supported by video and audio material from our own personal corpora, carried out within the Crossroads project.^[1] We used video as a default for data collection (DuFon 2002). We believe that in all research settings, we have to a certain extent a partial and temporal interpretation of any given event we are studying. As researchers, we encounter limitations in accessing certain information (e.g. due to differing linguistic repertoires) and are influenced by subjectivity and our sociolinguistic perception of the given communicative event as described by Kleinschmidt et al. (2018) and sociolinguistic patterns we might (temporarily) be subject to (Labov 1972). Furthermore, we also view the change of interpretation and analysis throughout “socio-space-time” (Silverstein 2016: 186) and change of values and knowledge as important points to be taken into account during the research process. Additionally, we may be subject to biases or psychological effects such as emotions, stress or aims, which significantly influence our interpretations. To mitigate this, we incorporate others’ perspectives in the analysis of the multilingual settings in question, as will be demonstrated throughout this article. In the following section, we introduce conceptual definitions of multilingualism, linguistic non-understanding and how space aligns with these.

2.1 Multilingualism and linguistic non-understanding

As poststructuralists, we follow a holistic approach on language and the use of language throughout our work, which is the base for the definition of multilingualism we use. We perceive multilingualism as an omnipresent phenomenon that describes the fluid application of linguistic repertoires in different (social) spaces through oral, signed, written and non-verbal expression. Hence, from our point of view, every repertoire user is multilingual as all individuals master the art of aligning their linguistic availability to space, needs and purpose (Blommaert and Backus 2013; Canagarajah 2018). Repertoire users can activate various parts of their linguistic repertoires, which we take to mean accommodating linguistic abilities like languages, styles, registers, and lects, but also connect our abilities to express ourselves to experiences, social, cultural and common knowledge, emotions and individual objectives (Busch 2012, 2015]; Cangarajah 2018). Furthermore, the linguistic repertoire^[2] may also encompass knowledge of language as a named and possibly bounded entity. By this we mean how an official language, e.g. French is taught in formal educational settings. Widespread standardised, named languages are systemic constructs that were deliberately created at specific times in history, serving various purposes within specific geographical contexts, often linked to the process of nation-state building (see e.g. Otheguy et al. 2015; Storch 2016), though many of the named languages in the Casamance region have not been standardised. Therefore we, as researchers, recognise named languages as relevant parts of repertoire users’ daily interaction, however, we rather focus on the analysis of the actual linguistic behaviour and the manifoldness of possibilities of interpretation that provide us with a deeper understanding of language use (Weidl 2022). From this point, we then consider what relevance named languages may have (if at all), in addition to how people name language(s) and their linguistic practices, e.g. whether referring to Joola Eegimaa or Joola, and their associated meanings, e.g. for identity purposes (Goodchild 2018). Our triangulation method (detailed in Section 3) coupled with our understanding of multilingualism above, enables an in-depth investigation of all languages and linguistic practices used in certain spaces and their users’ interpretations, which may transcend strict borders of named languages to encompass a broader linguistic spectrum of non-understandings.

We take as our point of departure in our reflections of linguistic non-understanding, the concept as proposed by van Hest and Jacobs (2022). They focus on spaces of linguistic non-understanding from a researcher’s perspective and the relationship to the researcher’s linguistic resources. They define spaces of linguistic non-understanding as:

[i]nstances in which the researcher, while collecting and/or analysing data, encounters (stretches of) interaction which unfold in languages outside their own repertoire. Whenever research participants converse or communicate in a language that is unintelligible to the researcher, this seems to form a (temporary) obstacle in the face of gaining understanding of the situated practices.

(van Hest and Jacobs 2022: 20)

Even though we try to move beyond assumptions of a priori (named) languages, and recognise the multiplicity and fluidity of repertoires, there are clear parallels between this definition and our research contexts, where people’s repertoires diverge and overlap to varying degrees dependent on their life histories and linguistic trajectories (see Goodchild 2018; Weidl 2018). In addition, our respective repertoires diverge and overlap in different ways with each other and people in the research context, therefore it was likely during the research that we would encounter communicative events where we would not understand all the practices in question. Here it is worth pointing out that non-understanding may also occur because of a lack of shared cultural understanding, even if language(s) are shared (King 2023). LNU can be seen as similar to what Lorette (2023: 658) refers to as “knowledge gaps”, that is challenges to understanding which arise due to linguistic and/or cultural unfamiliarity. We want to point out again that linguistic non-understanding also occurs between participants, and is therefore an important analytical point to consider when studying multilingual linguistic practices, but in this paper, we focus predominantly on the researchers’ perspective, whilst demonstrating how to integrate other perspectives and analysis on our data. Whilst the above has mainly focused on the idea of linguistic non-understanding, in the following we examine the spatial aspect of the concept of spaces of linguistic non-understanding.

2.2 Spaces of linguistic non-understanding (LNU)

Spaces of linguistic non-understanding, as defined above by van Hest and Jacobs, are instances when the researcher does not understand the interaction, as they do not share the linguistic repertoire of participants (they may also not share cultural backgrounds with participants, see King 2023). Even if in the moment of interaction, they do not understand the content, the spaces of LNU do however influence how the researcher understands the event overall. They further specify that “spaces of LNU are temporally dynamic” (van Hest and Jacobs 2022: 22) meaning, they change throughout time and space. Additionally, in their work, instances of LNU occurred during analysis, where for example missing information or “blank gaps” were left in the transcript for non-understandable portions of communication (van Hest and Jacobs 2022). In the future, after data collection, it is crucial for researchers to edit and fill in spaces of linguistic non-understanding. The researcher may do this, e.g. by integrating further perspectives, or by getting translations, to aim for deeper understanding.

In the following, we would like to complexify the concept of spaces of LNU (see Section 4), with a focus on space, which is itself multifaceted and layered. The different understandings of space and related concepts such as chronotopes (defined below) have been taken up within sociolinguistics to recognise the situatedness of communicative interactions in physical space and time. We thus believe that by focusing on the spaces of LNU, we can also demonstrate how and why LNU might come about, as exemplified in our analysis in Section 4.2.

Space has received a renewed focus in some areas of sociolinguistics, moving away from previous associations with the realm of dialectology (Johnstone 2009), towards recognition of it being an inherent part of embodied communicative practices, and that using translanguaging as a theoretical approach, alongside study of people’s linguistic repertoires, requires a revised investigation of spatiality in order to attempt to capture the full complexity of communication (Canagarajah 2018). Lefebvre (1992: 11) defines three different types of space: the physical space, the mental space and the social space. Importantly, physical spaces take on meaning through use, thus spaces “[⃛] correspond [⃛] to a spatial practice that they express and constitute” (Lefebvre 1992: 16). This is also the case for social spaces where (in our research) people interact and practices take place. Furthermore, the fact that space is an expansive concept is particularly useful and aligns with flexible analyses, such as translanguaging. Massey (2005: 9) introduces “space as the product of interrelations; as constituted through interactions”.

When thinking about the multiplicity of possible spaces in our work and how these relate to our focus of studying multilingual linguistic practices and people’s perceptions of those, it has been further helpful to use the concept of “sociolinguistic space” (Juilliard 2016; Liebscher and Dailey-O’Cain 2013). Sociolinguistic space encompasses both Lefebvre’s (1992) and Massey’s (2005) understandings of space, and is further expanded as a theoretical concept to include sociolinguistically salient phenomena and thus includes communication situations, interpersonal relationships, varieties and/or languages, linguistic usage, etc. Therefore, a sociolinguistic space is a theoretical space which occurs in interaction, in a physical space and a social space, i.e. between people, to investigate what (sociolinguistic) features might be available as resources to be used in a communicative context. Importantly, sociolinguistic spaces are expansive, may occur simultaneously and are related to other (sociolinguistic) spaces, whether they are latent or manifest (Juillard 2016: 97). This also entails examining why certain resources are (not) used, for example.

One other spatial concept which has been a useful analytical construct, is that of the chronotope (see Goodchild forthcoming). Chronotope is a notion taken from Bakhtin (1981) which recognises that time and space are inseparable. A chronotope is a representation of a given (or imagined) space and time. Karimzad (2020: 108) defines chronotopes as “semiotized images of times and spaces populated by certain social types that organize social relations and semiotic practices”. These could even be imagined future chronotopes, and people may refer to certain expectations or norms of behaviour i.e. how they acted or were expected to act at a given time and space (Karimzad 2020). Thus, chronotopes have the potential to influence how people interpret their (linguistic) behaviour and perceptions thereof in the present.

For our consideration of space(s) in connection to LNU, the physical space, social space, sociolinguistic space, as well as chronotopic representations of time and space are highly significant. All have the potential for creating LNU in the following ways: 1) The physical space, e.g. a village and physical objects, which could mean not understanding the linguistic practices that happen in a particular space or e.g. written language on a billboard; 2) social space, the people involved in an interaction and not understanding the relations between them and their intentions; 3) the sociolinguistic space, not understanding the sociolinguistic resources used in an interaction, e.g. because repertoires do not align; 4) chronotopes, by assuming that norms and expectations of linguistic behaviour at one space and time will be replicated in future times and spaces may create LNU.

4.1 Contextualisation of the data: triangulating video data analyses

The data presented here is an excerpt from a lengthy collaboration with people living in or regularly frequenting the compound of an established Baïnounk Gubëeher family, in the village of Djibonker, Senegal. The geographical space illustrated below has frequently been characterised as dominated by Baïnounk Gubëeher, which can be traced back to the male founders of the households and their descendants (Weidl 2018: 141–143) and is ascribed with rich social, cultural, linguistic and historical meaning (Canagarajah 2018). Overall, the village presents a high degree of societal and individual multilingualism, with various languages utilised for communication alongside Baïnounk Gubëeher, which serves as the patrimonial language of the village (see Goodchild 2018; Weidl 2018), the identity language traced back to the male founder of the village (Lüpke 2018). The physical space in the first data example (see Section 4.1.1) defies any monolingual characterisation, as the use of languages becomes fluidly adapted to contextual nuances, thematic content, objectives, and repertoire users present. Furthermore, no repertoire user within this space has reported to be monolingual.

Conducting this research, we follow strict ethical rules, all participants provided informed consent for audio and video recording during their daily activities and were clearly informed about data collection. Recording devices could always be switched off by anybody present and data requested to be deleted at a later stage (see also Weidl 2018). Employing a participatory research methodology, Weidl (henceforth MW in data example) became integrated within the household, actively engaging in daily routines and recordings. This approach was essential to foster an environment where all involved felt at ease contributing to the research process and to gain a comprehensive understanding of the complex layers of linguistic and cultural complexity (Bodó et al. 2022).

Following the triangulation method, the data analysis presented below integrates multiple perspectives: the speakers’ insights into daily conversations and language usage, the observers’ viewpoint represented by local research assistant, and the researchers’ reflections on initial and subsequent interpretations of the data in different times and spaces (Weidl and Goodchild forthcoming). The data are expanded through extensive knowledge of all of the individuals’ self-reported linguistic repertoires, reports on their daily language use within different spaces and with various interlocutors as well as attitudes and ideologies towards languages, mainly collected through interviews, conversations and self-analysis, ​where individuals reflect on their own linguistic behaviour, abilities, emotions, experience and beliefs (see also Weidl 2018).

Throughout data collection and analysis, researchers navigated various linguistic, social and cultural challenges, encompassing non-understandings across languages, cultural nuances, differing shared common-knowledge, and misunderstandings within social contexts. To address these challenges we established close collaborations and the integration of perspectives. However, some of the non-understandings might never be fully resolved as interpretations differ and views change over time and with new knowledge gained. Gaining a better comprehension of the context, linguistic repertoires and conversations proved highly complex; as such, we invite readers to navigate through a space of linguistic non-understanding as we incorporate numerous perspectives on different facets of the data along the progress of this article.

4.1.1 Context and participants

On the recording, showing the building site above (see Figure 3), three people are visible – TS04, a construction worker from a nearby village, LOG who is from Guinea Bissau and married into the family and IPS, born in the region’s capital, who is a regular guest and friend of the family, at that time living in the closest neighbouring village. MW is in the household as a researcher, however well integrated in day-to-day life. JS, the only person not present during data collection, is a research assistant responsible for the annotation, transcription and translation of the video data. He was born in Djibonker and lived there and in neighbouring villages most of his life. JS was recruited as an administrative secretary and research assistant in the Crossroads project^[4] due to the fact that he was a Master’s student in sociology with a broad knowledge of languages and cultures represented in the area, IT skills and a great organisational talent.

Figure 3:

Data collection setting.

Self reported linguistic repertoires: ^[5]

TS04: 6	Bayot, Joola Kaasa, French
LOG:	Njago, Baïnounk Gubëeher, Balante, Joola Fogny, Joola (?),7 Kreol
IPS:	Arame, Kreol, Wolof, Joola Fogny, Sose, French, Bayot, Joola (?)
MW:	German, English, Wolof, French, Baïnounk Gubëeher, Joola Banjal, Finnish
JS:	Baïnounk Gubëeher, Joola Banjal, Wolof, French, Joola Fogny, Joola Kujireray, German, English

1. 6

   TS04 reported only three languages, however, MW conducted only one interview with him and repertoire users tended to represent an identity that they considered to be appropriate in the given situation. With all participants, some languages were only reported after longer cooperation due to trust that was established (see also Goodchild forthcoming) but also deeper reflections on their own linguistic biography, often triggered by questions that were asked.

2. 7

   In this region, repertoire users speak many different Joola varieties, which are mutually intelligible to varying degrees. Neither the researchers nor the transcribers felt competent to distinguish between the varieties; therefore, we decided to either use the expressions used by the speakers themselves or refer to the language as ‘Joola’, which actually encompasses several languages and varieties, or could represent a translanguaging practice (Goodchild 2018).

The data set was collected on a day when only TS04 and IPS were working on the building site (see Figure 3) and LOG and MW were sitting close by most of the time, preparing food. LOG’s husband as well as his brother (for whom the house is being built) were away that day which made LOG the responsible consultant for decisions about the building site.

The three repertoire users of the below excerpt report to share parts of their linguistic repertoires, with preferences to use certain languages. MWs linguistic repertoire also corresponds to some parts, however, our cultural and contextual common knowledge varies more strongly than among the other three speakers. For analysis, the observer was included, his repertoire is again relatively close to the ones of the repertoire users, he is also very familiar with the social and cultural space and knows two of the speakers.

As multilingualism researchers, we anticipated encountering rich diversity in our data, along with interactions or circumstances that might be hidden, forgotten or ignored from perception, analysis and even transcription (see also Blommaert and Rampton 2011, 2016]). Through our framework, the inequalities of linguistic understandings and resources are addressed to a certain extent, facilitating fluent interpretation and co-construction of meaning (Jungbluth 2016). Furthermore, we analyse the linguistic data going beyond named languages, which means that we hold the repertoire users’ view on language(s) and semantic interpretations as equally important to any more broadly used standards (Otheguy et al. 2015).

In the analysis presented below, we place particular emphasis on delineating the diverse levels of LNU as perceived by the researchers. We present the strategies employed to navigate these complexities within the framework of fluid multilingual data analysis. Moreover, we invite readers to envision the type of analysis that would have resulted had we not integrated the diverse perspectives on the data.

4.1.2 Audiovisual data excerpts, transcriptions and translations

Given the complexity of the situation at hand, the transcript below also represents several layers of interpretation. Next to the line numbers (01–20), the speakers and the addressee of each utterance are indicated, followed by the transcription and translation from the observer’s perspective and, if applicable, an additional interpretation from the researcher’s perspective. Underneath each line, there is a translation into French provided by the observer and an English translation by the researcher. This translation to English had to be based on the French translation provided by the observer if the researchers did not have the languages as part of their repertoires. However, if the original language was understood by the researcher, the translation was based directly on that, and therefore might not align with the ‘expected’ translation from French to English for speakers of these languages. This further clearly shows that individuals translate and interpret the same words and phrases in varied ways, and even the same person needs to adapt translations and interpretations across different times and spaces e.g. according to changing linguistic repertoires. Unlike in previous publications using English translations only, we believe that omitting this step of translating French to English can result in significant loss of content, hence we aim to minimise LNU and misinterpretation. To highlight the languages used in the transcriptions following the observers’ and researchers’ perspective Joola is black, Wolof bold, French underlined and Kreol gray. Any additional information describing relevant actions (for this analysis) in the video are indicated in extra lines, corrections for some of MWs translations at different points of time, were added between square brackets. Lines 5, 7, 15, and 16 show empty spaces in the table. In these instances, the observers did not hear or transcribe utterances and phrases that the researchers could hear in the recording; therefore, they were added by the researchers.

01: Data Example.

DJI260116MWa_cut02a (04:26-00:04:54 and 03:33-03:48)

During data recording sessions, MW took notes and then transferred smaller parts of the recordings to the observers, who transcribed. MW then reviewed the transcriptions and translations, and questions were discussed with the observers and the speakers, with a focus on recognising that multilingualism is just one aspect of the complex diversity that we seek to comprehend (see e.g. Heugh 2021). While researching and analysing multilingually, through critical self-reflection, we have found that not only do erroneous transcriptions and translations lead us to draw inaccurate conclusions, but also that our aims and objectives in conducting the research significantly influence our interpretation of the data (Costley and Reilly 2021; Weidl 2022).

The LNU for the researcher occurred and evolved on several levels in this example. Firstly, during data collection, MW was excluded from information conveyed in ways that were not part of her linguistic repertoire, as well as from certain cultural and social knowledges shared by others. Later, with the assistance of transcriptions and translations, the process of meaning-making proceeded for the researcher; however, some parts of the conversation remained (and still are) ambiguous. In this regard, we endeavour to incorporate additional linguistic and interpretative dimensions into our analytical practice as researchers (Costley and Reilly 2021) to enhance linguistic comprehension and enable multi-perspective accuracy. Furthermore, MW seems to have inadvertently created linguistic non-understanding (LNU) herself through a process of repeatedly working on the same data for different purposes. Additionally, all participants involved appear to reinterpret certain sections differently at different times and in different spaces.

Given the depth of our investigation and the abundance of data collected regarding this brief example, not all interesting facets of the data can be discussed here. The main discussion revolves around the level of a window or hole in the wall, involving TS04 and LOG, who primarily use Joola varieties to communicate. At the time, MW spoke only a few words in Joola and primarily used Wolof in the household, which generally facilitated some understanding of the context. Only days before this data collection, MW interviewed TS04, who preferred to be interviewed in French and for example reported not to speak or comprehend Wolof at all. During the interview, MW and the observers anticipated that this preference likely reflected strong identity attitudes and ideologies (a widespread fear of a prevalence of Wolof, see e.g. Keese 2016) rather than describing his proficiency in Wolof.

In the Joola part of the conversation, MW was excluded from comprehension on the spot. However, in her notes, she already mentioned that it seemed to trigger laughter in LOG and IPS if TS04 used Wolof (from the researcher’s perspective), and even before receiving the transcriptions, the video recording made clear that the discussion was about a window. Later in the transcriptions, MW’s LNU was complemented and she understood that in line 03 TS04 (temporarily) forgot the expression for ‘to be dark’ in a Joola language, and therefore, after a pause, used Wolof after a very clear gesture showing the closing of window shutters. This made LOG and IPS laugh out loud, and later they explained that they found it amusing how TS04 insisted on not speaking Wolof in front of MW but then couldn’t avoid using it. This is a clear example of a situation in which we as researchers not only influence the interpretation of the data after collection but also the communicative interaction itself, specifically, the choice of language considered to be the ‘correct’ one in the given space.

In 03, JS transcribed the phrase “o nane sire maa pan lëndëm” (French: il dit si c’est ce niveau il fera sombre) and as a response in 04 “ude wa ban lëndëm?” (French: quoi il fera sombre? [ça ç’est quoi, il fera sombre?]) which was understood differently by MW, with a differing meaning of the phrase. In 03, MW hears “o nane sire maa dafa lëndëm” which not only adds a Wolof grammatical structure but also changes the temporal aspect of the sentence from a future (Joola: pan; English: ‘it will be’) to a present construction (Wolof: “dafa”, English: ‘it is’). Again, “dafa” was perceived by MW in 06 and not by the observer JS. In 04, JS transcribed “ude wa ban lëndëm”, again a mixture of Joola and Wolof, which puzzled MW, as she hears “tu wax quoi, lëndem?” (English: ‘you say what, dark?’) in a mixture of French and Wolof, and set off a lengthy discussion with JS. In the discussion, he said that his translation to French is incorrect, and that “ude wa” is closer to “li lan la” in Wolof (E: ‘What is it’) than “quoi” in French. Our discussions about the data always happen in a highly multilingual fashion, trying to find the best fitting translation in other languages. Trying to resolve the issue, we had a short query with LOG, who said that both are possible, meaning this phrase will remain unresolved if our aim is to clearly determine languages. This example shows that interpretations vary even after lengthy considerations, and that people have different interpretations of the same situation. What is clear is that the use of Wolof was marked and triggered a reaction, which would not have been comprehensive without rich background knowledge of the situation.

In lines 10 and 12, IPS, who had moved and was no longer captured by the video, articulates thoughts concerning fuel while repeating the lexeme gazoïle in Portuguese based Kreol, which although not visually apparent, is accompanied by a distinct pouring sound discernible in the data file. JS transcribed the term as gazoïle in both the original utterance and subsequent transcription, employing an adapted spelling of “gasoile” or “gazole”. Previously, MW translated this term as ‘petrol’; however, contemporary reflection prompts her to consider ‘fuel’ a more apt designation. Upon reanalysing the data for this publication, MW was struck by the presence of fuel at the construction site on that particular day. Recollecting the circumstances surrounding the recording, MW recalls the absence of any machinery or vehicles requiring fuel, thus leading to a temporary non-understanding. Only upon thorough review of notes from that day did MW uncover a brief notation indicating her bringing palm wine along, a locally much appreciated beverage often sold in gasoline cans. Consequently, the audible discourse and subsequent commentary by IPS pertained to the act of pouring a drink rather than fuel-related matters, which changed the understanding. Interestingly, this LNU only occurred later, in a different context (several years after data collection at a desk in a European university), whereas the situation was initially clear and comprehensible during data collection, and likely remained so for a considerable period thereafter. However, as MW’s awareness of events evolved over time, what was captured in the digital space proved insufficient for the initial understanding.

It is further interesting that pan/ban was written differently in 03–05; however, it refers to the same lexeme according to the observer who did not interpret the changes in writing as a mistake, but rather as an adaptation of the Joola varieties used by (or attributed to) the speakers. TS04 is perceived by JS as a Joola Fogny and Bayott speaker, languages in which speakers often use unvoiced plosives (p, t, k) where other Joola languages use voiced plosives. LOG was rather analysed as a speaker of another Joola variety (Joola Eegima), which tends to use voiced plosives (b, d, g), and was JS’ description of the choice of writing. This explanation was not surprising as MW shared this knowledge about Joola languages, however, the ban could also have been interpreted as Wolof and could also mean ‘which’ when translated from Wolof. Furthermore, at least for MW the distinction between unvoiced and voiced plosives is not always clear in this recording.

Other instances of LNU included missing expressions in the transcriptions, such as mhm (05), voilà (07), force (15), and force de (16), as well as, as mentioned before, missing lexemes in Wolof and French from the researcher’s view (03, 06, 12, 15, 16, 18, 20). In line 15 and 16, the expression forcé (de) is used, which MW transcribed as forcé and not forcer (with the same pronunciation in French) since MW is familiar with the term being used as an imperative expression (similar to aie forcé in the past imperative) rather than in the infinitive. The expression is used (with a different semantic meaning to standard French) to convey ‘to be forced, to be obliged to do something’ with an additional Wolof adverb, expressing reinforcement. JS missed these expressions, and in a discussion with him, he explained that he didn’t want to work on the Wolof phrases too much, as MW was ‘the professional’ in standard Wolof and could correct the phrases herself. These additions to the transcript were crucial for the analysis; however, one should not overlook that MW can only perform these additional analyses for items that are part of her linguistic repertoire, potentially resulting in the omission of aspects in other languages.

A further interesting example is line 20, where JS transcribed sa jëkkër gi, which was double-checked by MW since the wrong noun class determiner gi is used in Wolof. Whereas IPS speaks a Wolof variety close to the standard of the north, JS speaks a southern variety in which the noun classes are diminished. JS explained that he did not hear bi, which is the general noun class used in the south for (nearly) all singular nouns, and therefore thought that gi would be the correct noun class. Listening to the data again, JS and MW, however, agreed that IPS used ji, which would be the noun class also used in standard Wolof. Additionally, JS also did not transcribe IPS’s slip when he said sa jaber, ‘your wife’.

Many of the lines would today be translated differently by MW, who not only improved her French skills since the first translations were done but also realised that, by working continuously on the same file, small adaptations and proofreading changes might have been made to the English translations, resulting in a falsification of the results. For example, in line 03, MW seems to have translated the French phrase il dit si c’est ce niveau (⃛) out of context as ‘he says if it’s this level (⃛)’, whereas it became very clear that the (more) correct translation would be ‘it means if it’s this level (⃛)’, both being possible translations in English of the same phrase in French. However, in line 09, a negation was lost, and since I did not use the French translations for a while and my Joola knowledge was not sufficient, this mistake went undiscovered for a while, leading to LNU and a loss in the meaningfulness of this phrase, which was caused by MW herself.

4.2 Integrating future spaces of linguistic non-understanding

In line with recent developments in researching multilingually, in this section we investigate the dynamics of being a multilingualism scholar and engaging in reflective practice through turning an analytical lens to our own multilingual repertoires and multilingual fieldwork practices (e.g. Holmes et al. 2022; Phipps 2019; Rolland et al. 2023). The example we use here is based on Goodchild’s (2018) PhD project. The fieldnotes for this study totalled over 30,000 words and could be considered to have been created using translanguaging practices. Medina Riveros and Austin (2022: 294) encourage “researchers to develop and perform a welcoming mindset toward their own and their participants’ multilingual resources”. In researching multilingually, and during the process of note taking, Goodchild did not pay strict attention to named languages or language borders, although some might identify a predominance of French and English in the notes, and items which could be identified with other named languages, such as Wolof. Yet in 2023, and six-to-eight years after the original notes were made, a problem became apparent – Goodchild no longer understood some of the fieldnotes, and thus sought a deeper reflection and investigation.

At the time, Goodchild had also been working on a paper concerning linguistic biographies and changing repertoires across time and spaces (Goodchild forthcoming), reflecting on how participants’ linguistic resources may be forgotten through non-use in their present times and spaces, but could remain latent to be reactivated or not. Goodchild and Weidl regularly have joint online working sessions, and during one of these Goodchild brought up the example below as an instance in the notes which she no longer understood, thus invoking another perspective to try and overcome an instance of LNU which had arisen. The instance in question was the phrase Ma kaay on panneau. In Figure 4, this note is shown in the context of messages sent during the online working session in question. Goodchild also provided the surrounding notes for context, although they did not help in understanding the referent of the note. From the context of the note, it seems that during field work Goodchild interpreted ma kaay as Wolof, and then noted that wouldn’t be said in Wolof (Ça se dit pas! – On dit ma dem mapis [mais] pas ma kaay! ‘that is not said! – you say I’m going [not] I’m coming!’). Because in literal translation ma kaay means ‘I’m coming’, but that is generally regarded as ungrammatical or infelicitous and most users would say ma dem, which literally means ‘I’m going’. Weidl confirmed this interpretation. Yet, it was still unclear what the whole note referred to, as last year Goodchild interpreted panneau to be a verb, due to the presence of on, i.e. the 3.sg pronoun in French and due to the French in the surrounding notes (note these are Goodchild’s interpretations of language(s) used and named). Goodchild considered whether panneau could indeed be a noun meaning ‘sign’ or whether there was a verb missing or other missing information, as the note still did not make sense.

Figure 4:

LNU field note, Zoom work meeting between Goodchild and Weidl.

Later Goodchild was reviewing photos taken when documenting the linguistic landscape of one of the research areas and remembered that the note did indeed refer to a sign. Then Goodchild found the following photo (Figure 5) and messaged Weidl (Figure 6), after realising that the on was meant to be the English preposition ‘on’ and not the French pronoun on:

Figure 5:

Photo of stop sign with Makaay written in the dirt.

Figure 6:

Message sent from Goodchild to Weidl on WhatsApp.

Goodchild had transcribed makaay from the road sign with a space as ma kaay, where on the sign there was not one. We found it interesting that someone would mark the sign in this way, as although it is a temporary way to write a message, it was in the dry season and could be expected to be there until the wet season many months later. Looking at the date of the photo, December 2015, led to the final analysis of the sign, many years, linguistic practices, and spaces later. In Djibonker, Brin and neighbouring Casamance villages, there is a great variety and fluidity of repertoires which are reflected in non-standardised writing practices (Weidl et al. 2022). Although standardised orthographies and practices exist, often associated with named languages, e.g. French and Arabic, these are often the model base with which to write in other Senegalese languages, by using their sound-grapheme correspondences. After further discussion concerning the physical space of the sign, which was on a busy road in the village of Brin, many people would in theory pass by. As the photo was taken in December 2015, Weidl suggested that the text Makaay could stand for ‘Macky’, i.e. Macky Sall who was President of Senegal at the time and had proposed a constitutional referendum which was held in March 2016. Many years later, in a different place, with a new perspective, Goodchild had reached a plausible level of understanding of the fieldnotes and object in question.

Having begun fieldwork in a state of what Phipps (2013) terms “linguistic incompetence”, Goodchild had also begun by focusing on e.g. objects in physical space as a way to situate herself and her developing understanding (Phipps 2019: 40), moving away from incompetence towards capability, in some language(s) practices, and as a researcher (Phipps 2013). Therefore, at the time, Goodchild felt capable of making a field note concerning the non-understanding of the sign, based on a developing linguistic capability, which actually served to further compound the non-understanding of the item to be analysed. Although at the time, Goodchild must have considered the notes enough to be later understood, whether they might be considered to be, or not, in line with Victoria’s (2022: 220) recommendation when working multilingually to make “detailed and meticulous field notes and transcripts”, Goodchild forgot to create a link in the NVivo project between the photographic data and the field note, thus resulting in a first instance of linguistic non-understanding, i.e. what the note referred to.

The second instance of non-understanding occurred through no longer understanding the multilingual linguistic practices in the note itself. As Medina Riveros and Austin (2022: 294) put forward above, researchers should practice an open mind to their multilingual resources. At the time of fieldwork, Goodchild certainly tried to work in this fashion. From analysing this reflective example in a future space of linguistic non-understanding, it became apparent that it would be beneficial to exercise this welcoming, perhaps as well accepting-forgiving, mindset regarding the changes in her linguistic repertoire and multilingual researcher capabilities. Accepting the (renewed) state of linguistic non-understanding and returning to being a language non-knower and given that analysis is an ongoing process, Goodchild recognised that it would be necessary to integrate other perspectives to move towards understanding, thus called upon Weidl for help.

In this future virtual space, and at desks in Norway and Austria, we attempted to seek a repertoire users’ report on the data and note in question. But in this instance, the repertoire user was absent, and not known to us, nor was the context of the language use known. We did not know who wrote Makaay on the stop sign, nor why. Thus, signage in public space had become a space of LNU (van Hest and Jacobs 2022: 22). Because of the temporal and spatial dislocation involved in analysis for the researcher, a full triangulation of perspectives (according to the model in Section 3 above) therefore proved impossible, but nonetheless another perspective was beneficial to jointly work towards linguistic understanding, which only came about through understanding the physical space-time of the photograph of the documented stop sign and by integrating the plurality of perspectives on writing practices.

In the above, numerous spaces and times of linguistic non-understanding coincided which ultimately resulted in a joint analysis through integrating different perspectives and a certain understanding of the example above:

1. A physical space of LNU. A material object in physical space (Lefebvre 1992; van Hest and Jacobs 2022) was subjected to a linguistic practice, the writing on the stop sign, in the village of Brin which was not understood by Goodchild at the place of data collection.

2. A social space (Lefebvre 1992; Massey 2005) of LNU. At the time and space of data collection, the author was (and remained) unknown and the social motivations to write Makaay on the stop sign were not understood.

3. A sociolinguistic space (Liebscher and Dailey-O’Cain 2013; Juilliard 2016) of LNU. Because of the space where the data were collected (the village of Brin), Goodchild interpreted Makaay as ma kaay and to be an (incorrect or unusual) Wolof phrase. In creating fieldnotes, a sociolinguistic space was created where Goodchild’s repertoire was used for translingual communication with her future self.

4. Chronotopic (Bakhtin 1981; Karimzad 2020) LNU. Actual multilingual practices in a previous time and space, that is a research trip to Senegal in 2015, and an imagined future chronotope of a researcher who has the same multilingual repertoire to be able to understand the notes, do not align, creating LNU.