Translanguaging as a resource for meaning-making at multilingual construction sites

: In this article we investigate spoken professional interaction at construction sites in Sweden, where workers from Poland, Ukraine and Estonia are temporarily employed as carpenters, ground workers and kitchen installers. We study how the workers use resources associated with different languages and how these resources are mobilized along with embodied resources for meaning-making. The analysis aims at investigating what social space the workers construct by going between or beyond different linguistic structures, as defined in the theory of translanguaging. The study is based on Linguistic Ethnography and Conversation Analysis is used for close analysis. We focus on instances of translanguaging, such as Swedish-sounding institutionalized keywords, practices of receptive multilingualism and the search for communicative overlaps in repertoires. The findings from busy construction sites show that the stratifying aspect gives some workers a voice in the organization, while others remain silent. Hence, it is primarily professionals functioning as team leaders, who talk to different occupational categories and use resources associated with different languages. The data provide an opportunity to investigate the theory of translanguaging and its transformative power in relation to professional settings that are linguistically diverse, but also strictly hierarchical.

Translanguaging as a theory represents a shift in the ideology of language. Rather than seeing languages as discrete systems upheld by linguistic criteria, languages are considered social constructs. Blackledge and Creese (2017: 252) explain that "[a] translanguaging lens proposes that, rather than making decisions about which 'language' to use in a particular social setting, people have a linguistic repertoire from which they select resources to communicate." A repertoire is an accumulation of all linguistic and semiotic resources at a speakers' disposal (Busch 2012) and it is acquired over the course of a life trajectory, as individuals "take on board bits of any language available in strategically diverse ways in order to achieve a (localized) communicative function" (Spotti and Blommaert 2017: 172). In professional settings, such as construction sites, repertoires are related to the professional register, i.e., linguistic resources are socially organized in relation to the specific aims and tasks relevant for the profession (Pennycook 2018). For instance, we can observe how a Polish carpenter working in Sweden learns how to name certain tools in Swedish and includes them as professional institutionalized keywords (e.g. Franziskus and Gilles 2012) into his repertoire among, inter alia, professional expressions for dimensions and spatial relations.
Translanguaging as a theory also challenges monolingual norms (Wei 2018). The theory criticizes earlier assumptions of bilingualism as full competence in two languages, claiming that speakers may draw on any kinds of resources that are useful and accessible to them, with varying degrees of fluency. A broader linguistic repertoire provides a larger space for acting socially and translanguaging can even transform power relations, as new configurations of language practices generate new subjectivities, understandings and social structures (Wei and Lin 2019).
Our focus settings, construction sites, are different from those considered in most earlier studies on translanguaging (for a contrastive discussion on school settings see e.g. Creese and Blackledge 2010;Paulsrud et al. 2017). In our study, communication is subordinated to manual work and most professional tasks are solved with minimal verbal interaction. Another important difference is that construction sites tend to lack explicit language policies, which are customary at schools. Wei and Zhu (2013) note, however, that the notion of translanguaging is particularly relevant to studies of the multilingual practices of transnational individuals and groups as they move across space and time: "It enables us to show how everyday practices and identities are profoundly rooted in the developmental trajectories of the communities to which the individuals belong, and how they constantly shift, develop and transform." (2013: 520).

Translanguaging, multilingualism and manual work
Only recently, scholars have begun to show interest in translanguaging as a resource in individuals' professional repertoires. These studies point out that professional multilingual practices gain meaning in relation to the objectives, participants, settings and interests in a particular context (e.g. Franceschi 2017; Räisänen 2018). For example, in a study of city market interaction, Blackledge and Creese (2017: 266) demonstrate how the salesperson "tries out the word not typically associated with his background", echoes the customers' language, deploys a strategically simplified version of English and uses transcultural gestures, such as pointing and measuring with his hands, for achieving communicative goals. Similarly, at a cross-border supermarket in Luxembourg, studied by Franziskus and Gilles (2012), two colleagues speaking German and French rely on a limited passive competence of their partner and speak their first languages, which is considered a reduced form of Translanguaging at multilingual construction sites receptive multilingualism (Ten Thije et al. 2012). In addition, institutionalized keywords associated with other socially defined languages are observed in the pragmatic context of the market (Franziskus and Gilles 2012: 63). Studies of multilingualism at construction sites are gaining ground, although not necessarily from a perspective of translanguaging. Studies of larger sites in the sector demonstrate that manual workers are organized in teams recruited from limited geographical areas and that cross-linguistic contacts are rare (Gonçalves and Kelly-Holmes 2020a;Matous 2011, 2015;Kraft 2017). Kraft (2017), as well as Handford and Matous (2015), show the importance of a language broker in these settings, a person who enables communication between different groups and conveys instructions from the management to the workers. As a consequence, cross-team communication is taken care of by persons in certain positions at the construction sites (Söderlundh et al. 2020), while others only speak to individuals within their own team, often recruited from the same country. In fact, talk may be scarce overall, possibly because of the nature of the work tasks and due to loud noise. Interactionally, professional communication at construction sites features abundant deictic expressions and extensive use of embodied resources (Handford and Matous 2011). Yet, there may be room for multilingual interactions and employment of a broad linguistic repertoire, as demonstrated by Theodoropoulou (2019). In her ethnographic study of blue-collar workers from India at a site in Qatar, she demonstrates that workers may adopt reciprocal strategies and multilingual and multisemiotic communicative repertoires to accomplish mundane activities. In this article, we will chronicle similar multilingual and multisemiotic occasions, arguing that they provide an opportunity to develop the theory of translanguaging spaces in the blue-collar professions.

Data and method
Data were generated as part of a three-year project aimed at investigating professional language practices among individuals commuting to Sweden from Eastern Europe (mainly Estonia and Poland) for manual work within the construction sector. Poland takes the top-position when it comes to the number of posted workers in Sweden (Batsaikhan et al. 2018). Estonia is also a country from which many workers commute for work, although first and foremost to Finland but also Sweden. The project adopts Linguistic Ethnography as the overall methodological framework, a methodology that aims at "tying ethnography down" with concrete examples of interactional data and "opening linguistics up" through rich ethnographic analysis and interpretation (Rampton et al. 2004). Data for the present study consist of ethnographic observations of three construction sites and video recordings of actual, naturally-occurring interactions involving workers from Poland, Estonia, Sweden and Ukraine. The videos were recorded with small GoPro cameras that were easy to carry around in the complex work settings, such as when climbing stairs under construction (site large below) or filming from the edge of a deep hole in the ground (site medium below). In most cases, we were two researchers at the site, with different linguistic competencies: Estonian and Russian (Keevallik), Polish (Weidner), Swedish (Keevallik, Kahlin, Söderlundh), English (all four). A translator has been engaged for transcribing the recordings of Ukrainian workers. Our skills as speakers of different socially constructed languages were crucial for establishing trust at the sites. The study is approved by the Swedish national ethic committee and all participants gave consent for the collected data to be used for research purposes. To ensure anonymity, all names have been replaced with pseudonyms.
The three sites were chosen due to their transient characteristics and variable size. However, the sites also represent different types of construction work: building residential houses, repairing a sewer pipe and installing a marble countertop in a kitchen. The first site is the largest one, with more than one hundred workers from different countries and the work there is stretched over several years. The second site is a middle-sized one, with a team of four Polish workers and two Swedish craftsmen. The third is a small site, with a small number of teams from different countries and companies. We observe an Estonian installer of kitchen countertops, who knows no other person at the site and works alone (see Table 1 for an overview of the collected data). As demonstrated elsewhere (Söderlundh et al. 2020), the size and organization of workers into teams influence whether they will collaborate and talk to workers recruited from other countries or not. Teams are recruited through subcontractors and they often seem to come from Translanguaging at multilingual construction sites the same region in Poland, Ukraine, Estonia etc. At site large and medium, team leaders are the ones involved in interactions with individuals outside of their team, and it is also predominantly team leaders who mobilize resources associated with different languages for meaning-making in our recordings. This makes the size of the workplace an important aspect in our study.
We target interactions involving workers recruited from different countries, subsequently investigating these interactions for sequences where participants mobilize resources conventionally associated with different languages. To document how the resources are mobilized, we use conversation analysis (CA), a qualitative method summarized in, e.g., Sidnell (2010). Given its solid anchoring in actual interaction, it can be deployed for documenting linguistic realities in complex multilingual environments and the interactants in situ, capturing whatever resources they use for sense-making here and now. This is in line with the overall endeavor of translanguaging, to respect and to nourish whatever resources people find useful for meaning-making. Furthermore, it is specifically developed to analyze multimodal aspects of interaction, which is why we use the transcription system by Mondada (2014). Multimodal aspects, such as gaze, space and touch, are essential devices of sensemaking in manual work (Wagner 2018). The moment-by-moment analysis of how translanguaging practices emerge and are acted upon by co-workers enables us to anchor our discussion in the participants' own perspective.

Translanguaging practices at the sites
In what follows, we demonstrate how resources associated with different languages are mobilized along with embodied resources for meaning-making. As we are interested in what social and professional spaces the workers construct by going between or beyond different linguistic structures, we also pay attention to the local interactional function of these resources. We move from the larger to smaller work sites. First, we illustrate a rather common phenomenon at the sites, the use of professional and institutionalized keywords associated with Swedish. Second, we analyse how the co-present workers rely on partly overlapping repertoires as well as display receptive multilingualism. Besides work tasks, these linguistic skills may be used for managing interpersonal relationships, which is our third focus of analysis.

Site large: residential buildings
The organization of the large site is hierarchical, with a Swedish management at the top, subcontractors and team leaders in the middle, and manual workers spread out at the site. As mentioned above, team leaders have a special task in this larger organization as they regularly function as language brokers. The Ukrainian team leader, called Vanko, is involved in several examples from the site, and we will show how his position in the hierarchy is one where mobilization of linguistic resources associated with others can have transformative power, and where linguistic flexibility is essential.

Professional keywords
At the large construction site professional keywords identified by the Swedishspeaking researchers as Swedish are a resource regularly mobilized in English lingua franca conversations, especially by the team leader Vanko when interacting with the Swedish management. We recorded more than 10 items: Some of them are generic construction terms, such as regel 'square bar', bänk 'countertop' and flytspackel 'screed', while others require specific institutional knowledge, such as UE-möte 'subcontractor meeting', fogkille 'seal guy' (the guy installing the fire seal). These keywords seem to be a resource in an active professional register used between speakers with different repertoires. Interestingly, the workers recruited from Ukraine also use the Swedish word flytspackel 'screed' in their internal communication, which is otherwise mostly in Ukrainian: "вони положили кусок мазоніту і залили flytspackel 'they put a piece of masonite and poured screed'. It is highly unlikely that these institutional keywords would be useful outside Scandinavia. Rather, this can be considered evidence of a place-bound linguistic accommodation and an example of how the linguistic repertoires of the commuting workers expand and change over time, according to the situational communicative needs that they encounter (see Spotti and Blommaert 2017).

Translanguaging for managing interpersonal relations
While the use of terminology may be triggered by the necessity of precise reference to locally specific items, resources associated with other people may also be mobilized for social purposes. In Example 1 below, a code switch (as described by García 2009) happens at the closing of a longer instruction by the Swedish manager Peter to team leader Vanko, in which Vanko and his team are assigned to drill holes between the floor levels. In this case, the switched item is not a term but an assessment, used to positively evaluate the manager's report, which has been flattering to the Ukrainian team. It is them and not "the plumbers" who will be assigned the task. After Peter's summary, Vanko uses a phrase associated with Swedish and says mycket bra 'very good'.
Translanguaging at multilingual construction sites Example 1: Mycket bra 'very good'. Peter = Swedish manager, Vanko = Ukrainian team leader. Assessment in Swedish, line 06 and 07 in bold.
The assessment mycket bra 'very good' (line 6) functions as an acceptance of the instruction order, and it is as such also an invitation to close the sequence. Assessments typically occur in evaluations of task sequences in institutional interactions (Lindström and Heinemann 2009) and are used to indicate termination of the encounter and to coordinate the transition from one task to another one (Antaki 2002;Lindström et al. 2019). The use of linguistic resources associated with Swedish is here playful, accompanied by a smile, amounting to a multimodal positive evaluation of the announcement. Thereby the participants orient to the phrase mycket bra as a switch of code as well as a friendly suggestion for a closure. Nevertheless, the sequence involves transformative and empowering aspects, as Vanko employs resources that are associated with his superior and receives positive feedback (through a repetition and a smile). The phrase is repeated by Peter (line 7), which serves as a confirmation of the negotiation as finished but likewise as an appreciation of Vanko's choice of words.
In sum, at this highly complex work site, particularly Vanko, being the leader of a Ukrainian team, practices translanguaging by using linguistic resources associated with his coparticipants. The practice involves social and transformative aspects, as it builds rapport to the Swedish management via linguistic accommodation.

Site medium: sewage repair
The project at site medium is restricted to a deep hole in the ground, with pipes to be repaired. All co-present participants overhear or take part in conversations, notwithstanding the surrounding noise from the excavators at the site. The main participants are two Swedish excavator operators and a team of four Polish workers. However, this construction site is also visited by short-term participants, such as an electrician (Emil in the transcripts below). One of the excavator operators, called Berra, is also the supervisor responsible for the site. Wacek, the Polish team leader, has a central role at the site functioning as a language broker between Berra, other subcontractors and the Polish team. He takes part in all the identified instances of translanguaging at this site.

Professional keywords
Also at this site, professional keywords associated with Swedish are mixed into conversations in English. The words are nouns for tools or work materials such as gummiring 'rubber ring' or ringlyft 'ring lifter' (a special kind of hook that can be attached to the excavator's arm). In Example 2 below, we demonstrate how the latter word, ringlyft, is used in a conversation between the manager Berra, the team leader Wacek and the electrician Emil. Emil is in the hole, trying to dismount a pump located inside a concrete ring. Emil needs an easier access to the pump and asks Berra for something to lift the heavy concrete ring up with (line 1) so that the pump can be dismounted on a solid ground. In response, Berra first suggests a ringlyft as a possible tool (line 2) and then selects Wacek to fetch the tool from the pickup (lines 4 and 5).
In this example, Emil uses an imprecise expression (line 1), whereas Berra utters the established professional term in Swedish when addressing Wacek in English (line 4). The directive to Wacek to fetch the ring lifter is accomplished via multiple resources: he says the Swedish word, points at the pickup and looks directly at Wacek. This multimodal action succeeds in eliciting an embodied response by Wacekhe heads towards the car.

Professional competence and receptive multilingualism
Occasionally, the team leader Wacek engages in interactions in Swedish, which is a language that he is not overtly associated with at the site. In these cases, Wacek either responds verbally in English or produces an embodied action that addresses the issue. This is an illustrative case of receptive multilingualism and translanguaging in which monolingual norms are transcended (see Wei 2011Wei , 2018 as two separate codes are used in the same interaction. In Example 3, Emil's utterances at lines 1-6 are designed to be directed to Berra, not Wacek, with his gaze and body turned towards the excavator, and Swedish as the medium of interaction. The concrete ring, just lifted from the hole, is hovering above the ground, and Emil suggests that they need a square bar to put underneath the concrete ring before it is laid on the ground. Emil repeats his suggestion several times, but receives no response from his selected addressee, Berra. Finally, Wacek self-selects and produces a response instead of Berra (line 7).
When Emil repeats his suggestion several times (line 1, 2, 3) he also adds details. With no response forthcoming from Berra, Wacek takes the turn (line 7), looks straight at Emil and produces a candidate response: "a forty-five"? which is partially repeated by Emil in overlap (line 8). The expression "a forty-five" is a professional term from the carpenter's register, and an institutionalized way of naming wooden bars in construction that refers to their dimension. Thus, Wacek solves a work-related problem by responding to a first action that was not directed at him, and which was produced in a language that is not normally associated with him (Swedish). Wacek's apt response provides evidence for his multilingual competence and receptive knowledge of (some) Swedish, accompanied by his professional vision (Goodwin 1994) that allows him to interpret and act upon what needs to be done. Wacek is thereby breaking the monolingual norm, where instructions to him or the team are usually produced in English and demonstrates that he has a broader linguistic repertoire at his disposal.
Professional vision is commonly relied on at all the sites. It accompanies and enhances receptive linguistic skills, while participants also display mutual understanding through multimodal actions, thus achieving translanguaging. At the onset of the fragment presented below in Example 4, all participants, the electrician (Emil), the manager (Berra) and the team leader (Wacek), are positioned around the concrete ring. The interaction concerns a pump that Emil is removing from the ring. Wacek and Berra are closely monitoring the unfolding work.
As the transcript reveals, Emil (line 1, 3, 6, 7) and Wacek (2, 4, 5) use resources associated with Swedish and English respectively, without marking them interactionally as being two different codes. The sequence documents a typical pattern for spoken interaction during manual work at this site, as the participants are oriented directly to the ongoing task (notice the deictic expressions line 3, 5, 6) rather than to the linguistic preferences of the co-participants. Emil formulates a question in Swedish (line 1) and Wacek answers in English while simultaneously pointing at the screw (line 2), thus demonstrating his understanding of the activity at hand. In line 4-5, Wacek amends the location of the screw and Emil is now able to identify it, which he acknowledges (line 6). In his response Emil demonstrates his repertoire of linguistic resources, as he starts the turn in Swedish ja::: 'yes' and ends it in English (d/the::re). However, the pronunciation is a mixture of Swedish and English sounds, thus being an example of translanguaging that goes beyond rather than between separate codes: there is a Swedish sound e/ä (which we as native speakers interpret as being associated with a variety spoken in the region of Dalecarlia in Sweden) but English th-sound where Swedish would have a d-sound (där 'there'). Hence, Wacek and his co-participants transcend monolingual norms and employ multiple communicative resources for accomplishing the task.

Translanguaging for managing interpersonal relations
Also at this site, we find an instance of code switching in the closing of on-tasksequences (see Example 1 from site large). The evaluative closing exchanges seem to offer brief opportunities for translanguaging, where the linguistic resources normally associated with the interlocutors are used for a social function, presumably to build solidarity. In Example 5, Wacek and a Polish worker, Adam, are working down in the hole. Berra is standing on the edge, looking down. At the end of the sequence, where they start planning the next steps, Adam says dobra 'good', a word associated with Polish, which breaks away from the medium in the preceding interaction. Adam's choice of a linguistic resource is mirrored by Berra (line 10). This is a rare case in our data where we hear linguistic resources associated with the countries of the migrant workers being used by other persons. The code switch is accompanied by other closure-implicative behavior: pauses, bodily orientation and gaze. Berra agrees to the closing before initiating another sequence. Wacek, in turn, responds to Berra's dobra with a smile (line 11).
In sum, the examples from the site demonstrate how the workers practice translanguaging for accomplishing both professional and social aims. Wacek, the leader of the Polish team, is involved in all the cases of translanguaging in our collection, which reflects his intermediary position and puts spotlight on the multiple linguistic resources that can be used for social and professional aims.

Site small: countertop installation
At site small we target an Estonian worker Paul who is mounting a marble countertop in a kitchen under construction and works alone. However, the marble stone is heavy to manipulate, which is why he needs to ask for help from unfamiliar coworkers. A further complication arises when the pre-drilled socket hole in the marble does not match the position of the actual socket in the wall and an electrician needs to move it.

Professional keywords
When chatting with the researcher (in English), the countertop installer claims to recognize several words associated with the Swedish carpentry register, written on the wall for cupboard installers. The words include spis 'stove', disk(maskin) 'dish(washer)', skåp 'cupboard' and ho 'sink'. This qualifies as an instance of receptive multilingualism, although in written form and asynchronously, as Paul has commuted to Sweden for several years.

Translanguaging in information seeking
The unexpected problem with the mismatch of socket and pre-drilled hole needs to be solved but there is no electrician in sight. However, there is a carpenter working in the same room and through walking closer to him, Paul makes visible his intention to engage him in talk. Paul approaches him first in lingua franca English but after receiving an indistinct sound and a head shake in response (line 2), he offers to speak Russian instead, explicitly naming the language.
Paul asks in Russian whether Kuba speaks "Russian" and in response Kuba claims to speak "Polish". Paul repeats this answer in Russian, thereby demonstrating a multilingual competence and perfect understanding of the answer. As the linguistic systems of Polish and Russian are closely related, Kuba's ta, (a colloquial form of tak 'yes' in Polish), is apparently also understandable to Paul who does not display any comprehension problems with the exchange. He goes on to ask about Kuba's Russian skills and this time the answer is 'a little'. Paul repeats the answer in Russian and then uses the keyword jelektrik ('electrician' in Russian) to put forward his concern. In response, Kuba utters a complex turn in Polish (lines 12 and 13), which Paul receives with a mhmh 'uhuh', possibly because also the negation word nie and the verb form widziałem ('haven't seen') are reasonably similar in pronunciation in the two linguistic systems. We have thus witnessed a fluid use of resources from three different languages, identified and named by the speakers themselves, that have resulted in a coherent conversational sequence between the two workers. They started by searching for overlapping repertoires but ended up exchanging information by each sticking to their own linguistic repertoires that worked by proxy. This we consider as translanguaging in its most fluid form, as the speakers go both between different linguistic systems and beyond them. Indeed, they demonstrate knowledge of multiple linguistic resources, including testing and adjusting utterances in relation to the repertoire of the co-participant. Naming certain languages seem to be a help on the way, although the conversation as a whole transcends monolingual norms.
In sum, we observe how the Estonian installer of countertops mobilizes multiple linguistic resources in search for overlapping repertoires when talking to a Polish carpenter at the site. His repertoire helps him create a social space, in which he can ask for help to solve a professional problem.

Translanguaging and social space in a language marginal context
Considering the relatively extensive number of recorded hours, we can conclude that workers recruited from different countries rarely talk to each other at multilingual construction sites. The larger sites are multilingual in the sense that they recruit individuals speaking different languages, but the organizational principle of teams rather than individuals arriving from different parts of the world keeps the speakers of different languages effectively separated. In fact, most manual workers have limited contact with workers outside their own team, and only communicate with the Swedish management via team-leaders. Most workers at the sites are socially constructed as monolingual speakers of Polish, Ukrainian etc., and language ideologies at the sites prioritize (monolingual) interaction with linguistically similar others. The combination of hierarchical workplace organization, monolingual norms for interaction and work norms of efficiency means that there are few social spaces for power negotiations at the sites, especially for migrating, manual workers. Accordingly, there are few possibilities for the workers to mobilize resources from other languages as a tool for renegotiating identities and making room for new voices (see Wei 2018), as they lack a channel for communication with superiors. The lens of translanguaging theory has been helpful in detecting how these contextual circumstances function together at construction sites, and how they influence the migrating, manual workers' social relations at work. Our analyses demonstrate that primarily professionals at specific positions use resources from multiple languages, and primarily individuals who need to talk to different occupational categories. First, this applies to individuals working alone, such as the Estonian installer at site small. He needs a broader linguistic repertoire and linguistic flexibility in order to recruit help from workers with only partially overlapping repertoires or discuss upcoming problems with the site boss in a lingua franca. Second, it applies to the team leaders at site medium and large: Wacek and Vanko. They function as links between the management and the workers, and they talk to different categories of employees. They work with their hands and feet as all the workers, but also with their ears and mouths to a higher degree than others, and they mobilize a broad repertoire of linguistic resources to achieve social and professional aims. In relation to earlier studies of team leaders as language brokers (e.g. Kraft 2017), we could, from the perspective of translanguaging, add that team-leaders are in a hierarchical and social position that gives them latitude to create a professional and social space where they can transcend linguistic boundaries and monolingual norms. This space has advantages, as it may lay the ground for upward mobility at the sites, as well as provide opportunities for language learning.
We showed how studies of language marginal contexts, such as construction sites, have potential to contribute to the theory of translanguaging. In our study it became obvious that translanguaging practices cannot influence power relations per se. Rather, the transformative power is visible in specific spaces (or contexts), involving participants at different hierarchical positions, such as in the examples with team-leaders and their managers, and when the participants make an upward or downward move in the (locally constructed) hierarchies via their choice of linguistic resources. Thus, the transformative potential of linguistic performances is only evident in certain contexts, where the superiors are exposed to the transcending of norms. It is likely that hierarchy and speakers' local social status are important factors in relation to the transformative power of translanguaging in many sectors of work life, but the factors may be especially visible in language marginal settings where very few manual workers have access to spaces where it is possible to re-negotiate identities and make new voices heard.
The study furthermore contributes to the understanding of translanguaging as a resource in professional interaction among transnational individuals and groups, as they move across space and time. Firstly, institutionalized keywords were adopted by all categories of workers at the sites: by migrating manual workers, team leaders and managers. Keywords are a fundamental ingredient in a professional repertoire, and a resource that apparently shifts, develops and transforms depending on the individuals' professional trajectories across countries and linguistic regions. Often being nouns, they can be easily mixed into interaction, and, as such, they seem to be an early articulation of a new identity of being a construction worker in a certain space, at a certain time. Secondly, we noticed how translanguaging can function as a resource in the search for overlapping repertoires, including testing and adjusting utterances in relation to the repertoire of the co-participant. Individuals working alone seldom have time to wait for a translator or language broker at the busy construction sites. Communication is subordinated to manual work and most professional tasks can be solved with minimal verbal interaction, in combination with professional knowledge and vision. Accordingly, the readiness for going between or beyond different linguistic structures and systems, along with embodied resources for meaning-making, is indeed helpful at the international construction site, especially for individuals working alone and for the team leaders.

Conclusions
In this article we looked at professional interaction at construction sites, searching for situations where workers use resources associated with different languages when planning or carrying out manual work. The analysis aimed at investigating what social and professional space the workers construct by going between or beyond different linguistic structures and systems, as suggested in the theory of translanguaging. We found that language plays a minor role at workplaces for most migrant workers due to the very nature of their work tasks. Hierarchical organization, long distances between colleagues or altogether isolated work situations mean that the "voices of others" (Wei 2018: 24) are seldom heard in the construction sector. The EU free labor movement legislation creates situations where workers with different linguistic repertoires gather at the same construction site, but the very organization of the workplaces and the characteristics of the work tasks hold workers and their linguistic resources effectively separated. We focused on a range of exceptions to this generalization, mostly involving team leaders who due to organizational reasons and depending on project size make use of a broader repertoire in order to accomplish their professional tasks, but also regarding individuals working alone. Workers at these positions create professional and social spaces, where they can transcend linguistic boundaries and monolingual norms for achieving professional and social goals. The different linguistic realities of individual workers documented in this study underline the importance of context-sensitive analysis of language practices and demonstrate the fruitfulness of combining ethnographic data with interactional analysis when searching for how resources associated with different languages are used in a mobile working life.