Conceptualizing industrial workplace learning: an information systems perspective

Purpose – The purpose of this paper is to identify the constituent parts of learning in the manufacturing work contextandunderstandwhythesepartsarekeyinthelearning of theemployees. Design/methodology/approach – The data was collected from two sources: a literature review of the Information Systems literature to establish an initial picture of what learning in relation to digital technologies entails and in-depth interviews with engineers in the automotive industry whose knowledge- intensive work is exposed to substantial digitaltransformation. Findings – The authors ﬁ rst identi ﬁ ed three constituent parts for learning: change, re ﬂ ection and deliberation. When the authors cross-checked the initial ﬁ ndings through in-depth interviews with the engineers, it was found that these three themes trigger learning through three different mechanisms, that is, balancing newness, ﬁ nding point of reference and organizing actively. Thus, the ﬁ ndings of this paper extend beyond a categorical identi ﬁ cationofwhatconstituteslearningtoalsoillustratewhylearningentailstheseconstituentparts. Research limitations/implications – This paper implies that progressive learning requires active organizing of learning stages. The data is limited to the review of the Information Systems ﬁ eld. The authors have also only focused ontheautomotive industry astherepresentative sectorinthemanufacturingindustry. Practical implications – Applying the model of progressive learning can be a primary way to actively plan and organize learning opportunities for employees. This is key for supporting learning culture in organizations that are exposed to continuousanddisruptive changes. Social implications – A signi ﬁ cant part of social sustainability is based on sustainable employability and feelings of contentment at work. This paper is an attempt to highlight how sustainable employability can be achieved by providing effectivelearningopportunitiesat work. Originality/value The originality of this paper emerges from two sources. First, the authors conducted the literature review and in-depth interviews by devising innovative methods because of the challenges of identifying when (informal) learning has occurred at work. Second, the authors owe the in-depth interviews to the ﬁ rst author ’ s extensive familiarity with the automotive industry and the knowledge and rapport acquired through her prior longitudinal research on the engineers ’ work. It was this background that allowed the authors to ﬁ nd out when these engineers were about to leave the ﬁ rm because of discontent about their competencedevelopment. This paper cannot be an example of workplace “ learning ” as its major part illustrates how focusing on the characteristics of IT artifacts would essentially lead to the development of new theoretical perspectives. Over time , necessity causes change in the employees ’ understanding of their organizational contexts, and in the organizational control over how practices need to be arranged. [ ... ] The account of incidents are as they occur during the natural ﬂ ow of work. No tensions are lifted and discussed in a way as to consciously overcome problems and plan a change in the way things are being done.


Industry and workplace learning
Organizational and structural features of a sector have a profound influence on developing cogent workplace learning measures (Stroud and Fairbrother, 2008;Billett, 2014). Manufacturing industry, the secondary large-scale industry with labor-intensive and heavy investment requirements, is encumbered more immediately by issues of capital flow than improvement of life quality (Schmiede and Will-Zocholl, 2011). Constant need for bringing in heavy capital to keep the industry going has turned the limits of the organization into a financial one (McKenzie, 2002). This means that the organizational processes, structures, and cultures tend to adapt towards the stakeholders' short-term values, rather than the experts' long-term strategic values (Reardon, 2010).
It is not just capital flow hurdles that treat human expertise lightly. The informatization of the industry and its associated globalization processes also alter the nucleus of human work and redefine its related competences (Baase, 2013;Winter et al., 2014;Tilson et al., 2010). The unfolding of Industry 4.0, as a methodology to turn machine-dominant manufacturing into digital manufacturing (Oztemel and Gursev, 2020), is characterized by the connected world of different enterprises that jointly develop technological projects, build collaboration scenarios and use cases, and coordinate combination of products and services into novel business opportunities (Shamim et al., 2016). This connected world consists of heterogenous enterprises that cooperate on one level (e.g. building hardware platforms) and compete on another level (e.g. offering distinct end-user services). The coalition of heterogeneous or competing businesses signifies that development requirements and work conditions are defined by multiple enterprises each with their specific and often diverging goals and provisions (Winter et al., 2014;Oztemel and Gursev, 2020). Therefore, the ubiquitous informatization of the manufacturing industry calls for learning new collaboration and development practices.
Additionally, what characterizes novelty and innovation is continuous change and loosely defined standards, use cases and, therefore, development requirements that respond to the fast-changing and unexpected future requirements (Oztemel and Gursev, 2020;Shamim et al., 2016). That is, accommodating digital innovation calls for learning how to continuously adapt to fluid work conditions and project requirements. In other words, informatization of the manufacturing industry has consequences for learning at work, as: much of the technological development work is novel and unprecedented; work structures and guidelines are inherited from multiple contexts; and these multiple contexts are loosely defined and fluid, as they tend to be responsive to the fast-changing and unexpected future requirements.
The informatization of the manufacturing industry, thus, prompts us to realize appropriate training strategies that respond to not only the new but also the continuously changing work conditions (Liao et al., 2017;Shamim et al., 2016). To meet the requisites of a new work life, reports on competence development and changes at work have re-emerged in recent years (Susskind and Susskind, 2015;Brynjolfsson and McAfee, 2016;Segal, 2018;World Economic Forum, 2018). Besides training new skills and competences, there is equal emphasis on supporting the upskilling and retraining of the incumbent workforce (European Commission, 2021) and transformative competences of the workforce (OECD, 2019) triggered by the informatization of industry. However, despite the increasing reliance of Industry 4.0 on knowledge work and the employees who hold this knowledge, the workforce in manufacturing industry struggles to define their learning needs in a focused way (Rangraz and Pareto, 2020;Blayone and VanOostveen, 2021).
Similarly, the extant literature on workplace learning underlines that even when workplace learning opportunities are available, they include more service-oriented and overgeneralized procedures rather than the actual and active organization of learning processes (Stroud and Fairbrother, 2008). We argue that, to contribute to the realization of progressive learning opportunities in this sector, there is a need to address the learning needs of the manufacturing industry, i.e. the type of learning that corresponds to the specific conditions of today's manufacturing industry as opposed to general workplace learning concepts and directions. By learning needs, thus, we refer to the appropriate training strategies and organization of work in a way that fosters lifelong workplace learning in correspondence with the digitalization of the manufacturing industry (Liao et al., 2017). To do so, we propose that it is pivotal to a) identify what constitutes learning in the manufacturing industry and b) shed light on the relationship among its constituent parts. Our research question is, thus, "What constitutes learning in industrial work contexts?" To answer this question, we first conducted a literature review to provide an initial understanding of what constitutes learning in industrial contexts. We then conducted semistructured interviews with industrial practitioners to cross-check our findings. The interviews were valuable in highlighting how the different identified constructs play out in an industrial context. In light of this, the originality of this paper does not lie in identifying learning constructs. Numerous well-received models and definitions of workplace learning have already contributed to our understanding of what constitutes learning at work in general (Illeris, 2011;Bratton et al., 2004;Marsick and Watkins, 2003). However, the function of these models and definitions is to create abstract and general conceptual frameworks rather than illuminate specific learning needs. Identifying the specific learning needs of a sector relies on a concrete analysis of empirical data obtained from that sector. The objective of this paper is, thus, to establish a concrete analysis of learning needs specifically in an industrial context. The rest of the paper is structured as follows. First, we explain how we conducted the literature review and cross-checked our findings through in-depth interviews. Then, we present the results with examples based on both the review and the interview data. The results are subsequently discussed, and the paper ends with research implications and limitations.

Method
The study of learning at work has traditionally been translated into studies of competence/ skill at work and affiliated with disciplines such as human resources, pedagogy (Murawski and Bick, 2017), management (Sandberg and Pinnington, 2009;Sandberg et al., 2017) and workplace learning (Fuller and Unwin, 2011;Illeris, 2011). However, with the focus of this paper being on the context of the industry as affected by Industry 4.0, we intended to review articles that understood the role of digital technologies as central in shaping today's industrial environment. We did an initial search in the Journal of Workplace Learning, looking for articles that had investigated the issue of learning in the context of the manufacturing industry. Although Workplace Learning would be a logical outlet to look for articles about learning at work, a quick analysis of the industrial papers in this journal revealed that many of these papers had not dealt with the characteristics of the industrial context in depth. These articles had usually selected the manufacturing industry, as, for instance, it was an important sector to a specific country's economy (Coetzer, 2007) or it presented the need to develop certain skills among immigrant workers (Duval-Couetil and Mikulecky, 2011), certain learning challenges among mature-aged workers (Smith et al., 2010), various methods of producing manufacturing technology (Jalonen, 2016) and specific management styles (Wallo et al., 2021).
To stay closer to the challenges of working in relation to the informatization of the manufacturing industry, we decided to conduct our review in the field of Information Systems (IS). One reason for doing so was that, with the increasing focus on the pervasive role of digital technologies at work, IS has emerged as a new interdisciplinary field to deal with issues of managing various aspects at work in relation to the emerging technologies (Wade and Hulland, 2004). In addition to being interdisciplinary, IS is also an applied discipline that uses, redefines and potentially redevelops the theories developed in other disciplines (Keen, 1980;Vessey et al., 2002). Hence, the literature review would not only reflect the foundations of a wider range of related disciplines but also would provide a mutual ground for the Workplace Learning field to benefit from the practice-driven nature of the IS discipline.
In selecting the search keywords, we found that unlike the workplace learning articles that are familiar with the key term "workplace learning," articles in IS could have discussed learning at work without necessarily using the terms "workplace learning" or even "learning." It must be mentioned that we view workplace learning as a type of "competence development" at work which entails all that it takes (e.g. knowledge, skills and abilities) to perform in a given work context (Illeris, 2011). Given this perspective on workplace learning as competence or the capacity to perform work-related tasks (Billett, 2014;Sandberg, 2017), we looked for a keyword that could more closely capture the learning that occurs in everyday work practices. The established keyword that considers any workplace reality (such as learning) as a production of everyday practices (Pickering, 2001;Leonardi, 2015) and is quite close to "work" and "workplace"was "work practice." This strategy concurred with the recent views on workplace learning that emphasize the importance of uncovering the kind of learning which occurs in "everyday participation at work" (Fuller and Unwin, 2011, p. 50), rather than focusing on direct and measurable instances where learning is the objective in a top-down manner (e.g. courses or competence development programs). In IS, the explicit discussion of work practices usually entails commenting on work objectives and the way subjects (individuals and teams) collectively achieve those objectives. Additionally, our initial search with the keyword "learning" did result in a few hits; however, a quick review of the papers revealed that these articles did not necessarily include detailed discussions of workplace learning.
Additionally, we classify our literature review as a hermeneutic literature review in which the objective is to establish a dialogical interaction with the literature to continue interpretation, questioning, and critical assessment (Boell and Cecez-Kecmanovic, 2014). This type of literature review calls for highly intellectual and creative research activities that seek originality rather than a highly structured search process with an emphasis on replicability (Boell and Cecez-Kecmanovic, 2014). In other words, our review is not a systematic literature review with the goal to exclusively map a particular literature and provide "replicable and unbiased" or "ultimate" answers (Boell and Cecez-Kecmanovic, 2014, p.261). We believe this type of literature review does not hold up in practice and claiming to conduct such a review countermands our goals for initiating or expanding a focused debate around "industrial" workplace learning. Thus, although our search process is structured and the selected search outlets and keyword are relevant to the study objectives, they are not exclusively relevant. Other search strategies can expand and deepen our mapping and classifications which are only aimed at providing the basis for critical analysis. The search phases of our hermeneutic literature review are summarized in Table 1. The search was conducted through Scopus and included top journals selected by the Association for Information Systems (eight top journals known as the senior scholars' basket of eight).
A total of 39 articles with the keyword "work practice" in "abstract, title keywords, body" (2000-current) were checked manually for their relevance. The initial relevance criteria included: The articles had to have mentioned "work practice" at least once in their abstract/ keyword/title or the main body of the article. Mention of the keyword "work practice" in the references would not qualify the article for inclusion. The articles had to be empirically grounded.
The articles had to deal with work within an industrial context.
Thus, articles which dealt with non-industrial contexts such as school settings, hospitals, call centers, public administration offices or environmental studies were excluded. In the end, 20 of 39 articles included all the above-mentioned criteria and, thus, constituted the basis of our analysis.

Data analysis
The articles were read and summarized carefully in tables based on their authors, year and journal of publication, theory/concept, methodology, context, findings, and possible links with learning/competence/skill. Figure 1 exemplifies the tables developed in this round of data analysis. The next step in a literature review such as this one would have been to categorize the findings based on a previous conceptualization (Webster and Watson, 2002). However, a conceptualization of workplace learning, as discussed previously, did not exist. We, thus, had to develop our own strategies to conceptualize learning in the manufacturing industry in four steps: (1) Developing commentary: We first developed a commentary on the pertinence of each article to the subject of workplace learning based on our own interpretation.
To write these commentaries, each of us had to ask ourselves whether we considered the article to present an instance of workplace learning and why we thought so. To complement this step and to establish a comparative criterion, we also wrote commentaries on articles that we thought did not present instances of workplace learning and why we thought so. Each researcher developed their own commentary independently, and once this round was complete, the notes were detached from the tables and were scrambled to be coded randomly by each researcher.
(2) Coding the commentary: The notes were coded with the aim of comparing researchers' personal understanding of the concept of workplace learning. As both authors have had considerable experience in conducting research in the field of workplace learning, we thought this strategy would help uncover their tacit understanding of the workplace learning research. In this step, we each closely read the notes independently aiming to find repeated thought patterns that would allow identifying themes we commonly associate with workplace learning. Each researcher found two sets of themes, labeled as "FOR Workplace Learning" and "NOT Workplace Learning." These two sets of codes were also compared. This way, we were able to both select themes that we commonly agreed on and then verify them by commonly agreeing on what was not considered to be workplace learning. In the end, three major themes were identified. These themes include: 1. change, 2. reflection and 3. deliberation.
(3) Industrial verification: As the review allowed for an indirect approach, that is, analyzing learning in cases where learning had not been directly in focus, we attempted to validate the findings through a more direct approach. The main reason for cross-checking the findings directly was that the first author had already conducted a three-year ethnographic fieldwork study at a Swedish automobile manufacturer. This gave us substantial contact and familiarity with the practitioners' work in an industrial context. While the previous fieldwork is not directly discussed in this paper, the aggregate data has provided indispensable contextual information for the focus of this paper. For example, through her prior research work, the first author was aware of the extensive change that was digitally transforming the work of the engineers. Similarly, the first author had observed feedback sessions where the engineers reflected on their work processes and their general progress at work. These sessions were called retrospective sessions in the dominant Agile software development processes. However, the first author's prior fieldwork had revealed that despite change and reflection being a consistent part of the engineers' everyday work, they had continuously reported dissatisfaction with their learning progress at work. Thus, although by analyzing the reviewed literature we understood "change" and "reflection" as important themes and constituent parts for learning, we were not quite sure if, how and when these constituent parts necessarily lead to learning.
Therefore, to make sense of the discrepancies found in the interpretation of the reviewed literature and the first-hand experience of the industrial practitioners, we interviewed the same group of engineers (six people) whose work the first author had followed for years. All the engineers were members of a development team whose industrial knowledge-intensive work included developing autonomous drive functionalities over the past four years. Although the number of interviews is modest, a sample of six interviews can be sufficient to develop "meaningful themes and useful interpretations" (Guest et al., 2006, p. 78). This is especially the case when a homogeneous group of participants is selected purposively (Guest et al., 2006).

(4)
Conducting the interviews: We conducted behavioral event interviews (McClelland, 1973;Spencer and Spencer, 1993) which are used in competency/skill identification studies. Behavioral event interviews are widely used in competency identification studies (McClelland, 1973;Murawski and Bick, 2017) and are useful in identifying competencies/learning points by examining critical past events. This strategy is particularly relevant given our competence view of workplace learning which assumes learning as entangled in everyday practices and non-measurable through any particular performance (Fuller and Unwin, 2011).
We selected this team of automotive engineers, as their work was marked by extensive efforts to develop complex digital technologies, including digitizing a substantial amount of previously hardware-based components inside the car to develop autonomous drive functionalities. Their work would be emblematic of how industrial workplace learning is emerging when the industry is engaged with digital technologies on a large scale. As the team was working according to the Agile software manifesto, the team members' roles were not rigid. Rather these members would define work requirements, roles and processes among themselves in a bottom-up way. However, despite holding their niche expertise, team members worked based on learning and communicating expertise from and to each other. The team members were multi-cultural; however, they all had obtained educational degrees in Sweden and were close in previous experience and age. Through the first author's prior fieldwork study, in-depth familiarity and rapport was already established with this group of engineers.
In conducting the interviews, as it is often difficult for interviewees to self-evaluate in terms of what counts as learning, we asked them to describe one event where they felt they had learned something and one event where they felt they kept working without learning something considerably. This round of data collection resulted in approximately 270 min of audio-recorded interviews which were transcribed and coded subsequently. When we analyzed the interview material, the previously identified themes (change, reflection and deliberation) recurred. However, we also identified three new overarching themes: 1. Balancing newness; 2. Finding point of reference; and 3. Organizing actively.
It must be mentioned that in developing the commentary on the literature review and coding the commentary, we had followed an inductively oriented thematic analysis (Rapley, 2016;Braun and Clarke, 2006). In these two rounds of analysis, we performed two steps in thematic analysis that include familiarizing ourselves with the data and then abstracting the themes into more overarching themes (Rapley, 2016). In analyzing the interview data, we were initially open to all the themes that we could identify. However, we were also curious to understand how these themes would relate to the previously identified themes (change, reflection, and deliberation). Hence, in the later analysis of the interview data, we followed a more deductively oriented approach (Rapley, 2016;Braun and Clarke, 2006).
In the deductive analysis round, we intended to understand how change, reflection, and deliberation play out in the engineers' work context and how and why they relate to the newfound themes. To do so, we applied the bracketing technique which Lee (2017) describes as a way to avoid taking the interpretation of a statement for granted. In other words, bracketing refers to putting one's belief and the validity of one's interpretations into brackets to suspend what they already know and re-investigate their belief. This is where we asked ourselves, "if change is a constituent of learning, and if the engineers' work entails high degrees of change, then why are they dissatisfied with their learning process?" Figure 2 illustrates the iterative bracketing process through which we tried to bridge the three JWL themes of change, reflection, and deliberation with the themes balancing newness, finding points of reference, and organize actively.
Through bracketing, we juxtaposed our interpretations of the review data with the data from the interviews on how change, reflection, and deliberation could lead to workplace learning. We eventually noticed that the ways through which change, reflection, and deliberation can lead to learning may vary in contexts where degrees of uncertainty and exploration is higher because of more extensive engagement with digital technologies (innovation). For instance, as illustrated in Figure 2, change can no longer lead to workplace learning by "triggering diversion from old routines" (finding based on review data). When change is a constant reality of everyday work (interview data), it can only be effective if, through some mechanisms, one can balance the uncertain and the unknown or, as the engineers put it, "balance the newness" (interview data). Bracketing allowed us to notice that balancing newness, finding point of reference, and organizing actively can be categorized under the previously identified themes of change, reflection, and deliberation. These themes explain how, when, and why change, reflection, and deliberation can be effective in learning in the context of the manufacturing industry with high degrees of digitalization. In other words, balancing newness, finding point of reference, and organizing actively can be understood as mechanisms through which change, reflection, and deliberation can lead to workplace learning in such contexts.

Results
As mentioned in the previous section, three workplace learning-related themes were developed that could assist us in achieving an initial conceptualization of industrial workplace learning. We present and exemplify these themes below by excerpts taken from the review commentaries as well as quotes selected from the interviews.

Change
As we reviewed the articles, we noticed that when some technology-related change emerges in the workplace, it is likely that the employees try to find ways to cope, adapt or benefit Industrial workplace learning from changes. That is, technological change unsettles the well-established routines and practices and can, therefore, trigger learning opportunities. The reviewed articles entailed ample example of how the implementation of technologies at workplace created significant changes in the practice of the employees. One of the early examples refers to an article that discussed the changes of work practices because of introducing smartphones for a more immediate and smooth communication at work. Smartphones facilitated instant communication and longer accessibility of the employees which in turn disturbs the limits of working hours. As the employees try different ways to solve the problem, they realize that the meaning of communication and working hours is changing because of the introduction of technologies and that balancing work and life is turning into a new skill that they need to seriously think about: Here, employees change their practice or attach new meanings to well-established practices to improve their working life. In doing so, certain competences appear. At first, they try to gain control of their workload by ignoring their ever-present smartphones. As the strategy does not work, they change their way; rather than trying to eliminate the smartphones, they introduce two smartphones; one for work, one for life outside work. By developing a certain skillthe skill of balancing work and personal lifesmartphones are no longer associated merely with work. Rather, smartphones are employed to connect the employees to the intended part of life on and off.
In the interviews, as well, change or newness emerged as a manifest example of opportunities for learning. The learning opportunities seemed to be even intensified for the engineers as developing autonomous driving functionalities entails high degrees of digital innovation (e.g. developing new products and services through cloud communication and connection possibilities). With the increasing research and development projects, the work of the engineers relies majorly on new practices and collaborations. The engineers mentioned that rather than an occasional incident, change and newness was a constant situation at work. However, they emphasized that, dealing with constant newness and rapid changes often could lead to frustration in terms of learning. As they explained, engaging in (digital) innovation implied that opportunities for exploiting and benefitting from established knowledge areas become narrow, and engineers are often required to explore new solutions on their own. Constant newness and lack of guidance, thus, can create obstacles when it comes to workplace learning: There are kind of mixed feelings around it [learning triggered by change], you know, you have moments that you feel this energy, this motivation to try and explore more even though you have no guidance, and then at some point, you feel a bit frustrated, because you're not getting anywhere, and are making very small progress. The downsides of dealing with constant newness and change become highlighted once we consider reflection, that is, the next important theme in relation to workplace learning. As explained in the next section, reflecting on what and how we have always done something is vital for developing one's knowledge and skill. However, with the speed of change and diminishing of established practices, there is little time for reflection.

Reflection
This theme referred to some level of reflection or evaluation of how work was being performed and whether it was the optimal way to continue working. The commentary JWL excerpt below was written on an article on service co-creation where the authors had adopted an action research methodology. They had consequently been reflective both on the way the work was practiced for offering plausible designs and on their own research methods and phases to guarantee a credible design. This constant reflection meant that the researchers as well as the study participants had to repeatedly negotiate and learn from their mistakes to understand how practices could be accommodated in design. With its extensive reflective approach, this article was a vivid example of extensive workplace learning both for the employees and the action design researchers: The research/work in this paper assumes a completely reflective and evaluative air from the very beginning. The goal is to consciously experiment and evaluate ways for service development. The researchers/coaches comment meticulously on the ways they had conducted the work/research; on whether it's been successful or not; and on what could possibly be a relevant explanation for the success/failure of the methods applied [. . .]. Therefore, the paper with its "meta-processing" approach towards experimenting, developing, and evaluating applicable work practices is a vivid example of learning through iterations and consideration.
For the engineers too, iterative processes and methodologies entailed substantial learning opportunities. However, the engineers explained that their development projects entailed high degrees of innovation where past experiences and well-established routines are de-emphasized. They commented that lack of established routines and past criteria posed a real challenge for making sure that they were developing the relevant technical competences. Apart from the diminishing well-established paths, the increasing speed of change from one project requirement to another did not leave the engineers with enough time for reflection on past experiences. This was frustrating to the point that the engineers believed delivering project requirements is not the same thing as learning, as it would not add to the depth of their knowledge: The term, technical depth, if you know about it, you focus a lot to solve the problem, and in the end a lot of things are just hanging there, because you're taking a lot of shortcuts to solve the problem, and these are not really acknowledged at the end of the project, you just jump to the next project and it's really overwhelming if you never get to stop and sort them out.
As the speed of change and number of R&D projects with different development requirements and working dynamics increased, the engineers felt they needed points of reference that would help them reflect on their work more immediately and in situ rather than in retrospect. They mentioned that, rather than reflecting on "How are past things still relevant," they needed to reflect on "How is what we they are doing now relevant in a larger context." The engineers, for instance, mentioned how industry standards and protocols or working with more experienced people in their teams provided them with more immediate points of reference to reflect on their work in situ and at short intervals. As protocols and industry standards provide ways to compare and evaluate the technical processes through several iterations, following them provided a sense of compatibility with what one needs to know about the industry, thereby a feeling of personal growth and learning: [When you work with industry standards and protocols] the feeling is like that you are not being left behind. Because you don't want to be in your own universe where you think you're working in a very good way and then in a future circumstance when you need to, for example, change companies, suddenly, you have this shock [. . .] then there is this confidence or a feeling that now, my personal skills are also going to improve, and I won't be in my own world or universe so to speak.
Lack of time and the increasing speed of change in project requirements and work practices are not only problematic for the ability to reflect on work, but as the next section explains, they also challenge the possibility to deliberately plan and coordinate learning paths.

Industrial workplace learning
Deliberation This theme refers to developing a deeper and fundamental understanding of how and why things happen the way they do and how and why change needs to happen. This theme was particularly important, as we noticed that sometimes change can occur over time without employees having actively and consciously formed a deeper understanding about the changes in their practice. Thus, although some sort of learning might occur in adapting to the emerging situation, yet no substantial learning opportunities are provided without deliberation. The example below shows how in dealing with technology, changes occur over time and without any in-depth consideration of the course of events: This paper cannot be an example of workplace "learning" as its major part illustrates how focusing on the characteristics of IT artifacts would essentially lead to the development of new theoretical perspectives. Over time, necessity causes change in the employees' understanding of their organizational contexts, and in the organizational control over how practices need to be arranged.
[. . .] The account of incidents are as they occur during the natural flow of work. No tensions are lifted and discussed in a way as to consciously overcome problems and plan a change in the way things are being done.
When analyzing the articles, we noticed that, although deliberation requires some form of reflection to lead to the development of a deep and fundamental understanding, it is not necessarily the same as reflection. Unlike reflection that usually occurs on more immediate issues, deliberation entails a longer and more careful consideration of that paths taken. The engineers, too, saw this long and careful consideration of the situation as something that exceeds the immediate evaluation or reflection on a situation at hand. While reflection might produce insights in the mind of the observer to cope with the flow of the things as they happen, it does not necessarily result in a deliberate planning of the longer path one needs to take. For engineers too, deliberation required a long brooding over how things have or might turn up in practice: For the engineers, however, the speed of change in project requirements, collaboration needs and team structures created challenges for forming a deeper understanding of the situation. According to the lead engineer, with over ten years of experience in the industry in various countries, such careful consideration and intentionality needed to unfold in the attentive organization of learning opportunities. However, actively organizing mindful approaches to learning required substantial amounts of time for deliberation and exchange of ideas. When asked to elaborate on an event which he had experienced as a great learning opportunity, the lead engineer adamantly lamented the lack of such opportunity at his current workplace. The lead engineer believed that the last learning opportunities he could remember had taken place in his prior workplaces years ago. According to the engineer, the centerpiece of all those substantial learning experiences was careful deliberation and organizing of the learning process: The difference was, we really had this streamlined approach of training for each and every individual. So, we had this competence matrix, and then guys from us would review each and every line we'd written in the software, we had very strict controls, because they wanted to teach us. And then there was a very careful approach towards increasing the competence because they really wanted to see the India office delivering something on our own, which we did in a few years [. . .] just because of the processes my manager put in at that time.

JWL
Thus, deliberation is not merely a longer or deeper form of reflection and evaluation of the situation. It is the formation of a deeper understanding about the direction of things which is then translated into the way learning approaches are organized actively with respect to the work conditions. However, as the engineers explained, fluid work conditions can potentially compromise deliberation, as the subject of deliberation is constantly shifting. Actively organizing learning approaches step by step, for instance, would be more challenging when team structures, and therefore, levels of expertise and combination of knowledge areas change form project to project.

Discussion
We have identified three major themes of workplace learning in the context of industry: change, reflection, and deliberation. Through the analysis of articles as well as in-depth interviews, we found that responding to technological change corresponds to the prime trigger of learning occasions. Industry 4.0, the fourth industrial revolution, incorporates the latest technologies to boost production, offer novel functions, and remarkably improve prices (Yoo et al., 2010). Ubiquitous digital technologies with their affordances to provide extensive connectivity have the possibility to transform the design, production, and distribution of industrial products in shorter cycles. These technologies, thus, call for a faster rhythm of reskilling as well as life-long occupational competence development (Flores et al., 2020). Both the reviewed articles and the engineers discussed extensively how the continuous changes brought by technology exposed them to the need to explore new things constantly.
In the midst of this, reflecting on what has changed and what courses of action need to target those changes seemed to be extremely consequential. For instance, assuming that being exposed to change would necessarily push the engineers to explore new ways and, thus, improve their knowledge led to feelings of frustration or disgruntlement (Reardon, 2010). Four of the six engineers whom we interviewed were about to leave the company at the time of the interviews. On most occasions, the engineers mentioned that they were left to their own agency and willingness to learn. Similarly, the extant literature on occupational learning either emphasizes the importance of individual agency or the importance of contextual factors and, thus, fails to understand how professional development is linked to the jobs the employees are responsible for (Emiliani, 1998;Migliore, 2015).
The engineers believed that the company had once detected the type of change it needed to go through; however, it had failed to continuously and iteratively understand how those changes affect the engineers' day-to-day practices. Being left to their own devices, and in the face of fast-changing work conditions, the engineers lacked the necessary time and tools to reflect on their work in retrospect, and develop deeper technical skills. In the absence of retrospective reflection, they mentioned that, working with more experienced people, or with technical protocols and industry standards would have provided them with more immediate checkpoints to reflect on their work on-the-go. However, without these immediate reference points in most of their prior projects, the engineers had experienced unnecessary loss of time and trial-and-error rounds. Their learning, thus, could be considered defensive and regressive rather than skill-oriented and progressive (Stroud and Fairbrother, 2008). Working with protocols and standards had instead become the reference points which reassured the engineers that their actions were in accord with the expected expertise in the industry anywhere outside their current workplace. Hence, the protocols and industry standards were mentioned as checkpoints that equipped the engineers with immediate reflection tools to evaluate their actions in shorter intervals.
While reflection includes thinking about what we say or do and evaluating the outcome or possible outcomes for a brief period of time (Emiliani, 1998), deliberation appears to be a more methodical process. Deliberation, as our results show, consists of what Emiliani (1998) considered to be a more "labored and detailed" analysis process during the course of a longer time to formulate a particular understanding (p. 31). While successful reflection can uncover useful lessons learned and, thus, provide direction (Emiliani, 1998) for the present issues, our results indicate that it does not necessarily lead to the formation of a fundamental understanding of how learning should be planned and executed. Rather than a one-time reflection, deliberation entails a series of reflections and combined with intentionality, as well as focused and streamlined approaches to respond to change. Unlike the protocols and industry standards which provide checkpoints for the prevalent tasks, as the lead engineer exemplified, a substantial competence development is the result of a streamlined and organized approach to build and increase competence step-by-step. The constituent parts of learning and their indicators are summarized in Table 2.
Although each and every constituent part of learning in this study is consequential, our analysis proposes that without deliberation, change or even reflection provide only regressive forms of learning at best. That is, in the absence of mindful and coordinated approaches, learning becomes principally a matter of meeting narrowly focused and immediate workplace-related needs and situated problems (Stroud and Fairbrother, 2008). Based on these constituents, we propose the following conceptualization for workplace learning, keeping a manufacturing context in mind: Workplace learning constitutes the formation of a fundamental understanding of change through iterative reflections and a mindful approach to react to its consequences. The change in understanding is a result of passing time and not an active approach JWL Theoretical implications Multiple recent and comprehensive models have presented holistic views and categories for the required workplace competences and learning paths. However, these views and categories are often generalized and lack information on how the categories relate to each other (Flores et al., 2020). For instance, the identified themes such as change or reflection are mentioned in several well-received models of learning such as the adaptive and developmental learning model by Ellström (2011) or sequence of learning actions in an expansive learning cycle (Engeström, 2011). What is missing, however, is the way change and reflection are interrelated in contexts that are not marked by well-defined objectives and processes. An example is how Ellström's (2011) model separates adaptive learning from developmental learning and maintains that adaptive learning at work occurs when employees perform routine-based actions and developmental learning occurs when, upon the emergence of a problem, employees deviate from the routine-based actions and engage in reflective actions while questioning the old ways. Although Ellström justly points out that, in reality, the well-separated categories of action and learning are fuzzier than the model shows, adaptive learning and developmental learning remain essentially separate categories of learning built upon separate categories of action.
The results of this study, however, show that developmental learning cannot be so neatly separated from adaptive learning. In fact, lack of routines and well-defined processes had created serious impediments in the engineers' learning process. In other words, the study results show that rather than being separate categories of learning, adaptive learning constitutes developmental learning. The objective of this study, however, is not to dispute the validity of the categorical learning models. On the contrary, our literature review exemplifies the validity of these categories on a general level once more. It is based on the insights provided by these models that we argue for the importance of theorizing the relationship among these well-established categories for the specific conditions that govern the manufacturing context.
For instance, this study proposes that given the characteristics of the manufacturing industry and its digitalization, change, reflection, and deliberation result in developmental learning under certain circumstances and through certain mechanisms. These categories and their relationship are presented in Figure 2. As the figure illustrates, although our literature review revealed change as an important trigger of learning, the engineers made us aware of the paradox of continuous newness inherent in their industrial sector. The need to balance the constant newness in their work pushed them to seek re-assurance in more experienced people or in structured ways of working present in industry standards and protocols ( Figure 3).
That is, the constant changes in the manufacturing industry create the need "to balance newness." This need, in turn, impels the engineers to seek "points of reference" to iteratively Industrial workplace learning reflect on the appropriateness of the technical processes. Finally, the need to identify points of reference continuously and iteratively requires suspending the individuals' subjectivity and agency (Emiliani, 1998) and organizing streamlined approaches towards competence development. In other words, the prevalent reliance on subjectivity and individual agency in the workplace learning literature (Migliore, 2015) needs to be balanced by external and streamlined approaches. However, the mere existence of formal/informal learning opportunities is no longer an adequate factor for learning. Instead, providing learning services should be replaced by organizing learning stages accumulatively (Stroud and Fairbrother, 2008;Rangraz and Pareto, 2020). Hence, while change, reflection, and deliberation are the constituent parts of workplace learning, we argue that "balancing newness," "finding points of reference", and "active organizing" of cumulative learning are the mechanisms of progressive learning (Figure 2). These mechanisms extend the literature on workplace learning beyond a categorical model of learning and illustrate the mechanisms through which such constituents play a role in learning in contexts that are defined by high degrees of innovation and volatility. As explained in the next section, focusing on learning mechanisms, rather than learning categories, highlights that although the identified categories of learning are already well-established and recur in different contexts, their meanings and implications change in specific learning circumstances.

Implications for practice
With the informatization of the manufacturing industry, and the increasing digital innovation in products and services, the number of R&D projects is increasing. The model of organizations is shifting from production firms to development firms with very different logics and strategies for learning. This is important because the decision to allocate resources to workplace learning is highly related to management's view of what is desirable in terms of supporting the firm's operational model (Ellström, 2011). For instance, learning in production organizations is based on exploiting the existing knowledge areas and making sure that the employees adapt to the best practices and established ways of doing things that lead to optimized production. Development organizations, on the other hand, promote learning activities that are based on exploring new areas to afford new possibilities and opportunities for ongoing development (Ellström, 2011). This is problematic, as learning for employees happens as a result of a balance between these two types of learning. Adaptive learning happens when employees adopt the established best practices, while developmental learning happens when a problem occurs, and employees reflect on the old ways to find a new way to solve the problem. However, with digital innovation, ongoing novelty finds primacy to ongoing continuity, change becomes a constant, routines and established way of doing things diminish, fluidity in internal and external collaboration increases, the structure of teams becomes fluid, and there is little time to think, observe, and exchange ideas to form deliberate learning plans.
Thus, adaptive learning on top of which developmental learning can be offered shrinks. Change is no longer a trigger for learning; it is the new everyday routine that clashes with the way learning occurs. Once the past and established practices shrink and innovation and the speed of internal and external collaboration increases, there is little time to reflect on past practices in retrospect. Reflection becomes a matter of finding immediate reference points that quickly review what is being done at the moment. Finally, deliberately designing learning activities to offer opportunities for competence development relies so much on the values that are remote in time and, therefore, less definitive. This is hard to cope with depending on the employees' years of experience and position (Ellström, 2011). Apart from the remote values and lack of time, the constantly changing structures of the teams affect the team members' relationship and trust which are key in planning feasible workplace learning plans (Eraut, 2011). In summary, the mechanisms through which change, reflection, and deliberation can be effective elements in workplace learning transforms with the increasing digitalization of the manufacturing industry. The significance of capturing this transformation is that workplace learning should be understood with respect to not only the conditions of the manufacturing industry but also its temporal conditions. That is, with the speed of change brought by increasing digitalization of the industry, learning approaches need to be evaluated and revised in shorter intervals.
However, our study is limited in the context of the reviewed literature. We merely selected the IS literature to establish an initial picture of how learning is addressed in the manufacturing workplace. We did so, as we believed that the IS literature with its focus on digital technologies in the workplace provides a closer picture of the characteristics of the manufacturing industry today. Future research can focus on other contextual specifications and extend this review to the research domains that more closely represent those specifications. We hope that this type of research contributes to the work environments where sustainable employability is valued through the active organization of learning opportunities. Industrial managers can arguably benefit from practical learning models that: are developed with exclusive attention to the specific conditions of the industry; and provide guidance by summarizing not only categories but also overarching mechanisms through which workplace learning objectives can be realized.