Learning in and from projects: the learning modes and a learning capability model

Abstract The notion of ‘project delivery’ is well embedded in and across the management and organizational sciences literature – generating a narrative that reflects and recognizes the instrumental nature of projects and programmes in strategy execution. Project management, as a distinct and well-established body of research enquiry, has increasingly sought to focus our attention on the impacts of complexity, risk and uncertainty in projects; the corollary being a desideratum to strengthen our theoretical understanding of how insight and learning from projects may influence improvements to organizational efficiency. The wider literature suggests that organizational learning remains a challenging proposition, particularly in the context of organizations operating in environments of high complexity. In this paper, we enhance the conversation on organizational learning through a series of case studies, generating evidence of thirteen ‘learning modes’. The paper proposes that mature organizations tend to exhibit a greater number of learning modes and that there is a tendency to capture and socialize knowledge with a greater emphasis on the context of the learning situation rather than the learning artefact in isolation. The empirical evidence gathered in this paper forms the basis of a capability model, characterized by the thirteen modes of learning. The model intimates that learning occurs, and is more effective, when knowledge and information are enacted in practice through the learning modes which form a nucleus of the organizational learning capability. The research concludes with a 'call to action' that emphasizes the strategic importance of learning practices and routines in project oriented-organizations.


Introduction
Projects and programmes are unique, impermanent actions typified by varying degrees of risk, uncertainty and complexity; these characteristics in themselves suggest that projects may not, in isolation, provide useful antecedents for organizational learning (see Rolstadas 1994). In this paper we argue that project oriented organizations should seek to develop 'learning capability' in their routines and practices as a means of developing new capabilities, encouraging innovation and maintaining financial sustainability. Thus, the desideratum to connect project learning to wider organizational practices such as strategy development, performance management, risk and knowledge management and corporate governance management practices is proposed.
Learning from projects and project management practice continues to deliver a rich vein of academic enquiry in the literature yet the majority of these endeavours adopt a 'sender/ receiver' approach (Hartmann and Dor ee 2015)even where a systemic organization-wide approach is taken to learning (e.g. Duffield and Whitty 2015). McClory et al. (2017) conceptualize the lessons-learned process by demonstrating how knowledge gathering methods fit within organizational routines; in this paper we advance those propositions by arguing that knowledge transfer and learning is a complex process, irreducible to simple material or signal transfer.
Our intentions lie in the domains of knowledge management, learning and dynamic capabilities. These lenses enable us to build on practice theories and in particular structuration theory (see Giddens 1984) and advance previous studies on organizational learning in the context of project studies (see S€ oderlund 2010; Byrne 2014a, 2014b;Davies and Brady 2016). Our conceptual framework seeks to connect the organization-wide structure, including organizational capabilities, resources and knowledge assets with the micro-level activities and 'episodes of learning', leading to the extension and creation of new knowledge and capabilities and ultimately, enhanced business value. Thus, the objectives of this research are: 1. To investigate the modes of learning in projects and programmes across a sample of organizational settings, and 2. To investigate the learning mechanisms (modes) observed in those project-based organizations that may lead to successful project capability development We draw on data garnered from participant organizations situated in Europe, the United States, Asia and the Middle East. The results uncover thirteen modes of learning from which, a learning capability model is proposed, representing a continuous learning-in-practice phenomenon that is enduring throughout the project lifecycle. This model describes a meta-capability and not only a process, in that it encompasses resources and behaviours in addition to processes.

Learning and knowledge management
It is entirely plausible that organizations perceive knowledge creation as fundamental to achieving and sustaining competitive advantage. A range of perspectives in the extant literature explore the mechanisms and processes associated with knowledge creation, utilization and transference. The traditional view classifies knowledge into distinctive and separable components: the first being explicit knowledge that is relatively simple to codify and transfer, the second being the more complex (arguably ephemeral) tacit knowledge that is acquired through cognition and is difficult to codify and transfer (Polanyi 1966;Nonaka 1991). This view advanced a general understanding of knowledge and remains influential today, particularly in practice but also in research (Zhu 2006;Tsoukas 1996). This taxonomic perspective of knowledge (Tsoukas 1996) builds on a positivist epistemological position and overlooks the complex social side (Easterby-Smith and Prieto 2008). However, some critics argue that this taxonomy does not provide a full understanding of the mechanics of knowledge (John Seely and Duguid 2001;Orlikowski 2002;Tsoukas 1996;Tsoukas 2009). Others view knowledge as a transferrable asset, for instance, Lin, Geng, and Whinston (2005) develop a framework for knowledge transfer from the sender to the receiver through a medium; similar to signal transmission in telecommunications. We argue that knowledge transfer and learning are more socially complex processes that cannot be reduced to material or signal transfer.
An alternative stream of thought 'sought to understand the nature of organizational knowledge through making analogies between organizations and the human brains on the one hand, and the organizations and the individual minds on the other' (Tsoukas 1996, 13). Here the different types of knowledge (tacit and explicit) are seen as inextricably linked. Tsoukas (1996) argues that tacit knowledge is inseparable from other forms of knowledge and that it 'is not made up of discrete beans which may be ground, lost or reconstituted' (p. 14), rather it is the basis for all types of knowledge, and that explicit knowledge always draws on a tacit component. Tsoukas (1996) argues that 'no set of rules can ever be self-contained, complete. Thus we are led to the conclusion that every act of human understanding is essentially based on an unarticulated background of what is taken for granted' (p. 16). Orlikowski (2002, 249-273) builds on this view to develop the concept of knowing in practice arguing that knowledge is not static but 'an on-going social accomplishment, constituted and reconstituted as actors engage the world in practice' (Orlikowski 2002, 249). For her, this reciprocal relationship between knowledge and practice implies the necessity to investigate both together, as she puts it: 'It suggests there may be value in a perspective that does not treat these as separate or separable ' Orlikowski (2002, 250). John Seely and Duguid (2001) also suggest that the practice perspective solves the paradox of the knowledge dichotomy of what they refer to as sticky and leaky knowledge. Ollus et al. (2011) argue that key stakeholders need to jointly participate in learning process to ensure collaborative alignment and to have common understanding throughout the project. Nonaka (1991) promotes the distinction between explicit and tacit knowledge, this is broadened in Nonaka and Toyama (2002) to an alternative theory, drawing upon Giddens' (1984) concept of the duality of agents and structures. In this alternative theory, Nonaka and Toyama argue that knowledge creation is a 'dialectic process where new boundaries are created through the dynamic interaction between the agents as well as between agents and structure' (p. 9). Similarly, Tsoukas (2003) argues that tacit knowledge can only be observed when it is enacted and put into practice. He explains that 'tacit knowledge cannot be 'captured', 'translated', or 'converted' but only displayed and manifested, in what we do'. Similarly, Al-Busaidi and Olfman (2017) argue for the importance of considering 'human factors' as they directly and significantly affect knowledge sharing.
In the previous paragraphs, we discuss the trends in organizational knowledge literature and the recent developments in linking knowledge-to-practice theories. Some authors have also addressed organizational knowledge and its relationship to other phenomena, such as dynamic capabilitieswith its origins in the resource based view of strategy. For example, Easterby-Smith and Prieto (2008) show that the literature on knowledge management and dynamic capabilities has acknowledged their symbiotic relationship, in that the creation and development of dynamic capabilities actually lies in the ability to create and transfer knowledge. Winter (2003) and Zollo and Winter (2002) argue that dynamic capabilities can be developed through the process of deliberate learning activities. Similarly, Nielsen (2006) identifies knowledge management activities that support the development of dynamic capabilities. He argues that the concept of dynamic capabilities can be understood from within knowledge management activities and advises that managers need to focus on knowledge management in order to operationalize dynamic capabilities.
Whilst project management has developed apace in recent decades to the extent that is a recognised capability, research continues to illustrate the challenges of learning from 'doing' projects (Newell and Edelman 2008). Typically, projects are characterized as temporary, multidisciplinary initiatives with unique outcomes. Learning in projects is thus influenced by temporal externalities, where emphasis lies in schedule compliance rather than developing long-term project management and delivery capabilities (Mainga 2017). An important research study on learning in project management is the empirical work of Prencipe and Tell (2001) resulting in the notion of three broad 'landscapes' for learning: explorer, navigator and exploiter which, respectively, draw on communication between people, basic information technology and advanced information technology. Their work builds on Zollo and Winter (2002) where the argument is that dynamic capabilities are shaped by three learning mechanisms: first is the experience accumulation, second knowledge articulation and third is knowledge codification. Zollo and Winter propose that experience accumulation is a semiautomatic process that is established through 'deliberate investments in knowledge articulation and codification activities' (p. 339). In their view, all three learning mechanisms must exist and interact in practice to build new capabilities. In another study, Newell and Edelman (2008) refer to learning in projects and the ability to transfer knowledge as a dynamic capability, since it is concerned with changing the current routines.
From the above discussion, we infer the following three premises: 1. Knowledge can only be observed and studied from within the practice in which it is instantiated. 2. Agents add their own understanding and knowledge to their actions when they follow rules or knowledge that is codified. 3. Learning is essential in the process of developing organizational capabilities.
Premise 1 includes all types of knowledge and argues that knowledge can only be observed in action. Premise 2 reinforces this point by arguing that what is traditionally referred to as explicit knowledge is non-static in nature; rather individuals add their different knowledge bases to it and hence they may have different perceptions about it and hence act differently.

The conceptual framework
In order to sufficiently investigate patterns of learning in project management and the mechanisms that lead to development of project management and organization capabilities, a conceptual framework based on theories of practices is required. The framework allows scrutiny of the social structure (Giddens 1984) of the organization and the activity configurations (Regn er 2008) allowing the possibility to investigate how learning happens and how organizations reconfigure their practices based on the results of learning.
The initial theoretical position of this research builds on the work of Easterby-Smith and Prieto (2008) and the structuration theory of Giddens (1984). Easterby-Smith and Prieto (2008) propose that learning is considered as the 'central mechanism' that links dynamic capabilities and knowledge management in an organizational context. This conceptualization is useful in providing a theoretical basis to the pilot-study. Results from the pilot study indicated the need for further research on the learning mechanisms that lead to the development of dynamic capabilities. For the purposes of more fully understanding those learning mechanisms, a revised conceptual framework was developed, as presented in the following paragraphs, which builds on the concepts found in the theory of structuration developed by Giddens (1984), especially those concepts relating to the characterization of agents as being knowledgeable and reflexive. Giddens (1984) considers social practices as the main driver and observable unit in social science where he argues that 'the basic domain of study of the social sciences, is neither the experience of the individual actor, nor the existence of any form of social totality, but social practices ordered across space and time' (Giddens 1984). In the modified theoretical framework we separate, for analytical purposes, between agents (knowledgeable and reflexive) and the prevailing social structures. The research framework also builds on the concept of activity configurations developed by Regn er (2008).
Regn er (2008) compares and contrast the strategy-as-practice approach with the dynamic capabilities perspective and how they can complement each other. The root of strategyas-practice is found in social theories (e.g. Giddens 1984;Bourdieu 1977) with the main intent of looking at strategy formulation and formation as an ongoing social practice (strategizing). Whereas, the dynamic capabilities perspective is rooted in evolutionary economics and is concerned about the firm level performance. Regn er (2008) develops a framework with 'activity configurations' as the unit of analysis and illustrates that 'activity configurations that involve specific combinations of certain actors, socio-cultural contexts, cognitive frames, artefacts and structural properties, besides diverse practices, are a more useful unit of analysis since they emphasize the significance of inter-linkages and interdependencies among these in the process of strategy formation over time'. In the framework, Regn er (2008) separates the social structure of the organization from agents and treats 'activity configurations' as an interaction between the two. This view was central in the development of our theoretical framework, as the research takes both a micro-practice and a system-wide perspective. This helps investigate the learning mechanisms at the temporary project level and their interaction with the permanent organization.
In the following section, we discuss the theoretical background of the research followed by discussion of the proposed conceptual framework and how it was informed by theory from the available literature.
The proposed framework suggests a practice perspective. The central intent is to separate, for analytical purposes, the structural properties (rules and resources) that govern the activities of doing projects and the related organizational activities in wider sense (such as strategy) from the actual actions of actors (learning and reconfiguration); in other words to distinguish between what organizations have (structure and resources) and what people do. This is important for developing the research methodology and for guiding the data collection.
The framework is shown diagrammatically in Figure 1. The upper part of the diagram represents the tangible elements. These include the formal knowledge components (explicit) that are codified as organizational assets, the formal knowledge management system (if one exists), and the current capabilities of the organization. In addition to this, the structure also includes all the resources that are possessed by or available to the organization from external sources.
The lower part of this framework contains the activity configurations (Regn er 2008). The argument here is based on the first premise that knowledge can only be observed and studied from within the practice in which it is instantiated; the agents' knowledgeability and motivation for action (Giddens 1984) are placed within the activity configuration, as a process of learning and reconfiguration of practices. This reflects the second premise introduced earlier.
The lens of investigation draws upon Giddens's structuration theory (1984); the actions are outlined as activity configurations (see Regn er 2008), implying a collection of actions that coalesce to form an observable set of activities that may be witnessed (and analysed) by the researcher. These activities represent project management practices (and routines) and related activities such as strategy formulation, portfolio management and organizational knowledge management. In this paper we argue for the connection of project management to the wider organizational activities and establish a basis for a more holistic and systemic recognition of the inter-relationship of projects and the broader corporate environment (Grundy 1998;Morris and Jamieson 2004).
Each episode of an activity configuration represents a small increment of time, like a snapshot, encapsulating a set of activities resulting in change. The social structural properties include the organizational structure, polices, and procedures. Feedback from activity configuration episodes, can lead to learning, which initiates changes in future activity configurations, and also to changes in the structure. The empirical data supports the view that this iterative process of learning and change leads to the development of new organizational capabilitiesreflecting premise three mentioned above.
The reverse arrows between the structure and activity configurations in Figure 1 represent this perpetual bi-directional feedback. It is referred to as the 'duality of structure' by Giddens (1984), signifying that the structural properties are both medium and outcome of the agents' activities. As opposed to Easterby-Smith and Prieto (2008), in this model the learning aspect and tacit knowledge are placed within the activity configurations part of the model. This accords with the perspective of 'knowing in practice' developed by Orlikowski (2002).
Following the above explanation of the conceptual framework, it is useful now to take a more detailed look at its two major parts: 1. Structure: This includes the rules, resources and dynamic capabilities. Another important aspect of the structure is the knowledge management as a system and explicit knowledge; whereas knowledge of how things are done in practice (traditionally referred to as tacit knowledge) is part of the activity configurations, as explained below. The reasoning for this is that tacit knowledge can only be observed when instantiated in action. 2. Activity configurations: Are the human actions within the organization. They include the activities of actors in performing strategic and project management related activitiesmore specifically, they include the existing and newly reconfigured activities that are the outcomes of the interacting elements in the upper part of the diagram in Figure 1, which are primarily responsible for the reconfiguration and development of new capabilities required to meet changing environments. The activity configurations also include the process of learning or knowing in practice.
The assumption here is that tacit knowledge is an active element of 'knowing' and is inseparable from action (Orlikowski 2002).
The developed conceptual framework connects the organization-wide structure including capabilities and resources with the activity configurations to allow investigation of learning patterns and mechanisms in project based organizations taking into consideration the organizationwide practices.
However, it is important to remember that people do not always act in a rational way based on the rules of the social structure, but their actions can also be influenced by their motivation for action. For instance, Tommelein et al. (2014) build on Schiefloe (2011) to develop what they refer to as the "Pentagon model" for analysing project organization performance. In the model, emphasis is on the formal and informal structures, social relations, networks and culture in project organizations and how they can shape actions and behaviours of actors. Rolstadås et al. (2014) point out that social relations represent an informal structure that defines relationships among actors such as friendships, alliance and conflict. Whittington (2010) discusses the tendency of agency to follow one social system or do otherwise and argue "everybody has some sort of social power". In this sense, actions that follow the norms of the social system or otherwise will have consequences based on the prevailing rules and at the same time gradually change the social structure in a duality based interaction (Giddens 1984).
This section has outlined the theoretical background to the research, briefly exploring the ontological and epistemological considerations that apply to our assumptions regarding the nature of knowledge and learning mechanisms within the project management and organization-wide related activities. This explanation of the theoretical orientation of our research is supported by a discussion of the key literature from which it has been developed. The resulting conceptual framework for this research has then been described, with an explanation of its relationship to the literature. This framework was used to shape the methodology for our empirical research, which is outlined in the next section.

Research methodology
The methodological approach seeks to complement the exploratory and inductive nature of the study. Our epistemological orientation draws predominantly from practice theories, and in particular, Giddens' structuration theory (1984), which suggests that a micro-practice perspective while paying attention to the wider organizational activities is suitable and appropriate to the nature of the data.
The methodology is characterized by a qualitative casebased approach, applied through a 'micro-practice' lens. By taking both a micro-practice and a system-wide perspective the research benefits from the collective learning capacity across multi-organizational domains and levels. This approach lends itself to finding deeper and richer learning episodes than would be possible if a more narrow perspective was used. An analogy here would be the iceberg model used by Bosch et al. (2013) to explain the study of evolutionary learning using a systems approach.
In comparison, traditional project management is historically limited to using variations of the familiar lessons learned process to enhance and improve project management practice (Thompson 2005). However, research has shown that this approach has suffered from a lack of effectiveness (e.g. Shokri-Ghasabeh and Chileshe 2014; Goffin et al. 2010) that may be related to its application within the relatively tight boundary that project management processes often operate in. Once we open up the process of learning to a more systemic analysis, greater opportunities for understanding issues and problems arise since our field of view increases to encompass more of the organization. This allows issues of higher complexity, spanning organizational internal and external boundaries, to surface and creates additional awareness and richness. A more parochial view may still identify, address and fix the symptoms of a problem but without a deeper understanding of the broader systemic causes will be of temporary and limited value. This represents a lost opportunity for deeper root cause analysis. Accordingly, we have designed the methodology used in this research to span the learning in projects and related organizational activities in organizations in a bid to open up new avenues of analysis to allow the research to uncover systemic learning mechanisms and processes.
In the theoretical framework we propose activity configurations as the unit of analysis (Regn er 2008), where the process of formation of (activity configurations) is influenced by the wider organizational conditions. This involves collective activities, including diverse actors from within, and external to, the organization drawn from the prevailing social structure. The expectation was that 'learning episodes' would reveal themselves within the activity configurations portrayed by the interviewees.
The methodology is underpinned by the conceptual framework and draws from the approaches developed by Regn er (2008) and Easterby-Smith and Prieto (2008). It was chosen to help with the identification of information to be collected and analysed (Miles and Huberman 1994) and for the identification of complex and causal relationships. Yin (2009) is firm in the need for a theoretical proposition for case studies before the collection of any data. The case study interview protocols and database coding structure were created according to recommendations for establishing construct reliability and validity (Yin 2009). More specifically the interview protocol design was guided by the structure and relationships in the conceptual framework.
The research seeks to explore the role of learning capabilities and dynamic capabilities in the projects. Specifically, the researchers hope to shed light on how these capabilities interact with other capabilities of the organization to facilitate balanced and dynamic changes that ultimately leverage competitive advantage. A qualitative approach is preferable to a quantitative approach: qualitative research (Creswell and Poth 2017) enables identification of the contextual factors that relate to the phenomena of interest, facilitates the description of complex phenomena situated and embedded in specific contexts, and is useful for studying a limited number of cases in depth. It can also yield a much richer and more detailed picture.
A total of 47 interviews were collected from 23 organizations in 6 countries in Europe, the Middle East, Asia and North America. Organizations included a mixture of private and public sector organizations; large, medium and smallsized organizations. Data were collected from a variety of job roles, levels and projects in the organizations studied. The process of finding interviewees varied depending on the relationship established between case organization and the researchers. For some organizations, the main contact helped to identify and get approval for a cross-section of people to be interviewed. For others, especially smaller organizations, it tended to be based on personal contacts and networks. This purposive selection of samples is important in qualitative research, especially where the target sample is small compared to the population (Miles and Huberman 1994).
A limitation of this research and possible sources of error can be in the diverse nature of organizations studied belonging to diverse cultures and legal frameworks, which limits the characterization and classification of organizations. Another limitation lies in our subjective classification or organizational maturity rather than using an attested maturity model.

Data analysis
In this section, we present the analysis of data. The interviews were transcribed and coded (using the Nvivo 11 software package) following the structure described in Table 1, which includes all variables found in the conceptual framework ( Figure 1). Consequently, a series of data tables at an increasing level of abstraction were constructed as described below.

Level 1: data analysis tables
The results generated from the interviews within each organization are presented in a single table format; each represents a unique organization and is based on a template structure (see Table 1) similar to the conceptual framework described in the previous section of this paper. By way of introduction, each table is accompanied by a brief description of the organization and a brief summary of the key issues emerging from the interview data. The size and structure of each data analysis table reflects the number of interviews conducted and the richness of the data.

Level 2: data analysis cards
Following the initial data analysis, a second level of data examination and analysis was performed in order to identify specific learning episodes within each organization; the purpose being to reveal learning practices that were enacted and the modes of capabilities, which the organizations developed as a result of these. A total of 60 data analysis 'cards', each one representing a discrete learning episode (activity configuration), were developed using the template shown in Table 2.

Level 3: summary analysis table
A third level of analysis was performed, at a higher level of abstraction, in order to synthesize the emerging themes derived from the learning episodes identified across all 23 case-study organizations. The resulting analysis is presented in a summary table that considers: (i) the learning episode code, (ii) type of business, (iii) size of business, (iv) maturity of business, (v) trigger area, (vi) enabling/dynamic capability, (vii) locus of new capabilities, (viii) learning practices and (ix) learning modes. It is worth emphasizing that it was the learning modes which held the most pertinent information associated with the objectives of the research. The end of the analysis identified a total of thirteen distinct learning modes. These are discussed at length in the next section, which gives rise to Level 4 of the data processing and analysis work.

Level 4: cross-case analysis
An in-depth cross-case analysis completes the data processing and analysis work. In the cross-case analysis, each of four sets of learning mode is used as a basis of comparing the results of the different cases study organizations, with a reflection on the type of capabilities associated with each of them.

Learning modes
As discussed in the previous sections, data was analysed to find learning practices that lead to new organizational capabilities or the modification of existing ones. From the data, we identified clear and discrete learning modes, which represent a summation of the learning practices identified in the research. Figure 2 represents the final set of thirteen learning modes discovered by this study. We do not suggest that this set is exhaustive as other modes of learning can still be discovered by further research in the future. Rather, they form the nucleus of the learning model ( Figure 3) and are the active components of learning capability.
The first set of learning modes are related to codification of knowledge and how codified knowledge is used in practice. They are as follows:

Codification of individual knowledge
This refers to the situations where the organization realizes the need to develop standardized processes, procedures, policies, etc. The data shows that, in this mode of learning, the organization recruits qualified employees or uses existing employees to write the documents needed to establish standardized processes and policies. These documents are essentially codified artefacts of corporate knowledge, which are then embedded in the governance mechanism of the organization at different levels. These learning artefacts shape and direct the behaviour and practices of employees. Data from interviews show that this mode of learning usually happens in less mature organizations.

Practice to codification
In this mode of learning, organizations deploy a range of tools and techniques to capture or codify lessons from routines, which can later be put into practice. The simplest way is similar to capturing tacit and explicit knowledge in classical knowledge management (KM) systems or by using the knowledge to develop or improve processes, procedures, policies, etc. The data indicates that high maturity organizations also capture information about the context of situations and the social traits and personalities of the people involved. For example, capturing lessons learned within the context of the prevailing culture and affected people. For instance, a senior officer of a high maturity organization one of the cases says about lessons learned: We have them all [lessons learned] available, and we have them all tied to personalities that were available Another manager declares: Knowledge is contextual, it is about the situation, you can work a project effectively in one place and you can come to our organisation and you can fail because of the dynamics, the politics, the people and the nature of things are done differently.
The data shows that this learning mode in its simplest form is followed by high maturity and moderately mature organizations, but high maturity organizations tend to place greater emphasis on the context of learning situation and the people involved. The type of capabilities developed as a result of this mode of learning span from document control to global project management and finding a balance between process rigour and fast decision making.

Codification to practice
As discussed in the first learning mode, codified or sanctioned knowledge can be in the form of a process or a policy developed by experienced or knowledgeable people for other people to read and follow in practice; in this way some of the knowledge is transferred from one to another. Another form of learning can be by way of transferring lessons learned and case studies collected from practice in a documented format (e.g. in a knowledge management system). The challenge here is how to transfer maximum knowledge, meaning how to maximize the throughput of learning from an existing or past situation, via documentation, to learning and onto improved future practice in a new location or context. Naturally, knowledge will fade when passing through each step in this knowledge transfer process.
The second set of learning modes is contextual learning which includes: understanding context and research.

Understanding context
Evidence from the interviews demonstrates that mature organizations are acutely sensitive to and aware of context; this mode of learning is dominant where organizations enter (or seek to enter) new markets and where they desire to appreciate the legislative and cultural landscape and how this may impact upon their strategy.
The interviews provide a useful example, where a senior manager opines the necessity to understand the cultural differences in the various offices of the company around the world: We always jog between the company spirit and at the same time trying to preserve the local culture. So this is a very difficult trade off to manage but that's why we have chosen IPMA; the choice of certification for IPMA for example is based on that This mode of learning is usually associated with the development of global project management capabilities, the capability to balance control and governance with flexibility and the capabilities to penetrate and develop new markets. The data obtained from the interviews provide evidence that organizations which place an emphasis on contextual learning are more mature in these types of capabilities. In this regard, understanding context or contextual learning can be considered as being a dynamic capability in its own right, which can generate new globally-oriented capabilities.

Research
This mode is usually associated with the capability to collect required information such as contract laws and legislation in different countries. Examples of organizational capabilities associated with this mode of learning include the capability to enter new markets or engage in international contract management. For example, one of the companies studied while entering a new market realized risks that can arise from the different contract laws they will have to use. Because of that, the company decided to establish a legal department, which contributed to improving its capabilities in global project management and international contract management. Another company studied in the research conducted a big data search and developed fifty new documents related to standards and procurement conditions. From the data, one of the important capabilities of a high maturity company, which is a global development and manufacturing company, is to understand the contexts of different markets in relation to customer requirements and tastes. The company gathers extensive information from different countries, including the different standards and regulations. This capability is also linked with another capability, which is the capability to develop products for different markets with more common elements and less customization. A manger says: You have to make a decision in your design where to differentiate and where to keep the same … very few products in the world appeal to everyone in the world, take for example the iPhone, I don't know how they do that! From the above, there is evidence that the company is aware of the need to further develop the capability of developing products that appeal to everyone in the world. Several capabilities are necessary to do that. The capability to do market research and research about standards and regulations is also vital.
The third set of modes is related to organizational wide practices beyond the context or project management, including:

Learning from external sources
Interviews show that less mature organizations learn more from external sources than internal learning compared to mature ones. External sources of learning in less mature organizations are mostly through learning accomplished though external consultants, whereas mature companies adopt a more systematic benchmarking approach.

Employee feedback
By this, we mean employees offering feedback on practice or routines for the purpose of improvement; this can manifest in the form of employee suggestions. Employee feedback can be a source of learning and reconfiguration of practice, leading to new capabilities.

Quality management tools
There is some evidence from the data that the use of quality tools and techniques can offer good avenues for learning. Examples of these include root cause analysis, inspection and value engineering. An example of this mode of learning can be seen in one of the case studies, which is a design and construction company for large infrastructure projects; due to delays in projects, they established a routine for senior managers to regularly visit and inspect their construction sites. This new routine opened up many opportunities for learning and troubleshooting and fast problem solving.

Performance management
Similar to the use of quality tools, there is some evidence from the interviews that performance management frameworks, such as the 'Balanced Scorecard' (see Kaplan and Norton 1992), can also help in knowledge sharing through learning about the organizational successes and issues. Organizations need to develop a performance management system that helps achieve their strategic objectives and use the right measures and indicators that can lead to better learning. This is in line with the findings of Fouch e and Rolstadas (2010) where they describe performance measurement in projectized organizations 'as a core process and holistic basis for the control of single project life cycles as well as experience transfer and organizational learning in a longer term multi-project perspective'

Training and education
This mode is simply the traditional class learning in education or training courses.
The fourth set is related to the prevailing culture of the organization, including:

Innovative thinking
Fostering and encouraging innovative thinking can be a rich mode for generating new ideas, especially ideas that do not follow traditional ways of thinking and learning. An example can be seen in Case H, a telecommunications operator, at the time of the financial crash, when businesses were losing and shrinking, this company spotted an opportunity for growth and decided to expand their network to improve service quality and increase market share. One of the interviewees described this situation by saying: During the financial crash we decided to invest as we saw an opportunityeverybody else was pulling back … We invested in the infrastructure, you know, networks for better quality and that gave us a competitive advantage, and enabled us to capture the market share.

Leadership led
In this mode, top management who give direct instructions on the improvements to be made, which are then followed by employees to improve practices, drives learning. This mode is predominantly observed in moderately or less mature organizations where top management are the catalysts and drivers of learning as a means to achieve improvements.

Face-to-face
Linked to the previous discussion of 'practice to codification' and 'codification to practice', a more commonly used learning mode in practice is face-to-face; this is widely used by mature and moderately mature organizations. Mature organizations seek to find venues for more informal ways for face-to-face interactions to occur between employees, even globally within global teams. This mode of learning was illustrated within the research by practices such as encouraging discussion during review meetings and in-company training. This mode intersects with many other modes and practices of learning, for instance all sorts of meetings can facilitate face-to-face discussions and learning. Among the many benefits of teamwork is that it is also a way of sharing face-toface knowledge. An interviewee explains: I think the more powerful one is informal; you need to have people, so we learn from people to people whilst another manager states: The project delivery profession was set up as a means of creating informal knowledge exchange opportunities.
The above quotes emanate from participants situated in high maturity organizations and appear to indicate a greater recognition amongst such organizations, than seen in less mature ones, of the need for informal and face-to-face learning modes.
What is noticeable is that mature and moderately mature organizations show evidence of more modes of learning compared with less mature ones. It is noticeable that less mature organizations have a greater tendency to learn from top managementleadership led learningand from individuals. Analysis of the interviews shows that maximum learning happens in the more practice-based modes of learning. It also shows that for an organization to be a mature leaning organization, they need to invest into finding venues for learning in practice to happen with emphasis on the modes of 'practice to codification' and 'codification to practice'. The challenge here is to maximize the throughput of putting learning into practice. This is of high importance in the management of projects, where the organizations and individuals involved often work under pressure of multiple constraints. Therefore, project management methods and processes need to find new ways to facilitate the maximization of learning in practice. This may require new practice learning oriented processes or possibly a new knowledge area, yet undefined, in the major project management bodies of knowledge. Such practice oriented learning processes, once recognized and accepted, will lead to new roles for new members of the project team or additional roles for existing team members.

A learning capability model
The results of this study, which are based on a micro-level perspective, have produced empirical evidence of the preference of a practice-based approach to the understanding and enactment of learning mechanisms in organizations. The main premise is that learning takes place within the sphere of people that are interactingindividuals or groupswith the social structure of the organization, including the existing knowledge, resulting in a new knowledge base and improved capabilities and business performance. This perspective resonates with the notion of duality of structures proposed by Giddens (1984), a construct he called structuration, which describes how knowledgeable agents draw from social structures in producing and reproducing social systems. Here we refer to the 'knowledgeable agents' in the general sense of structuration theory, not to people with greater knowledge in a particular field. In this section, we present a general model for learning capability in organizations. Figure 3 illustrates the learning capability model. This model does not represent purposeful events of learning in isolationsuch as lessons learned discussionsrather it represents a continuous 'learning-in-practice' phenomenon that occurs before, during and after project delivery. It describes a meta-capability (and not simply a process), in that it encompasses resources, behaviours and processes. We argue here that the actions of the leaders of organizations can foster or hinder this learning capability. Hence, for learning capability to be developed and to function effectively, leaders need to develop and maintain conditions which are cognisant of the internal and external context including stakeholders (Harrington et al. 2016  visible failure events, which attract top management attention. Failures with big impact engage top management in the analysis and investigation of reasons and remedies. By contrast, organizations can easily miss the opportunities of learning from successes and good practices in other parts of the organization. If these successes are not captured and institutionalized across the organization, this represents a failure to learn. Hughes et al. (2016) refer to this as 'Postmortem process' in their study to investigate factors for project failure. Chipulu et al. (2019) give a dimensional analysis recognizing the non-symmetrical relationship between assessments of success and failure in projects. Here we argue for both success and failure, big or small, as inputs to the learning process. In this way, learning is a continuous process for individuals and organizations. There is strong evidence from this research that successful and mature organizations put huge emphasis on capturing and understanding the internal and external contexts of learning situations. For example, understanding the market they operate in, including, for instance, its legal, cultural and commercial aspects. In the internal context, organizations need to understand the culture and norms of their organization. Data shows that mature organizations invest in capturing these elements of context, including the personality traits of the people involved in learning events. A third input to the learning mechanism is the existing capabilities as the aim of learning is always to improve current practices and capabilities.
The central part of the model (learning modes) represents the device of learning including the thirteen learning modes presented in Figure 2, where we argue that learning happens in a reflexive fashion between sensing and reflecting and by interacting with the modes of learning shown in the central part of the figure. The construct of sensing and reflecting is at the heart of Schon's reflective practitioner construct (Sch€ on 1983), which has been studied extensively in healthcare contexts (see . Later, we argue that being a reflective practitioner is a critical element of an individual's learning skills repertoire that facilitates effective learning in a practice-based learning environment. With regard to the 'sensing' element, we are not only referring to a macro representation of 'sensing market and technological opportunities' as argued by Teece (2007Teece ( , 2009), but we extend that to the micro-level representation of an organization or individual sensing for a need for improvement or change in process or in practice.
Sensing here can come from top management or an individual on the shop floor and needs to be followed by a reflection on activities and experience (Giddens 1984). Only then can structural changes take place and new or improved capabilities develop. Employees at different levels can formulate ideas and opinions about how the organization does its business, but only when these ideas are considered and acted upon can they contribute to performance and bottomline improvements.
This cycle of sensing and reflecting requires mediating policies and cultures that enable it to operate and create value. This can be difficult to establish and typically requires the intervention of senior leadership in creating the essential conditions. In the centre of the learning cycle are the learning modes described in the previous section. We do not argue here that this is an exhaustive or inclusive list of modes, rather these are modes observed in the course of this research and that further studies are desirable in identifying supplementary modes of learning.
The modes of learning are the seeds for learning; they operate in both directions in the cycle of 'sensing' and 'reflecting'. For instance, sensing a need for improvement in a technical aspect can trigger a 'training and education' mode. In addition, as a result of 'training and education' employees develop new competencies and can identify or sense a need for change or improvement. 'Reflecting' is the element of acting upon 'sensing' where an individual or a group within the organization takes steps towards finding solutions or alternatives for possible changes. This leads to the 'Reconfiguring (make changes)' step, where actual changes take place in the form of codified improvements or change in the norms and culture, paving the way for improved or new capabilities.
For the above to take place, we identify the following conditions that are required to be in place for an effective learning capability: 1. Leadership commitment 2. Developing and facilitating diverse modes of learning 3. Create opportunities for employees from different parts of the organization to meet and share knowledge, especially from different parts of the world in the case of multinational companies 4. Balance between rigour and freedom 5. Embedding learning in project management methodologies and processes As mentioned above, there is a major role for leadership in the development of learning capability in an organization; this leads all the other four conditions. There is evidence that mature organizations follow diverse learning modes to facilitate maximum learning opportunities. They also seek to establish informal conversations as a powerful learning opportunity for employees to learn from each other. An outcome from this research is the need for balance between rigour in following processes and policies versus freedom of decision making and adaptation. This is of high importance for the proposed model to operate as illustrated in the central part of the diagram. For people to sense, reflect and reconfigure their actions; the right balance needs to be found. An employee who is forced to rigorously follow processes will not sense and reflect effectively. Similarly, too much freedom from very loose processes or a lack of processes, especially as an organization grows, can lead to chaos. The last condition identified above is to embed processes and roles in project management methodologies to encourage and facilitate the learning process. We argue that this creates a need for new learning-based processes and roles that are not yet included within the published project management bodies of knowledge, but which should be considered and certainly be the subject of further research and experimentation.

Conclusions
This paper argues for the intensification of practice-based learning mechanisms in project delivery. The research identifies thirteen learning modes that are situated at the nucleus of a novel learning capability model. The results suggest that project management governance, methods and processes should seek to embrace new methods and methodologies to enable the maximization of learning in practice. This may require new, practiced based and more socially oriented processes for learning or possibly a new knowledge area, as yet undefined, in the major project management bodies of knowledge. Such new learning orientated processes, once recognized and accepted, may generate new roles, requiring novel capabilities, for members of the project team and its wider network of stakeholders.