An in-depth analysis of programming in the Swedish school curriculum—rationale, knowledge content and teacher guidance

The term curriculum is often defined as the complete outline of education, the totality of the learning experience. That is, the term encompasses not only official curriculum documents but also the philosophy of education as well as methods for teaching, learning and assessment (Priestley, 2019; Sundberg, 2007). An even broader approach also includes non-governmental materials used by or designed for schools, such as textbooks. In this study, a narrower view of curriculum is applied and equates it with current official documents published by the school authorities. These texts stipulate fundamental values and overall goals for the education system and provide aims, content and knowledge requirements for individual subjects. For Swedish compulsory school, these texts are collated into a formal written curriculum which teachers interpret and transform into actual teaching (Linde, 2012) and thereby “shape the ‘what’, ‘why’, ‘how’, ‘whom’ and ‘whose’ of knowledge” (Adolfsson & Alvunger, 2018). The significant importance of the curriculum for teachers is well documented (Adolfsson & Alvunger, 2018; Lundgren et al., 2004; Wahlström & Sundberg, 2015). The formal written curriculum is accompanied by non-regulatory, complementary texts called commentary materials. They are intended to provide teachers with a more elaborate information about a given syllabus or a particular theme, for example school digitalisation.

Swedish compulsory school curricula have historically followed the northern continental curriculum tradition. Here, specific subjects constitute the core of the curriculum, unified by a set of broader educational goals “anchored in a philosophy of democracy as a collective and social concept” (Wahlström & Sundberg, 2018, p. 8). The northern continental tradition has generally entrusted teachers to exert their professionalism while transforming curriculum intentions into educational activities (Mølstad & Karseth, 2016). This curriculum model assumes that teachers are adequately trained and thereby have the capacity to make informed interpretations of the texts and to design teaching activities in accordance with what has been perceived.

Research has shown that realising reforms is a complex matter (Cuban, 2013; Fullan, 2009) and several key factors that affect reform implementation have been identified (Altrichter, 2005; Finger & Houguet, 2009; Fullan, 2001; Ryder, 2015). Fullan (2001, p. 72) organise these factors into three main categories: (a) Characteristics of change, (b) Local characteristics and (c) External factors. The present study concerns factors related to the characteristics of change, one of which is clarity. Clarity is considered a “perennial problem” (Fullan, 2001, p. 76) of reforms in the sense that issues related to clarity are present in most studies of reform implementation. Clarity is a notion related to the “quality of being coherent and intelligible” (Oxford Dictionary) and “freedom from indistinctness or ambiguity” (Dictionary.com). Hence, a message with a high level of clarity is conceived by its intended audience as distinct, comprehendible, coherent and unambiguous.

This does not mean that a message with a high level of clarity has to be simplistic in its nature and to be understood by “anyone”. Lack of clarity with respect to goals and means of implementation severely reduces the chances for intended and sustained change in practice. Fullan (2001) notes that while clarity is essential, it is difficult to determine and define what clarity means in concrete terms. As a result, false clarity may occur, characterised by an oversimplified interpretation of a reform message where vital and more profound aspects of the new policy are overlooked or misinterpreted and therefore excluded from implementation. According to Fullan, a situation of false clarity could arise if a teacher decides to rely too heavily on a new textbook¹ to cover necessary aspects of the new policy. That is, the intended reform may not be met to a sufficient level without understanding that the policy may “represent a system of practices” (Spillane, Reiser, et al., 2002) and require changes in beliefs and work processes. False clarity may also occur when a teacher recognises the message of a reform at a superficial level and dismisses it by claiming to already meet the requirements. If so, the teacher may disregard more fundamental aspects of the reform related to, for example, the novel content, teaching strategies and cognitive processes.

Fullan (2001, p. 77) concludes: “the more complex the reform, the greater the problem of clarity”. That is, not all reforms can be easily communicated (Altrichter, 2005). Hence, when studying teachers’ response to educational change, the characteristics of the innovation itself must also be taken into consideration.

The new programming policy has certain specific characteristics that require attention. For example, the complexity of learning programming must be acknowledged when investigating conditions for the reform implementation. Learning programming is generally considered to be a cognitively challenging process (Grover & Basu, 2017; Mladenović et al., 2018; Pea & Kurland, 1984; Qian & Lehman, 2017). Robins, Rountree, & Rountree (2003) summarise literature findings up until 2002 and describe challenges with designing programming education in higher education contexts. The conclusion is that “the average student does not make much progress in an introductory programming course” (Robins et al., 2003, p. 156). More recent research suggests that learning difficulties of novice programmers may be related to unrealistic expectations of what students should be able to achieve after an introductory course, not to the complexity of the subject itself (Luxton-Reilly, 2016). Nevertheless, both earlier and more recent research shows that introductory programming courses for adult learners still are facing a variety of challenges in terms of both learning and teaching. Even though the failure rates of introductory programming courses may not be as bad as its reputation (Watson & Li, 2014), the key issues still remain: how should teaching be designed and what content should be focused on to improve the learning situation for students (Crick, 2017; Luxton-Reilly et al., 2018; Medeiros, Ramalho, & Falcão, 2018; Pears et al., 2007). Moreover, the literature exploring teaching children programming lacks breadth (Crick, 2017; Kjällander & Petersen, 2016). Most educational research about programming has been carried out in higher education contexts (Crick, 2017).

Passey (2017) has presented six arguments that concern both societal and individual aspects which he refers to as economic, organisational, community, educational, learning and learner arguments. The economic argument concerns the necessity for schools to provide young people with the ability to contribute to growth of society and meet the needs of future employers, to become employable. Here it is argued that technology proficiency, such as programming, is becoming increasingly important. The organisational argument is related to the economic argument and raises the importance of collaborative skills. CS jobs are not generally one-man projects but often requires the ability to work with colleagues and peers. A community argument deviates from the urgency of CS from an individual or commercial perspective and instead recognises the potential role of CS in activities where schools interact with the community, for example, while doing study visits or working on socio-scientific issues (SSI). The community argument also identifies the relationship between formal, non-formal and informal learning and how CS can be used to satisfy community groups’ special interests and needs, for example, for a group of bird watchers. The educational argument recognises ubiquitous aspects of technology in society, and that elements of CS can contribute to the schooling of aware, digitally competent, citizens. The educational argument also recognises the need for lifelong education. The learning argument furthers the educational argument by being more elaborate about what fundamental skills and competences CS education supposedly facilitates, such as problem solving, creativity and logical thinking. This argument also encompasses ideas of moving away from ICT training to a more elaborate, in-depth approach to technology, such as incorporating programming or other elements of CS. The learner argument acknowledges the individual learner’s perspective in the sense that by incorporating CS initiatives, schools can provide opportunities for children to get interested in technology and CS and offer learners with special interest an opportunity for specialisation. Rationales for levelling a gender imbalance within technology education and careers are subsumed under the learner argument.

The framework for programming knowledge, designed by McGill and Volet (1997), represents knowledge types that together “form the basis of programming knowledge” (p. 283). It combines categories of knowledge derived from literature of educational computing and from cognitive psychology into a single conceptual framework. From the field of educational computing, three types of interrelated knowledge were identified:

From cognitive psychology literature, McGill and Volet incorporated three knowledge types:

By combining the two categorisations of knowledge types, McGill and Volet conceptualised a programming knowledge framework with four interrelated, but distinct knowledge types (Table 2): (1) declarative-syntactic, (2) procedural-syntactic, (3) declarative-conceptual and (4) procedural-conceptual knowledge.

Learning programming requires the learner to understand (type 1) and apply (type 2) specific language rules and features (correct syntax), understand the semantics of specific constructs and principles (type 3), for example, what a function does and why, and how such knowledge can be applied to solve programming problems (type 4), for example, design a program to approximate the value of π. The fifth dimension of programming knowledge, strategic/conditional knowledge (type 5), refers to the combined use of knowledge categories 1–4 to “effectively design, code and test a program that solves a novel problem” (McGill & Volet, 1997, p. 285). This is expert level knowledge and not only encompasses the knowing of how but also why, where and when this knowledge can be used.

The framework was originally developed as an instrument to diagnose learning deficiencies amongst novice adult programmers. However, the authors suggest that the framework is a potentially beneficial instrument for “designing appropriate instruction in introductory programming” (McGill & Volet, 1997, p. 294). As addressed in the Background, introductory programming courses (for adult learners) have been shown not to lead far in terms of students’ programming knowledge. One reason is that syntactic knowledge has received too much attention at the expense of other knowledge types (McGill & Volet, 1997; Robins et al., 2003). A “complete” learning experience would account for all types of programming knowledge represented in the framework.

The documents were selected for analysis based on the following criteria:

To deduce what texts that complied with the criteria listed above, the formal written curriculum was first scoured for instances of the word programming. This process confirmed that programming is primarily tied to two subjects: mathematics and technology, and to a lesser extent to the civics subject. Subject syllabuses form part of the overall formal curriculum but can also be regarded as separate curriculum documents, as in the case of the present study.

Programming as subject content is linked to the push for digital competence. Therefore, in this study, the subject-generic curriculum text (formal written curriculum minus subject syllabuses) was scoured for any instances of digital competence and digitalisation to collect information of potential value for the study. To better understand and analyse the intended curriculum, official complementary materials were added to the sample. Table 3 lists the eight documents selected for analysis.

Programming was added to the curriculum as part of an overarching reboot of the school digitalisation initiative, hence a complementary text that address cross-curricular and overarching digitalisation aspects is included in the sample.

The Swedish formal written curriculum for compulsory school (Lgr11) is the main legislative document which declares the main purpose and principles of basic education. In 2017, programming was added to the curriculum as part of a major curriculum revision. The curriculum document is divided into five parts, of which parts 1, 2 and 5 are within the scope of this study. Parts 1 and 2 present Fundamental values and tasks of the school and Overall goals and guidelines, whereas part 5 contains syllabuses for the individual subjects that children study in compulsory school.

A subject syllabus is structured into three sections, Aim, Core content and Knowledge requirements. The aim section in a syllabus consists of a text between ½-1 page intended to represent a concise and comprehensive account of the subject’s main characteristics and what knowledge pupils should develop. The core content is divided into three sections based on the three key stages in Swedish compulsory school: grades 1–3, 4–6 and 7–9. These sections are further divided into working areas which present the subject (core) content that should be covered during each 3-year period.

The selected documents have been examined by using qualitative text analysis (Kuckartz, 2014) with the purpose to discern ideas and meaning and thereby create an understanding for the textual content that goes beyond a summary of the obvious (Esaiasson, 2017). At an overall level, the procedure included the following steps:

After the relevant main documents had been identified (step 1), the analysis commenced with a close reading (step 2) to identify all meaning units, that is, words and sentences, relevant for the purpose of the study. Sentences which contained the word programming, or any derivatives of the word stem were identified. The process further required vigilance for occurrences of related terms (e.g. coding). Each instance was assessed in relation to its context and when necessary, complimentary preceding and/or succeeding sentences, and in a few cases full paragraphs, were included in the data corpus to maintain context.

Next (step 3), the identified text excerpts were collated in separate plain text files and thereafter subjected to distant reading (step 4). Distant reading (Moretti, 2000) made it possible to explore global features such as such as frequency and inter-relations of words. For this study, distant reading was utilised on both the subset of data extracted from the documents during pre-processing, as well as on the full documents. The plain text files compiled in the pre-processing were uploaded to the online distant reading tool Voyant Tools. Once the text files had been parsed in Voyant Tools, common Swedish stop words³ were filtered out prior to the application of additional filters and visualisation methods.⁴ The integrated tools Cirrus, Contexts, Collocates and Links were used to obtain a visual understanding of keywords and their contexts. Cirrus, for example, provided a word cloud of the most common words of the corpus. In a subsequent analytical step, the documents (Table 3) were parsed through Voyant Tools in their original form. This was primarily done to obtain basic statistics (e.g. word count) and a topological map, that is, a visual understanding of the documents in their entirety and thereby open for detecting any contributing information and intertextual relations.

The final step of analysis (step 5) was carried out three times, once for each research question. For the first research question, a set of keywords (e.g. develop, ability, understanding and knowledge), established during close and distant reading of the data set (steps 2 and 4), were utilised to identify arguments for the inclusion of programming. Once a sentence, or a combination of interlinked sentences were assessed to provide an argument, it was registered as an argument statement. For example, “through teaching, pupils should develop a general understanding of programming and how it can affect society” (Skolverket, 2017a, p. 8).

The analytical procedure for the second research question involved a repetition of the process of parsing through the data set sentence by sentence, this time to identify meaning units that revealed details about the properties of programming knowledge. The meaning units were thereafter assessed against the conceptual framework of programming knowledge ((McGill & Volet, 1997), to investigate if programming as knowledge content is represented in such a way that it was possible to identify what types of programming knowledge the curriculum intends to cover. Each unit was interpreted with good intentions to avoid an overly narrow understanding of the inherent intention, while it was equally important not to over-interpret its meaning. Table 4 illustrates two examples from the analytical process.

The third research question was answered by following the procedures for text analysis to identify meaning units which convey the intent to, for example, reinforce a particular matter or provide advice for the organisation of teaching, such as “In visual programming environments, for example, pupils can ‘drag and drop’ predefined graphical elements to compile their programmes. In everyday speech, this is often called block programming” (Skolverket, 2017b, p. 17). Four categories of guiding statements were discovered during analysis:

In total, 118 statements were identified for all three research questions, of which 27 were argument statements, 15 programming knowledge statements and 88 guidance statements. Twelve guidance statements were linked to two categories and were accounted for twice.

In general, subject syllabuses in the formal written curriculum are in its concise form sparse on providing arguments for content choices made. While the cross-curricular parts of the formal written curriculum (document 1, Table 3) do not specifically mention programming, they reveal information about broader aims and goals with compulsory schooling. The programming content in the curriculum is linked to these aims and goals through the digital competence framework (as previously outlined). The analysis identified 27 statements in four of the examined texts: seven in the cross-curricular parts (document 1), ten in the cross-curricular commentary material (document 2), eight in the commentary materials for mathematics (document 6) and two in the commentary material for civics (document 8). Amongst these statements, four of the argument categories (Passey, 2017) were identified: economic (3 statements), educational (18), learning (14) and learner (1). Neither the organisational nor the community argument could be identified in the data. Ten statements were assessed to combine arguments, mainly the related educational and learning arguments. The educational and learning arguments were most common.

The economic arguments are few and cautiously raise the importance of understanding society's digitalisation from an economic perspective. The Swedish school should prepare its pupils to “live and work in [a] society” (NAE, 2018, p. 7) where “algorithms and programming are used in many areas and professions” (Skolverket, 2017b, p. 8).

The educational arguments presented in the examined texts emphasise the role of technology in the society by arguing that it is necessary for pupils to learn how to “act in a complex reality with a vast information flow, increased digitalisation and a rapid pace of change” (NAE, 2018, p. 7). It is further argued that digital technology and programming are used to “influence course of events and public debate” (Skolverket, 2017b, p. 8) and that it is therefore important to “develop a critical and responsible approach to digital technology, to be able to see opportunities and understand risks and to be able to evaluate information” (NAE, 2018, p. 8) both for the individual citizen and the development of the democratic society as a whole.

The notion of problem solving stands out as a significant aim with programming in school. The learning arguments link digital competence and programming knowledge to problem solving. Compulsory education aims to stimulate pupils’ “creativity, curiosity and self-confidence as well as their desire to translate ideas into action and solve problems” (NAE, 2018). The learning arguments provided in the examined texts link digital competence and programming to creativity, critical thinking, innovative ability and entrepreneurship and seem to assume that such skills can be fostered through increased use of learning technologies and programming. In addition, the quote encompasses the learner argument by focusing on the process of the individual, as connoted through words like creativity, self-confidence and curiosity.

The positioning of programming in the curriculum and commentary materials is accounted for in three sections below. The first section concerns the representation of the notion of programming in the curriculum materials, while the second section addresses the function of programming in each subject. The third section presents an analysis of statements by applying the knowledge framework of McGill and Volet (1997).

There is no explicit definition of the term programming in any of the examined subject syllabuses or corresponding commentary materials. However, in the cross-curricular commentary material, a number of practices for problem solving are listed in an attempt to describe the nature of programming:

Programming includes writing code, which has great similarities with general problem solving. This includes problem formulation, choosing solutions, test and retest, and documentation. However, programming should be seen in a broader perspective, which also includes creativity, control and regulation, simulation and democratic dimensions (Skolverket, 2017a, p. 10).

There are nine statements in three different subject syllabuses (documents 3, 5 and 7 in Table 3) that specifically address programming. These statements are not isolated but constituents of established subjects and specific subject contexts. The contexts facilitate the process of interpreting and understanding the meaning of the statements. For example, the statements in the technology syllabus are part of the working areas of technological solutions and working methods for technological solutions and should be understood in those contexts. The aims sections of all three syllabuses emphasise domain specific conceptual understanding. For mathematics and technology, problem-solving ability is a desired learning outcome. The role of programming in mathematics is explicitly made visible in the syllabus which states that “through teaching pupils should be given opportunities to develop knowledge in using digital tools and programming to explore problems and mathematical concepts” (Skolverket, 2017d, p. 1). This statement illustrates that programming plays a supporting role in school mathematics, to be used as an assistive tool for mathematical endeavours. A similar picture of programming as a tool emerges in the technology syllabus. Three out of four statements (see no. 5, 6 and 8 in Appendix 1) ascribe programming to the role of an assistive instrument, to be used for control and regulation of technological artefacts. No explicit connection is made between programming and problem solving, even though the technology syllabus emphasises that problem solving is a significant part of the nature of the subject.

The cross-curricular commentary material provides a broadened perspective of the outcome of programming activities. Pupils should develop a “general understanding of programming” (Skolverket, 2017a, p. 8), and also “basic programming skills in addition to knowledge at a systematic level” (Skolverket, 2017a, p. 11), for which a specific language (teacher’s decision) may be used. The “general knowledge about and experience of programming” (Skolverket, 2017a, p. 8) is also intended to be used to acquire other, however, unspecified, knowledge. The technology syllabus provides no explicit details regarding the significance and role of the technology subject for the development of “basic programming skills”. Instead, the commentary material for technology refers to mathematics as “where pupils can learn about the programming fundamentals” (Skolverket, 2017c, p. 13). There is a corresponding statement in the commentary material for mathematics which declares that programming fundamentals are part of algebra and should provide pupils with the “opportunity to develop an approach to and knowledge of programming” (Skolverket, 2017b, p. 15). The wording “can learn” in the commentary material for technology (see quote above) creates some ambiguity regarding how mathematics and technology relate to each other when it comes to programming. Without reading the syllabus as well as the commentary material for mathematics, a technology teacher will not with certainty know whether pupils will learn the “fundamentals of programming” in mathematics or not. An intriguing follow-up question is what these fundamentals are in terms of programming knowledge, which leads to the analysis of statements by using the knowledge framework by (McGill & Volet, 1997).

A total of 15 statements were identified to contain information about programming knowledge (for details, see Appendix 1). As a single entity these statements seem to connote intentions that cover both the declarative and procedural knowledge dimensions. However, most statements did not prove to be accurate enough to determine whether they aim to cover syntactic and/or conceptual knowledge types for the particular context. There are three statements in the mathematics syllabus that together are claimed to constitute the ‘fundamentals of programming’ but it is difficult to interpret what this means in concrete terms given the scarcity of information the statements provide (see statements no. 1–3 in Appendix 1). Besides the frequent use of sequential instructions in terms of ‘algorithms’ and ‘step-by-step instructions’, there is little additional input provided through the commentary materials for mathematics, except a lone statement which mentions repetitions (see statement no. 11 in Appendix 1) in a context that makes it reminiscent of a programming concept.

Three of the statements in the technology syllabus (see no. 5–8 in Appendix 1) seemingly require the development of procedural knowledge, while one was assessed to be linked to declarative knowledge. Neither statement provides any details in terms of syntactic or conceptual knowledge. The commentary material for technology provides no details in terms of domain specific programming knowledge.

The formal written curriculum and its subject syllabuses provide teachers with information about subject aims, content, learning outcomes and knowledge requirements. Guidance in terms of teaching and learning activity design is not provided. Instead, in alignment with the northern continental curriculum tradition, the teacher is entrusted to understand context and terminology and exert professional teacher knowledge in the process of transforming curriculum guidelines into concrete educational activities. The commentary materials are written to complement the written curriculum with the intent to elaborate on content choices made, and thereby provide a better understanding of the curriculum or a specific subject syllabus. Therefore, also these texts were subject to scrutiny, despite their non-regulatory status, and that it remains unknown to what extent and how these texts impact teachers’ transformative work.

Statements that were classified to this category generally express cautious guidance towards, for example, cross-subject collaborative work. Examples of such guidance is solely provided through the commentary materials. The commentary material for technology suggests a joint effort with mathematics about “mathematical algorithms when it comes to programming” (Skolverket, 2017c, p. 17). In the same material for mathematics, it is suggested that the technology subject cater mathematics with ‘system level knowledge’ and an arena for practical work which could complement a more theoretical approach to algorithm design in mathematics. This corroborates with the statements made in the technology syllabus (see no. 5,6 and 8 in Appendix 1) which can be interpreted in such a way that, in technology, programming means less theory and more practical work.

Another example of guiding instructions for how teaching can be organised is expressed in terms of a statement in the cross-curricular commentary material. Here it is suggested that teaching should start from concrete and personally relevant situations and proceed to more comprehensive and complex technical systems (Skolverket, 2017a). This suggestion seemingly adds information to the commentary material for technology, which provides some suggestions about how to begin with programming, such as using tangible devices, programmed to “move in a certain way” (Skolverket, 2017c, p. 17).

Potential cross-curricular collaboration between mathematics and civics is elaborated upon in the cross-curricular commentary material. It is suggested that a collaborative effort on the design and effects of algorithms can give pupils a broader and a deeper understanding of how “they can be affected by and influence their surroundings” (Skolverket, 2017a, p. 11). However, this potential cross-subject collaboration is not mentioned in neither of the civics texts nor in the mathematics syllabus. In the mathematics commentary material, civics is merely referred to as provider of knowledge about societal effects of programming.

The cross-curricular commentary material gives an example of an educational argument and explains that programming is surrounded by continuous development and that programming languages can be changed or replaced. For this reason, the curriculum texts “do not mention specific programming languages” (Skolverket, 2017a, p. 8), but the text concludes that a specific language should be decided for in order to make sure that teaching provides the children with a “general understanding of programming” (Skolverket, 2017a, p. 8). In addition, the mathematics syllabus stipulates that programming in mathematics should be conducted in “different programming environments” (Skolverket, 2017b, p. 17) of which one should be a visual programming environment.

There is little room for explanatory amplifications in the subject syllabuses. Such information is instead provided through the commentary materials. In the commentary materials for technology and mathematics, the reader is provided with some insight into the notion of visual programming environments and related basic terminology:

In visual programming environments, pupils can, for example, "drag and drop" predefined graphic elements to compile their programmes. This is often called block programming in everyday speech (Skolverket, 2017b, p. 17).

The cross-curricular commentary material contains a similar, but differently phrased explanation:

In a visual programming environment, the code is represented by graphical elements that build a program. This is often referred to as block programming in everyday speech, but that term is also used in computer science contexts where the meaning is slightly different (Skolverket, 2017a, p. 2).

The latter also refers to the domain of computer science, but without any further explanation of what the differences in meaning are. While the subject syllabuses are written upon the assumption that the reader holds enough professional knowledge, the commentary materials can be perceived as written for the less initiated. The explanations border to being superficial at times, and whether these statements contribute to the understanding of the subject syllabuses is unclear.

The commentary material for mathematics adds information to one of the syllabus statements (no. 2 in Appendix 1) by giving a short explanation of what an algorithm is and can be used for (giving instructions to computers) and then suggests that these algorithms can be used to calculate the mean, or number sorting. The comment concludes with an organisational reference to the technology subject and the suggestion that mathematics provide the theory, while practical work could be done in technology.

In the commentary materials for technology, an explanatory statement elaborates upon a core content statement (no. 8 in Appendix 1) and suggests that control and regulation can be approached by working with tangible devices, programmed to avoid obstacles. Another example elaborates on one of the syllabus statements (no. 7 in Appendix 1) and suggests a teaching activity around electronic systems “that are programmed for something desirable to happen, such as turning on and off lights, or turning on and off or making sounds” (Skolverket, 2017c, p. 13). Overall, the examples provided in the commentary materials are approximately at this level of detail. A brief suggestion of a possible action is given, but often with little or no specifics about the value of the example in terms of knowledge construction.

The analysis identified four arguments, of which the educational and the learning arguments are the most frequently used. These arguments emphasise the role of programming as a means for achieving broader, interdisciplinary learning goals, such as problem solving, creativity, critical thinking and general proficiency with digital technology. The modest use of economic arguments and the absence of organisational arguments suggest that transnational education policy trends with focus on employability needs of the future economy (Wahlström & Sundberg, 2018), have made little explicit impact on the Swedish curriculum. The choice of arguments and how they are used seem to consolidate the Swedish curriculum’s bond to the northern continental curriculum tradition. However, in a broader sense, the 2017 Swedish education reform follows transnational educational policy movements that draw attention to the shaping of critical thinking, problem solving and autonomous citizens (Wahlström & Sundberg, 2018).

Neither of the governing subject syllabuses provide a definition of programming. The text that most closely contains a description is the cross-curricular commentary material. The subject-based focus of the northern continental curriculum tradition presupposes that teachers exert their professional teacher knowledge (Shulman, 1986) to unpack the meaning of the curriculum statements and make informed decisions about content, methods and evaluation (Mølstad & Karseth, 2016). In other words, teachers are expected to understand the concepts, practices and contexts of their subjects, and this is a feasible reason for why there is no definition of programming in the subject syllabuses. Regarding the commentary materials, however, it may have been helpful to provide more details about how the NAE defines programming.

The analysis of programming as knowledge content was made with the support of a programming knowledge framework designed by McGill and Volet (1997). As shown in the results section, the framework could be used to classify statements to a certain degree. It was possible to make distinctions between procedural and declarative knowledge, and in a few cases, also between syntactic and conceptual knowledge. Two of 15 statements could be fully classified according to the framework’s knowledge types. The result suggests that the level of detail in the statements is low in terms of programming knowledge. This reinforces the picture that programming is not considered as a knowledge domain in itself, but instead seen as a tool for exploring and solving mathematical problems and for controlling technological objects. The low level of detail in terms of programming knowledge may be a consequence of curriculum traditions as well as the design and context of the reform. The choice to present programming as a tool may also be a consequence of transnational policy trends which to a greater extent acknowledges the broader notion of CT (Bocconi et al., 2016a, b).

In 1984, Alan Kay argued that the computer, and consequently how it is given instructions, “is not a tool, although it can act like many tools” and described it as a “metamedium”, with “degrees of freedom for representation and expression” (Kay, 1984). The idea of programming as a powerful, adaptable tool is reasonable in a greater societal context, as argued for in the curriculum texts and elsewhere (Guzdial et al., 2019). However, there is an inherent complexity related to learning this tool that cannot be underestimated. Deliberate practice is necessary, mere access to a Ferrari will not make you a race car driver.

The commentary materials contain four categories of guidance statements written to assist teachers in their teaching, namely methods and organisation, explanations, examples and learning progression In terms of methods and organisation, the guidance cautiously proposes a coordinated interdisciplinary approach to programming where each subject (technology, mathematics and civics) should contribute with their pieces to the puzzle. However, this intent is not made clear in the separate subject syllabuses. Instead, any information on this topic must be picked up from the optional commentary materials. While mathematics seems to be the expected facilitator of a coordinated approach, the commentary materials contain limited information on how such cross-subject collaboration can be organised.

The progression statements cautiously propose a shift in operative tools as the children expectedly progress to a state where programming becomes a ‘learning tool’ (see statement no. 15 in Appendix 1). There are no explicit reasons given to why it is desirable to start with a visual programming environment and later move to text-based programming. Instead, the progression statements seem to imply that knowledge progression comes automatically if the complexity of the chosen tool increases. That is, the more complex programming environment, the more advanced the learning outcome would be. Hence, learning progression is seemingly reduced to a matter of characteristics of programming tools, rather than of growth of knowledge and skills. Guidance on learning progression in terms of gradual advancement of programming knowledge is left for teachers to decide. This issue seems to cause concern amongst teachers who believe pupils run the risk of encountering the same or similar content in several grades during an indefinite transition period, as most teachers and pupils lack previous experience of programming in teaching (Vinnervik, 2020).

There are aspects related to the programming discourse that are addressed very briefly or not at all in any of the texts, for example, learning outcomes, assessment and programming languages. The absence of more elaborate and direct suggestions to how teaching could be organised is likely an effect of the northern continental curriculum tradition, in which teachers are entrusted to make their own knowledgeable decisions about methods and tools. Even though the NAE generally provides assessment support materials through a separate website,⁵ the absence of guidance around programming and assessment in the commentary materials is unexpected. Programming is new to many teachers and knowledge assessment is closely tied to teaching and can be challenging. It is also noteworthy that the assessment website does not provide any material specifically related to programming.⁶ Deliberate attention to assessment is important as Grover and Pea (2013, p. 41) remark: “Without attention to assessment, CT can have little hope of making its way successfully into any K–12 curriculum”. Although their study is about the integration of the broader concept of CT, their conclusion is relevant for the Swedish context. As suggested by Åkerfeldt, Kjällander and Selander (2018), there is a risk that programming becomes focused towards drill and practice. Therefore, the broader, interdisciplinary and more difficult to assess (digital) competences could be overlooked.

In terms of programming languages, the NAE refrains from giving explicit guidance besides requiring that different programming environments are used. Using a programming environment also means choosing a programming language. Research suggests that programming languages that emphasise semantics over syntactics are easier to use for beginners, and such languages potentially makes it possible to give greater attention to programming principles and ”the problem” to be solved (Mladenović et al., 2018). However, the characteristics of an appropriate beginners' language is under debate, both for adult learners (Kunkle & Allen, 2016) and for younger learners (Kölling & McKay, 2016; Mladenović et al., 2018). Moreover, languages designed for younger learners may elicit different or new effects on learning, for which teachers have to be prepared (Swidan et al., 2018). Within the northern continental curriculum tradition, expecting teachers to make knowledgeable decisions about methods and tools, such as programming languages, may not be remarkable. What is remarkable, however, is that the question of programming languages is omitted from commentary materials, especially since it is known that most school teachers have little or no background in computer science. The absence of a discussion about programming languages and environments in the commentary materials reinforces the picture of programming mainly being considered as a tool for teaching and learning and that the unique characteristics of programming in itself are downplayed.