We focus on novice programming processes in secondary education. In Germany, prospective high-school CS teachers often face the problem that they have only very little access to actual school classrooms during their teacher preparation programmes. They only learn about teaching programming from a very abstract point of view in lectures or seminars, usually with a focus on research rather than on practice. Once they arrive in schools as graduated teachers, they are often faced with completely new situations and rst have to learn how the pupils learn, program, debug, and think. To bridge this gap between university education and actual school practice, we propose a new methodology that we are currently integrating into our curriculum. Our main approach is to activate re ective processes in prospective high-school teachers by rst having them watch video-material from actual programming lessons at school, and then giving them exercises that let them re ect on the material they have just seen, e.g. by asking them about possible misconceptions, or letting them speculate about the pupils' thought processes. The goal of this article is to develop a theory (supported by learning theories) on how such videos and accompanying exercises should be composed in order to activate deep re ective processes. We believe that our methodology is general enough to be applicable to a wide range of programming processes, independent of actual programming languages or paradigms, age of the pupils and students, or other external factors.
Programming is an important topic of computer science courses in school. It is not only the primary experience and fundamental activity of computer scientists, but also enables the demystication of the machine. It allows, for example, to move from being a consumer of software to become a creator thereof. Therefore it is important to provide learners with high quality programming lessons.
To ensure this quality, it is mandatory to properly educate teachers about
how students learn [
An important part of this knowledge is the understanding of student
behaviour, particularly what students do while coding. Its importance comes from
the value that this knowledge can provide in dierent ways. Luxton-Reilly et
al. [
While the literature review of Luxton-Reilly et al. [
Because of their qualitative nature we mainly use qualitative content analysis
[
The last two research questions are highly dierent in nature and require a
dierent approach. The main dierence is that we want to develop, analyse and
rene our own instructional approaches in our university courses for prospective
high school teachers. To accomplish this, we use a design-based research setting
[
Reection is generally agreed to be important for professional development
[
In this paper we describe our framework in detail. In Sect. 2, we dene some terms that have been used liberally above. In Sect. 3, we discuss the concept of reection and its signicance to our project. In Sect. 4, we present the considerations serving as the basis for the video material we use. In Sect. 5, we describe our course design. Finally, in Sect. 6, we give an outline on which further steps are required to achieve the long-term goals. 2
Because our project takes place on multiple levels of the educational system,
it is reasonable to dene how we refer to dierent actors and entities. To this
end, we use the nested tetrahedron model [
(a) Didactic Tetrahedron [
(b) Nested Didactic Tetrahedron [
The didactic tetrahedron identies four components as important parts of learning processes in institutionalized learning settings (see also Fig. 1a). The rst dimension ( L) refers to the persons who experience the learning process. The second dimension (T ) references the actors who facilitate the learning process. The third dimension is called subject matter ( SM ) and describes what the learners are learning.
The dimensions mentioned so far constitute the classic didactic triangle. What turns it into a tetrahedron is the last dimension ( A), which considers any thing that serves a didactic purpose. Its importance arises from the idea that human actions manifest themselves as interactions with the world, or rather artefacts therein. (See below for concrete examples of artefacts.)
We will now model the subject of the rst two research questions (RQ1 and RQ2), which refer to the school level.
Learners (L) Our main focus lies on pupils in higher secondary education, while we do not exclude other learners categorically. The important aspect with respect to our analyses is that the learners should only have little prior programming experience and work with a text-based programming language.
These learners will mainly be referred to as pupils.
Teachers (T) The role of the teacher is carried out by the regular CS teacher of the class. However, since we focus heavily on processes happening during learning, the teacher plays a rather minor role.
Subject matter (SM) We focus on programming in general and require the learners to be active. Any task that asks for programming related activities lies in that scope, while, for example, a lecture on syntax would not. Artefacts (A) The artefacts that the pupils will interact with depend on the concrete tasks they are working on. In general, they will always create or modify a source code, using a programming language and programming environment of some kind, using input devices such as keyboard and mouse. This means that we do not focus on unplugged activities, even if they involve working on program code, or on programs running on external hardware like microcontrollers or roboters. 2.2
With RQ3 and RQ4 we introduce the university as a second level at which we want to analyse learning processes. Since this second level might cause terminological confusion, we use an extension of the didactic tetrahedron as a conceptual framework to describe the remaining parts of our project.
Since the learning processes at university level are similar to the ones at school, the didactic tetrahedron from Sect. 2.1 would be sucient. However, now the complete learning processes at school level will act as the subject matter at university level. This means that the tetrahedron from Fig. 1a acts as the vertex ’subject matter’ of the tetrahedron at university level.
Fig. 1b illustrates the relationship of the two tetrahedra and conceptualizes the learning processes at the university level. The dierentiation between the two tetrahedra is accomplished by adding a leading U to the labels. We can now describe the aspects of our project that are related to teacher training. Learners (UL) Our focus lies on university students with the goal of becoming school teachers. These learners will be referred to as students.
Teachers (UT) The learning processes at university level will happen during computer science teacher training courses. Because of that the teachers (UT) are the course instructors.
Subject matter (USM) Our main goal at university level is to enable the students (UL) to reect (see Sect. 3) on their teaching and the pupils’ learning processes. To achieve that, we want them to rst gain knowledge about pupils’ learning processes and their associated diagnostics.
Artefacts (UA) The main artefacts that the students will use are videos showing pupils programming and exercises related to those videos. Sections 4 and 5 will explain those two artefacts in greater detail. The nal goal of our project is to provide a digital learning platform that includes both the videos and the exercises. 3
Like a computer programmer, a teacher needs to nd and x problems in his
or her instructional approaches. The task of debugging them requires the ability
of what is called reection . And since assuring the ecacy of one’s teaching
activities is an important part of a teacher’s competencies, the development of
the ability to reect on pupils’ learning processes should be an integral part of
teacher training curricula. As ClarÆ [
To nd a suitable denition, we take a look at the very roots. According to
Dewey [
From this quote we derive the three major foci that we associate with reection. They are (a) beliefs and (supposed) knowledge, (b) grounds, or rather reasons, and (c) derived conclusions.
Additionally, for (a) Dewey dierentiates between descriptions and
interpretations [
We want the students’ to reect on all aspects of the pupils’ programming processes. Beliefs and knowledge (a) will therefore be addressed by a detailed description and interpretation of the pupils’ actions. A possible statement could be: The pupil tries to declare a variable by typing the words ’Int’, ’init’ and ’Init’, before giving up. That statement contains an observed action (typing of words) and its interpretations (attempt of variable declaration; resignation). Based on the given statement, a proposition for an underlying reason (b) might be: The pupil seems unaware of the correct keyword for integer variables and case-sensitivity, since he inputs words similar to ’int’ while varying capitalisation. A conclusion (c) might be the proposition of an action that should be taken, like: The pupils should be in some way instructed on the importance of correct notation of keywords. Another possibility is a description of an alteration in the personal point of view on something: Until I saw the pupils struggle, I didn’t think that syntactical rules could be a challenge. Finally, even general statements of (hypothetical) facts might serve as conclusions: This example shows that the syntax of a programming language is in no way trivial to learn.
All the given examples would be interpreted as indicators of reective processes. To summarize what we mean by the term reection , it can be described as a reconstruction of the scientic investigation of the programming processes. However, we do not require the students’ analyses to be as rigorous as our own, but rather a small scale version thereof.
It is still open how such reective processes might be fostered and how to
support them. One answer might be suggested directly by Deweys and Schns
works: ClarÆ [
To be able to perform such confrontations, we use video recordings of programming pupils. How these videos are composed and recorded will be described in the next section. 4
In the previous section we argued that we want to confront the students with their incoherent understandings in order to achieve reection. Based on our research focus, these confrontations will be about viewing pupils’ programming processes. We identify the following properties that the material (USM) should full.
Authenticity: Inauthentic situations would be perceived by the students as staged. This would directly resolve any incoherences in the understanding, as the students could possibly just argue that the pupil’s perceived behaviour can be attributed to the articial situation.
Informativeness: The material needs to enable the viewers to derive reasonable and useful conclusions for themselves. Material that does not allow for statements on important topics cannot be used in education on that topic. Therefore our material in question has to provide referenceable actions of the pupils.
Anonymity: As there should be as few restrictions as possible on the usage and dissemination of the material, we want it to be practically completely anonymised so that it does not allow for the identication of the pupils.
These properties are inuenced by the Critical Incident Technique [
To achieve these goals, we decided to use videos as the primary medium. The main dierence to vignettes is that our videos show real work rather than hypothetical stories. Nonetheless, we use the term video vignettes to refer to the combination of our videos and exercises.
Videos can meet the demands mentioned above if certain conditions are ensured. To assure authenticity we prefer recordings of pupils’ regular working environments (classroom equipped with computers) over laboratory settings. We believe that this is already a quite strong criterion for authenticity. We further keep the setting for recording as little invasive as possible (see last paragraph of this section).
A trade-o exists between the demands on informativeness and anonymity, as anonymisation in its essence means the exclusion of certain information. An example is the removal of facial recordings, which grants non-identiability, but at the same time prohibits observations of the pupil’s emotional state. Therefore, a compromise has to be made between the demands on informativeness and anonymity. We try to accomplish this compromise by replacing the information lost through anonymisation with supplementary information. For example, the information lost by excluding facial recordings is partly reintroduced by adding video recordings of the pupils’ input devices, hands and notes. These might enable the students to get a hint of emotions if they are intense enough to result in motor responses of the hands. Additionally, that view also indicates actions like pointing at the screen, which might be useful as well. Since the learners (L) often work in pairs, they are practically required to communicate verbally with each other, so we also add transcribed versions of the audio recordings.
The screen recordings can be realized in two ways: either by software running on the system that is being recorded, or by additional hardware. Since the videos will be recorded in schools and the specic features of the computer systems (hardware and software) can be quite dierent from school to school, we ruled out a software solution. The hardware solution does not require any software to be installed or run on the recorded systems. It consists of a video grabber that is installed between the computer and the screen and intercepts the video signal before passing it through to the screen. Altogether this provides the exibility that is needed to record in unknown settings.
The nal aspect of our videos is another supplementation used to support the
interpretation of what is seen in the videos: eye-tracking data. As Przybylla [
In the rst design cycle we had the challenge that there were no videos available
that we could use. We dealt with that issue by incorporating video production
into the course. This led to a design containing six steps for the students to
complete:
1. Read literature on misconceptions [
The approach has been evaluated using a written survey after course completion as well as qualitative observations during the course.
Observations First, we describe our qualitative observations during the course. Our rst nding is that video production bound a lot of time in rather technical activities, like learning how to handle the recording devices or how to edit the recordings. During that time the students neither learned anything about pupils’ work nor were they encouraged to perform reective thought processes. We therefore agreed to remove video production from the course design.
During the analysis session students had to look for interesting situations in the videos. The term interesting situation had not been dened explicitly in order to nd out which situations the students would perceive as interesting, but was highly inuenced by their prior reading of literature on misconceptions. Apart from this, we would have liked them to identify other diculties as well, but this did not happen. We concluded from this that students do not automatically reect in the sense of Sect. 3 and need more ne-grained instructions. Empirical Evaluation Next, we describe the insights we got from the survey that we conducted after the course had been completed. The survey was a voluntary online questionnaire. The students participated after their term papers were graded. Seven of the eleven course participants completed the survey. In the survey we asked them about their opinion on the relevancy of the course topics for teachers. We also asked how the course inuenced those opinions and whether they learned to perform the corresponding actions (e.g. to draw conclusions based on pupils’ conceptions). Additionally, we asked what activities should be removed, done less of, done more of, or be added to the course.
First, we present the reported changes in opinions during the course. Regarding the knowledge of pupils’ typical conceptions and the ability to draw conclusions, about half of the participants stated that they see them more important than before the course. The other half answered that their opinion had not changed. Therefore we investigated what opinion that group had before, and found that they already strongly agree on the importance of the concepts in question.
Approximately the same distribution of opinions can be found in regards to identifying pupils’ conceptions: Half of the participants see the topic more important than before and now totally agree on the importance for teachers. The other half did not change their opinion and does not totally, but mostly agree on the importance. However in this analysis we see a single student that did not change his or her opinion and mostly disagrees on the importance. Investigating that student’s other answers showed that he or she seems to see no value in the programming process. We assume this because he or she agreed strongly on the statement that videos cannot help to grade the nal products in a fairer way. Additionally he or she does not agree that the work process should be observed and taken into account during grading.
This opinion raises the question if we should address the signicance of the programming process explicitly during further development of our tasks. To answer that question, we investigated whether there were more students with a similar opinion. In fact we found another one that gave similar answers to those questions. In contrast to the rst student, this one agrees that teachers should be able to identify pupils’ conceptions by observing their work processes. This student seems not to regard the process as an integral part of programming. We will emphasize the role of the programming process more in our future tasks.
Finally we present suggestions that the participants gave for improving the course. The rst comment was that the classroom visit was perceived as great, because it gave insights into future practice. It was suggested to do multiple visits and ideally do them in multiple classes. These statements support the need for a connection between the theoretical aspects of teacher training and future practice. However, classroom visits take a lot of time, which is limited during courses at the university.
Another participant reported that he or she did not feel able to draw conclusions from pupils’ conceptions ((c) from Sect. 3). He or she asked to include examples of possible conclusions and explanations on how to draw them. Since this is a rather creative and especially situational process, case studies in the form of video recordings might be particularly helpful to that end. We will use frequent group discussions as the main tool to foster this process. We expect that the discussions will allow review and enrichment of suggested conclusions by other participants.
A similar point was addressed by a dierent participant. He or she asked to include the discussion of assistive measures and how to conduct interviews with pupils. He or she motivated that with a lack of guidance on how to deal with misconceptions. We will integrate those topics into the course. While there were presentations on how to address misconceptions, that information seems to not have been received well, as the next suggestion will show.
One participant suggested to summarize the information from the presentations of the other students. While one might argue that for a university student it should be obvious that they should take notes by themselves, we will take note of the fact that it seems not to be that obvious from the students point of view. An explanation might be that the students do not perceive the presentations of their fellow students as important with regards to their own learning goals of the course. In future iterations we should at least explicitly tell the students that we expect them to learn the topics from those presentations. 6
We introduced a new methodology to foster prospective programming teachers’ reection processes. Our approach is based on video vignettes showing pupils work on programming tasks, accompanied with appropriate exercises to encourage reection.
Future work includes both work on school and university level. Our goal on school level is to build a larger repertoire of videos, thereby contributing to an understanding of how pupils program. The long term vision for this collection would be to identify and categorize components of programming processes that allow to induce hypotheses.
At the university level we follow our design research setting of conducting
and analysing design experiments. The students’ (UL) work processes and term
papers will be analysed similarly to those of the pupils with qualitative content
analysis [
This work is supported by the Bundesministerium fr Bildung und Forschung under Grant No.: 16DHB2130.