reflect sequences, parallel procedures, and loops). Flow-based Programming (FBP) is a visual programming paradigm in which processes are represented as boxes connected by arrows [33]. Its flowchartlike architecture simplifies the transition for students from conceptualizing their ideas to converting them into executable code [15]. Thus, it facilitates the rapid development of functional applications by individuals with limited CS background [1,22].

Several studies have shown that professional development (PD) workshops alongside job-embedded support, can positively impact teachers' perceptions, technical knowledge, and attitudes towards integrating CS and CT into existing disciplines [21,35]. Our previous studies have explored the impact of a flow-based music programming environment on students, particularly how it enhances their attitudes and engagement with programming [31,42]. However, there is still a lack of empirical research on how teachers implement CS integration lessons in their classrooms, the challenges they encounter, and the impact of this integration on students' perceptions and attitudes toward learning computing concepts and related subjects. Elementary school teachers typically have no or limited experience with CS and can feel insecure about implementing CS activities [25,34]. It is crucial to understand how teachers implement flow-based music programming to reach children and investigate the potential impact of a music-making FBP platform on students. This work expands the research to the teacher side.

In this report, we describe our experiences engaging teachers and students with no or limited prior programming experience through a flow-based music programming environment. We introduced a platform called M-Flow and an associated ten-lesson curriculum, with the intervention spanning from November 2023 to May 2024. The curriculum included classroom programming activities where children worked in groups, using M-Flow to create sounds and learn basic programming concepts. We evaluated the intervention from both the teacher and student aspects by conducting a sequential mixed-method evaluation. To understand how teachers integrate flow-based music programming into classrooms and what challenges they face, we collected multiple sources of qualitative data from three teachers, including workshop videos, focus group videos, and debriefing videos after a semester of classroom implementation. We also gathered data on attitudes and self-efficacy from 75 students taught by these teachers to see what impact flow-based music programming has on students. This report offers valuable insights and references for the broader implementation of programming education in K-12 settings, particularly at the primary school level.

In recent years, integrating CT and CS education into the primary school curriculum has gained significant attention, particularly through STEM integration. Preparing students at an early age for CS education is crucial as early exposure can foster interest and foundational skills essential for future learning [2]. Instead of introducing CS as a new, required discipline, researchers and elementary educators often explore ways to weave computing concepts into existing disciplines. This integrated curriculum development is collaborative, with CT concepts and practices often incorporated into science and mathematics [7,47]. While most curricula integrate CS into STEM education, incorporating CS into music and art for K-12 students has shown promising results in creating a more inclusive and engaging learning environment [48]. Among the various approaches in K-12 Arts+CS education, musical flowcharts are a great method to teach computing concepts such as sequences, loops, and conditionals. Through musical flowcharts, students can make connections between different musical sections and understand how they converge to convey a song's message, enhancing students' CT skills, including decomposition, pattern recognition, generalization, and abstraction [16].

Toward Integrating CS into Learning

Besides investigating the impact of integrating CS and CT into existing disciplines on student outcomes, it's also important to investigate teachers' experiences and perceptions toward using such tools.

Studies have examined teachers' perceptions of integration through PD workshops. For instance, a study [39] investigated elementary teachers' perceptions of integrating CT into their math and science teaching. Despite widely acknowledging challenges such as limited exposure to CS, time constraints, and the difficulties of addressing high-level CT thinking in developmentally appropriate ways, teachers were able to draw strong connections between CT and both mathematics and science instruction. Similarly, [7,8] found that K-8 teachers achieved significant gains in technical knowledge and held a positive attitude regarding confidence and motivation toward CS and its integration after receiving support and resources in PD workshops. However, there remains an empirical gap in investigating how teachers implement CS with music integration in their classrooms, the challenges they face, and how this integration affects students' perceptions of and attitudes toward learning computing concepts and subject matter.

3 Methods

M-Flow is a FBP tool developed by our team for creating music and sound compositions. Shown in Figure 1, users can drag blocks into the canvas, insert sound recordings into them, connect them with arrows to form sequences, loop them to generate rhythms and manipulate them in various ways. M-Flow offers three different difficulty levels, each featuring distinct functional blocks. In total, there are eight different blocks. An initial version of M-Flow was made public in July 2022. As of July 1, 2024, over 890 teachers and students have registered for accounts on the M-Flow platform. More information can be found on the M-Flow website and other papers describing the tool [32,42].

A ten-lesson M-Flow curriculum, aligned with the Computer Science Teachers' Association's (CSTA) standards for CS, has been developed based on the M-Flow platform in collaboration with the Chula Vista Unified School District. The current version of the curriculum is available on our team's website. For details about the curriculum development process, see [42].

Table 2: Student Survey Items

To evaluate the outcome of flow-based music programming on teaching and learning, we utilized the instruments mentioned above as well as video recordings for data collection. For teachers, in addition to pre-workshop questionnaires and focus group (one hour), we recorded a four-hour workshop video and a two-hour-and-twenty-minute post-classroom implementation debrief video for analysis. For students, pre-and post-survey data were collected on their experience, self-efficacy, interest and willingness, and identity towards engineering and programming.

For teachers' qualitative data, we used thematic analysis to understand how they implement the curriculum in the classroom and what challenges they have. Thematic analysis is a broad, encompassing approach to uncovering recurring patterns within qualitative data [3]. We use this approach as it allows for an in-depth exploration of teachers' experiences, attitudes, pedagogical strategies, and challenges in the intervention. For students' survey data, we used descriptive analysis and t-tests to measure if there was a change in students' attitudes and self-efficacy.

We analyzed the workshop video (four hours), the focus group (one hour), and the debrief videos (two hours and 20 minutes) using thematic analysis to find out how teachers integrate M-Flow into their classrooms and the challenges they face. We identified 104 relevant script pieces with 17 different codes. After refining, aggregating, and removing codes unrelated to the integration, we found three themes with seven sub-themes for teacher integration of flow-based music programming and two themes with five subthemes for challenges teachers face in classroom implementation.

4.1.1 Theme 1: Teaching Strategies. The teachers applied various teaching strategies to integrate flow-based music programming into their classrooms. One of these strategies was collaborative teaching and learning (sub-theme 1.1). For instance, teachers T1 and T2 mentioned that they would collaborate on preparing lesson materials and share their experiences: "After we've done a few lessons, we'll kind of know the rhythm of it. And then you and I can talk about it like this is where we're at. (T2, workshop)" Besides, the teachers tried to create a collaborative learning environment for students.

The teachers paired all students for programming tasks and clearly defined roles within the collaboration (T3, debrief). They also provided specific times for role switching, which effectively helped students complete their learning tasks:" We'll partner students up and assign roles. One student will be the driver and the other will be the director." (T3, workshop) Teachers also planned and successfully ensured a student-centered learning environment (sub-theme 1.2). The teachers viewed themselves as facilitators by providing clear direction and exceptions, allowing students to explore independently. For example, T3 said in the interview that I see myself as a facilitator, as a facilitator, I aim to be clear and direct with expectations and directions from the beginning. I will not only provide these verbally but also display them clearly. This ensured students' active engagement when using M-Flow in the classroom.

4.1.2 Theme 2: Scaffolding and Guidance. Another theme focused on Scaffolding and Guidance. This theme includes diverse learner support (sub-theme 2.1) and structured support (sub-theme 2.2). The first aspect addresses the needs of special students to enhance their engagement with M-Flow. Additionally, from the teachers' perspective, the second aspect reflects on the importance of structuring support and direction provided to students. 3). In the workshop, teachers expressed concerns about certain students, such as language learners or students with disabilities. For example, they worried about how these students would engage with and explore the M-Flow platform effectively:" my students who with a disability, at least for my class, they just kind of don't participate as much or they're just kind of in their own world." (T2, workshop).

However, in the debrief, teachers did not mention concerns about student engagement. Instead, they discussed challenges students faced during independent exploration, such as difficulty understanding tasks, coming up with solutions, and handling multiple tasks simultaneously:" Some students were really focused on that emotion. So they're trying to figure out a sound that goes with it. They were just kind of walking around and couldn't come up with one." (T2, debrief). 4.1.5 Theme 5: Pedagogical challenges. The teachers also faced pedagogical challenges. Due to their limited knowledge of CS and programming (sub-theme 5.1), they were concerned about their ability to answer students' questions and provide deeper instruction. For example, T3 mentioned in the debrief that I was kind of lost in that translation where all of a sudden the kids were doing a lot more and like what would just happen there? So I just kind of like I didn't know. So I'm like, alright, you figure it out.

Another pedagogical challenge the teachers described as the tension between planning strategies aimed at promoting both scaffolds for some learners and student creative expression (sub-theme 5.2). For example, teachers used scaffolds such as sentence frames to help ELLs express themselves in class, but they indicated that this support may have limited the creative expression of other students, leading teachers to question the necessity of such strategies. T2 said in the debrief that "I don't know if the frames themselves provide that authentic conversation or lend themselves to this type of lesson, what we're doing a reading lesson or something and you're talking about a text, it's easier to kind of refer to a sentence. But this (programming learning) is a little bit different, because everyone's doing the discretions are different." This issue also appeared in pair programming, where in some pairs, one child took on all the exploration tasks while the other completely stepped back.

The overall description column in Table 3 shows the survey results from students who completed the survey. According to the presurvey results, 39% of the children responded that they did not have any programming experience, and 46% said they had minimal programming experience, resulting in a total of 85% of participants with no or minimal prior experience. Data from 61 students could be matched between the pre-and post-survey.

The paired t-test in Table 3 is from students who answered Kind of or Yes to item #1 in the pre-survey, indicating that they know what programming means (n=36). We can see a significant increase in students' experience (#1 to #3) and self-efficacy (#4 to #6) after the classroom implementation. However, there was no significant change in interest and willingness, nor in the identity score. Note: **p<0.01, *p<0.05. The overall description shows the results of all the students who completed the survey. For the 36 students who had matched pre-and post-survey data, we conducted paired t-tests. Effect sizes were calculated using Cohen's D. For the 25 students who previously reported not knowing what programming meant, and thus did not have pre-survey answers for survey items #2 to #10, we analyzed their responses using one-sample t-tests to assess whether their attitudes were generally positive by comparing their scores to a neutral value (e.g., 2 on a 1-3 Likert scale).

For children who in the pre-survey indicated that they didn't know what programming meant (n=25), we run a one sample t-test, comparing their scores to a neutral valence. Results show that, after going through the curriculum, they had a positive experience and high self-efficacy (p<0.01 for #1 to #6). For questions 8 to 10 the results were not significantly different from the neutral value. Figure 1 shows screenshots of projects created by four children in lesson 7. Children created projects with different levels of complexity (e.g., number of nodes and types of blocks). For example, in Figure 1(c), the project is very structured, consisting of 11 blocks, which reflect the concepts of parallelism and loops in computing.

Previous research demonstrates that EarSketch [18] and TunePad [14] have been effective in engaging children, although the children were older, and the experiences relied on specialized educators with deep knowledge of both music and programming [37]. Flow-based programming offers an alternative that addresses these challenges. It is designed not only to be accessible to younger students but also to be implemented by non-specialized teachers. This adaptability is a key strength of our method, allowing for broader accessibility without the need for extensive technical training or background knowledge in CS or music. Furthermore, our focus on creating positive, early experiences aligns with research showing the long-term benefits of early exposure to CT and creative problem-solving [43]. Even though scratch [11,50] has music-making capabilities, limited research has explored its impact on students' attitudes or the extent of teacher training required for Scratch-based interventions. Moreover, a flow-based approach is more suitable for replicating complex musical structures. For example, in Scratch, generating bifurcating parallel structures requires the use of 'broadcasting,' which is relatively advanced. However, in M-flow, a single block can have multiple output arrows that simultaneously generate parallel sound streams, simplifying the process for both students and teachers [42].

The thematic analysis identified three themes regarding teachers' integration of flow-based music programming into classroom teaching. First, teachers employed strategies focused on collaborative teaching and learning, as well as student-centered learning. For their own collaboration, teachers planned to engage in team teaching, including discussions after teaching several lessons (T2, workshop) and dividing instructional responsibilities (T1, workshop). For students' collaborative learning, they are assigned roles such as director and driver (T2, T3 interview), which is beneficial for facilitating pair programming and enabling students to build on each other's ideas [49]. In student-centered learning, teachers acted as facilitators, primarily setting expectations and directions (T1, T3 workshop), allowing students to explore the M-Flow platform independently and engage in peer learning [24].

Regarding assessments, teachers mentioned that they would use rubrics and checklists to evaluate student learning (T1, T2, T3 interviews), along with verbal and reflective assessments by asking students questions (T1, debrief). Additionally, teachers discussed using presentations where students screen-record their work to showcase their learning (T2, debrief). Integrating methods such as reflective assessments and student presentations in programming classes aligns with research underscoring the benefits of diverse assessment techniques. These approaches offer deeper insights into students' comprehension and application of programming concepts while promoting fairer evaluation practices [12,27].

Some student-related challenges were recognized in this implementation process. Regarding student engagement, teachers expressed concern that some students, such as students with disabilities, might struggle with navigating a computer and might not participate, becoming absorbed in their own world (T2, workshop). Sometimes, students talked to each other but were not effectively communicating (T1, workshop). Additionally, some students faced decisionmaking difficulties, so teachers aimed to provide options whenever possible (T3, debrief).

Pedagogical challenges include teachers' limited knowledge of programming and M-Flow and restrictions brought by certain scaffolding and guidance strategies. T1 mentioned in the interview that a challenge for her is that students may want to do more with M-Flow than she can actually support. Sometimes, students did much more than the teacher understood, and the teacher had to let them figure it out independently (T3, debrief). However, there is a need for teachers to balance guidance with opportunities for independent learning [30]. A well-designed tool could support students' inquiry while allowing teachers to facilitate effectively [4,19]. Pair programming, although fostering collaboration, resulted in only one student having access to a computer at a time, and they might not switch roles, causing the other student to disengage (T1, T3 debrief). It noted the importance of role rotation in collaborative learning to ensure active participation from all students [19,20]. [44] emphasized the need for differentiated instruction to foster creativity. Some strategies used by teachers in this study are critical for certain students. However, teachers could offer other students the choice of using these strategies or not, thereby alleviating the restrictions imposed by these strategies.

The study revealed that flow-based music programming significantly enhanced students' programming experience and self-efficacy, particularly those without prior experience. Specifically, 85% of participants had no or minimal prior programming experience, and their experience and self-efficacy scores showed significant improvement after using M-flow. Despite these gains, there were no significant changes in students' interest, willingness, and identity scores. This is consistent with studies that show that career-related changes typically develop gradually, with significant shifts being rare during school years, particularly in STEAM fields [41]. Younger students often have broad, changeable interests and may lack the motivation or capacity to connect current experiences with future career goals [23]. At this stage, fostering positive early experiences might prove more crucial, as these experiences play a foundational role in future development [43].

These findings align with the literature, which suggests that creative programming activities, such as music programming, can effectively build computing skills and confidence among novice learners. For instance, [12] found that high school students' engagement in computational remixing with EarSketch increased interest and attitudes in computing. Similarly, [10] reported that high school students enhanced their creative expression through significant gains in computing attitudes and creativity while compositing and remixing songs in a programming environment, highlighting the positive impact of music-based programming on students' skills and attitudes. Additionally, [14] noted a platform for sound composition, supported CT through music, which increased student interest and engagement and provided the potential to broaden the participants in CS. Despite these positive outcomes, maintaining and enhancing students' long-term interest and identification with computing remains a challenge. Research by [9] indicates that while initial exposure to engaging programming activities can boost confidence and skills, sustained interest requires continued engagement and support. While students feel more competent, additional efforts are needed to sustain their intrinsic motivation.

In this report, we explored the impact of flow-based music programming on teaching and learning. After participating in workshops, discussions, and classroom implementations over six months, three teachers with no prior programming teaching experience completed M-Flow classroom instruction for 75 students. We described how teachers employed various strategies to integrate flow-based music programming into their classroom, specifically through collaborative teaching and learning, providing students with scaffolding and guidance, using multiple assessment methods. Student-related and pedagogical challenges were also found in the implementation process. For students, we found that flow-based music programming enhanced students' programming experience and self-efficacy. Future work will refine the curriculum and create a streamlined, PD program to reach more teachers and students, continuing to examine the program's impact on teachers and students.