Enhancing Computational Thinking of Deaf Students Using STEAM Approach

Abstract

Computational thinking (CT), an interrelation of skills and practices, is a crucial competency that empowers individuals to tackle logical problems, enabling them to overcome various challenges in their daily lives. To help Deaf students (those with hearing loss and using sign language for communication) enhance their CT, a STEAM learning program using a physical computing tool is proposed. The learning program composes four courses: learning concepts, implementing concepts, finding solutions to real problems and developing innovations. The program engaged Deaf students from 18 Deaf schools. It is geared towards boosting students’ CT and facilitating their capacity to devise technology-based solutions. The program measured students’ CT effectiveness based on the CT framework: concepts, practices, and perspectives. The measurement encompassed multiple-choice assessments for CT concepts, task rubrics for CT practices, and interview and invention observations for CT perspectives. The program concludes with participating in a science project competition, using a physical computing tool, called KidBright, to solve real-world issues by integrating science, mathematics, and art. After completing the learning program, Deaf students demonstrated an improved understanding of CT concepts, performing high-level CT practices, and expressing strong CT perspectives. These indicate that a STEAM learning program utilizing a physical computing tool can help Deaf students enhance their computational thinking.

1. Introduction

Computational thinking (CT) is indispensable for logical problem-solving, aiding individuals in addressing challenges of various scales in their everyday lives (Bocconi et al., 2016). CT is a problem-solving approach that draws on principles from computer science to address complex issues in various domains. It involves four key techniques: (1) decomposition, breaking problems into smaller, more manageable components; (2) pattern recognition, identifying similarities within and across problems; (3) abstraction, focusing on relevant information while ignoring unnecessary details; and (4) algorithm design, creating a step-by-step solution to address the problem (Tsarava et al., 2022). These foundational principles and practices of computer science are essential at all school levels to equip students for careers in the 21st century (Computer Science Teacher Association, 2017). Existing research provides evidence of the mutual benefits between computational thinking (CT) as a cognitive framework and STEM education (Li et al., 2020). A report, Charting a Course for Success: America’s Strategy for STEM Education, released by the White House in December 2018 emphasizes that fostering computational literacy is one of the four key pathways to advancing STEM education, which is “Build computational literacy through STEM education heavily imbued with computational skills and accessed through digital means” (Committee on STEM Education, 2018). As a result of this, there have been numerable attempts to include STEM/STEAM in education curriculums (KOFAC, 2012; Morales et al., 2019; Commonwealth of Australia, 2021) around the world in recent times. Art in STEAM encourages students to explore connections between science, technology, engineering, and mathematics with visual arts, music, literature, performance, and other forms of creative expression. It aims to cultivate well-rounded individuals who can approach complex problems with imagination, critical thinking, and interdisciplinary perspectives. In our paper, STEAM is defined as the integration of science, technology, engineering, art, and mathematics. The “science” component represents the knowledge and understanding of the natural and social world. “Technology” and “engineering” refer to expertise in computer science, while “mathematics” emphasizes the application of logic and reasoning. The “art” component highlights artistic expression, critical and creative thinking, as well as system design. The proposed learning program in this paper complements all STEAM components, particularly technology, engineering and art, to promote technology-driven innovation.

Since this paper focuses on nurturing technology-driven problem-solving skills, the contents of the learning program involve the fundamental knowledge of computer science, encompassing areas like coding, electronics, and automated systems. Teaching computer science falls under the technology and engineering components of STEAM. Educational tools in STEAM for teaching computer science aim to engage learners and enhance learning across diverse formats. Computer-based visualizations, one type of educational tool, use virtual models to enhance learning experiences (Kiraly & Balla, 2016; Weintrop & Wilensky, 2018; Lindberg et al., 2019). Scratch, as an example, serves as a visual programming tool tailored towards simplified coding for children, enabling them to create interactive stories, animations, and simulations (Dúo-Terrón, 2023). This process involves linking blocks to form sequential commands, predominantly to manipulate on-screen 2D sprites representing characters utilized by children to narrate their stories.

In addition to computer-based visualizations, the utilization of physical computing tools, such as educational boards and robots, has been implemented (Cvjetkovic & Matijevic, 2016; Loannou, 2018; Teiermayer, 2019). Physical computing, which combines software and hardware, proves to be an effective method for teaching various aspects of computer science (Przybylla & Romeike, 2014). Examples of physical computing tools are LEGO (2024), BBC micro:bit (2024), Arduino Education (2024), and KidBright (2024), which are equipped with sensors and/or actuators operable through coding (Louis, 2016; Alisher & Zafar, 2022; Rogers et al., 2017). Through these tools, children can engage in playful coding, developing automated systems and reinforcing skills in STEAM. Strategies for the effective utilization of robotics as an educational tool and its impact on students’ interests in STEAM-related subjects have been suggested by Ernest and Myint (2017), Kim and Kang (2021), and Kvaššayová (2022). From an educational perspective, the physical computing approach enables learners to develop presentable, evaluable, and discussable prototypes easily. There is clear evidence that educational boards and robotics significantly contribute to the development of CT compared to visualization methods, as students can apply their knowledge to real applications, providing technology-based problem-solving opportunities (Israel et al., 2015; Lu et al., 2022).

The growing interest in fostering computational thinking (CT) through STEAM education is extended to Deaf students as well. Recently, the cultivation of computational skills for Deaf students has concentrated on coding activities, such as challenging them to analyze and comprehend the functions of computer systems, identifying and rectifying errors, and refining their solutions. In 2018, Apple (2018) created the comprehensive “Everyone Can Code” curricula so students who are Deaf, blind, or possess other assistive needs can learn and write code using Swift. With teacher guides and lessons, students learn basic coding on iPads with Swift Playgrounds. Deaf Kids Code (2015) is another project that promotes technology and computing skills to middle school Deaf students. They experience coding using block-based programming and by creating robot projects. The Playful Learning Lab collaborated with Metro Deaf School (MDS) delivering an “Intro to Electronics and Computer Programming” course to Deaf students (Leininger et al., 2023). The project used Scratch as a teaching tool. Based on these mentioned projects, a range of tools including Scratch, Arduino, robots, and programming languages, such as block-based programming, Swift, and Python, were utilized. Developing CT skills for children is a challenge, even more so when their language skills are limited. Deaf students face challenges in verbal communication, with some being entirely non-verbal. Effective modes of communication for them include hand signs and facial expressions. The absence of auditory input presents hurdles in both reading and writing, particularly in subjects like science, technology, and coding (Moeller, 2015). Based on observations in special needs schools, Rusilowati et al. (2020) mentioned that the general restrictions of Deaf students are a (1) low understanding of vocabulary, (2) difficulty in comprehending abstract vocabulary, (3) low reading speed, (4) difficulty linking concepts of learning with life activities, (5) low retention of taught material, and (6) preference towards visuals or tangible objects. Based on these restrictions, the learning strategy for Deaf students should address their challenges by reducing reading intensity, enhancing their conceptual understanding through hands-on activities, and connecting learning concepts to real-life experiences. Therefore, block-based programming is a more appropriate approach compared to text-based programming such as Swift and Python.

Computational thinking involves more than just writing code. It requires developing problem-solving strategies, analyzing problems, and devising efficient solutions. To foster comprehensive computational thinking, it is important to supplement coding activities with other instructional approaches such as STEAM. Integrating the science, technology, engineering, art, and mathematics concepts through project-based activities guides Deaf students in inquiry, discussion, and problem-solving. This paper proposes a STEAM learning program using a physical computing tool, called KidBright, to enhance the CT of Deaf students. KidBright simplifies coding and facilitates the development of automated systems without requiring an extensive electronics knowledge (Tan-a-ram et al., 2022). By using block-based programming, Deaf students can avoid making a mistake in the structure or format of a code that prevents it from executing (syntax error) while remaining focused on concepts of coding. The same aspect can apply to the development of automated systems which allow students to explore a wide variety of automated systems for real-life problem-solving.

2. Methodology

2.1. STEAM Educational Tool

The proposed learning program utilizes the KidBright board as a STEAM educational tool. KidBright board (hardware and software open-source platform) is a small, easily portable microcontroller board. It has LED lights that can scroll messages or display pictures. There are two programmable buttons and one reset button. The board has built-in temperature and light level sensors as well as a real-time clock and a buzzer, as shown in Figure 1 (a. front view; b. back view). The device manager in the KidBright firmware further enables learners to use KidBright to create many automatic systems, such as automatic plant-watering systems, clock alarms, and automatic doors.

KidBright is programmed through the KidBright IDE (Integrated Development Environment), supported both in Windows and Mac, as shown in Figure 2. It utilizes the simple form of coding—block-based programming. The command set is created by dragging and dropping command blocks. After which, the command set is compiled into a machine code before being sent to the KidBright board for execution.

2.2. The STEAM Learning Program for Deaf Students

Learning is the dynamic process in which individuals gain knowledge through accumulating experiences garnered from interactions with their surroundings. While the STEAM approach holds promise for enhancing learning outcomes, certain learners necessitate specialized tools and support systems (Computer Science Teacher Association, 2017). Lee et al. (2011) proposed a three-stage progression, termed Use–Modify–Create, to engage youth in CT. Initially, learners engage with pre-existing models, before modifying them, then ultimately creating their own models. This approach aligns with educational theories emphasizing learner-centered, hands-on, or project-based methodologies. Building upon this framework, Dong and Jia (2021) introduced novel teaching strategies for teaching coding to Deaf individuals. These strategies encompass concept abstraction, task decomposition, independent thinking, and practice reinforcement. The implementation of these methods has led to a substantial Deaf student engagement and achievement in the course. The framework integrates decomposition to teach coding concepts and confirm the understanding with small-step task assignments, making it well suited for Deaf students. Therefore, this paper proposes a learning program that applies the Dong and Jia (2021) framework within the STEAM approach. The proposed learning program addresses challenges of Deaf students by using block-based programming while encompassing interdisciplinary learning, creativity, innovation, and real-world connections. The learning program is structured across four courses: learning concepts, implementing concepts, finding solutions to real problems, and developing innovations.

2.2.1. Learning Concepts

The program starts with learning coding concepts because it serves as a fundamental skill for technology-based problem-solving by promoting problem decomposition, algorithmic thinking, automation, experimentation, creativity, interdisciplinary applications, and critical thinking. Mastering coding empowers individuals to address a wide range of challenges in a technology-driven world. KidBright was used to introduce coding concepts which complement the technology and engineering of STEAM. It employs block-based programming, where commands are assembled by dragging and dropping blocks. Commands are executed on the KidBright board, displaying immediate feedback and allowing students to grasp command functions. For instance, displaying “Hello world” ten times results in the text scrolling across KidBright’s LED screen accordingly. Any logical errors in commands prompt learners to identify and correct them, fostering analytic thinking.

Learning concepts is a fundamental and prerequisite course for the later processes that apply the coding concepts to solve real problems. To ensure that this course is implemented successfully, the learning concepts process was preliminarily implemented with a small group of participants. The group consisted of 10 Deaf students, fluent in sign language and possessing prior knowledge of technology, who were selected by their computer teachers, along with their computer teachers (14 teachers). In the effort to teach coding to Deaf students, we encountered initial challenges. The foremost of these being the absence of a sign language vocabulary tailored for the coding instruction of the KidBright board. To address this, the group developed sign language equivalents for coding concepts. Another obstacle emerged for Deaf students, who often require materials beyond traditional written texts due to literacy limitations. In addition to developing sign language resources, these computer teachers contributed to the creation of the handbook specifically designed for Deaf students. This handbook offers video instructions accessible via 2D barcodes, supplemented by graphical contents aimed at enhancing comprehension through visual aids, available at https://www.kid-bright.org/2021/09/20/ (accessed on 20 February 2025). In addition to the handbook, the video tutorials in sign language are also available (see Supplementary Material) at https://www.kid-bright.org/deaf-kid/ (accessed on 20 February 2025).

2.2.2. Implementing Concepts

Once proficient in utilizing all KidBright commands and interfaces, students engage in challenging tasks fostering knowledge and logical thinking. They tackle predefined systems, employing skills acquired in the previous course. This course involves STEAM concepts as learners need to transform their imagination into a real system through systematic thinking, analytical thinking, and creativity.

2.2.3. Finding Solutions to Real Problems

To encourage computational thinking, groups of three to four students are formed. Groups identify real-life problems and devise technological solutions, cultivating innovation aligning with the STEAM approach that incorporates multiple disciplines to foster creative problem-solving, collaboration, and critical thinking. Each group conceptualizes ideas, drafts workflows, and designs solutions.

2.2.4. Developing Innovations

Groups of three to four students are formed to create prototypes for testing their conceptual designs, gaining deeper insight into development processes and promoting systematic thinking. Synthesizing knowledge from previous courses, students translate ideas into reality. This course not only drives them to overcome obstacles but also instills a sense of pride in their project accomplishments.

2.3. Measurement Framework

Brennan and Resnick (2012) used a framework consisting of three dimensions to measure learners’ CT achievements: CT concepts, CT practices, and CT perspectives. The CT concepts are designed to evaluate programming concepts (iteration, parallelism, etc.), while the computational practices are designed to evaluate the understanding of programs (debugging projects or remixing others’ work), and CT perspectives are designed to evaluate their perspective of themselves and of the world around them. Based on the computational thinking (CT) framework, Kong (2019) proposed several enhancements. These include using test designs in conjunction with task-based questions to evaluate CT practices, as well as designing and administering surveys to assess perspectives on CT.

Grover et al. (2015) wrote that “a system of assessments is beneficial to get a comprehensive picture of learners’ CT understanding”. In other words, there is no single evaluation component or method that can entirely measure learners’ CT achievements. Researchers, therefore, strive to develop both quantitative and qualitative approaches to evaluate learning outcomes. In terms of quantitative approaches, most of the studies measured learners’ understanding of CT concepts using test items, most of which were designed with multiple-choice questions in the programming context (Ruf et al., 2014). Task and project rubrics were also commonly used so that teachers could assess learners’ project work with scoring guidelines. Among the studies using qualitative approaches, interviews were the most common method of evaluating CT concepts. Interviews are conducted to understand the CT concepts that learners use to complete programming tasks or projects. Project analysis was another method used to evaluate learners’ development in CT concepts.

This paper used the CT framework proposed by Kong (2019) as the measurement framework to measure the learning development of Deaf students. The framework consists of three dimensions: CT concepts, CT practices, and CT perspectives. The dimension of CT concepts refers to the computational concepts that learners develop during coding, the dimension of CT practices refers to the problem-solving practices that learners demonstrate repeatedly in the assignment process, and the dimension of CT perspectives refers to learners’ understanding of themselves and their relationships with others and the technological world that they develop by expressing, connecting, and questioning during the project development. The CT framework evaluates the knowledge, skill, and attitude of the learners in applying their knowledge across disciplines to create innovative solutions aligning with the STEAM approach. For example, CT practices like reusing, remixing, and iterative processes align with the experimental aspects of STEAM projects. Similarly, CT perspectives on connecting and questioning encourage the exploration of technological and societal impacts, aligning with STEAM’s broader focus on addressing societal challenges. Based on the idea of Grover et al. (2015), our study used both quantitative and qualitative approaches to evaluate learning outcomes. CT concepts were measured pre-test and post-test with multiple-choice questions to identify Deaf students’ learning outcomes for three processes: learning concepts, implementing concepts, and finding solutions to real problems. CT practices were measured by task-based assignments to identify learning outcomes in the “implementing concepts” course. The CT perspectives were measured by interviews and invention observations. In the competitions at the end of the “developing innovations” course, every team presented their work with the framework of identifying a problem, elaborating on its importance and devising a technological solution.

2.3.1. Measurement of CT Concept

The assessment of CT concepts is conducted through multiple-choice questions to measure students’ learning outcomes. The content areas of the measurement included (1) repetition/loops/iteration, (2) conditionals, (3) sequences, (4) parallelism/concurrency, (5) mathematical operators, functions, and Boolean logic, (6) event handling, (7) initialization, and (8) scientific project concepts. Three sets of questions were developed for use as pre-tests and post-tests (same questions as pre-test) in learning concepts, implementing concepts, and finding solutions to real problems courses. All questions correspond to the functionalities of KidBright command blocks and the construction of automatic systems for real applications, ensuring an alignment with CT measurement components. These questions are provided in

Table A1. The pre-test and post-test questions of the learning concepts course.

Questions	Choice A	Choice B	Choice C	Choice D
1. What is the icon of the KidBright IDE?
2. What does the image below indicate on the KidBright board?	Temperature sensor	Light sensor	Switch	All are correct.
3. Which of the following is an embedded board?	Microbit board	KidBright board	Arduino board	All are correct.
4. Which block set displays “Hello” on the screen?
5. Which block set increases the valve of x by 1 from 1–30 using the given blocks?				All are correct.
6. Which block set does correctly incorporates a loop command with the given blocks?
7. Which block set does not involve mathematics calculation?
8. Which block set represents a conditional command?				All are correct.
9. Which block set represents a comparison statement?				All are correct.
10. What does the block set below imply?	Show text “Pressed” on screen when switch 1 is pressed.	Beep sound is on when switch 1 is pressed.	Show text “Pressed” on screen once when switch 1 is released.	No right answer.
11. Which is the correct flowchart for the following tasks? ○ Winter begins. ○ Check the temperature. ○ If the temperature is less than 25 °C, wear a long-sleeved shirt. ○ If the temperature is 25 °C or higher, wear a short-sleeved shirt. ○ Winter ends.
12. List the steps associated with the flowchart.	1. Start Sunday morning. 2. Is it raining? 3. If no, wash the dishes. 4. If yes, do the laundry. 5. End the day.	1. Start Sunday morning. 2. Is it raining? 3. If no, do the laundry. 4. If yes, wash the dishes. 5. End the day.	1. Start Sunday morning. 2. If no, do the laundry. 3. Is it raining? 4. If yes, wash the dishes. 5. End the day.	1. Start Sunday morning. 3. If no, do the laundry. 3. If yes, wash the dishes. 4. Is it raining? 5. End the day.
13. What is the value of z?	50	60	70	40
14. Which statement correctly describes the flowchart for the Rock-Paper-Scissor game?	Reading the status of switch 1 is optional path of the flow.	Random number between 1–3 is optional path of the flow.	Scissor is optional path of the flow.	All are correct.

and

Table A2. The pre-test and post-test questions of the finding solutions to real problems course.

Questions	Choice A	Choice B	Choice C	Choice D
Which sensor can be used to create an automatic alcohol dispenser?	Infrared sensor for detecting objects	Reed switch sensor for detecting magnets	Gas sensor for measuring gas	Sound sensor for measuring sound
2. Which sensor can be used for an automated watering system?
3. Which block set is used for checking the status of connected external sensors?
4. What natural factors contribute to flooding, and what types of sensors can be used for a project?	Constructing houses that obstruct waterways—Ultrasonic and raindrop sensor	Heavy rainfall—Ultrasonic and raindrop sensor	Deforestation—Ultrasonic and gas sensor	Dumping garbage into rivers—gas and raindrop sensor
5. How do students connect the sensor to the KidBright board for an automatic alcohol dispenser project?
6. What is the range of equipment shown in the image?	Approximately 12 mm	Approximately 2–30 cm	Approximately 3–80 cm	Approximately 10–100 cm
7. What is the range of equipment shown in the image?	Approximately 12 mm	Approximately 2–30 cm	Approximately 3–80 cm	Approximately 10–100 cm
8. Which sensors should a student use for landslide and flood detection?	Infrared sensor and Reed Switch sensor	Vibration sensor and Water sensor	Gas Sensor and Sound sensor	Vibration sensor and Raindrop sensor
9. Which one is an output device?				All are correct.
10. Which sensors can a student use to create a project that helps villagers with agriculture?
11. Which sensors should a student use to create a wildfire detection system?	Gas Sensor and Flame sensor	Vibration sensor and Raindrop sensor	Infrared sensor and Reed Switch sensor	Gas Sensor and Sound sensor
12. In which region of Thailand are earthquakes most common, and which sensors can be used for detection?	Northeastern region–Infrared sensor	Northeastern region–Vibration sensor	Northern region–Vibration sensor	Northern region–Water sensor
13. Which of the following is a natural disaster, and which sensor can be used to provide an alert?	Robbery–Infrared sensor	Wildfire–Flame sensor	Traffic congestion–Gas sensor	Building collapse–Sound sensor
14. What is the resistance value between the 5VDC and the OUTPUT port for the automatic alcohol dispenser?	10 ohms	100 ohms	1 kilohm	10 kilohms
15. What types of sensors can be used to create an automatic door?				All are correct.

.

2.3.2. Measurement of CT Practices

Task-based rubrics were employed to gauge learners’ proficiency in these CT practices. The components assessed in our study encompassed (1) abstracting and modeling, (2) algorithmic thinking, (3) testing and debugging, (4) being incremental and iterative, (5) problem decomposition, (6) planning and designing, and (7) reusing and remixing (Kong, 2019). The assessment of CT practices was deployed at the end of the implementing concepts course. The criteria in

Table 1. Task rubrics of CT practices.

Criteria	Rating
1. Creativity. (related to component (1))	0.0 Low	3.0 Medium	5.0 High
2. System is well designed. (related to components (2), (5), and (6))	0.0 Not good	3.0 Normal	5.0 Good
3. Project works correctly. (related to components (3) and (4))	0.0 Does not work	3.0 Partially works	5.0 Fully works
4. Apply knowledge to complete tasks properly. (related to component (7))	0.0 Improperly applied	3.0 Properly applied	5.0 Well applied

Assessment components: (1) abstracting and modeling, (2) algorithmic thinking, (3) testing and debugging, (4) being incremental and iterative, (5) problem decomposition, (6) planning and designing, and (7) reusing and remixing.

were developed based on the Organization for Economic Co-operation and Development (2021) guideline, which includes relevance, coherence, effectiveness, efficiency, impact, and sustainability. However, impact and sustainability are not applied in the evaluation, as they fall outside the defined scope of the implementing concepts course.

2.3.3. Measurement of CT Perspectives

Kong (2019) proposed evaluating learners across three dimensions: (1) computational identity, (2) programming empowerment, and (3) perspectives on expressing, connecting, and questioning. The first dimension, “computational identity”, assesses the development of competencies in knowledge, attitudes, skills, and character. “Programming empowerment” refers to individuals’ experiences in creating and designing programs to address real-world problems, empowering them to confidently engage in the digital realm. The third aspect explores learners’ application of innovative thinking post-programming education, enabling them to contribute to society by expressing, connecting, and questioning digitalized environments.

CT perspectives used interviews and observations to assess students’ inventions during their demonstration at a science project competition. Measurement adheres to the criteria of the invention observation outlined in

Table 2. Measurement criteria of CT perspectives.

Criteria	Measurement Component	Rating (Ranging from 5 to 1)
1. Creativity in design.	(1)	New ideas 5—Not new ideas 0.
2. Invention functionality.	(2)	Completed 5—Not completed 0.
3. Invention complexity.	(2)	Complex 5—Not complex 0.
4. Relevance to real-life problems.	(2) and (3)	Relevant 5—Not Relevant 0.
5. Explaining concepts of their inventions correctly and answering questions.	(1) and (3)	Correct 5—Not correct 0.

Measurement components: (1) computational identity, (2) programming empowerment, and (3) perspectives on expressing, connecting, and questioning.

. The questions used for interviewing are listed in

Table A3. The interview questions.

Number	Question
1	What is your motivation for developing this invention?
2	What were the challenges during the development of your invention?
3	How did you encounter the challenges?
4	What are the potential improvements to your invention?
5	Is it a possibility to apply the concept of your invention to solve other issues? And how?

. Questions focused on the motivation behind the invention, the project’s development process, encountered challenges, problem-solving approaches, reflections on the project, and potential improvements. The criteria of the invention observation and questions were designed based on the Organization for Economic Co-operation and Development (2021) guideline.

2.4. The Implementation of the STEAM Learning Program

This paper seeks to assess the efficacy of an educational initiative employing a STEAM learning program using a physical tool to enhance the CT abilities of Deaf students, thereby fostering their ability to develop technology-driven solutions. A group of Deaf students (those with hearing loss and using sign language for communication), including both males and females with a prior knowledge of computers, participated in the program, details are in

Table 3. The demographic profile and attendance statistics of participants.

Courses	Males	Females	Previous Class	New Entry	Total
Learning concepts	42 (65.62%)	22 (34.37%)	-	64 (100%)	64
Implementing concepts	40 (59.70%)	27 (40.30%)	51 (76.12%)	16 (23.88%)	67
Finding solutions to real problems	35 (56.45%)	27 (43.55%)	45 (72.25%)	17 (27.41%)	62

. These students, enrolled in secondary school across 18 Deaf schools, were selected by their computer teachers based on their fluency in sign language. At least two computer teachers from each school, fluent in sign language, supported their students throughout the learning program. Additionally, two sign language interpreters, who contributed to developing sign language resources for coding concepts and the creation of a handbook, facilitated communication between instructors and students throughout the program. Together, the interpreters and computer teachers helped verify and reinforce students’ understanding of sign language.

The program was organized into three four-day courses (as in Figure 3), followed by a two-month development period, following the principles of the STEAM learning model and utilizing KidBright as the primary instructional medium. To ensure efficient learning, 18 Deaf schools were divided into three groups, each consisting of 6 schools. The learning program follows a sequential approach, progressing through three courses: learning concepts, implementing concepts, and finding solutions to real problems. All schools must complete each course before advancing to the next. After completing each course, teachers were encouraged to implement the training in the classroom using the course materials.

2.4.1. Learning Concepts Course

Deaf students took a pre-test to assess their existing knowledge prior to the course. Students completed the multiple-choice test using Google Forms (questions are translated to English and shown in Table A1), where each question was explained through sign language by interpreters. They were given approximately fifteen minutes to answer all questions. The same procedure was followed for both the pre-test and post-test in the implementing concepts and finding solutions to real problems courses. The primary objective of this course is to acquaint students with the functionalities of all command blocks within KidBright’s IDE. This included tasks such as displaying text, gathering data from onboard sensors, creating multi-tasking functions, and utilizing communication interfaces like WiFi and external digital sensors. Throughout the course, coding tasks were incorporated to reinforce understanding. By the end of the initial course, students were expected to proficiently employ all KidBright IDE blocks to control the board and interact with external sensors. Subsequently, students took a post-test (same questions as pre-test) to evaluate their acquired proficiency.

2.4.2. Implementing Concepts Course

In this course, students with experience with the command blocks of KidBright, either from the previous course or classroom training in their schools, can attend. Students took a pre-test and post-test to assess their prior and post-knowledge. In this course, students were introduced to the advanced communication interfaces of KidBright, such as connecting to external analog sensors and facilitating direct communication between KidBright boards. All groups from 18 Deaf schools engaged in the “Storytelling with Sensors” assignment, where they integrated automated systems into narrative scenarios aligning with the STEAM approach. For instance, in the retelling of “The Tortoise and the Hare”, the hare utilized an ultrasonic sensor to detect the tortoise’s presence and trigger an alarm to awaken itself, altering the story’s outcome. Students were encouraged to compose and embellish scenes which integrate artistic expression, design thinking, and system design. This then culminates in a presentation to the class, where their CT practices were evaluated.

2.4.3. Finding Solutions to Real Problems Course

Students once again took a pre-test (15 questions shown in Table A2) and post-test to assess their prior and post-knowledge. In this course, students were introduced to system design, scientific project concepts, and methodologies, leading to the construction of automatic systems for real applications. Groups identified real-life challenges, devised conceptual solutions, drafted designs, and constructed mockup prototypes with presentations showcasing each group’s work.

2.4.4. Developing Innovations Course

Following the course, a two-month “developing innovations” course ensued, during which students refined their prototypes, focusing on the concepts, functionality, performance, and preparation, for a science project competition. This course facilitated learning design, data collection, problem-solving, teamwork, and creativity skills. By the end of this course, the CT perspectives were evaluated based on the students’ innovations during the science competition.

3. Results

3.1. The Measurement of the CT Concept

Deaf students completed pre-tests at the start and post-tests (same questions as pre-test) at the conclusion of each course: learning concepts, implementing concepts, and finding solutions to real problems. These assessments were conducted to evaluate their understanding of the course contents in each learning course. The statistics of the pre-test and post-test scores are listed in

Table 4. The statistics of pre-test and post-test scores of participants.

	Mean of Students from Previous Course (%)		Mean of New-Entry Students (%)		Mean of Total Students (%)
Courses	Males	Females	Males	Females
Learning concepts
Pre-test	52.55	53.57	-	-	53.06
Post-test	72.61	70.13			71.37
Implementing concepts
Pre-test	34.71	36.47	40	41.48	38.06
Post-test	76.95	69.26	77.14	71.85	74.25
Finding solutions to real problems
Pre-test	33.84	35.95	27.78	32	31.76
Post-test	70.51	73.09	70.55	72	71.71

. The mean pre-test scores for two out of three courses were below 40 percent, indicating that Deaf students initially answered fewer than half of the questions correctly. However, after the learning program, their understanding of CT concepts improved, as reflected by the post-test mean scores rising to 72.61%, 74.25%, and 71.71% in the learning concepts, implementing concepts, and finding solutions to real problems courses, respectively. The table demonstrates that most Deaf students achieved a better understanding of the CT concepts after the training process. The results indicate a few differences between males and females as well as new-entry students and students from previous courses in both pre-test and post-test scores across all courses. The achievement in CT concepts of most Deaf students implied that the learning program helped students understand the coding concepts that complement technology and engineering in STEAM.

3.2. The Measurement of CT Practices

The CT practice was evaluated at the end of “implementing concepts” course. Students were tasked with completing project assignments in teams of four to five members. All members of each team were brainstorming the idea of the story and the design of the automated system, then assigning works to each member. Each member had participated in completing the final project and telling the story in class. As all members of teams were involved in brainstorming, developing, and presenting tasks, the evaluation of a team’s projects can be a representation of the individuals. A similar aspect is applied to the measurement of CT perspectives. Teachers evaluated the final programming projects based on the criteria outlined in

Table 5. The measurement results of CT practices.

Criteria	Rating
1. Creativity.	Low	Medium	High
Results in teams (in percents)	1 (5.56)	8 (44.4)	9 (50)
2. System is well designed.	Not good	Normal	Good
Results in teams (in percents)	3 (16.67)	13 (72.22)	2 (11.11)
3. Project works correctly.	Does not work	Partially works	Fully works
Results in teams (in percents)	0 (0)	13 (72.22)	5 (27.78)
4. Apply knowledge to complete tasks properly.	Improperly applied	Properly applied	Well applied
Results in teams (in percents)	2 (11.11)	15 (83.33)	1 (5.56)

. The average measurement score is 12.78 (63.9%), with a standard deviation of 3.0975. Videos showcasing the project assignments from the 18 teams are accessible at https://www.kid-bright.org/deaf-kid/ (accessed on 20 February 2025).

Based on the results of the CT practice measurement, nine teams achieved the highest scores in creativity. These teams created original stories inspired by their daily life experiences, such as how coding can enhance their quality of life. In contrast, eight teams adapted existing tales, and one team did not modify a story at all. All teams crafted scenes for their stories using a variety of materials, including paper, decorative items, sticks, and even costumes, effectively integrating art into technology and engineering. Only two teams earned the highest scores in system design. Their projects were well structured, featuring appropriately chosen sensors placed in optimal locations and a clear flow of tasks. However, most teams either used unsuitable sensors or failed to position them effectively for accurate detection. Despite this, all projects functioned correctly, indicating that the teams thoroughly tested and debugged their systems. The majority of teams successfully applied their knowledge of coding and sensors to complete tasks, with only two teams demonstrating improper applications. Deaf students excelled in creativity, showing strong artistic expression and a fair ability to apply their technical knowledge.

3.3. The Measurement of CT Perspectives

In accordance with the CT perspective concept, our paper evaluated students’ CT perspectives by assessing their inventions’ demonstration of problem-solving capabilities, the significance of their chosen problems, and their design and development processes during the science project competition. After the “developing innovations” course, students participated in the science project competition, where 28 inventions (teams) were showcased. The CT perspective measurement was performed through interviews and invention observations. Deaf students presented their inventions in sign language for 10 min and responded to questions from five judges in diverse fields such as programing, engineering, and teaching. The communication between students and judges was facilitated by interpreters. All team members actively participated in presenting their projects and answering questions, ensuring that the team’s scores reflected the contributions of every individual. Judges assigned scores ranging from 5 to 0 for each criterion in Table 2, categorizing them into three levels of CT perspective strength: strong, medium, or weak. The teams with total scores equal to or above 80% indicate a strong CT perspective, while total scores between 60 and 80% indicate a medium CT perspective. Other scores indicate a weak CT perspective.

The results showed that out of 28 teams, 15 (approximately 53.57%) demonstrated a strong CT perspective, 7 teams (25%) displayed a medium CT perspective, and 6 teams (21.43%) exhibited a weak CT perspective. Notably, the top three scoring projects were as follows: Intelligent System for Safety Alerts from Flooding in Tunnels and Underpasses, Smart Assistance Systems for the Elderly and Handicapped, and A Bicycle Alert System for Deaf Riders.

The Intelligent System for Safety Alerts from Flooding in Tunnels and Underpasses (illustrated in Figure 4) uses line notifications (a social media application) to alert train drivers during tunnel flooding. It also activates emergency lights and engages water-pumping motors. This innovative system integrates coding, embedded systems, electronics, and engineering to automate water-pumping and alert functions, addressing real-life problems with solutions grounded in scientific and mathematical concepts. The team created a detailed tunnel model equipped with sensors, a motor, and a KidBright board to simulate flooding scenarios and demonstrate the system’s functionality (watch the video demonstration at https://www.youtube.com/watch?v=HGG7dPUOjZE) (accessed on 20 February 2025).

The Smart Assistance System for the Elderly and Handicapped (shown in Figure 5) detects falls in bathrooms using an ultrasonic sensor array and automatically unlocks the door to facilitate rescue. This innovation leverages the fundamental principles of ultrasonic sensors, integrating them into a sensor array to enhance fall detection performance. It highlights the team’s innovation skills by combining concepts from science, technology, engineering, and mathematics. The team also built a bathroom model to demonstrate the system’s workflow, including a sensor array installation and the automatic unlocking mechanism (watch the video at https://www.youtube.com/watch?v=UQzFMvftmOw) (accessed on 20 February 2025).

Lastly, the Bicycle Alert System for Deaf Riders (depicted in Figure 6) was inspired by challenges faced during the team members’ cycling experiences. This system is equipped with an ultrasonic sensor and a microphone to detect approaching vehicles from behind and to alert the rider to avoid potential collisions. The team retrofitted a standard bicycle by attaching the ultrasonic sensor and microphone at the rear to detect both approaching objects and environmental sounds. This innovative and practical idea demonstrates the application of STEAM principles, effectively addressing real-world scenarios (watch the demonstration video at https://www.youtube.com/watch?v=melsPHziq6I) (accessed on 20 February 2025).

4. Discussion

The STEAM learning program is structured across four courses: learning concepts, implementing concepts, finding solutions to real problems, and developing innovations. All learning courses aim to integrate multiple disciplines and encourage students to apply their knowledge in a more comprehensive way. The program is based on project-based learning, which provides hand-on activities. Students learned coding concepts through assignments that enable a deeper understanding. The results show that most Deaf students excelled in creativity; for example, they crafted scenes for their stories in the “Storytelling with Sensors” assignment. Integrating “art” into the learning program not only stimulates their imaginations but also makes learning fun and engaging. STEAM learning offers a dynamic, interdisciplinary approach that can significantly enhance computational thinking (CT) in Deaf students. The advantages are enhancing hands-on learning, promoting holistic understanding, and fostering creativity and critical thinking. By combining hands-on activities, interdisciplinary learning, and collaborative projects, STEAM not only makes abstract computational concepts more concrete but also builds essential CT skills that are vital for problem-solving in the real world.

The physical board with block-based programming serves as a valuable learning tool for Deaf students. The block-based programming is more accessible to Deaf students compared to text-based alternatives, simplifying the coding process with visual command blocks. Second, it facilitates real-world applications of knowledge, nurturing technology-driven problem-solving skills. The KidBright physical board employs block-based programming to simplify the coding process for students. It encapsulates computer language commands within blocks, allowing students to drag and drop these blocks to create programs without writing computer code directly. While this approach makes coding more accessible, it has its drawbacks. The primary advantage is the ease of use and reduced complexity in coding. However, a significant limitation is that relying on block commands may hinder the development of coding proficiency in traditional programming languages. The mastery of computer languages is a critical skill for advancing in computer science education.

5. Conclusions

In today’s increasingly digital world, every nation needs individuals skilled in creative problem-solving across all age groups. Computational thinking (CT) provides a pathway to foster creativity and problem-solving abilities among youth. This study investigates the effectiveness of using the STEAM approach using physical computing to enhance the CT skills of Deaf students. The results demonstrated a significant improvement in CT concepts among most Deaf students. Regarding CT practice scores, all teams successfully developed functioning automated systems, highlighting that the STEAM learning program can help students translate knowledge into practical applications. The three example innovations showcased the students’ excellent innovation skills, demonstrating their ability to integrate multidisciplinary knowledge to solve real-world problems. Students can design a variety of automated systems, bringing their imaginations to life and addressing real challenges effectively.