Conversational agents (CAs) have become increasingly prominent in education, yet most research focuses on disembodied chatbots for higher education. Because visual representation is central for early childhood, we developed a web application with an LLM-based CA to examine how embodiment and visual appearance relate to children’s dialogic interactions. In an online study with 33 children ages 4–8, participants were randomly assigned to interact with a text-based CA, an embodied CA with a random avatar, or an embodied CA whose appearance children could customize. Children showed high verbal engagement regardless of CA type. Across interactions, children largely remained on topic, though playfulness, disengagement, and social connectedness varied. Interactions with custom avatar CAs more often included playful and disengaged behaviors, while customization patterns suggested that children created avatars resembling themselves or explored fantastical identities. This study provides initial insights into design mechanisms for child-centered pedagogical CAs.

Generative artificial intelligence (AI) has become a prominent presence in classrooms, yet relatively little is known about how students actually engage with such tools in authentic school contexts. This study examines more than 17,000 messages across 1,512 conversations from 484 middle school students using a classroom-based generative AI writing tutor. We extracted linguistic, cognitive, and interactional features, reduced dimensionality with principal component analysis, and applied clustering to identify conversation and student-level patterns of engagement. Results revealed five conversation profiles—ranging from directive to reflective dialogue—and four student profiles, including collaborators, transactional tool users, independent thinkers, and chatters. These patterns aligned with both pedagogical intent and individual orientation, underscoring that student–AI dialogue is heterogeneous but systematic. Findings contribute empirical evidence to debates about generative AI in education and provide a methodological framework for analyzing human–AI interaction in learning analytics

Integrating literacy and computational thinking (CT) can broaden computer science education participation, especially for multilingual learners. This study examined how the Computing and AI for All (CAIforALL) Act 1 Curriculum, an ELA-integrated Scratch-based CT curriculum, impacts coding attitudes among elementary students in predominantly Latine and multilingual districts. The curriculum integrates literacy strategies into CT instruction as proposed by the National Academies of Sciences, Engineering, and Medicine (NASEM) to support multilingual learners. We conducted a cluster randomized controlled trial with 1,325 students in grades 3–5 across 23 schools in two suburban districts. The treatment group used an ELA-integrated CT curriculum for a school year while controls continued with business-as instruction. Pre- and post-surveys measured five coding attitude constructs: confidence, interest, utility, perceived coding values of social circles, and perceptions of young coders. We estimated treatment effects using a two-level hierarchical linear model, controlling for student and classroom characteristics. Findings show no statistically significant differences emerged in overall coding attitudes between groups. However, students exposed to a year of ELA-integrated CT curriculum showed significant increases in coding confidence. The curriculum did not significantly affect other attitude dimensions. Findings suggest that an ELA-integrated CT curriculum can enhance coding confidence among elementary students, demonstrating the value of early computing exposure and integrated approaches.

Early literacy skills are crucial predictors of children's academic success.Dialogic reading-an interactive approach where adults and children engage in discussions about stories-has proven highly effective in developing these skills.However, many families face barriers implementing this practice due to time constraints, limited resources, or linguistic challenges.We present StoryPal, an

Peer Reviewed

According to an ecological affordances perspective, any static curriculum has a set of affordances, and differences in teachers, students, and the teaching environment change how those affordances are viewed and used. Therefore, teaching is a relationship between the curriculum, the teacher, and the students. As such, it is not only possible but expected that a teacher will diverge from the details of a lesson plan to better accommodate the needs of themselves as a teacher and their students as learners.

• Developed a bilingual conversational agent (CA) for parent-child shared reading • The CA boosted children's Spanish vocabulary learning via bilingual verbal engagement • The CA stimulated personally and educationally meaningful parent-child dialogue • Parents perceived the CA to be educational and culturally relevant Recognizing the challenges bilingual children face in school readiness and the potential of bilingual dialogic shared reading in improving language and literacy, this study investigates the use of a bilingual conversational agent (CA) to enhance shared reading experiences in home environments. While current CAs hold promise in fostering young children's learning, they do not typically consider the linguistic and cultural needs of bilingual children and rarely involve parents intentionally. To this end, we developed a bilingual CA, embedded within ebooks, to support children's language learning and parent engagement for Latine Spanish-English bilingual families. A week-long home-based study with 15 families indicated that the bilingual CA elicited a high level of bilingual verbal engagement from children, thereby promoting their vocabulary acquisition. It also stimulated meaningful conversations among parents and children. This study provides design implications for developing CAs for bilingual children and parents.

Abstract As generative AI becomes ubiquitous, writers must decide if, when, and how to incorporate generative AI into their writing process. Educators must sort through their role in preparing students to make these decisions in a quickly evolving technological landscape. We created an AI-enabled writing tool that provides scaffolded use of a large language model as part of a research study on integrating generative AI into an upper division STEM writing-intensive course. Drawing on decades of research on integrating digital tools into instruction and writing research, we discuss the framework that drove our initial design considerations and instructional resources. We then share our findings from a year of design-based implementation research during the 2023–2024 academic year. Our original instruction framework identified the need for students to understand, access, prompt, corroborate, and incorporate the generative AI use effectively. In this paper, we explain the need for students to think first, before using AI, move through good enough prompting to agentic iterative prompting, and reflect on their use at the end. We also provide emerging best practices for instructors, beginning with identifying learning objectives, determining the appropriate AI role, revising the content, reflecting on the revised curriculum, and reintroducing learning as needed. We end with an indication of our future directions.

The Latinx community is underrepresented in tech-oriented fields, which aligns with the lack of culturally relevant learning experiences in CS for Latinx youth, hindering their ability to conceptualize technology as a tool for transformation and vehicle for cultural expression. This study takes on a restorying approach at the elementary level, where 9 fourth-grade students engaged in focus group discussions over three days to generate prompts for a generative AI art. Through the lens of restorying, the prompts had students conceptualize a future with a focus on their Mexican-American heritage, local community, and technology. The study revealed that students associated their heritage with symbolic representations such as food and music, and characterized the community as a commercialized space while also emphasizing locations conducive to family-oriented activities. As a result, technology in community spaces was associated with consumerism. However, when envisioning a futuristic, transformed community, they made deeper connections between the role of technology in the community, making intricate connections between community improvements and technology-based solutions. This underscores the need for computing education to dedicate time for young learners to reflect on the role technology has on their current culture and community to make deeper connections.

The Latinx population is underrepresented in secondary computer science (CS) courses as well as in the STEM workforce. By exposing Latinx students to positive coding experiences and encouraging a growth mindset earlier in school, Latinx participation in CS may likely increase. This poster presents the results of examining the reflections upper-elementary students recorded in their journals after completing a structured Scratch coding project. Students were prompted to share problems they faced, how they responded to those problems, and what they were proud of in their project.

Bilingual children have unique needs for school readiness as they navigate between two languages and cultures. A supportive home language environment, where children are frequently exposed to language through conversation and reading, can positively impact their language development and prepare them for school. However, current conversational agents and e-books designed for children do not typically take into account the cultural and linguistic needs of bilingual children and do not involve parents. This project presents the development of a bilingual conversational agent and accompanying e-book, designed to support parent-child interactions and promote language development for Latinx Spanish-English bilingual children. Results from a user study indicate that the bilingual agent effectively engages children verbally and encourages parental involvement in reading processes. The study also provides design insights for creating conversational agents for bilingual children.

As computing education begins to grow in K-12 classrooms, it is increasingly important to understand the relationships between a student's coding attitudes and computing outcomes. While work has been done at the undergraduate and high school level, less is known about elementary students, and even less about Latinx students. We sought to explore the connections between Latinx elementary students' coding attitudes and computational thinking (CT) skills after participating in a year-long, remote computing curriculum in a public school district in Southern California. We collected validated measures on coding attitudes and CT skills for 164 fourth-grade students. We used structural equation modeling to analyze the relationship between the coding attitude constructs (i.e., confidence, utility, perception of coders, social influences, and interest) and scores on a CT assessment while controlling for demographic factors (i.e., gender, English language fluency, free or reduced lunch status, reading and math standardized test scores). In a model demonstrating acceptable fit, only math scores were a significant predictor of CT scores and coding confidence. Confidence was found to be significantly associated with the other attitude constructs except for perception of coders. Our results indicate that confidence plays a prominent role in student computing learning, especially at a young age. Efforts to better support Latinx students in computer science should include strategies that address students' coding confidence, such as developing positive perceptions of a student's learning process (e.g., failures as learning opportunities) in specific coding skills and practices (e.g., tinkering, debugging).

This study analyzed the relationship between computational thinking (CT) and coding attitudes of upper elementary students after exposure to a year-long CT curriculum. Using ordinary least squares regression and controlling for student demographics (i.e., gender and English Learner (EL) status), we found that CT skills were a significant factor in modeling coding attitudes, regardless of demographic controls. Interviews of the students unveiled an interaction between coding interest and social values in a debugging process.

Researchers and practitioners have worked to bring computer science to all students. However, CS is still an inequitable field. When developing curricula, we must account for the importance of connecting the lives and identities of historically marginalized students with instructional materials. We examine the knowledge and experiences that elementary (age 9-10) students drew upon to implement an open-ended programming assignment about themselves. We coded the Scratch projects of 189 students to investigate what aspects of their lives students reference, and how they use various modalities in Scratch to create digital media products about themselves. In our data, we found that the most common spheres portrayed were identity, hobbies and leisure, and interests. When expressing their identities, students narrated their experiences, expressed consciousness about their skills, connected to the user via their interests, and sometimes discussed equity. Additionally, our findings add to previous literature on Scratch as an effective platform for digital media production. In particular, we found that students used dialogue, images, sound, and the title to represent their spheres of influence through programming multimedia and interactive projects. This work has implications for the development of curricula and teaching guides for upper elementary students courses on computational thinking. Spheres of influence could inform researchers on the most influential aspects of students life in a specific context, providing elements for creating culturally relevant and interest-based materials and lessons.

Ying Xu, Dakuo Wang, Mo Yu, Daniel Ritchie, Bingsheng Yao, Tongshuang Wu, Zheng Zhang, Toby Li, Nora Bradford, Branda Sun, Tran Hoang, Yisi Sang, Yufang Hou, Xiaojuan Ma, Diyi Yang, Nanyun Peng, Zhou Yu, Mark Warschauer. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Video programs are important, accessible educational resources for young children, especially those from an under-resourced backgrounds. These programs’ potential can be amplified if children are allowed to socially interact with media characters during their video watching. This paper presents the design and empirical investigation of interactive science-focused videos in which the main character, powered by a conversational agent, engaged in contingent conversation with children by asking children questions and providing responsive feedback. We found that children actively interacted with the media character in the conversational videos and their parents spontaneously provided support in the process. We also found that the children who watched the conversational video performed better in the immediate, episode-specific science assessment compared to their peers who watched the broadcast, non-interactive version of the same episode. Several design implications are discussed for using conversational technologies to better support child active learning and parent involvement in video watching.

Background and Context. With increasing efforts to bring computing education opportunities into elementary schools, there is a growing need for assessments, with arguments for validity, to support research evaluation at these grade levels. After successfully piloting a 10-question computational thinking assessment (Assessment of Computing for Elementary Students – ACES) for 4th graders in Spring 2020, we used our analyses of item difficulty and discrimination to iterate on the assessment. Objectives. To increase the number of potential items for ACES, we created isomorphic versions of existing questions. The nature of the changes varied from incidental changes that we did not believe would impact student performance to more radical changes that seemed likely to influence question difficulty. We sought to understand the impact of these changes on student performance. Method. Using these isomorphic questions, we created two versions of our assessment and piloted them in Spring 2021 with 235 upper-elementary (4th grade) students. We analyzed the reliability of the assessments using Cronbach’s alpha. We used Chi-squared tests to analyze questions that were identical across the two assessments to form a baseline of comparison and then ran Chi-Squared and Kruskal-Wallis H tests to analyze the differences between the isomorphic copies of the questions. Findings. Both assessment versions demonstrated good reliability, with identical Cronbach’s alphas of 0.868. We found statistically similar performance on the identical questions between our two groups of students, allowing us to compare their performance on the isomorphic questions. Students performed differently on the isomorphic questions, indicating the changes to the questions had a differential impact on student performance. Implications. This paper builds on existing work by presenting methods for creating isomorphic questions. We provide valuable lessons learned, both on those methods and on the impact of specific types of changes on student performance.

Developing student interest is critical to supporting student learning in computer science. Research indicates that student interest is a key predictor of persistence and achievement. While there is a growing body of work on developing computing identities for diverse students, little research focuses on early exposure to develop multilingual students’ interest in computing. These students represent one of the fastest growing populations in the US, yet they are dramatically underrepresented in computer science education. This study examines identity development of upper elementary multilingual students as they engage in a year-long computational thinking curriculum, and follows their engagement across multiple settings (i.e., school, club, home, community). Findings from pre- and post-surveys of identity showed significant differences favoring students’ experiences with computer science, their perceptions of computer science, their perceptions of themselves as computer scientists, and their family support for computer science. Findings from follow-up interviews and prior research suggest that tailored instruction provides opportunities for connections to out-of-school learning environments with friends and family that may shift students’ perceptions of their abilities to pursue computer science and persist when encountering challenges.

With the rapid development of artificial intelligence, voice user interfaces (VUIs) capable of speech-based interaction are poised to support children’s language development by serving as their language partners. This paper reports an analytic evaluation of the currently available voice-based apps targeting young children to examine whether and how they incorporate evidence-based dialogue strategies that effectively support children’s learning. We found that, despite the fact that the current apps support a variety of language activities, most fail to carry out open-ended dialogue and provide extended back-and-forth opportunities, thus limiting their ability to encourage children’s language output and increase children’s language exposure. We discuss four design implications for developing VUIs that initiate dialogue and provide feedback in ways that better facilitate children’s language learning.

The demand for computational thinking (CT) problem solving abilities surge as every aspect of life becomes more dependent on complex digital technologies. Just as in math and language, a strong CT foundation needs to be established in early education in order for students to develop an instinctive CT perspective of the world. The urgent demand for CT instruction in elementary school quickly draws attention to the shortage of elementary school-level teachers qualified and interested in CT. Additionally, with a commitment to equity in the United States education system and knowledge of the high percentage of English language learning (ELL) students in schools, the obligation to create curricula that will provide access to CT knowledge, skills, and practices for elementary-level ELL students is loudly apparent. In response to these two needs, our team has adapted existing Scratch-based CT curriculum to support classroom teachers with minimal CT experience and to be more accessible to English language learners. The purpose of this paper is to share the framework that guided the curriculum adaptations, to describe the specific changes that were made, and to discuss discoveries made during the process. This journey may be helpful to anyone who is tasked with modifying a curriculum to meet the needs of novice content teachers and ELL students.

Although online courses can provide students with a high-quality and flexible learning experience, one of the caveats is that they require high levels of self-regulation. This added hurdle may have negative consequences for first-generation college students. In order to better understand and support students’ self-regulated learning, we examined a fully online Chemistry course with high enrollment (N = 312) and a high percentage of first-generation college students (65.70%). Using students’ lecture video clickstream data, we created two indicators of self-regulated learning: lecture video completion and time management. Performing a k-means clustering on these indicators uncovered four distinct self-regulated learning patterns: (1) Early Planning, (2) Planning, (3) Procrastination, and (4) Low Engagement. Early Planning behaviors were especially important for course success—they consistently predicted higher final course grades, even after controlling for important demographic variables. Interestingly, first-generation college students classified as Early Planners achieved at similar levels as their non-first-generation peers, but first-generation students in the Low Engagement group had the lowest average grades among students. Overall, our results show that self-regulation may be an important skill for determining first-generation students’ STEM achievement, and targeting these skills may serve as a useful way to support their specific learning needs.

As reliance on technology increases in practically every aspect of life, all students deserve the opportunity to learn to think computationally from early in their educational experience. To support the kinds of computer science curriculum and instruction that makes this possible, there is an urgent need to develop and validate computational thinking (CT) assessments for elementary-aged students. We developed the Assessment of Computing for Elementary Students (ACES) to measure the CT concepts of loops and sequences for students in grades 3-5. The ACES includes block-based coding questions as well as non-programming, Bebras-style questions. We conducted cognitive interviews to understand student perspectives while taking the ACES. We piloted the assessment with 57 4th grade students who had completed a CT curriculum. Preliminary analyses indicate acceptable reliability and appropriate difficulty and discrimination among assessment items. The significance of this paper is to present a new CT measure for upper elementary students and to share its intentional development process.

Joint book reading is a highly routinized activity that is nearly universal among families. Conversational agents (CAs) can potentially act as joint-reading partners by engaging children in story-related, scaffolded conversations. In this project, we develop a CA reading partner that incorporates components of effective conversational guidance (i.e., questions to stimulate thinking, specific feedback, and adaptive scaffolding) and examine children's interactions with this CA. We identify patterns in children's language production, flow maintenance, and affect when responding to the CA. We then lay out a set of affordances and challenges for developing CAs as conversation partners. We propose that, rather than attempting to develop CAs as an exact replicate of human conversational partners, we should treat child-agent interaction as a new genre of conversation and calibrate CAs based on children's actual communicative practices and needs.

Voice-based applications powered by conversational agents can potentially support young children's literacy development in informal settings. Yet, to realize such potential, designers must consider young users' typical communication and learning patterns. In this paper, we developed a framework of seven design dimensions across three aspects (i.e., learning content and goals, interactions and gamifications, and conversation design) that could influence the educational benefits young children receive from voice-based apps. We then used this framework to conduct a content analysis of 535 literacy-focused voice apps on the market to examine whether the prevalent design features of these apps meet the unique needs of young learners. Lastly, we discussed the implications of current design tendencies with the aim of encouraging future voice-based app designers to bridge the gap between research and practice.

Conversational agents (CAs) available in smart phones or smart speakers play an increasingly important role in young children's technological landscapes and life worlds. While a handful of studies have documented children's natural interactions with CAs, little is known about children's perceptions of CAs. To fill this gap, we examined three- to six-year-olds' perceptions of CAs' animate/artifact domain membership and properties, as well as their justifications for these perceptions. We found that children sometimes take a more nuanced position and spontaneously attribute both artifact and animate properties to CAs or view them as neither artifacts nor animate objects. This study extends current research on children's perceptions of intelligent artifacts by adding CAs as a new genre of study and provides some underlying knowledge that may guide the development of CAs to support young children's cognitive and social development.

Self-efficacy is an important construct for CS teachers’ professional development, because it can predict both teaching behaviors as well as student outcomes. Research has shown that teachers’ self-efficacy can be as influential as their actual level of knowledge and abilities. However, there has been very limited research on CS teachers’ self-efficacy. This study describes the development and implementation of an instrument that measures secondary school teachers’ self-efficacy in teaching computer science. Teachers attended a nine-week hybrid professional development program and completed the computer science teaching self-efficacy instrument. Confirmatory factor analysis validated the self-efficacy instrument, which can be potentially used in other CS education settings. The results also indicated that teachers’ self-efficacy in the content knowledge and pedagogical content knowledge dimensions of teaching computer science significantly increased from participating in the professional development program.

Informed by cognitive theories of learning, this work examined how students' self-reported study patterns (spacing vs. cramming) corresponded to their engagement with the Learning Management System (LMS) across two years in a large biology course. We specifically focused on how students accessed non-mandatory resources (lecture videos, lecture slides) and considered whether this pattern differed by underrepresented minority (URM) status. Overall, students who self-reported utilizing spacing strategies throughout the course had higher grades than students who reported cramming throughout the course. When examining LMS engagement, only a small percentage of students accessed the lecture videos and lecture slides. Applying a negative binomial regression model to daily counts of click activities, we also found that students who utilized spacing strategies accessed LMS resources more often but not earlier before major deadlines. Moreover, this finding was not different for underrepresented students. Our results provide some initial evidence showing how spacing behaviors correspond to accessing learning resources. However, given the lack of general engagement with LMS resources, our results underscore the value of encouraging students to utilize these resources when studying course material.

Representation and prediction of student navigational pathways, typically based on neural network (NN) methods, have seen their potential of improving instruction and learning under insufficient human knowledge about learner behavior. However, they are prominently studied in MOOCs and less probed within more institutionalized higher education scenarios. This work extends such research to the context of online college courses. Comparing student navigational sequences through course pages to documents in natural language processing, we apply a skip-gram model to learn vector embedding of course pages, and visualize the learnt vectors to understand the extent to which students' learning pathways align with pre-designed course structure. We find that students who get different letter grades in the end exhibit different levels of adherence to designed sequence. Next, we fit the embedded sequences into a long short-term memory architecture and test its ability to predict next page that a student visits given her prior sequence. The highest accuracy reaches 50.8% and largely outperforms the frequency-based baseline of 41.3%. These results show that neural network methods have the potential to help instructors understand students' learning behaviors and facilitate automated instructional support.

Group activities that use Google Docs for simultaneous collaborative writing and editing are increasingly common in higher education. Although studies show that synchronous collaboration can bring multiple benefits, such as enhanced productivity and writing quality, little is known about these writing practices in classrooms and their impact on students' writing. Using a mixed method approach, we conducted an empirical study that explores the different styles of synchronous collaboration in 45 Google Docs documents produced by 82 undergraduate students, and how students' practices affect the specific dimensions of the final text including quality. The results suggest that (a) out of four styles, Divide and Conquer style tended to produce better quality text whereas Main Writer had the lowest quality scores, and that (b) balanced participation and amount of peer editing led to longer texts with higher quality scores for content, evidence, but not organization or mechanics. Given these results, we suggest several design features for collaborative writing systems and propose guidelines for instructional practices.

Student clickstream data can provide valuable insights about student activities in an online learning environment and how these activities inform their learning outcomes. However, given the noisy and complex nature of this data, an on-going challenge involves devising statistical techniques that capture clear and meaningful aspects of students' click patterns. In this paper, we utilize statistical change detection techniques to investigate students' online behaviors. Using clickstream data from two large university courses, one face-to-face and one online, we illustrate how this methodology can be used to detect when students change their previewing and reviewing behavior, and how these changes can be related to other aspects of students' activity and performance.

We take an ecological approach to studying social media use and its relation to mood among college students. We conducted a mixed-methods study of computer and phone logging with daily surveys and interviews to track college students' use of social media during all waking hours over seven days. Continual and infrequent checkers show different preferences of social media sites. Age differences also were found. Lower classmen tend to be heavier users and to primarily use Facebook, while upper classmen use social media less frequently and utilize sites other than Facebook more often. Factor analysis reveals that social media use clusters into patterns of content-sharing, text-based entertainment/discussion, relationships, and video consumption. The more constantly one checks social media daily, the less positive is one's mood. Our results suggest that students construct their own patterns of social media usage to meet their changing needs in their environment. The findings can inform further investigation into social media use as a benefit and/or distraction for students.

In recent years, a number of low-cost laptops have been created for children's education, most notably the XO, developed by One Laptop per Child to embody principles of constructionist learning, and the ClassmatePC, designed by Intel to fit within and improve traditional education. We report on a series of field studies in Mexican elementary schools that deployed the XO or ClassmatePC. Although both devices are promoted as valuable for improving education in developing countries, our studies suggest that creating the social and technical infrastructures needed to sustain school laptop use is far more complex than what technology designers assume.

VizClass is a university classroom environment designed to offer students in computer graphics and engineering courses up-to-date visualization technologies. Three digital whiteboards and a three-dimensional stereoscopic display provide complementary display surfaces. Input devices include touchscreens on the digital whiteboards, remote keyboards, data gloves, and hand-position sensors.We use observations, interviews, and surveys to examine the pedagogical impacts of VizClass for teaching and learning computer graphics and virtual reality. Preliminary findings include positive student and teacher attitudes and greater learner engagement in after-class collaborations.