Designing Social Robots for English Language Learners
Across U.S. classrooms, more than 4.9 million English Language Learners (ELLs) face a persistent gap in instructional support. Only 2% of educators have formal training to address their linguistic and socio-cultural needs, leaving many students with minimal targeted language instruction—often less than 30 minutes per day. This shortage of resources and expertise has prompted researchers to explore innovative tools, including social robots, as potential aids in language acquisition.
Social robots have demonstrated the ability to increase motivation, engagement, and oral participation while reducing anxiety in language learning contexts. Studies in robot-assisted language learning (RALL) show gains in vocabulary, storytelling, and overall interaction when children work with robots designed as tutors, peers, or learning companions. Yet, for culturally diverse ELL populations, effective design requires more than technical capability—it demands cultural appropriateness and stakeholder collaboration.
An exploratory study engaged 95 ELL students from kindergarten to fifth grade, 39 parents, and eight educators in a participatory design process. Participants represented 18 home language backgrounds, with Spanish the most common. Using the Culturally Localized User Experience (CLUE) framework, the research team invited students, parents, and teachers to act as cultural informants, shaping the vision for educational robots in their schools.
Interactive group interviews used images of six robots—commercial models like Jibo and Nao, prototypes such as EMAR V1/V2, MIT’s Dragonbot, and Cornell’s Blossom. Participants labeled each image with emoji attributes representing positive traits (happy, smart, friendly) or negative ones (scary, boring). This method accommodated language barriers and encouraged visual, intuitive responses.
Students envisioned robots helping with partner reading, homework, and English practice. Some saw social benefits: “[The robot] can help and like if you don’t have a person to play with they can play with you.” Parents expressed enthusiasm, noting that robots might encourage shy children to ask questions without fear of judgment. One parent observed, “…just like my kids that are shy sometimes they have questions, but because of embarrassment they don’t ask, they keep those questions to themselves.” Educators echoed these points, highlighting the potential for robots to be “less intimidating” than human peers or teachers.
Preferences in robot appearance varied. Students and teachers gave the highest positive ratings to EMAR V1, while parents favored Blossom’s soft-bodied, animal-like design, followed closely by Nao’s humanoid form. Nao was most preferred by older students, while younger ones leaned toward Blossom. Parents valued designs that looked “real” and “friendly,” avoiding overly mechanical or unsettling features. Dragonbot, despite being designed for children, received the most negative labels across groups, often described as “scary” or “ugly.”
Concerns emerged alongside enthusiasm. Parents and teachers stressed that robots should not replace teachers, citing the irreplaceable human capacity for empathy. Privacy was another issue—some worried about data collection, surveillance, or hacking. A few parents feared increased screen time. Teachers cautioned against using robots as supervisory tools, wary of shifting classroom dynamics. Students’ concerns were more imaginative, shaped by media portrayals: fears of robots “messing everything up,” breaking objects, or even “exploding.”
Cultural and linguistic adaptability surfaced as a design priority. Educators wanted robots capable of understanding multiple languages, dialects, and accents to serve diverse classrooms. As one administrator noted, “It would have to be dynamic enough to hear and respond to regional dialects and not have a person constantly repeating themselves.”
The study’s findings align with prior research showing that human-like robots can foster empathy without triggering the “uncanny valley” effect, and that animal-like designs can be especially effective with younger children. The popularity of Nao in educational contexts suggests it may strike the right balance of familiarity and functionality. However, the success of any educational robot will depend on careful integration into classroom routines, respect for privacy, and sustained collaboration with teachers and parents.
For engineers and designers, these insights underscore the importance of participatory design in robotics for education. Technical capabilities must be matched with cultural sensitivity, stakeholder trust, and clear role definition. In the context of ELL support, robots are not just machines—they are potential social actors whose design and deployment could influence language development, confidence, and classroom equity.
