ProgSinais: uma ferramenta de feedback de programação em Libras com LLM e sinalizabilidade verificada

• Wyllgner França de Amorim UFPR
• Jáder Louis de S. Gonçalves USP
• Lucas Marques da Cunha UFRO

DOI: https://doi.org/10.5753/webmedia_estendido.2026.34807

Resumo

Deaf students learn to program by reading Portuguese as a second language, facing error messages in technical English and a computing vocabulary with no established sign in Libras. This paper presents ProgSinais, a web tool for Programming I exercises that runs the student’s code in an isolated container, classifies the error through deterministic heuristics, and uses a language model to write progressive hints in simplified Portuguese, presented in Libras with sign-by-sign synchronized highlighting. Signability is not requested from the model but verified: every model-generated hint passes through a gate that measures it, rejects and regenerates what would fall into fingerspelling, and incorporates the terms the avatar fingerspells at run time. The signability index estimated by the verifier rises from 0.25 in the interpreter message to 0.83 in the text delivered to the student, on a small and declared base, which characterizes the instrument and does not measure learning.

Palavras-chave: accessibility, Libras, sign language, programming education, language models, text-sign synchronization

Referências

Oliver Alonzo, Matthew Seita, Abraham Glasser, and Matt Huenerfauth. 2020. Automatic Text Simplification Tools for Deaf and Hard of Hearing Adults: Benefits of Lexical Simplification and Providing Users with Autonomy. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20). ACM, 1–13. DOI: 10.1145/3313831.3376563

Brasil. 2005. Decreto n. 5.626, de 22 de dezembro de 2005, arts. 14, 15 e 22. Diário Oficial da União.

Si Chen, Shuxu Huffman, Qingxiaoyang Zhu, Haotian Su, Qi Wang, and Raja Kushalnagar. 2026. Exploring the design of LLM-powered question generation for deaf and hard of hearing learners. Computers and Education: Artificial Intelligence 10 (2026), 100615. DOI: 10.1016/j.caeai.2026.100615

Enyo Gonçalves, Francisco Oliveira, Jessyka Vilela, Mariana Peixoto, and Jaelson Castro. 2015. Produção de Videoaulas de Programação em Java Acessíveis no Contexto de um Projeto de Capacitação Profissional para Pessoas Surdas. In Anais do XXVI Simpósio Brasileiro de Informática na Educação. SBC, 877–886. DOI: 10.5753/cbie.sbie.2015.877

Rafael Granada, Victor Cesário, Darcylene Domingues, Regina Barwaldt, Ricardo Rodrigues, and Cristiane Fernandes. 2017. Dicionário de termos de computação como facilitador no ensino de programação para surdos. In Anais dos Workshops do VI Congresso Brasileiro de Informática na Educação (CBIE 2017). SBC, 1049. DOI: 10.5753/cbie.wcbie.2017.1049

Majeed Kazemitabaar, Runlong Ye, Xiaoning Wang, Austin Zachary Henley, Paul Denny, Michelle Craig, and Tovi Grossman. 2024. CodeAid: Evaluating a Classroom Deployment of an LLM-based Programming Assistant that Balances Student and Educator Needs. In Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI ’24). ACM, 1–20. DOI: 10.1145/3613904.3642773

Mark Liffiton, Brad Sheese, Jaromir Savelka, and Paul Denny. 2023. CodeHelp: Using Large Language Models with Guardrails for Scalable Support in Programming Classes. In Proceedings of the 23rd Koli Calling International Conference on Computing Education Research. ACM, 1–11. DOI: 10.1145/3631802.3631830

Claudemir Lima, Elidiene Lima, Neiton Silva, and Jorge Correia Neto. 2021. Tecnologia assistiva e tradução para Libras: desafios da ferramenta de tradução automática de vídeos VLibras. Research, Society and Development 10 (2021), e385101220720. DOI: 10.33448/rsd-v10i12.20720

Mathias Müller, Zifan Jiang, Amit Moryossef, Annette Rios, and Sarah Ebling. 2023. Considerations for meaningful sign language machine translation based on glosses. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). ACL, 682–693. DOI: 10.18653/v1/2023.acl-short.60

Rose Niousha, Samantha Boatright Smith, Abigail O’Neill, J.D. Zamfirescu-Pereira, John DeNero, and Narges Norouzi. 2026. Misconception-Aware LLM Programming Tutor: Lessons Learned from Student-Tutor Interactions. In Proceedings of the 57th ACM Technical Symposium on Computer Science Education V.2 (SIGCSE TS 2026). ACM, 1455–1456. DOI: 10.1145/3770761.3777289

Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. BLEU: a Method for Automatic Evaluation of Machine Translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics (ACL). ACL, 311–318. DOI: 10.3115/1073083.1073135

Lianne Roest, Hieke Keuning, and Johan Jeuring. 2024. Next-Step Hint Generation for Introductory Programming Using Large Language Models. In Proceedings of the 26th Australasian Computing Education Conference (ACE ’24). ACM, 144–153. DOI: 10.1145/3636243.3636259

Ronnie E. S. Santos, Cleyton V. C. Magalhães, Jorge S. Correia-Neto, Leandro M. Queiros, and Guilherme Vilar. 2013. Trabalhando lógica de programação com portadores de deficiência auditiva: a experiência com a Linguagem Proglib e a IDE Hands. Revista Brasileira de Computação Aplicada 6, 1 (2013), 32–44. DOI: 10.5335/rbca.2014.3145

André Silva, Simone Ferreira, Tatiane de Sá, and Cláudia Cappelli. 2024. Use Plain Language and Unlock Efficiency in Automatic Translation to Libras by Avatars. In Anais do XXIII Simpósio Brasileiro sobre Fatores Humanos em Sistemas Computacionais. SBC, 870–885. [link]

Irvna Soares, Jorge Prates, Samir Murad, and Silvana Melo. 2024. Um Panorama das Tecnologias Inclusivas no Ensino de Programação para Alunos Surdos no Brasil. In Anais do XXXV Simpósio Brasileiro de Informática na Educação. SBC, 1279–1290. DOI: 10.5753/sbie.2024.242469