In modern organizations, agile methods have become key strategies for project development, promoting collaboration and adaptability within teams. These approaches optimize communication and cooperation, enabling effective responses to evolving environmental demands. However, collaboration relies on effective communication, coordinated action, and cooperative participation elements that are difficult to evaluate without automated methodological support. Meanwhile, the emergence of transformer-based natural language processing (NLP) models has enabled the identification of semantic features for analyzing communication and collaboration. This study presents a verbal intervention classification system based on natural language processing (NLP) techniques, utilizing multimodal learning analytics to transcribe audio into text and characterize interactions. The system, built upon DistilBERT and trained with manually annotated examples, identifies and categorizes interventions into five classes: question, answer, feedback, suggestion, and comment. The model has been implemented in a functional platform that visualizes results through an interactive interface, allowing facilitators of collaborative activities to analyze the dynamics of interactions and participant contributions, thereby supporting the continuous improvement of such activities.

The underrepresentation and low retention of women in STEM fields—particularly in software development—remains a structural challenge. For this reason, understanding how gender composition affects collaborative participation in agile teams, especially in contexts where there are gender-isolated participants, is crucial. Unlike traditional approaches, this work employs nonverbal communication analysis through Multimodal Learning Analytics, focusing on two key indicators: speaking time and bodily mimicry. The analysis was based on audiovisual recordings of 16 student teams in Computer Engineering, who participated in collaborative user story estimation sessions, both with and without the use of the Planning Poker technique. Teams were categorized based on their gender composition as homogeneous, balanced, or asymmetric. The results show that individuals who belonged to the gender minority within the team tended to participate less, both in speaking time and in nonverbal synchronization behaviors, suggesting subtle exclusion or the adoption of a peripheral role. Additionally, stronger mimicry patterns were observed before final voting, reinforcing their potential as early indicators of group convergence. This work provides empirical evidence on the impact of gender composition in agile collaboration dynamics and outlines new research directions for designing more inclusive teams in educational and professional settings.