A Mixed-Methods Study of Online Collaborative Learning Environments for Knowledge Building: Evidence from Five Universities in Hohhot

Online collaborative learning has become routine in hybrid university teaching, yet deep learning outcomes remain insufficient. This mixed-methods study investigated five universities in Hohhot using questionnaires (N=986), teacher interviews (n=20), and content analysis of 4,238 meaning units from six courses (triangulation design). Results show that basic platform infrastructure is adequate, but specialized tools for knowledge building are used by only 3.2% of participants. Collaborative activities cover 82.6% of courses, yet 67.9% are shallow “division–aggregation” cooperation. Interaction Analysis Model (IAM) coding reveals that deep constructive discourse accounts for only 16.8%, with testing/revision (4.1%) and consensus/application (1.5%) particularly weak. Teacher online guidance consists of 58.2% task instructions and disciplinary reminders, while probing and synthesis fall below 25%. Student overall satisfaction is moderate (3.42/5), with low scores for “ideas improved” (2.89) and “new understanding formed” (2.76). Independent-sample t-tests show that Mongolian-medium students report significantly lower perceived knowledge building than their Chinese-medium peers (2.61 vs. 3.12, p < 0.001, d = 0.52), indicating a strong association between language ecology and depth of collaboration. This suggests that improving online learning environments in ethnic regions must address structural differences arising from multilingual education. Operational recommendations are proposed in three areas: instructional system redesign, technological resource upgrades, and teacher professional development.