Learning proximal cyclic sighing buffers anxiety across learning moments and boredom across days.

Emotions during language learning fluctuate rapidly, but intervention studies often miss when change occurs. In a two-arm randomized ecological momentary trial, 84 university foreign-language learners completed a 20-day protocol comprising 5 baseline, 10 intervention, and 5 post-intervention days, with five assessments per day (6,281 observations). Forty-seven participants completed three daily 5-minute cyclic sighing sessions at fixed learning-proximal times (11:00, 13:00, and 15:00); 37 controls continued learning as usual. Dense within-person modelling revealed a dual-timescale pattern: anxiety was selectively lower at learning-proximal assessments during the intervention, indicating an acute relief window that was not evident at the end-of-day assessment, while across days anxiety increased in controls but remained near baseline with breathwork. Boredom showed weaker buffering and enjoyment no clear change. Exploratory network analyses suggested weaker heart-rate coupling to anxiety and enjoyment. Brief scheduled cyclic sighing may provide feasible just-in-time regulation during language learning.