Self-efficacy is an essential source of motivation for learning. While considerable research has theorised and examined the how and why of self-efficacy in a single domain of study, longitudinal research has not yet tested how self-efficacy might generalise or transfer between subjects such as mathematics, native and foreign language studies. The current study examined academic self-efficacy (two measurements 10 months apart) in three subjects (mathematics, native language, and foreign language) across students' first year at junior high school. Two studies were conducted, each including three schools (study-A: n=480; study-B: n=398) to support a test and retest of self-efficacy differences and interrelationships across the year of study. Analyses of self-efficacy change presented a general pattern of significant, small declines in students' self-efficacy for all three subjects. Longitudinal latent analyses indicated a consistent moderate predictive effect from foreign language self-efficacy to native language self-efficacy. The pattern of declines, while consistent with research in Western contexts is a source of concern. The transfer of self-efficacy from foreign to native language learning has potential educational and broader psychological implications.