A growth mindset is the belief that ability can be developed through effort and strategy, opposed to a fixed mindset, the belief that ability is a stable trait you either have or lack. Carol Dweck built the construct from her earlier work on implicit theories of intelligence: entity theorists treat intelligence as fixed, incremental theorists treat it as malleable. The popular claim, amplified far beyond the research, is that switching learners from the first belief to the second raises motivation, resilience after failure, and ultimately achievement.
Dweck's account is causal and specific. A learner who believes intelligence is fixed reads a low grade as a verdict on their capacity, so they disengage to protect self-worth. A learner who believes intelligence grows reads the same grade as information about effort or strategy, so they persist. The proposed lever in the classroom is process praise ("you found a good way to do that") rather than person praise ("you're so clever"), on the grounds that praising the trait teaches children their performance measures a fixed quantity. The original demonstration, Mueller and Dweck (1998), showed children praised for intelligence later avoided challenge and crumbled after failure relative to children praised for effort.
The gap between the popular claim and the data is the central fact about this construct. The decisive evidence is Sisk, Burgoyne, Sun, Butler, and Macnamara's two meta-analyses (2018, Psychological Science). Across 129 studies (162 independent samples, N = 365,915), the average correlation between a person's mindset and their academic achievement was about r = .10, meaning mindset accounted for roughly 1% of the variance in achievement. The companion meta-analysis of interventions found a weak overall effect, d = 0.08. The headline that changing beliefs reliably lifts grades is not what the aggregate data show.
The same null pattern recurs in large independent samples. Bahník and Vranka (2017, Personality and Individual Differences 117, 139–143) tested 5,653 university applicants and found mindset essentially unrelated to scholastic aptitude, r = −.03. Li and Bates (2019, Journal of Experimental Psychology: General), running careful replications with children in China, could change what children believed about intelligence but found those beliefs had no relationship to their grades or to grade change across a semester. Learners said the right things and performed no differently.
Intervention trials at scale tell the same story. The UK Education Endowment Foundation's Changing Mindsets effectiveness trial put the programme into 101 schools with 5,018 Year 6 pupils and found no additional progress in literacy or numeracy on national Key Stage 2 tests. One plausible reason: by then most teachers already knew the growth-mindset message, so the "treatment" added little.
The honest counterweight is that the effects are not zero, and they concentrate where theory predicts. Sisk found the intervention effect rose to d = 0.19 for academically at-risk students, and that low-SES and at-risk learners benefited more. The best-powered trial, the National Study of Learning Mindsets (Yeager et al., 2019, Nature), randomised a short online intervention across a nationally representative U.S. sample and found a real but small targeted effect: lower-achieving students gained about 0.10 grade points in core GPA and the share earning a D or F average fell by over 5 percentage points. Crucially, the effect held only in schools whose peer norms supported the message. So the defensible reading is narrow: a cheap, well-implemented intervention can nudge struggling students under the right conditions, not a universal driver of attainment.
Sarah Mercer and Stephen Ryan imported mindset into language teaching, arguing that beliefs about whether one is "a language person" shape engagement, persistence, and willingness to keep going after error. The intuitive pull is obvious, and mindset slots naturally beside the affective constructs in individual-differences research. But the import inherits the weakness above: the general-intelligence mindset evidence is thin, so a teaching claim resting on it stands on soft ground.
The more defensible line is domain-specific language mindsets. Nigel Lou and Kimberly Noels argue that beliefs about language aptitude specifically, not intelligence in general, are what predict L2 motivation, and they built the Language Mindset Inventory and a "meaning system" linking language mindsets to achievement goals, attributions, effort beliefs, fear of failure, and anxiety. Measuring the belief at the level of the behaviour it should predict is methodologically sounder than borrowing a generic intelligence measure, and the L2 findings have been more consistent. The distinction matters: "language mindset" is a more credible construct than the general growth mindset it descends from, and the two should not be cited interchangeably.
Treat growth mindset as a small, conditional effect, not a lever for attainment. Process praise and framing struggle as normal are low-cost and unlikely to harm, but they will not substitute for instruction, feedback, and practice, and selling them as a fix for low achievement oversells a near-zero average effect. If targeting anything, target the learners the evidence actually supports, struggling and disadvantaged students, and pair the message with classroom norms that back it, since the engagement gains evaporate where peers contradict it. For language classrooms specifically, beliefs framed around learning this language are better grounded than beliefs about general intelligence.