The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2778
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-12T17:22:29.750287Z
status: published
lens: g_space
sector: education
headline: The effects of generative AI on foreign language emotions: A meta-analysis
dek: Despite increasing attention to generative artificial intelligence (GenAI) in second language education, its effects on foreign language (FL) emotions remain unclear due to the lack of systematic evidence. In the current meta-analysis study, 43 effect sizes from 23 experimental and quasi-experimental studies (N = 3461) were synthesized to explore the effects of GenAI-assisted instruction on FL emotions. A three-level random-effects model revealed a large and statistically significant overall effect (g = 0.788, p…
gain_title: A meta-analysis of 23 studies with 3,461 learners found generative AI-assisted language instruction improved emotions more than traditional teaching (g = 0.788), with the largest gains in reduced anxiety and boredom.
problem_title: (none)
trace_subject: (none)
gain_reading: A meta-analysis of 23 studies with 3,461 learners found generative AI-assisted language instruction improved emotions more than traditional teaching (g = 0.788), with the largest gains in reduced anxiety and boredom.
gain_evidence: 43 effect sizes from 23 experimental and quasi-experimental studies (N = 3461) | a large and statistically significant overall effect (g = 0.788, p < .001) | larger improvements observed for negative emotions (anxiety and boredom) than for positive emotions (enjoyment)
problem_reading: (none)
problem_evidence: (none)
quick_read: A meta-analysis pooled 43 effect sizes from 23 experimental and quasi-experimental studies covering 3,461 learners. It found generative AI-assisted instruction produced a large overall improvement in foreign language emotions compared with traditional instruction (g = 0.788).

The gains were not even. Reductions in negative emotions such as anxiety and boredom were larger than increases in positive emotion such as enjoyment, which suggests AI tools may work best at easing the discomfort of learning a language.

Differences by technological, pedagogical and social features of the tools were mostly limited. An early signal favouring one or two sessions per week over more frequent use disappeared in sensitivity analyses, so it should be read cautiously.
limitation: The apparent advantage of 1-2 sessions per week became non-significant when studies with unspecified frequency information were excluded, so that finding should be interpreted cautiously.
tag: Evidence-backed gain
key_points: The analysis synthesised 43 effect sizes from 23 experimental and quasi-experimental studies with 3,461 participants in total. | A three-level random-effects model gave an overall effect of g = 0.788 (p < .001), described as large. | Emotion type significantly moderated results: anxiety and boredom improved more than enjoyment. | Moderator analyses based on the usefulness framework found limited evidence of stable differences across technological, pedagogical and social affordances. | Lower session frequency (1-2 per week) first outperformed higher frequency, but the difference became non-significant once studies with unspecified frequency were excluded.
rundown: The study, published in Acta Psychologica, set out to fill a gap in systematic evidence on how generative AI in second language education affects learners' feelings. It pooled studies that compared AI-assisted instruction with traditional teaching.

Using a three-level random-effects model, the authors found a large overall benefit (g = 0.788). They interpret this as AI reducing negative emotions and raising positive ones, with a stronger effect on anxiety and boredom than on enjoyment.

Moderator analyses grounded in the usefulness framework generally did not show stable differences across technological, pedagogical or social affordances. The session frequency finding lost significance in sensitivity tests, and the authors say it should be interpreted cautiously.
sources:
- peer_reviewed | Acta Psychologica | https://doi.org/10.1016/j.actpsy.2026.107336 | 2026-06-26
prev: 0000000000000000000000000000000000000000000000000000000000000000

SHA-256 hash

80aa7464ff8b0c00f2847dfc4226d373c337cee931452192631225cc37408fc3

Previous version’s hash

0000000000000000000000000000000000000000000000000000000000000000