Adults learning English in digitally supported courses who scored higher on emotional intelligence also tended to report stronger academic well-being and greater motivation to learn English, according to a survey study. The analysis also found that the association between well-being and motivation was stronger among learners who perceived more digital learning support. These were relationships in the data, not evidence that one factor caused another.
Researchers examined four measures: emotional intelligence, academic well-being, English learning motivation and perceived digital learning support. The question was whether these measures moved together, whether academic well-being formed an indirect link between emotional intelligence and motivation, and whether digital support changed the strength of the well-being-motivation association.
The pattern varied with digital support
Three direct associations stood out. Emotional intelligence had a standardized coefficient of 0.354 with academic well-being, 0.331 with English learning motivation, and academic well-being had a coefficient of 0.229 with motivation; all three probability values were below 0.001. In ordinary language, the measured relationships were positive: higher scores on one measure tended to accompany higher scores on the other. A standardized coefficient is a way of putting relationships on a common scale so their estimated strength can be compared.
The analysis also pointed to a possible route through academic well-being. The specific indirect association between emotional intelligence and English learning motivation was 0.081, with a standard error of 0.034, a probability value of 0.008 and an effect size of 0.033. The study treated this as a provisional association through well-being, not as proof of a confirmed mediation process. The indirect estimate also lacked conventional bootstrapped confidence intervals.
Perceived digital learning support was linked to how strongly academic well-being and motivation were associated. The interaction coefficient was 0.180, with a standard error of 0.047, a probability value below 0.001 and an effect size of 0.029. Descriptive simple-slope estimates rose from approximately 0.049 at lower support to 0.229 at the mean and approximately 0.409 at higher support. In practical terms, the well-being-motivation relationship was stronger among people who saw more support in their digital learning environment. That is an observed pattern, not evidence that adding support would produce a given change.
Even so, the model left most of the variation unexplained. It accounted for 12.5% of the differences in academic well-being and 24.3% of the differences in English learning motivation. The corresponding Stone-Geisser predictive-relevance coefficients were 0.126 and 0.244, a technical check reported by the researchers. The figures indicate that the variables in the model captured only part of what was associated with the outcomes in this sample.
A survey of adult EFL learners
The study focused on 435 adult learners in college and university English-as-a-foreign-language settings in Henan Province, China. Recruitment used convenience sampling, based on institutional accessibility, participant availability and willingness to participate. The measures used seven-point Likert response scales. Researchers analysed the cross-sectional survey with partial least squares structural equation modelling, a method for estimating several relationships in one model, using WarpPLS 8.0. Stable3 resampling with 100 resamples supplied the reported standard errors and probability values.
The reported measurement diagnostics met the study's criteria in this sample. Indicator loadings ranged from 0.778 to 0.880, while Cronbach's alpha ranged from 0.887 to 0.906. Composite reliability ranged from 0.917 to 0.930, average variance extracted from 0.689 to 0.727, and all heterotrait-monotrait ratios were below 0.85; full-collinearity VIF values were below 3.3. These checks support the reported measurement quality, but they do not establish that the scales work equivalently for college and university learners.
A pattern, not a cause
The main caution is the design. Because the measures were collected at one point in time, the study cannot establish which came first or whether emotional intelligence, well-being or digital support caused the differences in motivation. The indirect and interaction results also need testing with conventional bootstrapping using more resamples and confidence intervals.
College and university learners were not shown to be equivalent before their data were pooled. Convenience recruitment from Henan institutions also limits automatic generalization to other regions, educational settings, ages or cultures. The findings therefore describe associations in this adult EFL sample, not a confirmed way to increase motivation or a result that can automatically be extended to other groups.
Stronger evidence would require conventional bootstrapping with more resamples and confidence intervals, tests of measurement equivalence before comparing college and university learners, and replication beyond convenience samples in Henan.
Paper data and sources
Original title: Explaining English learning motivation in digital EFL environments through emotional intelligence, academic well-being, and perceived digital learning support
Authors: Guo D, Yao J, Al-Hashimy H
Journal/Repository: Frontiers in psychology
Status: Peer-reviewed
First online: 2026-10-10
DOI: Not available
Original paper · Full text