<![CDATA[This study explores the role of Informal Digital Learning of English (IDLE) in developing speaking expertise among English as a Foreign Language (EFL) learners within the present-day digital learning environment. Grounded in Krashen’s Input Hypothesis, Swain’s Output Hypothesis, and Long’s Interaction Hypothesis, the study presents the Digital Language Acquisition Ecosystem (DLAE) outline, which conceptualizes language acquisition as a dynamic interaction among learners, digital technologies, global communities, artificial intelligence, learner autonomy, and self-regulated learning. Employing a quantitative explanatory survey design, data were collected from 624 university students in Indonesia through stratified random sampling. Structural relationships among variables were examined using Partial Least Squares Structural Equation Modeling (PLS-SEM). The findings reveal that IDLE pointedly envisages speaking proficiency, while learner autonomy emerges as the strongest forerunner influencing both self-regulated learning and engagement in digital learning activities. Digital Engagement Quality was found to significantly facilitate the connection between IDLE and speaking proficiency, demonstrating that the effectiveness of digital learning hinge on more on the quality of engagement than on the duration of exposure. The proposed DLAE model explains 69.1% of the variance in speaking proficiency, demonstrating substantial predictive power. Furthermore, interactive digital activities, particularly YouTube and Discord, contribute more strongly to speaking development than passive forms of content consumption. The study concludes that speaking skill in the digital era is formed by a multifaceted ecosystem of formal and informal learning practices extending beyond the old-style classroom. The findings provide momentous theoretical contributions to second language acquisition research and real-world implications for language pedagogy, curriculum development, and educational policy in digitally mediated learning contexts.]]>
