Beyond Captions: Cross-Modal Transcript Support and Modality-Specific Comprehension in L2 Podcast Listening

Abstract

Written transcripts are common in digital second-language (L2) listening, but higher comprehension when text is available can reflect either genuine support for auditory processing or a broader cross-modal route to meaning that is not specific to listening itself. Distinguishing these accounts requires a direct statistical test of modality specificity rather than an informal comparison of separate significance tests. This study examined whether transcript-supported podcast listening was associated with differential improvement in listening and reading comprehension within a two-condition pretest-posttest design. The study included 60 participants divided equally between transcript-supported podcast listening (G1, n = 30) and audio-only listening (G2, n = 30). Parallel 24-item listening and reading forms were administered before and after an eight-session intervention protocol. Internal consistency ranged from KR-20 = .80 to .86 across the four forms. Primary analyses used baseline-adjusted ANCOVA with HC3 heteroscedasticity-robust standard errors, 10,000-resample stratified bootstrap confidence intervals, 10,000-permutation condition-label tests, Holm multiplicity control, and standardized effect sizes. Adjusted posttest performance favored transcript support for listening by 16.32 points, 95% CI [9.21, 23.44], p < .001, partial η² = .29, adjusted Hedges g = 1.22, and for reading by 15.24 points, 95% CI [8.49, 21.99], p < .001, partial η² = .27, adjusted Hedges g = 1.18. A participant-level mixed-effects model directly tested modality specificity; the Condition × Modality interaction was essentially zero (b = −0.005 standardized units, SE = 0.236, 95% CI [−0.468, 0.458], p = .982). Multivariate (Pillai = .432), gain-score, permutation, bootstrap, and repeated-measures sensitivity analyses converged on the same interpretation. The findings therefore distinguish a strong overall transcript-associated advantage from evidence of uniquely auditory remediation: transcript support was associated with substantial gains in both comprehension modalities, but the advantage was not detectably larger for listening than for reading. This pattern is most consistent with broad cross-modal facilitation rather than a modality-specific listening effect.