Shiraz University Teaching English as a Second Language Quarterly 2981-1546 45 4 2026 09 01 Language Choice in Pre-Speaking Discussions and EFL Learners' Oral Performance, Willingness to Communicate, and Self-Confidence 109 131 8854 10.22099/tesl.2026.56641.3522 EN Masoume Ahmadi Assistant Professor, Department of English Language, Farhangian University, Tehran, Iran Shahab Moradi English language and literature Dep., Faculty of Literature and Languages, Arak University, Iran Journal Article 2026 07 24 This explanatory sequential mixed-methods study examined how the language used in pre-speaking discussions (PSDs) relates to Iranian EFL learners' oral language performance (OLP), willingness to communicate (WTC), and linguistic self-confidence (LSC). Thirty-six male B2-level learners in a university speaking workshop took part. We implemented three PSD language modes across five intact discussion groups: L1-only (Turkish <em>n</em> = 7, Kurdish <em>n</em> = 6, Arabic <em>n</em> = 7), L2-only (English, <em>n</em> = 9), and mixed L1–L2 (<em>n</em> = 7). Over 10 sessions, each group completed a 10-minute PSD in its assigned mode before a 20-minute English-speaking task. We measured WTC, LSC, and OLP before and after the intervention. One-way ANCOVAs on post-test scores, controlling for pre-test scores, showed significant between-group effects for WTC, <em>F</em>(4, 30) = 4.56, <em>p</em> = .005, partial <em>η² </em>= .38, and LSC, <em>F</em>(4, 30) = 6.21,<em> p </em>= .001, partial <em>η² </em>= .45, but not for OLP, <em>F</em>(4, 30) = 1.84, p = .148. Exploratory within-group comparisons showed significant LSC gains in all four groups whose PSDs involved the first language (<em>g_av</em> = 0.34–0.55) and a significant WTC gain in the mixed-language group (<em>g_av</em> = 0.34); the English-only group changed significantly on neither affective outcome, and no significant OLP change was detected in any group. Interviews with 25 participants indicated that L1-involving PSDs were experienced as reducing anxiety, supporting brainstorming, and increasing readiness to speak. Because the groups were intact and participants were not randomly assigned, the findings are interpreted as associational rather than causal. https://tesl.shirazu.ac.ir/article_8854_821b2f9bc047e0af9f6aa9ed9b58c49f.pdf