What ELT Research Says About the Best Way to Improve Speaking Skills

Language teaching has a research base. It is genuinely empirical - researchers have been running studies on what makes speaking instruction effective for over 50 years. The findings are not always convenient, they don't always confirm what experienced teachers believe, and they don't always translate directly into practice. But they're real, replicable, and increasingly consistent.

The problem is that the research base is largely inaccessible to classroom teachers. It's published in academic journals, written in specialist language, and rarely summarised in a form that connects to what happens on Monday morning. This post is an attempt at that translation.

YapYapGo is a classroom speaking practice tool for ESL and EFL teachers built on research-supported principles. Here is what the research actually says.

Finding 1: Output drives acquisition, not just input

Krashen's Input Hypothesis (1982) dominated ELT for a decade with the claim that comprehensible input alone was sufficient for acquisition. Swain's Output Hypothesis (1985, 1993) challenged this directly: production forces learners to notice gaps in their grammar, access deep vocabulary knowledge, and test hypotheses about language that input alone doesn't require.

What this means: Speaking practice is not just practice for speaking. It is an acquisition driver. Students who speak more acquire language faster than students who only receive input. This research finding provides the strongest possible justification for maximising speaking time in every lesson.

Finding 2: Pair work is the most efficient format

Long and Porter (1985) documented that in teacher-fronted classrooms, each student gets approximately 30 seconds of speaking per lesson. In simultaneous pair work, this rises to 15+ minutes. The arithmetic alone would justify the shift. The research goes further: pair work produces negotiation of meaning (the back-and-forth that drives acquisition) at rates that teacher-fronted interaction cannot match.

What this means: The default format for speaking lessons should be simultaneous closed pairs. Whole-class speaking should be reserved for demonstrations, debrief, and specific pedagogical purposes - not the main activity.

Finding 3: Time pressure builds fluency

Nation's work on the 4/3/2 technique (1989) and De Jong and Perfetti's fluency training research (2011) consistently find that timed speaking activities - where students must produce language faster than their self-editing mechanism can interrupt - produce measurable gains in fluency that transfer to unpractised topics.

The mechanism is automatisation: language that is practised under time pressure shifts from controlled, effortful processing to automatic processing. This is what fluency actually is - not "speaking quickly" but "speaking without struggle."

What this means: Timed speaking activities should be a regular feature of speaking lessons. The timer is not a gimmick - it is the mechanism that produces the production conditions under which fluency develops.

Finding 4: Planning time improves output quality

Foster and Skehan (1996) demonstrated that even 60-90 seconds of pre-task planning time produces measurably more complex, fluent, and accurate output. The effect is particularly strong for lower-level students and for students with high language anxiety.

What this means: Always provide preparation time before speaking activities. Even one minute of silent thinking before a pair discussion improves the quality of what follows.

Finding 5: Varied interaction with multiple partners accelerates acquisition

Mackey (1999) and others found that interaction with multiple different partners produces more acquisition than extended interaction with one partner. The mechanism: different partners produce different vocabulary, different communication styles, and different negotiation-of-meaning episodes, each of which contributes to acquisition.

What this means: Partner rotation within lessons is not merely about variety - it has a specific pedagogical rationale. Students who speak with three different partners in a lesson develop more than students who speak with one partner for the same amount of time.

Tool tip: YapYapGo implements these research findings structurally: simultaneous pair work (Finding 2), timed activities with visible countdown (Finding 3), preparation time built into IELTS mode (Finding 4), and automatic partner rotation between rounds (Finding 5). A conversation topic generator provides varied content across sessions.

Finding 6: Corrective feedback timing matters

Lyster and Ranta (1997) and subsequent research found that immediate error correction during fluency activities disrupts the flow of communication and reduces output. Delayed, patterned feedback - addressing common errors after the activity, targeting patterns rather than individual mistakes - produces better outcomes for both accuracy and continued willingness to speak.

What this means: Don't correct during pair work. Circulate, observe, note patterns, and address them in the debrief. This is one of the most commonly violated research findings in ESL teaching.

Finding 7: The affective filter is real

Krashen's affective filter concept (1982) - that anxiety reduces acquisition by limiting the processing capacity available for language intake - has been extensively validated. Foreign Language Anxiety specifically predicts reduced speaking output and slower speaking development (Horwitz et al., 1986; Botes et al., 2020).

What this means: Classroom conditions that reduce anxiety are pedagogically important, not merely kind. Low-stakes pair work, predictable classroom procedures, error-tolerant feedback practices, and a classroom culture of psychological safety all contribute to the conditions under which acquisition actually occurs.

Finding 8: Task-based approaches outperform grammar-focused approaches

Spada and Tomita's 2010 meta-analysis found that explicit grammar instruction combined with communicative practice outperforms either alone - but the ratio matters. The research supports significantly more time on communicative practice than most syllabuses currently allocate.

What this means: Grammar instruction has a role, but it's smaller than most curricula give it. For speaking development specifically, task-based communicative practice should be the core of the lesson. See our full post on the case for more speaking and less grammar.

What the research doesn't say

The research base for speaking instruction has limitations worth acknowledging. Most studies are conducted with adult learners in Western academic contexts. Many findings don't translate directly to young learner classrooms or to contexts with very large classes. The research on what types of corrective feedback work best remains genuinely contested. And implementation always requires professional judgment that no research can fully replace.

But the core findings are robust enough to anchor practice: output matters, pair work works, timing matters, planning helps, varied interaction accelerates, and anxiety reduces. A classroom designed around these findings will produce more speaking development than one that isn't.

A classroom countdown timer and random student picker are the most practical tools for implementing Findings 3 and 5 consistently across every lesson.

---

Sources:

• Krashen, S. (1982). Principles and Practice in Second Language Acquisition. - Input hypothesis and the affective filter.
• Swain, M. (1985). Communicative Competence. Input in Second Language Acquisition. - The Output Hypothesis.
• Long, M. & Porter, P. (1985). Group Work, Interlanguage Talk, and Second Language Acquisition. TESOL Quarterly. - The 30-second-per-student finding and the pair work solution.
• Nation, I.S.P. (1989). Improving Speaking Fluency. System. - The 4/3/2 technique and time pressure in fluency development.
• Foster, P. & Skehan, P. (1996). The Influence of Planning and Task Type. Studies in Second Language Acquisition. - Planning time effects on output.
• Mackey, A. (1999). Input, Interaction, and Second Language Development. Studies in Second Language Acquisition. - Multiple partners and interaction variety.
• Lyster, R. & Ranta, L. (1997). Corrective Feedback and Learner Uptake. Studies in Second Language Acquisition. - Feedback timing and its effects on uptake.
• Spada, N. & Tomita, Y. (2010). Interactions Between Type of Instruction and Type of Language Feature. Language Learning. - The ratio of grammar to communicative practice.