The language school in 90 secondsOn YouTube ▸
Research on speaking · Updated: October 2026
English speaking practice: why talking matters most
In short: you learn to speak by speaking. Reading and listening build knowledge. Conversation trains you to find the words in real time while someone waits for your answer, and nothing else trains that. Studies on output, interaction and retrieval practice support this view. Below are the research, its limits, two quizzes, a reaction test and a calculator to try yourself.
17 sources · 2 quizzes · 200 ms test · speaking-time calculator
English testsHow good is your English?
Why you understand English but can’t speak it
You read English emails without trouble and you follow most meetings. Then someone asks for your opinion and your mind goes blank. On learner forums people describe this in the same words again and again: they “freeze”, they know a word but “can’t access” it, their English feels “clunky”. This is common: in the EF English Proficiency Index 2025, speaking was the weakest English skill in over half of the countries measured.
Understanding and speaking use the same knowledge in two ways. When you listen, the words come to you and you recognise them. When you speak, you have to fetch each word from memory yourself, put the words in order and say them while the other person waits. You can be good at the first and still out of practice at the second.
James Simmonds has taught English to adults for 24 years. In his video on simmonds.online he says that “soaking in a warm bath of English dialogue with the subtitles on” won’t help you present the latest figures to your multinational team on Monday morning.
“Remember, the brain is not a sponge, it’s a muscle, and it needs exercising.”Research has known this gap for a long time. In Canada, children in French immersion schools learned their school subjects in French for years. They understood the language almost like native speakers, but they kept certain grammar mistakes when they spoke, and they spoke far less in class than children taught in their first language. The linguist Merrill Swain concluded that producing language is what shows you what you cannot say yet (Woodfield, 1997).
The 200 millisecond test: how fast a conversation runs
In a study of conversations in ten languages, the pause between the end of one turn and the start of the next averaged around 200 milliseconds. Planning even a single word takes at least about 600 milliseconds (Holler and colleagues, 2015; Stivers and colleagues, 2009). Fluent speakers therefore plan their answer while the other person is still talking. You only practise that in conversation. Try it:
A question in English will appear in a moment. Click as soon as you could start your answer.
A game with written questions, not a measure of your English level.
What the research says about speaking practice
Speaking practice makes you better at speaking, as you would expect. A 2018 study suggests it can also help you understand. At the University of Wisconsin-Madison, Elise Hopman and Maryellen MacDonald taught people an artificial language in the lab. One group practised by listening. The other practised by speaking and got feedback straight away. Afterwards the speaking group did better on vocabulary, grammar and comprehension, even on tasks almost identical to the listening group’s own practice. In the authors’ words, producing language is “an incredibly strong learning experience (when the production involves generating the language yourself and you are provided with feedback).”
Talking with other people is well researched too. Researchers call it interaction: you talk with someone, ask what they meant and get corrected. Alison Mackey and Jaemyung Goo reviewed 28 interaction studies in 2007. Learners who took part in interaction did much better than learners who did not, and the difference showed more clearly on later tests than straight afterwards (Loewen and Sato, 2018).
Memory research points the same way. In a 2008 study published in Science, students learned 40 Swahili words with their English translations. A week later they recalled about 80% of the pairs they had kept retrieving from memory. Pairs that were dropped from the recall tests fell to about a third (36% and 33%), even when the students kept studying them (Karpicke and Roediger, 2008). In a conversation nobody hands you the word, so you practise recall all the time. People also remember words they said aloud better than words they read silently (MacLeod and colleagues, 2010).
The studies at a glance
| Finding | Study | Who and how many | Limit |
|---|---|---|---|
| Speaking practice with feedback also improves comprehension | Hopman and MacDonald, 2018 | Lab study, artificial language | Lab setting, not a real foreign language |
| Interaction shows large effects, clearer later than at once | Mackey and Goo, 2007 | 28 studies | 64% lab studies, vocabulary and grammar only |
| Recall beats rereading: about 80% against 33 to 36% | Karpicke and Roediger, 2008 | 40 word pairs, test after one week | Word pairs, not free speech |
| Gaps between turns average around 200 ms | Stivers and colleagues, 2009 | Conversations in 10 languages | An average; languages differ somewhat |
| Repeated short talks on the same topic build lasting fluency | de Jong and Perfetti, 2011 | 24 learners | Small sample |
| Weekly practice works as well as daily | Kakitani and Kormos, 2024 | 116 Japanese learners of English | Four training sessions |
| Online and face to face: no significant difference | Ziegler, 2016 | 14 studies | Online mostly text chat |
Myth or research?
Six statements about learning to speak. Which ones hold up?
Quiz 1 · question 1 of 6
“Watch enough series in English and speaking will come by itself.”
All answers with sources
“Watch enough series in English and speaking will come by itself.”
Myth · The immersion children in Canada had years of input and still kept mistakes when they spoke (Swain, 1985). In the lab, speaking practice helped more than listening practice (Hopman and MacDonald, 2018).
“Practising speaking also makes you understand the language better.”
Research · In Hopman and MacDonald’s lab study (2018), the speaking group also did better on the comprehension tests.
“In a conversation you have about a second before you need to answer.”
Myth · The gap between two turns averages around 200 milliseconds (Stivers and colleagues, 2009).
“Words you pull out of memory stick better than words you only read again.”
Research · About 80% against 33 to 36% after one week (Karpicke and Roediger, 2008).
“Online conversations do much less for learning than talking in the same room.”
Myth · A review of 14 studies found no significant difference between interaction online and face to face; online mostly meant text chat (Ziegler, 2016).
“Learners can feel which study method works for them.”
Myth · In the Swahili study, students’ predictions of how well they would do were unrelated to how well they actually did (Karpicke and Roediger, 2008).
Where the research has limits
“Matters most” is our view. No study has ranked the four skills against each other, and the research has limits you should know about.
- You still need input. The linguist Stephen Krashen argued that we acquire a language by understanding messages in it (summary and critique: Nguyen and Doan, 2025). Nobody learns to speak without hearing a lot of the language. The open question is whether input alone is enough for speaking, and the immersion children suggest it is not.
- Employers don’t all put speaking first. In a Cambridge English survey of 5,373 employers in 38 countries (2016), reading was the most important English skill in 11 industries and speaking in nine.
- Talking without feedback can repeat mistakes. The strongest results above come from speaking with feedback, from a teacher or from a task that shows you what went wrong.
- Many studies are short lab experiments. In Mackey and Goo’s review, 64% of the studies took place in a lab, and the effects straight after training were larger there than in classrooms.
- Anxiety is linked to lower results. Across 216 results from nearly 20,000 learners, those who felt more anxious tended to get lower scores (Teimouri and colleagues, 2019). The link doesn’t show that speaking more cures nerves.
So keep reading, listening and watching series. For people who already understand a lot, though, those hours rarely turn into fluent speech by themselves. Speaking practice does that job, and it is the part that is easiest to skip when you learn alone.
How much do you actually speak in a lesson?
The formula: (lesson length × share without teacher talk) ÷ number of learners. Set your own situation.
Your speaking time per lesson
7 minutes
Your speaking time per year (40 lessons)
280 minutes
Simple maths, not a study: the time the teacher is not talking, divided by the number of learners. Real lessons also lose time to reading, listening and pauses.
Four examples, based on 40 lessons a year
| Setting | Per lesson | Per year |
|---|---|---|
| School class: 30 learners, 45 minutes, teacher talks two thirds | 0.5 min | 20 min |
| Course: 12 learners, 90 minutes, teacher talks half | 3.8 min | 150 min |
| Small group: 6 learners, 60 minutes, teacher talks 30% | 7 min | 280 min |
| One-to-one: 60 minutes, teacher talks 40% | 36 min | 1,440 min |
For HR and L&D: the same sum shows how many minutes of speaking practice a training programme really delivers per person per year. More on our page about English for companies.
How to improve your English speaking: more speaking time
- Speak with feedback. Practise where someone notices your mistakes and helps you fix them.
- Repeat topics. Nivja de Jong and Charles Perfetti had 24 learners record three talks in a row, of four, three and two minutes. Some kept the same topic, others changed topic each time. Both groups got more fluent during training, but only the learners who repeated their talk kept the gain on later tests (de Jong and Perfetti, 2011).
- Practise regularly. With 116 Japanese learners of English, four fluency sessions one week apart gave similar gains to four sessions one day apart, measured 7 and 28 days later (Kakitani and Kormos, 2024). A weekly rhythm was enough.
- Online works. A review of 14 studies found no significant difference between interaction online (mostly text chat) and face to face (Ziegler, 2016).
- Say new words aloud, and try to recall a word before you look it up.
Start for free
Check your level with the placement test, practise with the daily topics on our English conversation page, or read why many people overestimate their spoken English (in German).
What the studies found
Four questions about the studies on this page.
Quiz 2 · question 1 of 4
In a lab study, one group practised a new language by listening and another by speaking with feedback. Which group understood the language better afterwards?
All answers with sources
In a lab study, one group practised a new language by listening and another by speaking with feedback. Which group understood the language better afterwards?
The speaking group · The speaking group did better on the comprehension tests too (Hopman and MacDonald, 2018).
A review of 28 studies compared learners with and without interaction. When was the advantage clearest?
On later tests · The effects were large and clearer on later tests than straight afterwards (Mackey and Goo, 2007).
Learners recorded three short talks in a row. Who stayed more fluent later on?
Those who kept the same topic · Both groups got more fluent in training; only the same-topic group kept the gain (de Jong and Perfetti, 2011).
Four speaking sessions: one day apart or one week apart. Which helped more four weeks later?
About the same · Both schedules gave similar gains 7 and 28 days after training (Kakitani and Kormos, 2024).
Live conversation classes on simmonds.online
On simmonds.online we run conversation classes built around speaking time. A native-speaking teacher leads a small live group at your level, from A2 to C2. Each class has a topic and a few useful phrases. Everyone speaks, and the teacher keeps the conversation moving, so nobody sits and listens for an hour. Between classes there is an optional AI practice partner.
€30 a month, first month €10. Cancel every month.
What happens in your first class
- A short, free level check places you between A2 and C2.
- You only see sessions at your level and book in your browser, with nothing to install.
- A native-speaking teacher leads the small group. Everyone speaks, and there is no homework.
Frequently asked questions
Why can I understand English but not speak it?
When you understand, you recognise words; when you speak, you have to retrieve them yourself and build sentences in real time. Gaps between turns average around 200 milliseconds, so practised speakers plan their answer while the other person is still talking. That skill comes from speaking practice.
Can you learn to speak just by listening?
Listening builds the knowledge you speak from, and you need a lot of it. On its own it rarely seems to be enough: the immersion children in Canada had years of input and still kept mistakes in their speech. Use both.
How often should I practise speaking English?
Regular practice matters more than long sessions. In one study, speaking practice once a week gave similar gains to practice on consecutive days. A weekly class plus a few minutes a day of saying new words aloud is a realistic start.
Is it a problem if I make mistakes when I speak?
No. Mistakes show you and your teacher what to work on next. The strongest results in the studies came from speaking with feedback, so practise where someone can correct you.
Does my spoken English fade if I stop using it?
Research on language loss suggests that speaking and recall fade faster than reading and recognition, and that the loss is fastest at first and then levels off (Schmid and Mehotcheva, 2012). Knowledge you learned well can last for decades (Bahrick, 1984).
Do apps and AI chatbots help you learn to speak?
They can. A 2026 review of 16 studies found a moderate positive effect of dialogue systems on speaking (Hou and Min). We found no study that compared them directly with classes led by a teacher. In our classes the AI practice partner is optional extra practice, and a teacher always leads the class.
Sources
- EF Education First (2025). EF English Proficiency Index 2025, press release.
- Holler, J., Kendrick, K. H., Casillas, M. and Levinson, S. C. (2015). Editorial: Turn-taking in human communicative interaction. Frontiers in Psychology, 6, 1919.
- Stivers, T. and colleagues (2009). Universals and cultural variation in turn-taking in conversation. PNAS, 106(26).
- Woodfield, D. J. (1997). Output and beyond to dialogue: a review of Merrill Swain’s current approach to SLA. The Language Teacher, 21(9). Reports Swain (1985).
- Hopman, E. W. M. and MacDonald, M. C. (2018). Production practice during language learning improves comprehension. Psychological Science.
- Loewen, S. and Sato, M. (2018). Interaction and instructed second language acquisition. Language Teaching. Reports Mackey and Goo (2007).
- Karpicke, J. D. and Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319, 966-968.
- MacLeod, C. M., Gopie, N., Hourihan, K. L., Neary, K. R. and Ozubko, J. D. (2010). The production effect: delineation of a phenomenon. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(3), 671-685.
- Nguyen, Q. N. and Doan, D. T. H. (2025). Beyond comprehensible input: a neuro-ecological critique of Krashen’s hypothesis in language education. Frontiers in Psychology, 16.
- Cambridge English (2016). English at Work: global analysis of language skills in the workplace, executive summary.
- Teimouri, Y., Goetze, J. and Plonsky, L. (2019). Second language anxiety and achievement: a meta-analysis. Studies in Second Language Acquisition, 41(2).
- de Jong, N. and Perfetti, C. A. (2011). Fluency training in the ESL classroom: an experimental study of fluency development and proceduralization. Language Learning, 61(2).
- Kakitani, J. and Kormos, J. (2024). The effects of distributed practice on second language fluency development. Studies in Second Language Acquisition, 46(3), 770-794.
- Ziegler, N. (2016). Synchronous computer-mediated communication and interaction: a meta-analysis. Studies in Second Language Acquisition.
- Hou and Min (2026). Dialogue-based computer-assisted language learning systems for second language speaking development: a three-level meta-analysis. ReCALL, 38(1).
- Schmid, M. S. and Mehotcheva, T. H. (2012). Review of foreign language attrition research.
- Bahrick, H. P. (1984). Semantic memory content in permastore: fifty years of memory for Spanish learned in school. Journal of Experimental Psychology: General, 113(1).
Last checked: 12 October 2026. Text: James Simmonds and the Simmonds team.
Keep practising for free
Free lessons, daily conversation topics and a placement test, all without signing up.
Get in touch
Tell us briefly what you need English or German for — we'll suggest the best format for your goals and schedule.
We typically respond within one working day.
Hannover
Schaufelder Str. 11, 30167 Hannover
Berlin
Kurfürstendamm 30, 10719 Berlin