Lingua
Speaking is the skill learners want most and practise least, and the reason is logistical rather than psychological. Reading needs a page, listening needs headphones, and speaking appears to need another human being with a free evening and the patience of a saint. It does not. Most of what speaking trains can be trained alone, and the part that cannot be is smaller than you think.
Key points
- Speaking trains three things: retrieving words under time pressure, moving your mouth accurately, and noticing your own errors. Two of the three work perfectly well alone.
- Recording yourself is what turns talking to yourself into practice, because it moves you from speaker to listener and the mistakes become audible.
- A real person is still needed for unpredictability and for the pressure of being understood. Aim for that once a week, not once a day.
What speaking practice actually trains
Retrieval. Finding the word you need in under a second. Recognising it on a flashcard is a different and much easier skill, which is why people who know four thousand words still stall in a shop.
Articulation. The muscles. Sounds that do not exist in your first language need physical repetition, and there is no silent way to acquire them.
Monitoring. Hearing yourself say the wrong ending and fixing it in flight. This is the last thing to develop and the one that makes speech sound controlled.
Alone, you can train all three. What you cannot train is conversation itself: the part where somebody replies with a sentence you did not expect.
Seven ways to practise without anybody
- Narrate what you are doing. Out loud, in the kitchen, in the car. Present tense, simple sentences, no self-correction. This is the cheapest retrieval practice there is and it doubles as a way of thinking in the language.
- Send yourself a voice note. Two minutes on a fixed question: what you did today, what you think about a film. Listen back a day later with a pen. You will hear three recurring errors, and those three are your syllabus for the week.
- Read aloud, then reread. Take a paragraph you understand. Read it once for words, once for rhythm. Reading aloud is the safest way to practise sounds because the content is already decided.
- Shadow a recording. Speak along with a speaker a beat behind them, copying melody and timing rather than individual sounds. See the shadowing technique for how to do it without turning it into gibberish.
- Answer questions out loud. Any list of interview questions will do. The trick is to answer the moment the question lands, with no rehearsal, since the delay is exactly what you are training away.
- Describe a photograph for sixty seconds. Speaking exams are built on this because it forces you into vocabulary you have been avoiding: colours, positions, guesses about people.
- Talk to a bot or an app that answers. Endless patience, no judgement, available at midnight. Useful for volume and for rehearsing a situation before you live it, as long as you remember that it will never misunderstand you in the way a stranger will.
| Method | Trains | Misses |
|---|---|---|
| Narrating your day | Retrieval, fluency | Accuracy, feedback |
| Voice notes reviewed later | Monitoring, self-correction | Real-time pressure |
| Reading aloud | Articulation, rhythm | Word retrieval |
| Shadowing | Prosody, listening | Composing your own sentences |
| Answering questions cold | Speed under pressure | Being misunderstood |
| Talking to a bot | Volume, situations | Genuine unpredictability |
How to review a recording without hating it
Everyone hates their recorded voice, so do it briefly and with a job. Listen once for content, once with a pen, and write down only three things: a word you reached for and did not find, a sound that came out wrong, and an ending you got wrong. Put the missing word into your vocabulary trainer, practise the sound five times, and let the grammar mistake go unless it repeats next week. Three items, not thirty. A list of every error is a list nobody acts on.
The part you cannot fake
Conversation is unpredictable, and unpredictability is the skill. A partner replies with a word you have never heard, changes the subject, misunderstands you and forces you to say it another way. That last one, repairing a misunderstanding, is what separates a person who has studied speaking from a person who can speak, and it only exists with another human.
So the pattern that works is not either-or. Daily solo practice keeps the machinery warm and builds the vocabulary of your own life. One real conversation a week, with a partner, a tandem exchange or a tutor, exposes what the solo practice has been quietly avoiding. Both are cheap. Neither works without the other, and speaking from the first day is easier when the daily half is already a habit.
In LinguaPair
Speak exercises give you a line to say aloud and check what came out against what you were asked for, which is the recording-and-reviewing loop compressed into ten seconds. The games can be played against Robo when your partner is asleep, so the retrieval practice does not stop when the other person is unavailable. The live translator has a solo mode: speak a sentence, see what it became, and keep the words. And the reason the product is built around two people is precisely the part solo practice cannot cover, as learning with your partner sets out.