Lingua
You can read the language. You did the lesson, you knew every word in it. Then a real person says one sentence and it arrives as a single long noise with a question mark at the end. The usual explanation is that they speak too fast. That is almost never what is happening.
Key points
- Measured in syllables per second, ordinary speech is not much faster than a learner's own. What changes is that words stop having edges: they link, shorten and disappear into each other.
- Not understanding has three separate causes, and each needs different practice: missing words, missing sound patterns, and missing speed.
- The fastest gain comes from listening twice to the same short piece, once blind and once with the text, rather than from listening to more new material.
They are not fast, they are joined
Fluent speech runs words together in ways no dictionary shows, a set of habits linguists group under connected speech. Sounds link across word boundaries, unstressed vowels collapse, and whole syllables vanish.
- English: what do you want to do is said as something like whaddaya wanna do. Nine words, four beats.
- Spanish: ¿qué es esto? becomes one run of sound, and final consonants before a vowel jump to the next word.
- Russian: здравствуйте is regularly cut to здрасьте, and сейчас to щас, in speech nobody considers sloppy.
- French: liaison attaches a silent final consonant to the next word, so les amis has a z that does not exist in either word alone.
- Japanese and Korean shift or drop vowels too, which is why a word you know from a textbook can be unrecognisable in a drama.
None of this is optional or careless. It is how the language is actually spoken, and it is the part textbooks teach least. The gap you feel is not speed. It is that you learned the written form of words and are listening for it.
Three different problems wearing one disguise
| What went wrong | How it feels | What fixes it |
|---|---|---|
| You do not know the words | You hear clear chunks and understand nothing | More reading and vocabulary |
| You know the words but cannot find their edges | You recognise the sentence the moment you see it written | Listening with a transcript, again and again |
| You know the words and can segment them, but not at that speed | You understand the first half and lose the rest | Volume: more hours of listening at a level you mostly follow |
The test is simple. Listen once, then read the transcript. If seeing the text makes everything obvious, the problem is segmentation, not vocabulary, and no amount of new words will fix it.
The two-pass loop
Take something short, between thirty seconds and two minutes, with a transcript. A podcast for learners, a scene from a film, one of the reading passages in your course.
- Listen blind. Once, maybe twice. Do not stop it. Note what you caught.
- Listen with the text in front of you. This is where the work happens: you see exactly which word was hiding inside the noise, and the brain files the sound and the spelling together.
- Listen blind again. It should now be almost transparent. That feeling of "how did I not hear that" is the pattern being learned.
- Say one sentence aloud, copying the rhythm. Not the pronunciation of each word, the shape of the whole line. This is shadowing in miniature and it wires listening and speaking together.
Ten minutes of this beats an hour of background listening. Comprehensible input works when it is just beyond what you already understand, and a transcript is what turns incomprehensible input into comprehensible input without waiting a year.
A ladder of material
Start where you follow most of it and climb one rung when that stops being effortful.
| Rung | Material | Why it belongs here |
|---|---|---|
| 1 | Audio made for learners, with transcripts | Clear articulation, controlled vocabulary |
| 2 | Interviews and podcasts by one or two speakers | Real speech, but planned and unhurried |
| 3 | Films and series with subtitles in the target language | Natural speed, plus visual context |
| 4 | Several people talking at once, comedy, phone calls | The hardest listening there is, in any language |
Most learners try to start at rung four because that is what they want to understand, then conclude they are bad at listening. See learning from films for how to use rung three without drowning.
What about subtitles in my own language?
They are a comfort, not a lesson. With them on, attention goes to reading and the audio becomes background noise. Use them to follow a plot you would otherwise abandon, then rewatch the scene with subtitles in the target language, or with none. The moment a learner switches subtitles to the language they are learning is usually the week their listening starts moving.
In LinguaPair
Reading passages in the course come with an AI voice-over, so every text is also a listening exercise with a transcript already aligned to it: read it, hear it, then hear it again with the text covered. Listen-and-type exercises make the segmentation problem explicit, because you cannot type what you cannot separate into words. Films let you bring subtitles from something you actually watched, and the karaoke mode plays a line at a time so you can loop the one that defeated you. And the live translator is the honest safety net for real conversation: when a sentence runs away from you, you see what it was, and the words it contained land in your vocabulary trainer for later.