Comprehensible Input: What It Is and How to Use It

Comprehensible input is language you understand most of, even when you don't know every word: a story you can follow, a podcast you get the gist of, a conversation where the situation fills the gaps. The idea behind it is simple: you pick up a language mainly by understanding messages in it, rather than by memorizing rules about it.

This guide covers where the idea comes from, how much you actually need to understand (researchers have put a number on it), what the research since the 1980s adds, and how to use comprehensible input at every level, from your first word to native-speed podcasts.

Comprehensible input is spoken or written language in the language you're learning that you can understand from the words you know, the context and the situation, with only a small part of it new to you. The term comes from the linguist Stephen Krashen, who argued that this kind of input is what drives language acquisition.

Three things make input comprehensible:

• Known words. Most of the words are ones you already know.
• Context. Pictures, gestures, a familiar topic or a story you're following let you work out the rest.
• Support. Slower speech, a transcript, or a translation you can check when you're stuck.

Input that's too hard is noise: you hear sounds but take nothing in. Input that's too easy is comfortable but teaches you little. The useful zone is in between.

Is it the same as immersion? Not quite. Immersion means surrounding yourself with a language; it only becomes comprehensible input when you understand most of what you hear. Move abroad knowing a few words and you'll get plenty of input, but at first very little of it is comprehensible.

Krashen set out his theory in Principles and Practice in Second Language Acquisition (1982). Three of its ideas still shape how people talk about learning languages.

Krashen described ideal input as i+1: your current level (i) plus a little more. You understand the message thanks to what you already know and the context, and along the way you pick up the next bit of the language.

Krashen separated acquisition, the unconscious process that builds the language you can use without thinking, from learning, your conscious knowledge of rules. In his view only acquisition leads to fluent use; studied rules mainly let you check and correct yourself.

Krashen argued that anxiety, low motivation and low self-confidence work like a filter that keeps input from sinking in, so relaxed, motivated learners take in more of what they hear and read.

Krashen's strongest claim was that comprehensible input, taken in with a low affective filter, is what actually causes acquisition, and that speaking practice and studying rules don't drive it. That's the most debated part of the theory, and we'll come back to it.

Researchers have measured what "mostly understand" means. They measure lexical coverage: the share of words in a text that you know. In a classic study, Hu and Nation (2000) estimated that readers need to know about 98% of the words in a fiction text to understand it adequately without help. That's roughly one unknown word in every 50. At 90% coverage, one unknown word in ten, few readers in their study understood the text adequately.

Schmitt, Jiang and Grabe (2011) tested coverage between 90% and 100% and found that comprehension rises steadily as coverage rises. There's no point where understanding suddenly appears, but 98% is a sensible target for reading without help. For listening, the bar is lower: van Zeeland and Schmitt (2013) found that most listeners understood spoken texts adequately at 90% coverage, although results varied widely, while at 95%, one unknown word in 20, comprehension was good and more consistent. That makes 95% the safer target for listening.

Treat these numbers as a guide, not a law. A 2023 replication by Kremmel and colleagues found that even at 98% coverage many readers still fell short of adequate comprehension, and that the kind of text and the kind of questions asked made a difference. Still, the overall finding is clear: reading material where you know fewer than 90% of the words is usually a struggle, not input you can learn from comfortably.

How many words does that take? Nation (2006) estimated that, in English, 98% coverage needs about 8,000–9,000 word families for written texts and 6,000–7,000 for spoken language. You don't need that many to start: material for learners simply uses fewer, more common words.

Take a page or a minute of audio you're thinking of using, and count the words you don't know in a stretch of about 100:

• 0–2 unknown words: good for reading or listening on your own. You'll follow it and can start picking up the new words from context.
• 3–5 unknown words: fine for listening, or for reading with support such as a glossary, a transcript or a quick translation.
• 6–10 unknown words: hard going. Many people can still follow audio at this level, but it's tiring, and reading it on your own won't work. Add support or go a level down.
• More than 10: too hard for now. Come back to it later.

A second check: when you finish, can you sum up what happened in a sentence or two? If you can, the input was comprehensible.

Few researchers dispute that you need a great deal of input to learn a language. The debate is about whether input alone is enough. Here's what the evidence shows.

Merrill Swain studied Canadian students in French immersion programmes, who heard and read French at school for years. Their understanding of French came close to native speakers', but their speaking and writing lagged behind. In 1985 she argued that having to speak or write makes you pay attention to how you say something, not just what it means, which listening and reading alone don't force you to do.

Michael Long (1996) argued that conversation matters because you can ask, repeat and rephrase until the message gets through. That back-and-forth makes input comprehensible and draws your attention to the words and forms you're missing. A meta-analysis by Mackey and Goo (2007) found that interaction helps learners acquire both vocabulary and grammar.

Krashen held that studying rules adds little to real fluency. A large meta-analysis by John Norris and Lourdes Ortega (2000), covering dozens of studies, found that explicit instruction leads to substantial gains that last, although most of those studies measured them with tests rather than free conversation. Studying a grammar pattern won't make you fluent, but it can help you notice that pattern when it comes up in what you hear and read.

Critics such as Barry McLaughlin (1987) pointed out that Krashen's key terms, like the line between acquisition and learning, are hard to define and measure, which makes the strong form of the theory difficult to test.

The picture today: comprehensible input is necessary, and it's the foundation everything else rests on. On its own, though, it isn't enough. Speaking, feedback and a little focused study make it work better.

The aim is the same at every level: lots of input you mostly understand, pitched a little above where you are. What changes is the material and how much support you need.

When you don't know a single word yet, nothing is 95% comprehensible on its own, so you need support: pictures, slow and clear speech, very short sentences, the same everyday words again and again, and a translation you can check when you're lost. Keep sessions short and repeat the same material until it feels easy.

• Graded readers and learner podcasts made for your level.
• Listen to the same short recording three or four times: first for the gist, then for details, then along with the text.
• Try narrow input: stay with one topic, such as food, travel or work, for a while, so the same words keep coming back.

• Learner podcasts at natural speed, and YouTube channels made for learners.
• TV series with subtitles in the language you're learning, ideally ones you've already seen in your own language.
• Short native texts on topics you know well.

• Native podcasts, films, news and novels.
• Aim for a few unknown words per page, not dozens, and look up only the words you need to follow the meaning.
• Mix accents, speeds and registers so your ear gets used to more kinds of speech.

At any level, a little every day is easier to keep up than a long session once a week, and it keeps the same words coming back.

• Material that's too hard. If you're guessing at most sentences, you're tiring yourself out rather than learning. Run the self-test and step down a level.
• Only listening in the background. Audio that plays while you do something else sinks in less than audio you actually follow. Give at least part of your listening your full attention.
• Never repeating anything. Things you hear or read only once rarely stay with you. Each time you re-listen or re-read, you notice more.
• Looking up every word. Look up only the words you need to follow the meaning, and let context carry the rest.
• Waiting too long to speak. Understanding is the foundation, not the finish line.

You don't need to wait months. A practical order that fits the research: input first, so you have something to say; then short, simple conversations that show you what's missing; then more input that fills those gaps.

Many learners who rely on input alone run into the same problem: they understand far more than they can say. That isn't a sign the input failed. It's the signal to add speaking. Start with topics you've already heard and read a lot about, because that's where you have the words. And if native speakers still sound too fast, here's why you can't understand native speakers yet, and what helps.

You don't need special software to get comprehensible input, but some kinds of material make it much easier:

• Graded readers: books written for a level, with a controlled vocabulary.
• Learner podcasts: slower, clearer speech on everyday topics, often with transcripts.
• YouTube channels for learners: many teach with pictures and gestures from the very first episode. Search for "comprehensible input" plus your language.
• TV and films with subtitles in the language you're learning: once you're past the beginner stage.

Erla is built on the same idea, and this is how it works in the app. Each listening task starts with the text hidden: you listen first, then reveal the sentences and tap any of them for a translation and an explanation. Short stories with read-along audio and videos with subtitles add reading and watching, and when you're ready, you can talk it through with an AI teacher (part of PRO). There are four levels, starting at From Zero, and the free plan includes listening tasks, stories and videos every day.

1. Find one source at your level, using the self-test above.
2. Use it every day for a week, re-listening to and re-reading the same material.
3. Add a short conversation once you can follow most of it.

Erla has a separate app for each of its 22 languages. To pick yours, see learning Spanish, learning French or all languages.