Comprehensible Input: What Krashen Got Right, and What Input Alone Misses

Comprehensible input is language you can understand, with the help of context, that sits a little beyond your current level. Research backs the core of Krashen's idea: understanding lots of input builds comprehension and vocabulary. For speaking, though, input is necessary but not sufficient. What you can still say later depends on also producing language yourself.

Speech-language pathologists (SLPs) build language the same way: plenty of input first, then support to say it. These are learning strategies borrowed from speech therapy. Learning a language is not a disorder.

What is comprehensible input?

The term comes from Stephen Krashen's input hypothesis, set out in two books (Krashen, 1982, 1985). In the standard summary, we acquire a language by understanding messages, mostly without noticing the grammar. Consciously learning rules is a separate process, in his account.

The ideal input is written i+1: your current level, plus a little more. Context and pictures fill the gap. In Krashen's view, speaking emerges on its own once enough input has been understood.

Krashen later reviewed spelling and vocabulary research as further support. He argued that reading builds both, and that the evidence fit his hypothesis better than its rivals (Krashen, 1989). That review argues for his own position, so read it as a case, not a verdict.

What the evidence supports

Reading builds reading

Extensive reading means reading lots of easy, interesting text. A meta-analysis of 34 studies (N = 3,942) found d = 0.46 against control groups and d = 0.71 from pretest to posttest (Nakanishi, 2015).

A second meta-analysis of 49 studies (N = 5,919) found small to medium effects. Adults showed larger effects than children or teens (Jeon & Day, 2016). In both, the outcome was reading proficiency, not speaking.

Input teaches words, but slowly

Meeting a word again and again matters. Across 26 studies, the number of encounters had a medium relationship with incidental vocabulary learning, r = .34 (Uchihara, Webb & Yanagisawa, 2019).

Still, most words met in passing aren't learned. A meta-analysis of 24 studies found these pickup rates (Webb, Uchihara & Yanagisawa, 2023):

You need about 95% of the words

How much do you need to know before input is "comprehensible"? In one study, non-native listeners heard informal spoken stories with known-word coverage controlled. At 95%, their comprehension was fairly good and much less variable than at 90% (van Zeeland & Schmitt, 2013).

The authors estimate 95% coverage takes about 2,000–3,000 word families. That's far fewer than the 6,000–7,000 often cited for 98% coverage. Listening to simple stories is within reach early.

Structured input can carry over to speaking

Processing instruction is practice in interpreting input, like working out who did what from a sentence's grammar. In one study it improved both comprehension and production. Traditional instruction, explanation plus output practice, improved production only (VanPatten & Cadierno, 1993). These were sentence-level tasks, not conversation.

Where the critics pushed back

The acquisition–learning split. An early critique questioned the methods behind the research supporting Krashen's model. It proposed a simpler explanation based on how people process information (McLaughlin, 1978). Krashen replied the next year. Another early critique followed (Gregg, 1984).

Input doesn't show you what's wrong. Lydia White made four objections to the input hypothesis (White, 1987):

• Some grammar develops from inside the learner, whatever the input.
• Krashen overstates the value of simplified input.
• The idea of relevant input is too vague.
• Input explains how you add rules, not how you drop wrong ones. Hearing correct sentences may not show you what's ungrammatical.

White kept an input hypothesis. She wanted it tightened, not thrown out.

Immersion isn't enough for accurate speech. French immersion students understood a great deal. Yet one analysis described their spoken French as a "fossilized interlanguage," with transfer from English (Lyster, 1987). Lyster blamed the assumption that a second language is picked up just like a first. That was a position paper, not an experiment.

Speaking does its own work. In a think-aloud study, learners who noticed a problem while producing language changed what they said. That pushed them into more grammatical processing than comprehension needs (Swain & Lapkin, 1995).

The synthesis: input wins early, production wins later

A meta-analysis of 30 studies compared comprehension-based and production-based grammar instruction (Shintani, Li & Ellis, 2013). Both had large effects. The details are what matter:

The early comprehension advantage came mostly from processing instruction. A later meta-analysis of 33 studies agreed in part (Shintani, 2015). Processing instruction beat production-based instruction for understanding. For producing, the two were equal, and production-based instruction did better when both groups got the same explanation.

These are grammar-instruction studies, mostly in classrooms. A solo learner with an app is a different setting. But the direction is consistent: input first, then speak.

How to use input for speaking

1. Choose input at about 95% known words

A quick check: in 100 words, you should meet about five or fewer you don't know. Here's the same story moment at two levels:

If the harder line has more than one or two unknown words for you, start with the easier one. Move up when it feels easy.

2. Then speak about it

Close the text and retell the story out loud. Retelling turns understood input into output, and output is where the long-term gains for speaking showed up.

3. Ask for modified input, in conversation

Pre-simplified text isn't the only way to make input comprehensible. Long argued that adjustments in conversation, like checking and repairing, matter more than simplified speech (Long, 1983). In one study, learners understood directions best when they were repeated and rephrased during interaction. Simplifying them in advance made no significant difference (Pica, Young & Doughty, 1987).

A good partner slows down only when needed: shorter phrases, pauses at meaning breaks. That's the coach's Slows down move.

4. Negotiate when you don't understand

Say "¿Cómo?" or "Pardon?" Asking for clarification is what makes input fit you, not some average learner. More in negotiation of meaning.

For teachers, tutors and SLPs

Treat input and output as one loop, not two methods. Give a story at about 95% coverage, then ask for a retell. Make input comprehensible through interaction, by repeating and rephrasing when a learner signals trouble, rather than only by simplifying in advance.

FAQ

What does i+1 mean? It's Krashen's shorthand for input slightly beyond your current level (i). Context, pictures and background knowledge help you understand the "+1."

Is comprehensible input enough to learn to speak? No. Input reliably builds comprehension and vocabulary. On delayed tests of what learners could produce, production-based instruction did better.

How many words do I need to understand spoken stories? About 2,000–3,000 word families, according to one estimate. That gives you roughly 95% coverage of informal spoken narratives, where comprehension was fairly good.

Does watching TV count as comprehensible input? It can, but it teaches words slowly. In one meta-analysis, viewing had the lowest pickup rates: 7% on immediate tests and 5% on delayed tests.

Was Krashen wrong? Partly right. Understanding input is essential, and the evidence for reading and vocabulary is solid. The weaker part is the claim that speaking emerges from input alone.

Your next step: Try a story retell in Edgewise. You look at a picture story, then tell it out loud, and the coach talks with you about it. Your first retell sets your level. Edgewise scores story structure and sentence building, and every round ends with your next piece: one specific thing to add next time.

Join early access

Related: What 40 years of language acquisition research says about learning to speak · Negotiation of meaning

Sources

1. Gregg, K. R. (1984). Krashen's monitor and Occam's razor. Applied Linguistics, 5(2), 79–100. https://doi.org/10.1093/applin/5.2.79
2. Jeon, E.-Y., & Day, R. R. (2016). The effectiveness of extensive reading on reading proficiency: A meta-analysis. Reading in a Foreign Language, 28(2), 246–265. https://doi.org/10.64152/10125/66901
3. Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon. No DOI.
4. Krashen, S. D. (1985). The input hypothesis: Issues and implications. Longman. No DOI.
5. Krashen, S. (1989). We acquire vocabulary and spelling by reading: Additional evidence for the input hypothesis. The Modern Language Journal, 73(4), 440–464. https://doi.org/10.1111/j.1540-4781.1989.tb05325.x
6. Long, M. H. (1983). Native speaker/non-native speaker conversation and the negotiation of comprehensible input. Applied Linguistics, 4(2), 126–141. https://doi.org/10.1093/applin/4.2.126
7. Lyster, R. (1987). Speaking immersion. The Canadian Modern Language Review, 43(4), 697–713. https://doi.org/10.3138/cmlr.43.4.697
8. McLaughlin, B. (1978). The Monitor Model: Some methodological considerations. Language Learning, 28(2), 309–332. https://doi.org/10.1111/j.1467-1770.1978.tb00137.x
9. Nakanishi, T. (2015). A meta-analysis of extensive reading research. TESOL Quarterly, 49(1), 6–37. https://doi.org/10.1002/tesq.157
10. Pica, T., Young, R., & Doughty, C. (1987). The impact of interaction on comprehension. TESOL Quarterly, 21(4), 737–758. https://doi.org/10.2307/3586992
11. Shintani, N. (2015). The effectiveness of processing instruction and production-based instruction on L2 grammar acquisition: A meta-analysis. Applied Linguistics, 36(3), 306–325. https://doi.org/10.1093/applin/amu067
12. Shintani, N., Li, S., & Ellis, R. (2013). Comprehension-based versus production-based grammar instruction: A meta-analysis of comparative studies. Language Learning, 63(2), 296–329. https://doi.org/10.1111/lang.12001
13. Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive processes they generate: A step towards second language learning. Applied Linguistics, 16(3), 371–391. https://doi.org/10.1093/applin/16.3.371
14. Uchihara, T., Webb, S., & Yanagisawa, A. (2019). The effects of repetition on incidental vocabulary learning: A meta-analysis of correlational studies. Language Learning, 69(3), 559–599. https://doi.org/10.1111/lang.12343
15. van Zeeland, H., & Schmitt, N. (2013). Lexical coverage in L1 and L2 listening comprehension: The same or different from reading comprehension? Applied Linguistics, 34(4), 457–479. https://doi.org/10.1093/applin/ams074
16. VanPatten, B., & Cadierno, T. (1993). Explicit instruction and input processing. Studies in Second Language Acquisition, 15(2), 225–243. https://doi.org/10.1017/S0272263100011979
17. Webb, S., Uchihara, T., & Yanagisawa, A. (2023). How effective is second language incidental vocabulary learning? A meta-analysis. Language Teaching, 56(2), 161–180. https://doi.org/10.1017/S0261444822000507
18. White, L. (1987). Against comprehensible input: The input hypothesis and the development of second-language competence. Applied Linguistics, 8(2), 95–110. https://doi.org/10.1093/applin/8.2.95