Name: Is everyone WRONG about Stephen Krashen's "comprehensible input" theory?
Uploaded: 2026-01-20T09:34:08.000Z
Duration: 14 min 4 s
Description: Thousands of YouTube videos with English-Chinese subtitles! Now you can learn to understand native speakers, expand your vocabulary, and improve your pronunciation...

Basic passive voice

Video after video praises Stephen Krashen and his theory of comprehensible input.

Throw away your textbooks, forget about grammar.

But is everybody getting his theory wrong?

I'll explain why it's a little more complicated than people make it out to be, how solid it actually is, and

I'm Dr. Taylor Jones, and this is Language Jones.

I've seen people say that Chomsky is wrong about language and then two seconds later swear their intellectual allegiance to Krashen without realizing that not only does he know and get along with Noam Chomsky, not only does he pronounce his name correctly, but

his theories are within a Chomsky and framework.

What you're likely to see on YouTube is a claim that all you need to learn a language is a sufficiently large quantity of engaging comprehensible input.

Comprehensible means that you can figure it out, ideally that there's only one new word in a sentence or one new structure that you can get from context, and

By this interpretation, the only thing you need to learn a language is a lot of the language starting simple and getting more and more complex, preferably with something like pictures or video, and

But that's not exactly what Krashen meant.

And in his original theory, he actually specifically, specifically addresses the fact that that sometimes doesn't work.

So let's get into what his theory actually is and then decide how justified it is and whether we can make it work for us.

His writing covers not just the comprehensible input hypothesis, but also four other interrelated hypotheses.

If one proves wrong, it might destabilize the whole academic endeavor.

These need to be operationalized and tested.

He's working within a Chomsky and generative framework, and those assumptions are actually the first hypothesis.

They are 1 learning versus acquiring, 2, the linguistic monitor, 3, natural order, 4, comprehensible input, and

Stick around to the end and I'll add two components that nobody discusses about Krashen's theory, even Krashen himself, but

First is the hypothesis, or really assumption that there's a difference between learning and acquiring a language.

Learning is explicit, taught, and consciously accessible.

Declarative knowledge acquisition, on the other hand, is what children do.

They're not explicitly taught, say, how to make sure anaphors like himself occur in the correct grammatical framework with the correct dependencies.

Yet somehow they start to produce language we would consider grammatical, despite in many instances having never heard the exact sentences or sentence structures they use from adults.

That is, there appears to be a developmental stage, a mental module, as Chomsky might put it, where children simply start using finite grammatical structures to produce infinite possible language combinations in orders that we would consider grammatical.

Where Krashen argues with Chomsky is the critical period hypothesis.

There's a fair amount of evidence that if people start to learn a language too late, they will never attain native like fluency in grammar and phonology.

But it's really hard and more than a little immoral to test this.

He has stated over and over that he doesn't think we get worse at acquiring languages as we age even past 25, when the prefrontal cortex is done cooking and we start losing neuroplasticity.

On average, I'd like to acquire 100,000 subscribers.

So now is a great time to like and subscribe.

His second hypothesis is that of the linguistic monitor.

This is similar to Bill Labov's sociolinguistic monitor.

Shout out to Bill, one of my advisors in grad school.

The idea of the sociolinguistic monitor, put simply, is that we speak in a way that we think others will perceive as more correct when we're not stressed or agitated and have a lot of cognitive resources to put towards that.

The neuroscience wasn't as developed when he came up with this.

This is why sociolinguists will do reading tasks and polite conversation and then start getting into questions like tell me about a time you almost died.

Our job is in part to get people so worked up that they're natural, natural vernacular comes out because they're no longer monitoring how they speak.

He argues that we have a store of our target language that we have acquired somehow, and

that what we have learned merely functions as a monitor.

Basically, it has veto power to correct what we intuitively generate.

We'll come back to what happens when we're cognitively taxed shortly.

So I might know in a declarative sense that in French you use the future tense in both parts of a future conditional I will speak French when I will be in France instead of I will speak French when I am in France.

But I might only be able to generate such sentences when I'm monitoring my speech and I'm likely to produce a grammatical error until I've seen sufficient examples of sentences that use that structure correctly in French so that it just feels better to me his third hypothesis is about natural order of language learning.

There are some really fascinating patterns when children first learn language and they're cross linguistically.

Basically the same things like saying I know like it first and only later developing don't in imperatives first, then in sentences like I don't like that and only much later being able to consistently get negation correct in sentences with multiple clauses or conditionals, or so on.

This is a topic that deserves its own video, and I'll be making one soon.

Leave me a comment if you want to see that For a long time it's been known in the field of second language acquisition that in general, non native learners interlanguage their use of their target language while learning it follows the same evidently hardwired order of acquisition.

This goes for everything from negation to matching the appropriate gender on possessives to how questions are formed.

Krashen hypothesizes that this order of acquisition that non native learners tend to follow and which mirrors that of children learning their first language is natural and hardwired.

So why bother starting with declarative explanations of grammatical patterns that we know from experimental evidence?

Learners will likely not learn up front, especially if these patterns are basically impossible to acquire in that order.

It's not settled fact, but the experimental evidence is suggestive.

His fourth fourth hypothesis is what people refer to as comprehensible input.

That is, if you accept the hypotheses and these are not proven, if you accept that we do not learn, we acquire, and

that there is no strong correlation between age and ability to acquire, that is biologically innate.

And if you accept that what we produce is what we have acquired and what we have learned is merely a cognitively taxing conscious last check on the acquired language that wells up from our depths, and

if you accept that there is a natural order to acquisition such that it doesn't matter what order you were exposed to the language in, then he hypothesizes that the best approach is to get as large a volume as possible of, quote, comprehensible input.

Input just means your target language comprehensible in this case means that you learn at most one new thing at a time.

This can be new vocabulary in a sentence frame you don't know, or a new grammatical construction that you can figure out correctly because you know context and it's using all words that you already know.

It's not always clear whether he endorses translation and lookup and so on, or if he's implicitly endorsing an immersion only methodology.

The key is also, once you have your I +1, you need sufficient input, sufficient repetitions of that to acquire it.

So it's not just chasing the dopamine hit of a new word all the time.

I promised I'd get back to cognitive load.

The idea is that if you're experiencing negative affect, that is Negative emotions, all of this is harder.

This is required in his system because there's a lot of evidence that the first four on their own don't always work.

So his theory is that people get stressed or upset or in their heads or in their feels about language learning, and

This is also, by the way, the principle that Michel Thomas built his reputation on.

It's probably correct in some sense and that we know general learning is impeded by cognitive strain.

But notice this is necessary because even if we assume that each of the first four hypotheses are correct and do everything that that assumption would entail, the people fail to become proficient in their target language.

So the affect filter is a way of explaining away what happens when a large volume of comprehensible input fails to produce a language learner who is competent in their target language.

There are a lot of videos on YouTube that tell you not to study grammar and that quote, comprehensible input is all you need.

But they're leaving out a lot of the story.

A ton of content, even if it is engaging and even if it is at a level where you can figure it out from context, does not always produce a learner who can understand and meaningfully engage with a wide range of uses of their target language, or one who can produce comprehensible and correct language.

There's also an entire world of second language acquisition research out there, and

Subtitles ListPlay Video

Is everyone WRONG about Stephen Krashen's "comprehensible input" theory?

cognitive

approach

massive

context