Second Language Acquisition and Krashen’s theory

I bumped into a fascinating paper discussing three of Stephen Krashen’s hypotheses. If like me you read about Krashen’s theory on Wikipedia, you probably assumed it was outdated, apart from the importance of comprehensible input. These authors argue that subsequent research shows that he was in essence correct subject to qualifications of his hypotheses. So much for Wikipedia.

Click here for Krashen’s Theory discussed

Apologies if this is old hat here.


I asked for a summary of the text and this is what came out, is it correct?


Here are the pieces of information I extracted from your link:

  • The link is a research paper published on ResearchGate evaluating Stephen Krashen’s language acquisition theory forty years after it was proposed
  • Some of Krashen’s central hypotheses include: language acquisition occurs subconsciously through exposure to meaningful messages, acquisition requires meaningful interaction, acquisition is influenced by affective filters like anxiety and motivation
  • The paper aims to re-examine Krashen’s theory in light of more recent research and outline which aspects have been supported and which have been challenged over the past four decades
  • Areas the paper found support for include the input hypothesis (receiving comprehensible input slightly above one’s level) and the affective filter hypothesis (socio-emotional variables affect acquisition)
  • Areas which have received more challenges include the monitor hypothesis (learners rely primarily on acquired system not learned one for production) and the failure to distinguish acquisition vs learning
  • Overall the paper suggests Krashen’s theory was highly influential but incomplete, and more recent research has expanded on and refined some of his original hypotheses

Sort of. Here’s a summary of some key points.

His hypotheses were largely rejected at the time, but over the years as a result of research most linguists have come round to the idea that they are essentially correct subject to caveats.

Language is mostly learnt implicitly by exposure, not explicitly i.e. formal teaching.

The brain forms through massive exposure to input a complex representation of a language, not simple rules, and we often don’t know the rules consciously.

We need large amounts of comprehensible input with no more than 5% not understood.

Second language learning proceeds in a largely fixed order as per first language learning, and this is independent of any classroom exposure.

Language teaching in schools still revolves around explicit learning, due to the need to have a grading at the end. This is at odds with how we learn language, unlike subjects such as maths or geography.

The paper has changed the way I think about language acquisition. It is far more intuitive than I thought it was. The one part I have trouble with is explicit learning. That is because when I come across unknown English words or grammar, I look them up. This is of course as an adult with an already extensive knowledge of English acquired largely implicitly.

The above does not do justice to the paper, which I find fascinating, and I might have misrepresented some ideas. It does of course match what Steve Kaufmann and others have said.


Steve’s level in Russian. He learned it through input based methodology.

1 Like

Thanks for posting this paper. It was an interesting read.

This is the part that I have trouble with, too.

They seem to just define away the possibility of conversion of explicit knowledge to implicit knowledge. Both VanPatten and Krashen have acknowledged explicit learning and correction may help make input comprehensible, and that comprehensible input in turn can lead to implicit knowledge. This does not count as converting explicit to implicit, because they say that it is not the explicit definition of a word or grammar rule itself that becomes implicit knowledge, but something more abstract and complex.

This may be true, but it seems the message gets shortened to “explicit knowledge can’t be converted to implicit knowledge”, so don’t bother studying grammar or memorizing words. In reality that multi-step path from explicit knowledge to CI to implicit knowledge is a valid and well worn path to acquisition.

The research studies I have seen that purport to show that explicit learning is not effective in language acquisition, are really finding specifically that traditional classroom style explicit language instruction is not effective. I can’t see why this would remotely justify the conclusion that any and all explicit learning is not effective, and it appears that Krashen and VanPatten make room for that possibility.

Some other points:

  • LingQ itself is a explicit vocabulary teaching machine, optimized to give you a definition right when you need it to make input comprehensible.
  • There are studies that show spaced repetition systems enhance long-term retention and understanding of information. Many language learners have used these to make input comprehensible.
  • Most language learners who have learned a language to a high level, I suspect, have the experience of explicitly learning something and having it click right away, usable without accessing the “monitor” as described in the paper.

While I have read some studies on these topics, I certainly have not read them all, and I’d love to be corrected if someone has information to share that contradicts the above.


Yes, this is what I have always noticed too. They take one example of traditional classroom as the only way to use explicit learning and confront their theories. Which I find just a “marketing” way to push another “forced” idea.

What about if explicit learning, or even just grammar learning, it is used in a dynamic way and completely different way, matched to an input based methodology?

Why don’t use both inputs in an intelligent and smart way instead of pushing an idea against the other? Same for flashcards.

Of course, sometimes they say things like: “I’m not against grammar overall, just some bits here and there, but not as formal teachings”.
But they never say where, how, what it would be more efficient and so on because they don’t want to lose their “marketing” push on their overall idea.

Probably, the best would be to integrate everything in a way that it is not usually thought and that it should change dynamically depending on the type of languages (native and studied) and goals, and time available to achieve those goals.

Schools are a different problem and beast to tackle.


hiptothehop said: “They seem to just define away the conversion of explicit knowledge to implicit knowledge.”

My current thinking is that implicit and explicit knowledge are in essence the same, but the former can be far richer because it is sometimes too difficult to give and remember a complete explicit definition. As an example, if you learn that la mine is the pencil lead in French, I don’t think it matters how you learn it, it’s a well defined concept. I would argue that anyone who says you can’t learn such words explicitly is mistaken. That is how I learn words such as sparrow and chisel. If you learn that réclamer is to demand in French, you have a very shallow understanding of the word. How does it differ from demander, exiger et revendiquer? If you learn those words explicitly through listening, you learn the full meaning. Thus exiger is strong, demander is weak or politer, for example. The differences are of course more complex than that. That explains why a second language speaker may use words inappropriately e.g. an academic or rarely used formal word in place of the common one during casual conversation.

My take from this is that explicit learning can ‘prime’ your knowledge, so that when you hear a word, you know the basic meaning, but not the important subtleties, or not so subtleties. Then when you hear it many times in authentic speech, you start to understand the true meaning, This means that for non simple words, you cannot explicitly learn the correct and full meaning.

1 Like

Yes. Well said. I agree entirely with your post.

1 Like

Yes. It is interesting because I definitely feel like I’m on “team input”, as it were. It is just this point of essentially valuing explicit learning at or near zero, that doesn’t seem justified to me. It does appear in the OPs paper, they leave room for it when they say “The complex and abstract mental representation of language is mainly built up through implicit learning”

This is one area where I find discussions on this forum helpful, discussions about SRS, explicit study of collocations, etc.