I’m so thankful that I was born in a time before teaching reading using phonetics was replaced by insanity - and there’s a direct link to the Krashen theory of language acquisition. From Ken Goodman to Marie Clay and ultimately to Stephen Krashen.
Evgeny Eroshev does a great job of explaining how the US stopped teaching kids how to read (link below) and the link to Stephen Krashen. it’s in Russian, and here’s a summary of it in English:
The central takeaway is that relying solely on context and “guessing” leads to failure, and true mastery requires a balanced approach of conscious study, immersion, and active practice.
1. The U.S. “Whole Language” Failure
In the 1960s and 70s, U.S. schools shifted from phonics (learning letters and sounds) to the “whole language” method.
The Theory: Proponents like Kenneth Goodman argued children should learn to read “naturally,” like walking or talking, by guessing words from pictures and context rather than decoding syllables.
The Result: By the 1990s, a massive literacy crisis emerged; only 28% of fourth graders could read proficiently.
The Science: Brain scans and eye-tracking later proved that skilled readers do not “guess” from context; they fixate on every letter and mentally “sound out” words, even when reading silently.
2. Parallels in Language Learning
The source compares this educational failure to Stephen Krashen’s theory of language acquisition.
Input vs. Learning: Krashen argued that “acquisition” (subconscious immersion) is superior to “learning” (conscious grammar study).
The Flaws:
Inaccuracy: Students in immersion-only programs often understand speech perfectly but continue making gross grammatical errors for years.
The “Output” Necessity: According to the Output Hypothesis, learners only notice gaps in their knowledge when they are forced to actively speak or write.
Adult Advantage: Adults have superior analytical skills; ignoring these abilities in favor of “child-like” immersion is a disadvantage.
3. The “Ideal Circle” Strategy
To avoid these pitfalls, the sources suggest a three-step cycle for effective learning:
Intensive Learning (The “Hooks”): Start with conscious study of grammar and vocabulary using translations and videos. This creates “hooks” that help the brain catch information during immersion.
Extensive Immersion (The “Open Sea”): Read or listen to materials where you understand 93–95% of the words. This allows you to naturally reinforce what you studied during the intensive phase.
Activation (The “Output”): Practice speaking and writing early using templates and structures. This process reveals your weaknesses, allowing you to return to “Intensive Learning” to fill those specific gaps.
When I realized Children can take 7-8 years (even with 10-14hrs per day exposure to their native english language) before being able to read Harry Potter I stay away from any method that dogmatically promote learning like children as the only thing one should do.
My first main goal in language was to be able to consume native content (books, games, movies) and I don’t want to wait 7-8 years to do that. So when I notice some adults can start reading Harry potter within 1-2 years (learning the language at 1-4hrs per day) I chose to learn like an adult.
It has been known for decades that the Krashen theory is nonsense, and yet YouTube is full of people promoting it. The cynic might suggest that it provide a good sales technique for online tools that only provide input.
I wasted 18 months using comprehensible input with German, because I was told by ‘experts’ that it was the best way to learn. As far as I can see, Krashen plucked his theory out of thin air, and then promoted it as ‘based on science’. Every single hypothesis in his theory is incorrect, except for the obvious one, that learning is less effective when we don’t pay attention. His claim that language can only be acquired and not learnt has been disproven. His claim that we learn by hearing messages has been disproven.
As you indicate, adults have much better semantic memory and reasoning skills than children. They can use those advantages to learn faster than children. And through practice the words and structures that they have learnt are moved into procedural memory, and become second nature. Most sentences can be understood well if not perfectly without knowing all of the grammar. As you say, we have to explicitly study the grammar if we are to master it.
Hsingh makes a good point. Most people do not realise how much exposure children get to a language, and a child will have a much more limited vocabulary than an adult.
Yes, I am still working on German, thank you. I now use traditional methods with LingQ.
I initially wasted six months with Duolingo and then Babbel. They are junk. The creators use streaks and other nonsense to addict us to their products to the detriment of learning.
Duolingo was a great start for me with Russian. After that: 8 weeks intensive program at a US University (University of VA) where Russian was taught “old school” (grammar intensive). Since that time it’s been self study with the help of my Russian wife. Conversational Russian is finally taking off. Finally.
I have a degree in German. Came up the traditional way: 2 years in High School, three years in college and then one more year living in and going to college in Germany. Surprise, surprise! The tried and true methods worked. I was conversationally fluent the day I left Germany. I can still read and understand it reasonably well almost 50 years later. Surprisingly, I have not forgotten.
Sometimes I think the Russian study has refreshed those brain cells – pulled them out of moth balls. Меня удевило. что я помню всё ещё. Es ist überraschend, wieviel ich mich daran erinnere.
I wasted 18 months using comprehensible input with German, because I was told by ‘experts’ that it was the best way to learn. As far as I can see, Krashen plucked his theory out of thin air, and then promoted it as ‘based on science’. Every single hypothesis in his theory is incorrect, except for the obvious one, that learning is less effective when we don’t pay attention.
LeifGoodwin,
What is your tried-and-tested method? Can you provide more information about it?
How much grammar study do you do every day? How do you study it? And how difficult is your input level, if there is any? Do you listen to your target language a lot? Do you combine reading-Listening-Speaking activities together? What is the purpose of using Anki in your learning strategy? How much progress did you make with your learning method, let’s say, in German?
As a side question, for your French, which I believe is at a higher level, did you sit for any official exam that had tested your skills in all 4 language areas?
I don’t have a tried and tested method. I have techniques that are working better for me than the methods I used two years ago, in particular CI, Duolingo, Busuu and Babbel. Most online apps such as Duolingo are no more than glorified Anki and inadequate.
With German I change my approach as and when I find that a given technique is not effective, or I find a better one.
If you wish to learn about learning strategies, you would be better off reading undergraduate textbooks on linguistics and listening to interviews with linguists, rather than asking about the subjective experiences of one person who is bumbling along in his own imperfect way.
French is mainly input, in order to deepen my knowledge. I find input works best at a higher level, where unconscious learning is most effective. I know most words in podcasts, so the aim is to become more familiar with the way the French express ideas, and improve my input processing. I also spend time talking to myself, in order to practice pronunciation and grammar, and discover concepts that I cannot express, and need to research.
German is mainly working through short ‘stories’ such as the news in simple German in a particular day. Work through line by line, looking up words, figuring out the grammar, playing with the language e.g. changing the verb subject and tense, putting useful phrases into Anki. Sometimes I will write out sentences, trying to express related ideas, and checking with online tools. Then I listen to the story, without the transcript, stopping when I do not recall words. I also revise previous stories, listening to the audio, and looking up words or phrases when I do not recall the meaning. I think this approach is used by many self learners, it is similar to methods described by Alex Rawlins in his book. I learnt to ignore LingQ statistics, which encouraged poor learning. Slower is better, for me anyway. I need to do more output, and listen to more input.
I don’t use AI much, I probably should. I should also use a native French speaker for conversations. I have in the past spoken quite a bit with native French speakers, I’m not sure I gained much at my then B1 level.
French: I don’t do much grammar study, but I search online when something does not make sense.
German: I use Anki to drill structures e.g. cases and when I do not understand something in a video, I search online, and read the explanations. I don’t learn grammar tables, but I do drill verb conjugations, I find that useful. I don’t generally study grammar per se, rather when I do not understand a phrase, I deconstruct it, and work out how the various bits work together. I might also do some in depth research when I hit a new concept e.g. when I notice that some nouns gain an n at the end.
French: I listen to native level input.
German: roughly B1.
French: two hours a day, mostly listening, with one hour while walking or driving.
German: one hour study a day including maybe 30% listening. I also sometimes listen to podcasts without worrying too much about comprehension.
Yes.
In general I put phrases and not words into Anki. Research shows that we learn by making connections, and learning phrases rather than words creates far more connections, and shows us how to use the words in context.
German: I use Anki to store useful words and phrases in my memory. I also use it to learn grammar. It allows me to focus on specific aspects such as the case system. In effect Anki helps me drill the grammar through repetition. I find it helps me notice things that I would otherwise miss.
French: I use Anki to learn phrases that I would not come across often and to polish my knowledge. For example I can recall phrases such as l’acte de naissance and en garde à vue and get not only the correct words, but the correct prepositions. Anki primes my brain, and works alongside input, so that when I hear a word in the wild, I understand it and then a deep knowledge can develop. Sometimes only input can develop a deep knowledge.
French: I started out three years ago getting the basic gist of podcasts. Today I have no trouble with podcasts and videos on history, science, technology etc. I can repeat whole sentences after hearing them in podcasts. I struggle with films. Some I can understand, some fly over my head. A year ago I could not understand films at all, now I’m starting to understand some of them. I hope to have fluent comprehension of most non dialect films in a couple of years.
German: I cannot quantify my experience. I was making no progress with CI, even basic phrases were not sticking after 18 months using CI. I had previously (25 years ago) studied French and Welsh using traditional methods and made far more progress. It was very frustrating. Now with more traditional methods I find that words are starting to stick, and the case system is starting to make some sense. I struggle with prepositions, and often get them wrong, but I hope to develop an intuitive understanding over time c.f. French.
No. I would like to sit an assessment in both languages, but passing an exam, which is no more than a snapshot in time of competence, is on my to do list. However, my aims do not align with an exam, which tests general communicative competence and allows for a mediocre accent and grammar errors. I am not interested in C2 level language, maybe not even C1, but I do want a good accent, and a high competence in day to day language, both speaking and listening, with high accuracy. Some people aim to learn as many languages as possible to a B2 level. I aim to reach near native ability in French and German.
On an additional point, I found it hard to find suitable input in German. There are a lot of videos created by amateurs, many using AI voices. Some of these have grammatical errors, the vocabulary tends to be limited and they don’t cover enough subjects/situations. I currently use videos from professional media organisations such as Deutsche Welle, which are higher quality, and have a more diverse range of topics, and hence a wider range of vocabulary.
One very good technqiue that I use is to listen to a podcast or a video while reading the transcript. This trains the brain to recognise the words in speech, and accelerates the comprehension of the spoken language. Another good technique is to listen ‘like a baby’, just listening, allowing the language to sink in. Thinking often gets in the way of comprehension, and it encourages translation.
Lastly, speaking out loud is a good way to practice pronunciation. Just as an ice skater perfects a move by constant practice, so speaking helps build muscle memory. For example, I initially struggled with Germans love of consonants. With German I used online videos to learn the ich-laut sound, the other ch and the r were easy. One must listen a lot in order to develop a good accent, but practice is also needed.
Hopefully the above is of interest. I am of the view that there are many ways to learn a language, my approach is unlikely to be the best, and might not suit some people.
Let me know if you have thoughts, criticisms, suggestions. If you wish to explain your approach, that would be interesting.
I highly recommend using NotebookLM to create podcasts from the content you import. The AI voices are high-quality, the grammar is sound, and they use natural-ideomatic expressions as well. When imported into Lingq, the transcript is remarkably accurate. You can also give instructions such as “focus on x”, I have tried to tell it to use a certain level of the language, but it hasn´t worked extremely well.
Duolingo is definitely junk. A “game for the toilet” as one Youtuber once said.
But Babbel is not bad. It really gives you high quality content and explains grammar perfectly.
Yet I prefer Anki (Language on Fire for Russian) + LingQ + reading easy books.
And Duolingo is a waste of time. If someone mentions he was not able to learn a language using solely Duolingo and then gave up … well, that person definitely did not his homework in researching how language learning works.
I used Duolingo for almost a year at first. It helped me establish the habit of learning, helped me learn a lot of Russian words, and gave me a “taste” of Russian syntax. When I started learning Russian grammar, I had lots of examples in my head to consult — pretty much all of them from Duolingo.
The fact is that when I went off to the University of Virginia after that year mostly with Duolingo for an 8 week Russian intensive course I had one big advantage: I had a lot of vocabulary to work with.
My wife has used it for a long time with Spanish, and was quite impressive in Costa Rica in December.
In the end, I don’t think it’s any more or less effective than flashcards. I hope no one expects to become fluent using only flashcards.
Also, when Dr. Krashen talks about comprehensible input, he specifically mentions graded readers as an example of CI. In one of his online videos that I watched, he was talking about building a library of graded readers in different languages. I was not expecting some apps like Duolingo. The guy started learning German by reading simplified readers, and he began developing a natural feel for German. Here is the full article.
[What to read when you’re a beginner? | Antimoon]
You can’t draw any conclusions from one person’s rather casual experiment, with no independent assessment. That’s why I referred you to research in my earlier reply. There are also some interesting interviews with academic linguists that address the way that adults can and do learn languages. My experience might be of interest to some, but it’s only one person’s subjective experience. And after all, the well known polyglots all use different methods.
The Krashen theory is snake oil. The Natural Method was an honest attempt to apply the Krashen theory, but as the research has shown, large amounts of input is essential, but input alone is insufficient. As far as I am aware, not one single well known polyglot has mastered a language using input alone. Even Mr Kaufmann who markets LingQ on the back of the Krashen theory uses other techniques alongside input including talking with a language tutor. His earlier languages were all learnt at school or with one on one tuition.
Children have relatively poor semantic memory, and make use of procedural memory, with large amounts of input and output, usually with a ‘language partner’ e.g. mother. Adults have a much better semantic memory, and can counterintuitively learn a language faster than a child by using conscious study alongside input. There’s a reason the military and diplomatic language courses don’t use input alone and have a lot of classroom instruction.
As said, Duolingo can be used to memorise a base of vocabulary, but not learn a language. In my case with German, I got bored as they forced me to study every ‘lesson’ and the pace was incredibly slow.
As regards the Babbel course, I enjoyed it at the time, but subsequently realised that I hadn’t really learnt anything. They said it took me to B1 or B2, but I wasn’t even A2.
Incidentally, you earlier asked me a series of question, and I spent quite a bit of time writing a reply. I take the lack of an acknowledgement as a lack of manners on your part.
I see it was not directed at me, but maybe take that up directly with the one it was directed to and between the two of you, you can clear this up amicably. This site provides direct one-on-one “mail.”
That’s great! Seems, of course that all research is highly disputed. I love the Output Hypothesis by Merrill Swain, but naturally Krashen is reported to dispute this: https://en.wikipedia.org/wiki/Comprehensible_output. I’m looking through your response and wish I could find a link to the research you referred to.
As I understand the Output Hypothesis, one of its purported benefits is that it exposes “gaps.” That certainly has been the case when I put out Russian either in writing or speaking. It’s the best of all tests for me. Once exposed and brought into my awareness, I can do something about it.
I would not go to Wikipedia to learn about linguistics, it is not a reliable or complete source.
You can’t expect anyone to reference every claim they make, that’s not practical. But since you ask, I would refer you and anyone else to university level text books which summarise research in first and second language acquisition. A book I like is Approaches and Methods in Language Teaching by Richards and Rodgers. If you want a shorter summary, then 30 Language Teaching Methods by Scott Thornbury is very good. What I like about Thornbury is that he sees that most methods offer something of value that one can use when designing a language teaching/learning methodology. You could also look at some interviews with linguistics researchers on YouTube. The one with Dr. Norbert Schmidt is interesting, His main interest is vocabulary acquisition. He for example states that Krashen’s theory is discredited. There are other interviews in the same series that are also worth listening to.
It isn’t hard to see that the Krashen theory is nonsense. He makes a series of claims for which he provides no evidence. For example he states that a language can only be acquired and not learnt. He does not explain how to determine when something is learnt and when it is acquired, and hence it is untestable and unscientific. In other words, it is true because he says it is true. In fact neuroscientific research has shown that we can move information from declarative memory into procedural memory. Declarative memory is used for storing facts of the kind we learn in a chemistry lesson. Procedural memory stores information associated with carrying out a task such as riding a bike, or driving a car. If you search Declarative Procedural Theory, you should discover more. Thus we can learn words, store them in declarative memory, which is slow to access, and over time as we use those words they move into procedural memory, which is quick to access. This is sometimes called automatisation, and neuroscience provides an explanation. This incidentally tells us that we can use a tool such as Anki to learn words and phrases, but we won’t be able to use them naturally until we have used them often enough that they become second nature.
Incidentally, I would argue with your statement that all research is highly disputed, that certainly is not true. I should add that I spent eight years in physics research. It is certainly true that in many areas there are competing ideas, with insufficient evidence to decide which is correct. And of course some theories are highly contentious. However, there are many ideas in SLA that are generally agreed upon. For example, input is insufficient for learning a language, adult SLA is not the same as L1 acquisition, adults can learn an L2 to native level, though most do not acquire an L1 level accent, graded input is beneficial, words should be repeated across lessons or stories. and so on. I’m afraid you’ll have to watch the various interviews I mentioned earlier, and read text books to find sources for these remarks (or correct me if you find I made a mistake), as I have to go from memory.
There is interesting research carried out by Professor Stephen Pinker’s group on adult L2 competence which proves that most adults who learn an L2 do attain a native level understanding of grammar after ten or so years of immersion, I forget the details. I’m sure you can Google that. This research disproves Noam Chomsky’s claim that we have a Language Acquisition Device that disappears as we reach adulthood. Hence he claims that adults cannot learn an L2 to native level. Hence we have another researcher making an assertion with no evidence, but we should accept it because he says so.
Apologies for writing so much, but if you are having trouble sleeping, I might have provided the cure.
The youtuber Evildea started a by-the-book video CI experiment with Dreaming Spanish videos 11 months ago, and he reports every 50 hours:
He’s sticking to the method because it’s what his test is about, but you can feel his frustration sometimes, for example with not reading before xxx hours of watching videos.
