I will add to the last section that speaking three or more languages is not particularly rare, though it is relatively rare in England. A friend grew up in Malaysia, speaking Malay, English and two Chinese languages, as did her family. She subsequently learnt other languages including German. Another friend grew up in Andorra and speaks Catalan, French, Spanish and English. I’ve met countless Indians who speak several Indian languages plus English, I think that is common among the more educated Indians. Africans often speak several African languages plus English or French. I worked with a Berber who spoke Berber, Arabic, French and English. At university I came across many people who spoke three or more languages, they tended to be from well to do families. An Indian friend spoke many languages and could also speak English with a standard southern English accent, an upper class English accent, and an American accent. Italians often speak their local language as well as Italian, and sometimes English. Admittedly these people tend to learn their languages when young, but it does suggest that speaking multiple languages is within the grasp of most people, exposure to the languages - i.e. circumstance - being the key requirement. However you could argue that learning multiple languages as an adult is not the same, and it requires different mental attributes that are not possessed by everyone.
Some interesting points.
@LeifGoodwin With your list of “recognised techniques,” I note that some of those are actually used by “traditional” language classes, namely, testing (eg. translating sentences in line with the grammar rule of the day, which you receive feedback on if you were correct), and output (generally every lesson will have at least one speaking activity). Furthermore, these classes use didactical content, which generally results in the higher use of high-frequency words. Besides, isn’t Zipf’s Law just the “natural spaced repetition system,” if you actually expose yourself to any input?
The huge amount of input is generally lacking in “traditional” classes though, especially at the absolute beginner level. However, once you’ve passed the absolute beginner phase, the teacher is predominately only using the language to communicate to the class (i.e. input, which is designed to be understandable). The issue is that many people don’t get past the absolute beginner or beginner stage for the teacher to then start using the technique of explaining instructions in the language as a form of input.
Why then are “traditional” classes given such a bad rap, if they follow some of your “recognised techniques,” which are techniques confirmed by research as effective? Is it because they are still not efficient enough? Is the bad reputation of “traditional” language classes as a language learning technique actually justified? (Note: Do all countries even have a bad reputation for language classes? The Swedes, the Norwegians, and the Dutch all managed to learn English with the help of them, for example. @Dominic_Olofsson-Tuisku) Is the general judgement of “traditional” language classes being inefficient really just based on one’s experience in school many years ago (which, to put it blantantly, is probably the worst form of “traditional” language classes, mainly due to all the distractions created by large class sizes of children who don’t want to be there)?
I would assume mainly because there is a huge difference between how something is supposed to be taught and how it actually is taught. I was very lucky with my English teacher that I had the last five years in school. In another class they had an English teacher who only spoke German in class. So you have to be very lucky with the teacher you get.
Another aspect is that different pupils get different results, so even though the other pupils in my class shared the same teacher, not all of them were equally good. So the pupils themselves play an important role, too. The problem is that most people tend to either blame the others or to refrain to a lack of talent and so one first.
Of course, there are instances were the teacher might be to blame (see my above example) or where the system could be improved, and in some aspects others may have an advantage. If you are from an uneducated family you have to take higher hurdles then someone for whom that is not the case. But my experience is that most people don’t tend to reflect on what they could do better or to even ask why other peoples make better progress then themselves. And they just stick to that attitude til they’re old.
In my perception here in Germany schools don’t have the highest reputation, but language classes are usually not seen negatively. Those of us who are younger can to some extent at least communicate in English. The older ones here in Eastern Germany usually learned Russian as first foreign language, but even that sticked to them to some degree, even though it isn’t of much use in everyday life. So there isn’t the perception that we are lacking in that matter. German and especially Mathematics are a completely different topic, though.
Good to see you again @nfera. My take on “efficiency,” which I imagine is essentially speed in acquisition, though I suppose it could also be progress made per unit of serious effort, which might not map to speed (relative to some arbitrary unit of time), is that it is largely a function of intelligence.
Given my personal experience, my experience teaching my kids languages, my experience tutoring a handful of kids and adults, chatting with other language learners in real life, is that the key variable is intelligence. Intelligence just is the efficiency of consolidation of new information, or the ratio of instances seen to accurate generalization / systematization made.
Once your basic language acquisition method is pointed in the right direction (you are no longer doing obviously fruitless, time serving stuff), what would really make your efficiency go up in a noticeable manner is not tweaking the method to squeeze out tiny marginal gains, but magically increasing your IQ ten points.
I don’t mean this to be discouraging, cynical, or cheeky. The most truly impressive feature of a person on a language acquisition journey is not how quickly they learn but how persistently and hard they work, but on a theoretical level that’s what I think.
I agree with a lot of your post.
I explained myself rather poorly unfortunately. You’re right that classes often have exercises or tests to do in class, and that these can be very valuable. What I was referring to by testing, is testing after an interval. Thus a class might cover five grammar points in sequence, ten minutes on each, with an explanation and some exercises. Then at the end the teacher will do a quick test by asking some questions on all five grammar points. A teacher might even do a quick test on key points from the previous lesson. The aim is to bring the key points back into the students mind, to refresh their knowledge. This delayed testing has been shown to improve performance. No doubt it encourages students to pay attention in the lesson but it also aids retention.
Our schools might use this technique, I cannot comment.
I had to look Zipf’s law up. Of course we learn through exposure, as you say. However spaced repetition systems (SRS) use a different principle. Research has shown that long term memory storage improves markedly if the student is forced to recall a piece of information by regular testing over gradually increasing time intervals. There is something about the effort of having to recall a word just as you start to forget it that is more efficient at moving it into long term memory than simply passively hearing the word. Plus, less frequent words are less likely to be learnt ‘naturally’. Thus an SRS can aid retention of less common words. As I mentioned earlier, I use an SRS with phrases, it’s much more efficient in my experience. Words on their own are often useless, which is why some people find spaced repetition does not work and they go back to passive listening.
However, I have just remembered that I bought some flash cards which cover the English French GCSE requirements i.e. the school exam taken at age 16. So our schools DO use SRS.
And a good teacher will test students regularly by showing them words and phrases, and asking them what they mean, or to translate them into the target language.
I think this is a good point.
You are right that school language teaching often gets a bad rap:
Two years ago I restarted learning French at the young age of 59. Like many here, I watched YouTube videos, read the entire internet and concluded that comprehensible input was the only true religion, and Stephen Krashen the one true God, Of course anyone who praised traditional classroom teaching, which of course revolved around rote learning of grammar, was an evil sinner, to be pitied and ridiculed. After all, CI is a modern method based on science.
Experience and listening to others has changed my mind.
Like many in England, I spent roughly 2 hours a week, 38 weeks a year, for five years studying French at school. That’s about 380 classroom hours. That’s enough study time to reach a decent B1 level, and I think I would have succeeded had I worked hard. I didn’t because I was lazy, and not interested. But I still gained a solid foundation which helped me progress in later years. Of course they could be improved, there is not enough time spent studying and listening to the language, and of course not all teachers are excellent. I think we expect too much from school lessons, the motivation has to come from within.
I agree with @GMelillo that efficiency (language acquisition rate per unit time probably being the most useful form to refer to at the moment) is a mere trifle to the ability for people to persist in the learning process. As @Obsttorte mentioned in another post a while back, learning English in his school system in Germany entailed 4-6x the number of hours of @LeifGoodwin’s experience of learning French in the English school system.
There are large amounts of time in every school lesson where you are simply not studying, such as:
- daydreaming and not paying attention by the student, because the student doesn’t want to be there
- class-wide distractions from other students who don’t want to be there
- waiting for the teacher to come around and personally address you and answer your questions
- waiting for all students to finish the activity, which you finished earlier
- start and end times of the lesson, which involving preparation or packing up, not necessarily studying
As you can see, the amount of non-study time is significantly increased by large class sizes of 20-30 students.
Despite this, all this time-wasting appears to not matter much, if you can just keep doing it year after year. That is, the end result of having some form of language competency is eventually obtained. Perhaps the reputation of the respective school systems (bad rap in England and the U.S. vs. Germany, for instance) is merely based on the end results (basic competency or not) of which the largest determining factor is simply hours spent.
Time-wasting activities are all well and good (that is, begrudgingly accepted) when children are involved, because children have all the time in the world, whereas we are time-constrained adults. If we were to take classes again as adults, we would have sufficient advantages over those classes in the school system:
- nearly everyone wants learn (with a few exceptions) and actively signed up to the course themselves
- you are more likely to complete your homework due to the first point
- class sizes after often smaller (especially, if you actively find a course with a small class size or better yet get a private tutor for 1-on-1 classes)
If this is the case, as in not incredibly inefficient, why aren’t more of these Internet polyglots talking about classes and/or taking them? Is it a mere matter of convenience of studying at home, when and how they want, not necessarily efficiency?
@GMelillo The issue is ironing out the ‘basic language acquisition method’ to remove all the time-waste. How do we do this? How are we certain such activities are obviously fruitless? You’ve previously referred to using trial-and-error, adjusting your method based on evaluations every six or so months. Trial-and-error like this results in improvements to the method, but slowly, over the years. How do you judge that the latest adjustment to your technique was better than the previous implementation? Your children have the advantage that they can learn from your trial-and-error.
I seem to remember reading something, where it was still debatable on whether intervals performed better gradually increasing over other intervals, but I am not up-to-date on the research Has there been research to conclusively prove that increasing the intervals indeed does have a significant effect over other interval variants (eg. constant, etc.)?
Content as per Zipf’s Law is best considered constant intervals with a decent amount of randomness/noise.
The general opinion seem to be that spacing out repetitions does improve the effectiveness of the learning efforts, especially when combined with interleaving. One possible explanation is what is called repetition suppression: the neurons react less intense to a repeated stimulus. It is probably comparable to not feeling the clothes on your skin or not smelling the exhaust gases of the cars in your city because you somewhat accustomised to it.
Where there is no unison is on whether SRS as incorporated in Anki for example is of benefit (increasing the intervals of repetition). Unfortunately, the papers I found regarding that matter are locked behind a paywall, so it is hard to judge on the results.
But to sum it up: One should probably avoid repeating informations in a way were most of the repeated information is overly familiar to you. The learning effect seems to be higher if you are already uncertain on many things or about to forget them, especially in regards to long time learning. I guess that interleaving is a good way to achieve this. So using different sources that use different kinds of vocabulary and grammar, mixing reading with grammar studies etc…
Most of the texts I import into LingQ have a new word count of 15% +, simply because it is hard to find content with lower percentages that isn’t super easy. On top of that I have a similarly high amount of yellow words. So I get the spaced out effect due to my choice of learning material. In addition, I dedicate time for grammar and Hanja studies.
My personal impression is that I learn reasonable fast, although I have nothing to base this judgement on objectively. But from the list I’ve found (see below) I am early intermediate 2 after 1.5 years of studying Korean for 30-60 minutes per day as a native German who only knows some English and a bit of Spanish. My personal guesswork is I am somewhere at A2->B1 (CEFR). I did the TOPIK I on my own (yeah, reliable, I know 
) and managed to score around 80% of the points. I definetely can’t speak the language beyond some simple sentences, though, as I never practiced it. I don’t know if this means I have learned effectively or I am slow as a snail. As said, I don’t have any comparision to base my judgement on. I leave that judgement on to others.
The Wikipedia article on SRS indicates no concensus on fixed versus increasing interval spacing with time. One good practical reason to increase spacing is to allow the student to revise more items, assuming a fixed number of repetitions per day. Assuming no other differences, that alone would be sufficicient reason for variable spacing.
I use SRS via Anki, but generally only for sentences. I do sometimes see a problem associated with rote learning patterns. For example when I learn two phrases such as I can look after my little sister and Ich kann auf meiner kleine Schwester aufpassen, there is a danger that they are learned as abstractions dissociated from the meaning. Thus when I try to say Will you look after my dog, the verb aufpassen, or the preposition auf escapes me. The same issue can occur with grammar of course. My impression is that increasing the interval prevents rote learning, possibly because it forces the student to think more about the phrase. I think this is a danger of SRS, whereby you rote learn patterns not meanings.
I do wonder if what I need is good old fashioned exercises, translating random simple phrases to and from German. You know, as they do in school, and not in LingQ. 
Definetely. I remember back in the day when I used Duolingo that I could often recall the translation even though I haven’t yet read the sentence. I just recogniced it by its shape or length (I could always remember the longest sentence in a lesson). It’s not as if you wouldn’t learn anything at all, but it isn’t efficient for sure.
I guess that’s were AI can come in handy. You can easely ask it to create example sentences containing a specific grammar. Or use it for building up vocabulary. For example let it create a text or sentences using words that contain a certain word stem and share a connected meaning. And you can do it both ways, so English to German and German to English for example.
BTW. I don’t think that there is a contradiction between “as they do in school” and “as in LingQ”. One aspect of language learning in school is that you read a lot of texts. We read (usually simplified) versions of Shakespeare, “Death of a Salesman” (good book, btw) and saw movies and went to theatre (again, Shakespeare: Romeo and Juliet). And many students read books. If you go to a german bookstore you’ll usually find a section called “books for young readers” (not a translation, it’s written their in English) containing English books aimed at teenagers. I don’t know whether that is a thing in English speaking countries.
On the other hand, Steve Kaufman himself and many others in the forum have stated that they don’t restrict their studies on using LingQ or reading only. Not to mention that it is hard to tell how many LingQ users are actually active in the forum or how many of the people who successfully learned a language have done so using comprehensible input as their main learning method.
We are in agreement here, that determination and hard work are the key to successful L2 acquisition.
@ObstTorte and I discussed this a while back. I think he pointed out that Germans are exposed to a lot of English language content, especially films, books and music, and English is much more useful for a German. An English sales rep in China will speak English with the Chinese. A German sales rep in China will do likewise. Also English is probably fairly easy for a German, both are West Germanic languages, English grammar is simple, and both share similar phonetics with stress based timing and a love of diphthongs. An English speaker learning German has to deal with the delights of gender and a case system.
I assume classes are not more popular due to cost and inconvenience. And the cynic would point out that many internet polyglots are pushing their product, either an app or their videos, and not classes.
In my case, with French, it wasn’t too hard to work out a method that works. I started with apps such as Duolingo and Busuu but they weren’t taking me very far. I moved to LingQ in sentence mode, and doing the exercises, but it was not much better. Then I moved to far more listening, namely podcasts and videos for learners, and it was obvious that progress was much higher. It was daunting as I often understand no more than the gist of a podcast, but that gradually sorted itself out as my comprehension improved.
So in my opinion there are inefficient methods, and efficient ones, although I would rather use the term effective as I never really gained a feel for the language using Duolingo due to insufficient input.
I’m at an uncomfortable place with German, lower B1, and considering experimenting with new approaches.
There are some people who try something for a few months, get bored, and move on. Some continue, using a new approach, but get bored. I’ve made progress in several sporting activities, depite starting in my mid fifites, and being “bad at sports”. I went down several rabbit holes, sometimes making little progress, or even regressing, and endured a lot of bullying. There is so much I could have done better to achieve more in the same time. I had a not dissimilar experience with French. How do we avoid rabbit holes and time wasted ? Or is that part of life ?
How do you learn ? How did you find your learning strategy ? Is it efficient or simply effective and enjoyable?
As has been agreed on previously (I think) the efficiency of different method varies depending on the person. So I think one really has to make mistakes or waste time or however one may call it in order to understand what works and what not. In the end you may not only acquire a new skill, you also learn something about yourself. And the effort you had to invest will make you proud of what you’ve achieved.
If you would be more efficient to begin with, you would be similarly frustrated that you can’t be even more efficient. We are never really pleased with how good we are. But that is a good thing, as it makes us work harder … or die trying. 
I’m sure it’s possible to increase one’s efficiency in language learning. Consider relying strictly upon DuoLingo or traditional K-12 education to learn a language.
Not a bad place to start, but it won’t get you to fluency.
Those using LingQ are likely in the upper 20% of learners. I’m sure we can all be more efficient. Maybe throwing in more spaced repetition or systematic grammar study would help. Or sessions with tutors, etc.
But unless one focuses upon becoming a language learning athlete, I’m not sure it matters as much as sticking in for the long haul.
I’ve been a committed self-learner – autodidact – since I learned to read adult English fluently (as a native speaker). If I can read about something, I can learn it.
The main barrier IMO is maintaining interest. Just about everything worthwhile I’ve learned took years. When I failed, it was because I lost interest.
While I love to learn about the best ways to learn things, I pick and choose how I do it.
I optimize for enthusiasm and enjoyment. Without those I’ll lose interest and stop. It won’t matter how efficient I was learning. Unless I have a gun to my head – pass a test, get a degree, keep a job – I’ll move on to something else.
Maybe I’m not learning French as efficiently as possible. But if I had to do grammar exercises and spaced repetition according to some program I didn’t enjoy, I wouldn’t learn French at all.
I believe it’s highly subjective, but I also think that’s kinda part of the explanation. When I learned French I was extremely ineffective. I keep returning to it from time to time. It’s the language where I have the most known words, and I’m always surprised by how stupid and ineffective I was in it the last time.
But that’s neglecting that each time I’ve left French I’ve studied something more challenging and brought with me skills that have highly improved the way I learn.
I also notice, and I think it’s universal. That the more you learn, the less you need to try to get double the effect. Just reaching 5K known words in French was a trek akin to LOTR, but it also gets so much more enjoyable to read a book and see your first page on LingQ where there’s not a single unknown word. Where you understand everything perfectly and visualize it before you. The components are so much fewer so your brain I think has more power to concentrate on a singular thing. It’s also faster which means it’s less of a “humiliation” compared to your native language. You feel competent and confident. It’s also that the more volume you can push through, the higher your output is. It took me a full month of studying daily to read 100,000 words in French 2 years ago now. No school, nothing, now in Russian this year, I could read 100,000 words the same month where I was studying 80 hours a week. A month ago I could read 25,000 words in Afrikaans in like a day.
So yes, it is possible. Your brain reads so much better each year in your native language if you still use it. The same applies to all other languages.
Back in the late 60s R.A. Lafferty, a science-fiction writer with an off-kilter sensibility emerged. He’s hard to describe so I won’t. However, he did write a funny time-travel story that bears upon our current topic.
In “Thus We Frustrate Charlemagne”, a group invents a time machine and attempts to improve history by making subtle adjustments to the past. They hope to create a more efficient and advanced world by altering significant historical events and by giving their younger selves advice on how to be more productive.
Initially, their interventions appear to work, making the world a better and more advanced place. However, as they continue to meddle and provide advice to their past selves, it eventually backfires.
Instead of creating a better future, the accumulated changes result in their younger selves becoming increasingly disillusioned and overwhelmed by the constant pressure to be more efficient.
In the end, they give up.
I remember you mentioning before you found yourself going through French conjugation tables (and doing exercises?) as you were not satisfied with the speed with which input alone was teaching you the lesser frequent conjugations. With this comment and the above one I imagine the analogy of a wrecking ball coming in and destorying a house. Slowly, step-by-step, through trial-and-error, a new house is built. Upon stepping back to view the newly completed house, you realise that it looks exactly the same as the initial house, except with different colour windowsills.
That said, what I do on a daily basis for my language study looks nothing like a language class. It is almost entirely input, but not the ‘Stephen Krashen / comprehensible input’ way. As in, I use translations and/or dictionaries instead of only relying on context to understand unknown words and grammar.
However, I wouldn’t be surprised if you actually converged on the same or similar techniques to those used in language classes. They, too, went through a trial-and-error process over many years by many organisations to reach the techniques they currently use. Unfortunately, their results are mainly diluted by large class sizes and people (children) who don’t want to be there creating distractions. They are, however, quite holistic, in that they often insist on practising the four main skill areas, and you do a large variety of activities (which is much less fragile than relying on only one or two). For instance, on the rare occasion, when I actually don’t procrastinate touching my grammar textbook, I read the explanations and do the exercises (cloze exercises, conjugating verbs, translating sentences, reading out loud, comprehension questions, etc.) and find it very useful and efficient.
What is worth noting is that I consider language classes should be the baseline for efficiency comparisons, because it’s considered the ‘default’ method in our culture/s. It’s what you were previously using to learn a foreign language in school and it’s probably the first thought as how to learn a language by many people. However, when someone decides against going to a language class and instead branches out on their own to learn a language, they download DuoLingo or whatever the latest app/method in fashion is, and their efficiency actually drops. It’s actually lower than the default option! It’s only those who stay with it and use trial-and-error to get out of the valley of uselessness and inefficiency, do they actually increase it, and only with more tinkering do their methods end up surpassing the efficiency of a standard language class, aka the baseline.
@jt23’s point is interesting that he optimises for enjoyment, because he knows himself well enough to know that he’ll lose interest and give up otherwise. I wouldn’t say, I, personally, optimise for enjoyment. It’s an important factor, for sure, but not the only one I’m interested in. I would also like to get to a high level in my languages at a faster speed, so I can enjoy them more. As in, be able to read the great books, watch the classic movies, and chat confidently with native speakers.
As for the way I find my learning strategy, I’d say the number one way I do it is logical reasoning. This is especially the case with the background of knowledge I have attained about language learning, its stages, techniques, etc. over the years. This results in me making more informed and nuanced arguments than I did before. From this, I can consider if a particular adjustment to my technique or a new technique is worth my testing. Then, as @GMelillo, I try out the technique and after some period of time, I evaluate it. I’m not always systematic in this, but sometimes I am. The issue with these evaluations is sometimes there is no particular systematic way I can think of to measure the changes, as language learning is very complex, so often I have to go off some guess, ‘feel’, or intuition. This is especially the case, if you use the new technique with a new language (eg. consider Italian vs. Russian, where you have a new alphabet and much fewer cognates), but also occurs within the same language as you become more efficient with a language, the higher level you become. You somehow have to take these factors in account.
How do you effectively set up language learning trials (experiments)? How do you go about evaluating them? Does anyone have any examples of methods they’ve tested out and discarded as they think the methods didn’t live up to what they thought they would be? How did you come to this decision?
For example, I, too, like @Dominic_Olofsson-Tuisku have used reading speed as a proxy for efficiency for several of my evaluations. Recently, I was interested in seeing how much my reading speed fluctuates, so I took multiple measurements throughout the same book. I didn’t record any other variables, but I could observe the incredibly obvious affects of being tired and mood (affecting ability to concentrate).
As I mentioned, I optimize for enthusiasm and enjoyment. My enthusiasm drives me to learn French and try various learning techniques, some I make up for myself. It’s not just enjoyment.
I tend to emphasize the techniques I enjoy, which for me is reading intensively with LingQ, while taking notes. I review my notes but not with spaced repetition. I also make flash cards of verb conjugations, which I sometimes consult as I read, especially the Big Four – be, do, have, and go. I also listen to French music and watch movies with French subs.
My enthusiasm was high for year and a half. I really banged hard on French for that time. I steadily improved though I often felt my progress wasn’t up to snuff by the standards of some YouTubes I watched.
I’m not as enthusiastic now. I have other things I wish to attend to. But that’s OK. I feel like I’m over the hump. I can read a French novel in two weeks and I’m getting faster. French is now a discipline for me. I know it’s a matter of keeping my head in the language.
So these days I don’t worry about efficiency. I question how much language learners should worry about it.
One analogy I use for learning a language is solving a huge jigsaw puzzle.
Sure, there is a best strategy of starting with the corners, then the sides. But after that it’s a long slog and doesn’t much matter where you work or in what order.
Very informative.
For French I use essentially the same method as you. For one hour I listen. For another hour, later in the day, I listen, and study words and phrases I don’t understand. Some phrases go in Anki. I also do ten minutes Anki.
My current routine for German (lower intermediate) is 30 minutes YouTube videos for learners in LingQ, and 30 minutes reading and studying childrens picture books, with some grammar study and sentence construction. I also do 20 minutes Anki. I may move to more sentence construction, and maybe even stop LingQ. My German methodology has been changing quite a bit as I reassess study methods.
That’s an interesting point and it does make sense.
It’s a good point.
I don’t use any formal method to create and evaluate trials. My methodology is very subjective and unscientific, which is not dissimilar to your method. Thus I try something, and after a few weeks or a few months I should have an idea of its worth. My core metric for German is how well I feel that I am learning vocabulary, and to a lesser extend grammar. For French my core goal is to get my comprehension of spoken French to a near native level e.g. understand day to day conversations in native films,
Reading a transcript while listening to native French input markedly improved my aural comprehension in a few months.
Listening to a lot of native French input where I only understood the gist helped a lot, probably because it trained my brain to recall words quickly. Understanding words and grammar is one skill, being able to recall them fast enough to create meaningful messages is a quite separate skill.
Listening to French input that was barely intelligible due to the pronunciation rather than the grammar or vocabulary produced benefits with all input in a few weeks. I assume that is because it forced me to focus more, but that’s a guess.
For German I followed the comprehensible input approach for a year. My impression was that word acquisition was rather slow, and my knowledge was very passive.
I then tried listening to and studying simple German videos for learners and that worked better. Active study appears to help retention. However some YouTube videos are not produced by native speakers, and the creators are often amateurs and not experienced language teachers. Caveat emptor.
Recently I have started to experiment with reading German children’s book. Initial impressions are good. I can be sure that the German is correct and is a good representation of ordinary standard German i.e. not formal, not slang and not regional. The act of looking up words in a dictionary and writing them down on Post It notes seems to help retention, possibly because it requires active study. Pictures probably create more associations.
I’m also doing some German grammar study, and sentence building. I may increase the amount of sentence building using online translation tools as checkers. I do think output is required, though not necessarily with a teacher or a speech partner.
So in many repects my method for German is tending towards a language school approach. The methods I have tried and rejected are apps - Duolingo, Busuu and Babbel - and CI.
I must admit I haven’t tracked my reading speed. I know that in French it has increased markedly.
As an aside, I have recently been reading about the history of linguistics research into SLA. As @Obsttorte pointed out some while back, Krashen’s ideas are not new. Even as far back as the seventeenth century, there were experiments with a natural method of SLA. The grammar translation method became widespread in the nineteenth century but even then some people advocated for a more ‘natural’ method and such ideas were widespread in the first half of the twentieth century. Krashen brought these ideas together into a model, and added several dubious supposedly scientific claims. There are countless other models of SLA, many with a stronger evidential basis than CI, but for some reason Krashenism has taken hold among popularisers, especially on YouTube. The cynic might attribute this to shallow thinking, conformism and laziness. Krashen is exceptionally good at promoting his model. Many of us have seen the video in which he teaches the audience some simple German. I regard him as an excellent showman. Interestingly according to some sources, he has made a substantial amount of money from lecture tours, and acting as a consultant to government bodies that are responsible for SLA in schools, especially among immigrants. I cannot verify those claims. He also has numerous academic colleagues who promote his ideas. I would argue, that like Chomsky, his ideas have gained far more prominance than they deserve, due to the guru effect i.e. a very dominant personality. And this can have a detrimental impact on the field, as it directs research away from potentially more fruitful avenues of investigation.