LingQ: The Illusion of Learning

LingQ is one of the best ways to learn to recognize words and to see them in different contexts. For learning to actively recall them, there’s almost nothing here that could be useful for this purpose, what you need is either a proper SRS flashcard system, or another platform, such as Clozemaster.

3 Likes

This is absolutely right. Just yesterday I came across a word in German that I thought I absolutely knew, but here it had an entirely different meaning. I was surprised, but this happens in many languages - there are often multiple meanings for a word, and some are rare.

So I updated the meaning and moved the word one step back (from 5 - known - to 3 - partly known).

1 Like

LingQ has an integrated SRS system. In my view, it’s more flexible, customizable, and user-friendly than any other flashcard system I’ve come across.

Sure, but in my opinion its limitation is that it will only show you the same sentences you already saw in your lessons, so you might as well just read these lessons again, which is usually more engaging than just repeating flashcards. Or, keep reading new content and hope that the words you want to learn will appear again (for which, again, LingQ is absolutely the best). That’s why I mentioned Clozemaster, which shows the same word in many different contexts.

1 Like

Even Steve Kaufmann says in his videos all the time that he marks a word as Known by only recognising it in a single context while reading. Everyone has their different criteria for this. If you are choosing to use the LingQ SRS, you are choosing that as your metric for Known Words. Personally, I don’t like that as my metric, so I don’t use the LingQ SRS (I also don’t use SRSes in general for language learning).

Re-reading is a very powerful and useful strategy as a beginner, especially if you are limited on content. This is especially the case on LingQ, where re-reading is much faster than reading new content, due to the massive time-waste of creating lingQs. I’m not sure who this person is who wrote this book, but the fact they are giving you advice on how you should use LingQ is worrying. Are you sure they have even used the software? Do they really understand that you can get through 2x or 3x more content by re-reading over reading novel content, depending on exactly how you manage your definitions?

It maybe goes without saying that you shouldn’t only re-read. Variety is the spice of life. But the strategy has its place in the toolbox of various methods to use to learn a language.

If you don’t have a diagnosed psychological disorder which somehow affects your ability to learn languages, I doubt it. Were you a bottom percentile late speaker of English as a child? You seem to have mastered at least one language enough to write it very well and be able to read an adult non-fiction book. It’s a very common myth that people believe they are not “language people” or “maths people” or “creative people.”

1 Like

Sure, but flashcards work best if you have some context from what you’ve already read. LingQ gives this context. For another thing, you yourself said in your earlier post:

Again, LingQ has a proper SRS flashcard system. You seem to be arguing against yourself here.

The problem with other flashcard systems is that, very often, their only context is that of the user’s first language. LingQ’s SRS system adds the context of sentences the user has encountered before. This means that every flashcard has context from your memory, which means that LingQ’s SRS system is much more powerful than any other flashcard system that I know of.

I am not, but that’s because what I consider proper obviously differs from what you consider proper. Every learner has different needs and for me personally, the flashcard system offered by LingQ is not sufficient and definitely less engaging and efficient, compared to the other platform I mentioned.

2 Likes

In my opinion, Clozemaster is not a great platform for language learning. Sure, you’ll learn words, but at its heart it’s just a more advanced version of Anki, and as you said, we need to:

Unless it has added a reading component that makes it similar to LingQ, Clozemaster doesn’t permit that.

But sure, if you combine the reading system of LingQ with the SRS system of Clozemaster, you can have a system that works similar to LingQ. But in my opinion it is definitely not as engaging or efficient as LingQ. Also, IIRC (and it’s years since I used it so I may be getting confused), it relies on other users to create word lists, and sorting through these to find useful ones is a chore that makes Clozemaster very clunky and inefficient. That is the main reason I gave up on Clozemaster - I was spending more time finding and fixing the various user-generated word lists than I was using the flashcards.

You are quite right, but you are explaining what “known” means to LingQ, not to the people who use it. The crucial point is not to assume that “known on LingQ” means “part of my active vocabulary” (or, for that matter, not even necessarily “passively known in all contexts”)! It means precisely what you say - a word LingQ will show as white* without providing an (immediate) translation the next time you encounter it.

When I show LingQ to others, I always explain the “known words” as: “words I either immediately recognized in context, or else have recognized often enough in a row**.” I make no claim to “know” these words outside of that context. But surely, before I can “know” any one word out of context (whether actively producing or passively recognizing), I have to get to a stage where I can recognize it in context. That’s what LingQ provides me.

*worth noting that LingQ also considers words at level 4 (“learned”: shown as white but underlined) known.
** I use the status levels 1-4 + “known” to keep track of how many times I recognize the word without having to look it up: +1 if recognized, -1 if not recognized.

I have over 45,000 known russian words in Lingq. I actually know perhaps 2 or 3 of them, so you may have a point.

2 Likes

Another important vision that can help you is see the learning like a toolbox, where you can pick different tools like Anki, LINGq, books and stuff. Over time, you’ll realize that you will be capable of using other tools and resources to complement the LINGq. And while you’re doing that, you will really improve your capacity to solve problems too.

2 Likes

Let me give an anecdote.

I have been learning Italian for two years…I now do a weekly conversation class.

In that class, I find myself naturally (within reason) saying things and expressing myself. Outside of the class, I find it difficult to think of simple phrases and stumble upon words.

Oddly, when I’m in the class, I find it very difficult to remember English words.

I think there is something about allowing your brain to make a context switch and I think that explains why we have trouble, particularly at the beginning.

2 Likes

The book summarises the results of cognitive science research into learning. One author is a professional writer, his task was to make the book readable. The other two authors are researchers in the psychology of learning and memory. The research they cite consists of empirical studies in the real world. For example, studies show that if school children are tested regularly, their performance in exams increases significantly. That testing could consist of asking them questions about the previous thirty minutes of teaching, or asking questions at the start of a lesson which relate to the previous lesson. The testing encourages them to pay attention, but it also helps summarise key points and reinforce long term retention. Another trick is to mix up topics, rather than flood one topic. Study several points of grammar in one study period, rather than spend one period on one grammar point, another period on the second and so on. Students prefer not to mix topics because it is harder, but that difficulty is beneficial. Steve Kaufmann mentioned the book in a video, and it is excellent.

So yes studies show that simply re-reading is not an effective learning strategy. It needs to be accompanied by testing, and/or output, though output is a form of testing.

I find that re-reading is not effective, in German at least where I am low B1, and I have to add in testing. Of course my case is mere anecdote, though I’m glad that my subjective experience is confirmed by research.

The book does not discuss LingQ, but addresses learning techniques. And it’s not about quantity of input.

Certainly going over input again can be useful, but input alone is insufficient.

This is very true. That attitude can be very damaging.

2 Likes

I am aware that the testing effect is a highly replicated effect in cognitive science. I’d just say that the statement that “re-reading is not an effective learning strategy” is a blanket statement and has zero nuance to it. Several things which it fails to take into account:

  • The text was completely forgotten (eg. 20 years later) or close to it (which is much closer on the timescale)
  • Cognitive load (eg. if you have insane levels of cognitive load, like if you go through the first Mini Story in a completely unfamiliar language like idk Vietnamese, nearly all will likely be forgotten almost instantaneously. Your cognitive load is through the roof as a complete beginner in completely unfamiliar languages)
  • In language learning, there is consistent testing while reading. Within a single sentence alone, depending on your level, you may test yourself several times. Eg. “What does this word mean?” “Does this grammar make sense?” “Do I understand what’s being said in general (on a sentence-level)?” This doesn’t necessarily need to happen at a conscious level. I wonder how many studies have been done on re-reading vs. retrieval practice in language learning. Were they done on both beginners and intermediate language learners (re: cognitive load)?

To say you should never re-read in any circumstance and should always do retrieval practice instead has zero nuance.

Quantity of input is 100% important. If you are doing it right (i.e. the content is mostly understandable but not completely, and you are looking up definitions or you are comparing the sentence with its translation, and you are actually paying attention), then quantity of input = quantity of practice. As I mentioned above, you get huge amounts of practice retrieving definitions and analysing grammar structures over a single text, if done correctly. As I’ve mentioned on the forums before, my preferred method these days for this is with dual subtitles on YouTube or bilingual books, where you go through each sentence thinking, “Do I know every word/piece of grammar? If not, can I guess the unknown word(s) based on context? If not, scan the English translation to compare the difference.” Obviously, there are ways to use input less effectively or in a pretty much almost useless manner, such as listening to the radio while sleeping.

But not as good as retrieval practice, you are saying?

Depends on the goal, but if you want to speak, yes, 100%. Many Swiss Germans understand High German perfectly, for instance. They just fail to speak or write it with the ease and mastery of someone who has practised speaking and writing a lot. Furthermore, it’s doubtful that input alone is the fastest way to achieve perfect comprehension. But, as mentioned above, these days I prefer the sentence comparison approach of using input. It also has its limitations, but I like it as a way to increase the amount of vocabulary I can recognise (i.e improve my ‘passive’ vocabulary), as this still remains my biggest weakness for my goals, despite being my focus the whole time.

2 Likes

You have misquoted me and changed the meaning of what I said. Here is the complete statement: “simply re-reading is not an effective learning strategy.”

In other words, re-reading on its own is not effective.

Reading and listening only provide a limited degree of testing. When we read a paragraph, the words are already arranged in the correct order, and each component is already conjugated correctly. Yes, it does prompt us to translate the meaning of the components, and the phrase, but it does not force us to notice all of the features of the phrase e.g. word order. Thus we create a limited passive knowledge. A student can understand spoken German phrases without noticing the complete grammar. A proper test would, for example, ask the student to translate the phrase “This morning I had a shower” into German which is “Heute Morgen habe ich geduscht” and variations thereof. That forces the student to consider both the word order, and the conjugations of haben and duschen. The first time they try, they will probably make mistakes, and they will have to retry. Thus the testing highlights gaps in knowledge, improves noticing and forces appropriate practice.

In my German studies, when reading and listening, I now regularly test myself, by for example covering up the German sentence, and translating the English back to German. I also test myself by playing around with words and constructing new sentences, which are verified using online translation tools. I find it to be very beneficial, as expected from the studies cited in the book. Before when I only consumed input, I was not picking up the word order, and conjugations.

I said no such thing.

I don’t know the details of your method, but it sounds like you are throwing in some testing. In other words, you are not simply consuming input.

I’m not making a black and white statement. Obviously input is very important, and in my experience it becomes much more important at the B2 and C1/2 levels. However, what I am saying is that active study and testing add significant benefit. In the early stages the student will struggle to consume input, and study will dominate.

Now to be fair different people like to work in different ways. I know one polyglot who essentially crams grammar and vocabulary in the early stages of learning a language, and then moves over to more and more input as his knowledge develops. Some people would find that approach abhorrent. Other people claim to use mainly input from the word go. Whether it is the most effective approach is irrelevant if they have found an approach that suits them. I found that mainly input just did not work for me with German. But I’m not saying people must use my method. Too many polyglots, and I am not one, claim that their preferred method is the best, and that all students should do as they do. I suspect you will agree with me that that is nonsense.

I’m not sure Swiss German is such a good example, as it is sufficiently close to high German that it could be acquired relatively easily by a high German speaker.

My biggest weakness is also my vocabulary. If I could acquire German words much more quickly, I could move over to listening to a lot of input, and my skills would then snowball. Listening and reading are a great way to reinforce and automatise what one already knows. I know quite a lot of words and grammar, but I cannot understand and use them at natural speeds. The automatisation of that knowledge is crucial and takes time.

1 Like

You’re not understanding the point here, which is a very good one, and poses a significant challenge to Krashen-style pure input theories.

A lifetime of high-volume input is often not enough to acquire accurate productive skill in a particular language variety without specific focused practice and conscious learning. The case of Swiss German speakers illustrates it: most never acquire fully accurate High German, despite living all their lives in an input-rich environment. Their High German remains noticably foreign, with errors in pronunciation, grammar, and diction that directly result from a transfer from the Swiss German dialect. If input were all that’s needed to acquire good active production skills, that should not be possible.

(I am a Swiss German speaker, and a German teacher who’s worked on his High German for professional purposes.)

1 Like

I think I did understand. The point I was making is that a passive understanding would be relatively easy to obtain through passive exposure to Swiss German due to the relative closeness, even though Swiss German is unintelligible to a high German speaker with no previous exposure to it or a related dialect. Thus it was not the best example.

A passive understanding would be much harder to obtain through a lifetime of high volume input with a much more distant language. Several of Krashen’s coworkers speak fluent Spanish, which is relatively close to English, and thus easier to acquire.

Absolutely. That relates directly to and is consistent with my earlier posts. nfera has previously referred to the Krashen theory as balls, and I am in complete agreement with her.

1 Like

Nice response.

FYI: I’ve learnt basic Japanese using SR, but forcing myself to actively recall words, phrases and sentence templates. I also practiced a lot in real world scenario’s while I was living there, which also helped.

I just don’t see how reading (a la Krashen) and adding new words to a SP program primarily testing recognition can be as effective a methodology as active recall and practice speaking. Anyway, I do like LingQ and have now added reading to my Spanish study.

Oh and the last comment I made about my language learning aptitude was tongue in cheek.

3 Likes

I’m not sure what SP is? However, I’m not sure anyone can pinpoint a “most effective strategy”. First of all the most effective strategy is going to be one that keeps you motivated. If we assume active recall and practicing speaking is the most effective for now, if a language learner hates these things and doesn’t have sufficient motivation to keep them doing these things, then they will quit. Quitting is definitely not a good language learning strategy =)

It also may depend on what your goals are (and these can change by the way). If I want to be able to speak right from the start and get involved in conversations, then active recall and particularly with language islands and common everyday phrases and sentences probably are the most effective strategy. However, you will still probably hit a speed bump when someone asks you something that you don’t know the vocabulary for…or you can’t understand them because you haven’t been practicing listening. One’s goal may also not be to speak first, but be able to read novels, news, etc. and they might circle back later for being able to hold conversations.

Or, for me, the biggest obstacle for SRS was when I get enough vocabulary and reviews took up all my time. I began to not learn any new words. Or very few.

For me, I added far more vocabulary after the initial beginner phases when I started using LingQ. I think mostly because I would not be stuck in reviewing all of the words that I had difficulty getting to stick. With input via LingQ, I could simply move on to the next words that would stick. No SRS, no active recall. Thousands of words I am learning rapidly, mostly passively, but some actively as well.

Now, at a more advanced level for myself, I am starting to incorporate more SRS/active recall/language island approach to “fill in the gaps” so to speak. The words learnt with LingQ had slowed quite a bit and I sometimes was feeling like I’m learning all these advanced words (or not because they are uncommon), that I may never use. It would be better to actively work on the more frequent words that I might still struggle with and/or work on frequent words that I may never have come across due to the nature of the material I was consuming.

Anyway, not sure what the most ideal approach is, but it’s probably a balance of things as usual.

3 Likes

I should point out additionally, reading IS spaced repetition and active recall, albeit on a somewhat random timeline. Every time you see an unknown word in a sentence, you have a chance to practice your recall of the meaning. If you don’t know it, you look it up. You will also see the most common words over and over again and various spacing between events where you see it again. There’s your spacing. The benefit too, is you see the word in context which can both help in the meaning, and to connect words to other words.

3 Likes