Different slant on fluency

So fluency is a slippery beast. We frequently get asked on the forum “when am I fluent?”, “how many words means I’m fluent?” etc and a bunch of other different varieties of the same question trying to get a definitive answer.

I’m not going to answer that question today (because who can tell how long a piece of string is…) but something I just noticed when I just watched an Olly Richards video last night on youtube. For those who don’t know him, he is a Polyglot like Steve (so IMO he gets to have a legit opinion). Something he said made me think that I think we can answer the question “how do I know when I’m fluent?”

He was talking about the different artificial levels from the CEFR, A1, A2, B1, B2, C1 and C2 etc.
Something he said was (paraphrasing)… “when you get to B2 there is something that you will find… those who have studied only to B1 and then stop will rapidly find that they start to forget pretty quickly. But almost to a person, whoever learns to B2 finds that they have learned the language. i.e. it’s theirs for life, so even if they stop actively using it, it’s still there”.

So interestingly to me, I have Spanish to (at least B2), French to B1 (probably) and Russian to somewhere around B1 (for listening).

I don’t speak Spanish anywhere near as often as I used to. As in barely. But last week I had to take part in a Spanish only business meeting. It was as if I hadn’t skipped a beat.

Conversely with French, I got to intermediate (as measured by lingQ), then dropped it in early 2022. I tried to have a short conversation maybe a month ago at a party and the native speakers hit full velocity and I got zip. Nothing. Nada.
I could speak reasonably well (but probably mostly due to my Spanish) but I could hardly understand.

Those things combined with what Ollie said got me thinking. What B2 possibly means isn’t just the measure of the number of words that you know. Maybe it’s that the language is burned into your long term memory. Maybe fluency means the degree to which a threshold quantity of words/grammar is burned in. If it’s only partly burned in, you forget it. If it’s only partly burned in then you can’t recall quickly enough during conversations to bring it up. Maybe it’s all about the long term memory.

And if that is true, then some interesting observations come into play: I reckon that if you’ve gotten to the stage where you do not have to look up/translate then it’s burned in.
That’s a wishy-washy definition though.

So I’m going to come back to Steve’s words:
Anki (in my case) gives you an easily measurable way of knowing if you really know a word or a phrase. It’s this: if you have your word memorized it’s at 98% retention. Also, it’s likely already at a “mature” stage (i.e. it’s some months out in the future till your next repetition). So there is at least one scientifically measurable way of determining for vocabulary at least if you are fluent. That is to say: if you have memorized (as per Paul Nation) 7-8,000 head words to mature in anki (or equivalent in some other tool) then you have those words commited to the “fluent” part of memory.

^^^^ All of this is my opinion. Obviously others will have their own opinions.


I learned in my Motor Learning class in college that if you were to learn a skill like dribbling a basketball and you were to practice it every day for years it’s basically yours for life after enough consecutive practice sessions. You can get rusty, but you can pick it up again very quickly, and I assume it’s true for languages as well. I would think because it takes so long to get to B2 or C1 that it’s time spent with the language, like riding a bike. But if you didn’t put in enough initial time with the language then it’s probably much easier to forget.


Hi xxdb,

Unfortunately, I don’t have time for a longer discussion right now because I’ve got too much to do. Therefore only a few “food for thought points”:

  1. There is no absolute fluency even for native speakers, esp. in the larger languages (i.e., there are too many dialects, too many sociolects, too many specialized domains with their own jargon / expectations / norms, etc. and, in general, too many words they don’t know). There is only particular fluency.

  2. Thesis: You can’t be fluent in a domain (talking about sport xy, science xz, etc.) if you don’t know what you’re talking about - even as a native speaker. So the most important point in any conversation is: that you know your stuff. Often native speakers don’t - and that’s were you can beat them as a language learner!

  3. “then you can’t recall quickly enough during conversations” (xxdb): The problem with all “input-oriented” activities is that L2 recognition is much easier than L2 production. Thus, if you don’t practice L1->L2 retrieval in Anki and Co, you will still struggle when speaking and writing. And if you use flashcards the wrong way, i.e., as word equations without context info, it’s even worse…

  4. Little children (say 5 year olds) are fluent in their L1(s) with much less than “7-8,000 head words”, but, again, they can only be fluent in particular domains of their lifes.

  5. @cragiandrianos

“You can get rusty, but you can pick it up again very quickly”
As someone who spent his whole teenage years on a basketball court and wanted to become a pro , I agree - to a certain degree:
Yes, “re-learning” is faster than starting from scratch. But that’s not good enough for a real game. It takes still a long time until you’re really “game ready” - usually months until you reach your previous level. And after a certain age, say 35, you usually never reach your previous level again because the mind may be willing, but your body says: “Stop - do you want to kill me?”).

  1. The feeling of fluency has a psychological component as well: It’s the confidence you’ve developed that you’ll do fine in conversations with natives - even though that may not always be the case. And this self-confidence is the result of a lot of practice.

OK, that’s all the time I have this week for the LingQ forum.

Have a great day,


Peter, though I’d never met the term “particular fluency” before, I think that with (1) you hit on my initial thoughts when reading xxdb’s post, An 80-year old “hillbilly” in the deepest, darkest Ozarks considers himself “fluent” in English, and of course he is fluent in his everyday speech. But I could pretty easily say things that he wouldn’t understand in the least. However, since he does know English, and that’s what I speak, with a bit of additional vocabulary and with exposure to concepts hitherto unpondered by him, he could easily understand. Of course, it probably works the other way, I being unable to understand some of what he’d have to say.
Then consider accent and dialect. For an over-the-top example of that, try listening to this Cumbrian woman, already elderly when recorded in 1955, speaking English. I played the audio for my wife who could not even guess the language, much less the meaning. :slightly_smiling_face: The Disappearance (and Survival) of 'Thou' - YouTube


I think there is something of a boomerang in models people will use to try and categorize fluency and comprehension. We as humans seem to have some need to categorize and group things.

As a total mono-lingual “normie”, you might see everything outside your comprehension as fluent if it looks and sounds the part.

If you are interested and engaged, you might learn about CEFR and input and known word counts. You may find ways to objectively measure your proficiency. You might start to think, I am roughly B2 in a language, and I understand 90% what I hear when I watch YouTube videos. You may start telling people you are fluent in a language, or if you’ve reached a similar “level” in other languages you may even start saying you’re a polyglot.

Now the boomerang. This is where I reveal that “you” was “I” the whole time (though I have never referred to myself as a “polyglot”).

I reached a certain point where I threw out the need for measurements and categories and simply think, “Who can I understand, and who can understand me?”

To add some levels we can take the following as a model. I’m not certain who came up with this model, but I will describe them from the perspective of comprehension. This would work just as well for the ability to produce language.

  • Effortless/Easy - I would have to actively try to not understand. It is so automatic that regardless of distortions in the audio, changes in register, code switching, topic, whatever, I will understand it unless I let the sound go in one ear and out the other.
  • Comfortable - With minimal to no effort, I will understand what is being said. I may miss some details at first blush, but will be able to unravel my confusion without disrupting my comprehension.
  • Uncomfortable - I understand, but it requires active effort. It is like a radio that does not completely work. Some stuff is crystal clear, some… isn’t.
  • Incomprehensible - I do not understand.
    A single conversation with a person might go between all of these. They will vary not only on dialect/language, but register, domain/topic, environment, etc. etc. etc. If we are having a conversation and I say, “those mom spoilers are wild huh?”, you will be totally lost unless you have knowledge of what I am talking about.

These are also not precisely definable, I am comfortable because I feel comfortable speaking. I know when it’s not comfortable, because it feels uncomfortable.

So for me, to be fluent is to be able to have a conversation where both I and the other person are at least “comfortable” for the lion’s share of the conversation.

This ability is one you can lose over time. If you do not keep those neural pathways firing they will not remain. The length of time and ease of getting them back is probably directly proportional to how much they were burned in in the first place, but I know several people that absolutely cannot speak their native language in a way that is even “uncomfortable” with another human.

What I’d encourage is get to that point where you’re at least comfortable and then have something to talk about.


Yeah that’s the key point I was wanting to make: looking at it as “yours for life” and what the measure of that is.
The other points, while valid are discussing something else.

I wonder if dribbling has a bare minimum amount of practice you have to do every day in order to cross the threshold. I’m wondering if that might be the limiting factor in languages also: how much do you do each day. The one guy “chytran?” learned Mandarin to what sounds like a stupendous level, he did over 8 hours a day. I did something like 3-4 hours a day for the first six months then heavily slacked off. I maybe did an hour a day for the last year. Possibly not enough because by this point, I should in theory have crossed the 1,100 hours that the FSI says gets you to B2.


It definitely is how much you do/day. And I agree, there’s a tipping point where if you get a certain amount of exposure, everything starts to stick much more readily. It’s like your brain requires an overload to finally accept that this new thing you’re doing is important enough to get to work on.

If I had the time, I’d make sure that I’m hearing more TL than NL each and every day. I really don’t think hearing 2 hours of TL and then 12 hours of NL is all that efficient.

It’s kinda stating the obvious to say that doing something more leads to improvement, but I think it’s more important for language learning than pretty much anything else. Quantity (so long as we’re paying attention) is by far the biggest factor if you want to see real, tangible results on say a monthly basis.

Anecdotally, after a week+ of consuming more TL content than NL content, I get waaaaaaaaaaay more occurances of TL words and phrases popping into my head, and often before the NL equivalent. It’s actually frustrating at times because it can block a word/phrase in your NL whilst you’re speaking to someone.

That, to me, is a clear indication of how languages are learned, if you want to be able to use them spontaneously, without requiring effortful things like translating, which is the enemy of 2nd language “speakers.”


It’s interesting that thing where you have particular words in your native language blocked. It’s happened to me three times now. Spanish, French and then Russian. It seems to last a couple months. Maybe about six months in?
It’s a really wierd feeling - an actual mental block of words you know you know in your native freaking language and they just don’t come.

Also: I know you’re not saying what I’m about to say but I want to make clear the points it’s not just the total volume of hours, because for example an hour a day is better than 7 hours a week compressed into a single day. There seems to be a minimum that is required DAILY for it to stick easily. I suspect an hour a day isn’t it. As in, it is not optimal. I think it is more than an hour a day and less than 8 hours a day.


Only a short comment (I’ve got to prepare a webinar today):

“bare minimum amount of practice”(xxdb)
Excellent point! We really should focus on this “minimum effective dose” (Tim Ferriss), not on the insane amount of hours that native speakers have under their

Another aspect are diminishing returns:
It usually takes considerably longer to go from 80-90 to 100 percent than it did previously from 0 to 80-90 percent. This is true not only for native-like pronunciation, but also for many other aspects of SLA (or the acquisition of any practical skill in general).
So, perfection shouldn’t be our goal as language enthusiasts…

Have a great day



Yeah, it is strange. The other day I was trying to say “get his spirits up” and I got completely blocked by “animarle”, to the point where I had to come up with another way to say it. :joy:

Lets just say that doing more will lead to more results. I don’t wanna say we should all do 8+ hours/day, almost nobody has time for that, but if we could it would probably be ideal.

I’m still of the thinking that we need to let the brain know this new problem is a problem that needs solving. I’ve heard that in the bodybuilding world your muscles respond best when you shock them into action, and I feel like it’s the same thing with the brain, which, after all, is also a muscle.

We’re lazy by nature, and the thing that most kicks us into action is survival. Tricking the brain into believing the new language is crucial for survival does something special, I’m convinced that’s when the magic kicks in.


There was an article on Medium I saw at one point about someone who showed up in a French café and out their mouth came Chinese without the author realizing it. I tried finding it again, but no luck.

This is actually why I do not worry about forgetting words. I don’t suddenly get imposter syndrome that I am not actually competent in my native language because I couldn’t remember a word (I get that for entirely different reasons :D). I am a human and forget things all the time. The muscle to switch between languages correctly is one that takes time to build and hone as well. When I forget or can’t find words I know I that I knew, I simply look them up and know this the process working as it should.

1 Like

Fluency is more like the speedometer than a number on the speedometer.

Or, even better, more like the road than a mile stone on the road.

After all, you can be a native English speaker and never have known that milestones once referred to, and still can, the literal stones placed along side the path to mark distances. There’s always more to learn on any good path ahead.


Interesting that you chose a “road” as a metaphor for fluency, gmeyer.

Every time I think about languages (natural or artificial ones), I think about networks. And once networks grow and become constantly adaptive networks of networks of networks…, ,we learners can only master parts of them (-> if the networks adapt, we have to adapt. Thus, learning never stops).

This reminds me of River deltas, and even more so of the ever-changing nature of the Internet:

So, we have language families of language families (Romance, Germanic, etc. → Indo-European), then individual languages → variants of individual languages (e.g., Australian, American, British, etc. English) → regional variants / all kinds of sociolects related to families, groups, professions, etc.

And all these constantly adaptive networks of networks are deeply anchored in time - with linguistic layers upon linguistic layers upon…



Perhaps “road” is the micro view while “network” is the macro view.

Even in navigating a part of an ever-changing network, there’s a next step forward to always take.

Related, I used to run, a lot. Any given day that I ran, I considered myself “a runner.” Any given day I didn’t run, I considered myself “not a runner.”

Perhaps “fluency” is a bit like that too. Really, the most interesting question is what one does on any given “today.”


Fluency is, though perhaps, when you can understand and use those words in all their “wishy-washy” definitions. :slight_smile:

I heard somewhere that we’re rather human doings than human beings. Sounds logical. “Being” itself defines not the most important part of what we are.

Well, the left image is from the Japanese part of the Internet.
Therefore → not “typing”, but “タイプ” :slight_smile:

1 Like

When I’m asked the question I reply with what languages I currently “speak” (have conversations in) regularly.

I feel like the language is mine once I’m completely mobile in it. No, I’m not illiterate, but I highly prize the listening aspect of language learning.

Everybody either drives to work, takes a train, or they’re retired or work from home and they walk/drive to the grocery store or fitness walk every day.

With toddlers, I need that time (driving and walking) to listen to my languages.

I don’t have time in the day to shut myself in a room for long periods of time and read books or watch videos and look for visual cues because my listening is spotty.

Once I’ve reached that magical tipping point where I have an array of comprehensible podcasts to choose from for earbud listening during a fitness walk or bluetooth during a drive, I’m good to go. That language isn’t going anywhere.

1 Like