1. Forum
  2. >
  3. Topic: Duolingo
  4. >
  5. Fluency Score


Fluency Score

We received a bunch of questions about Fluency Score, so we figured it was worth another post! Thanks to GermanSoy, hughcparker, daughterofAlbion, WildSage, Sammoreland, and many others who asked questions and provided follow ups about Fluency Scores in our Ask Duolingo series.

Read about Fluency Score in our help section HERE.

Read our previous Forum post about Fluency Score HERE.

Here are some fast facts:

  1. Fluency Score is our estimate of your fluency in the language you're learning. Our goal is to provide as close to an accurate assessment of real-world language ability as possible, so finishing your tree won't get you to 100%.

  2. With Duolingo alone, you can achieve a fluency as high as 50-60%.

  3. Your Fluency Score will increase over time as you learn more words and strengthen your skills, but it will decrease if you don't keep up your strength.

  4. Fluency Scores are currently only available for those of you learning Spanish, German, French, Portuguese, Italian, and English.

  5. We are currently working on new systems that will let us more closely monitor the alignment of our courses with CEFR levels.

What does my Fluency Score mean?

Fluency measures your mastery of the vocabulary and grammar of a language and your ability to understand and produce the language.

What is it based on?

Duolingo estimates your fluency based on your progress through the skill tree and the accuracy of your answers. Because learning a language requires repetition, we also take into account how regularly you practice. Without continued practice, your fluency score will decline, based on our model of how the human brain forgets over time.

I finished the tree! Why is my score so low?

With Duolingo, you can achieve a fluency as high as 50-60%, which is equivalent to Advanced proficiency. Our goal is for this estimate to be as accurate as possible, so finishing your tree won't get you to 100%. We're constantly working to improve how much Duolingo can teach you, and we hope that providing this insight into your progress will be both informative and encouraging. Seeking out native speakers to supplement your Duolingo learning is a great way to up your fluency, too!

Wait, what Fluency Score? I see nothing.

Ah, you must be learning a language outside of Spanish, German, French, Portuguese, Italian, and English. Unfortunately Fluency Scores are not yet available outside of the languages listed above. But we are working on it, along with a new system that will let us more closely monitor the alignment of our courses with CEFR levels.

Will you share the exact algorithm?

The boss says no. But we can tell you that it is calculated based on:

-what words you know

-how important those words are

-how well you know them

-how likely you are to forget them

Your Fluency Score will increase over time as you learn more words and strengthen your skills, but it will decrease if you don't keep up your strength.

November 15, 2016



Still, it's only within the Duolingo environment - not the real world, where lots of words not used by Duolingo exist. Fine if you want to call it fluency within Duolingo, but it has nothing to do with real-life fluency. Coming from a university environment, you should know that you need to provide footnotes to prove your case.

November 15, 2016


If they were claiming it was only mastery of what's within Duolingo, then 100% would be achievable. It caps at 50-60% because they are claiming it's real life fluency.

November 15, 2016


Well, no test of any kind is really real world, is it? A test is always an estimation and test taken at that particular time. Never real world. Only the real world is the real world. That's true for the tests on duolingo as well as those international ones that give you your A, B, C level. Actually I prefer duolingo to those international tests, if it comes to that. Because you generally only take the international tests once and be done. But duolingo, if you keep using it, is a continued learning and testing. Just some thoughts...

February 1, 2017


No. It should not be called fluency. It's misleading and it upsets people. Word strength would be a more appropriate name.

November 15, 2016


"With Duolingo, you can achieve a fluency as high as 50-60%, which is equivalent to Advanced proficiency."

Sorry mate, but that's hilarious!

I have finished several trees here and just when I finished them the fluency score showed 60-65% in most of them (and 74% in one of them).

It is clear that my "real" level at that time, was NOT higher than A2 in any of those languages. So "Advanced Proficiency" (when you reach 60-65%) is just a dream, not reality.

An A2 level would be closer to the truth in my opinion.

By the way, those online CEFR tests don't work either, because I usually get C1 in Itailan and French (for example) and I hardly can converse in those languages just yet.

November 15, 2016


Sorry mate, but that's hilarious!

Agree, agree, agree.

A2 is pretty accurate, and the spoken score might be even lower, since (real) speaking isn't really taught or tested on Duolingo.

Greetings from someone who is actually C1 and C2 proficient in a couple of L2 languages, and has finished seven trees here.

November 15, 2016


This purely speculation so feel free to down-vote this if you want to. I think the problem with the fluency score is that it only measures words and not sentences. Duolingo's sentences continue to be very short even in the lower parts of the tree. I am talking about trees like German or Spanish here, not about Norwegian, which of course does not even have a fluency score. If Norwegian or Swedish had a fluency score, it should be higher than that of German for instance. But since only the competence of remembering words is measured, you would reach just as high a fluency score in German and in Norwegian, even though the Norwegian tree teaches more complex sentence structures. It is possible I have misunderstood something here and I am willing to admit that I am wrong should someone provide the necessary evidence. :)

November 15, 2016


Yeah, if it is just based on words then a count (of words in long term memory) would be more useful. It's how many people think about language learning. 'Oh yeah, I know 2000 Russian words' etc. I am not saying it is the best way to think about language learning, but it gives a rough idea of where you are on your language learning journey, it's transparent, and it's easy to understand what you have to do to improve it.

November 15, 2016


Since starting French in spring and Spanish in summer, I stumble now in native langage English to find words...as if my brain were not sure what source to draw from. I recognize French and Spanish words but my spelling and grammar are plunges into patterns and not something I can sort out naturally and reproduce when needed. I feel tired after a long practice session. Imagine the fatigue of living in foreign country with a low proficiency of 67!!!

November 13, 2017


Quite frankly, you should just scrap this fluency meter. It's useless at best and misleading for some people and you know it.

November 15, 2016


I agree with you that the name should be scrapped, but think that a scoring system is very useful when the goal is to Gamify language learning. So - come on everyone - what would everyone on here recommend for a replacement term?

November 15, 2016


I like it but it might be called something else and a measure of grammar could help also. Duo is bound to improve due to the creative sources and people working on it. It can help on the road to being bilingual but not the same as bi-cultural at ALL. Uniquely, a bicultural person living around mono-cultural people is not going to enjoy being understood at a depth they might seek. My brilliant, fun foreign friends who speak at most registers without much accent and understand American culture, grew up speaking two languages in their home country, they miss being understood as thoroughly as they are making the effort to understand, I've suspected.
Native American speakers are bicultural, but I know a few so immersed in two cultures they do not seem to know they are bicultural in some of their understandings. I can't get to their depth in their culture at times tho they seem to understand mine very deeply usually. It is always a two-way effort and struggle. But two human beings are in the same boat in the same culture given the complexities--even in a marriage. The material is different. The mystery is always there. It's to always language and culture. I have had pets who understood me as well or better than family members do, who could read my subtle thoughts and do face-saving with the other pets and me even. Language is a container for thoughts/feelings mostly. Some huge transmissions occur without it when trust is formed. Again, that is part of communication. Language skills without that are null, really. Pure robot stuff. An animal with all these skills---that has soul...and we hardly ever discover our own.

November 13, 2017


This will never align in any way with CEFR as long as you produce decontextualised sentences and fail to explain the normal way of expressing something in the context of everyday social interaction rather than word by word 'translation'. Nor will it represent any kind of fluency - certainly not whilst the program's command of both languages is so poor. For example, if you happen to be Dutch looking at the English course, you will be told that 'Please stop' is a legitimate translation of 'Hou alsjeblieft op!' but not 'Stop, please'.

November 15, 2016


I think a useful and simple measure would be percentage gilded/ percentage complete. A simple fraction would be easy to display and easy to understand. The top would go down but the bottom would only go up. It could be plotted over time if they wanted to show learning stats. They could call it Course Mastery or something similiar and it would be much less ridiculous than a measure that goes down as your fluency goes up. which it often does as you complete the tree and use other resources.

November 16, 2016


Like it. Good recommendations. I did hear a podcast with the founder maybe a year ago, who said that the business model moving forward would be related to enable cheaper than market cost price of English language certification. I wonder if that is the reason they are sticking with the term "fluency".

November 16, 2016

Learn a language in just 5 minutes a day. For free.