Some of you may have noticed a new piece of flair on your homepage. For those of you who can't see it yet, this is what it looks like: This is exactly what it sounds like: our estimate of your fluency in the language you're learning. It is calculated based on what words you know, how important those words are, how well you know them, and how likely you are to forget them. It 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. Since our goal is for this estimate to be as accurate as possible, be aware that 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. Happy learning!
This is currently an A/B test, so only half of you will be able to see it. It is also only currently available for those of you learning Spanish, German, French, Portuguese, Italian, and English.
they are saying the tree alone does not get you to 100 percent. but we all know that. after the tree you go on learning new things at every single session. now am at level 23, 24 in a few days, and every single session is learning. not repetitive. it is very very clever, this thing.
Fluency is primarily a measure of performance - how quickly and accurately information can be transferred.
More specifically, language fluency is the ability of an individual to listen, speak, read or write in an acquired language.
Since the Duolingo web interface has very limited ways of measuring performance in language, at best the "fluency" indicator is more correctly a crude measure of vocabulary knowledge - a necessary, but not sufficient, prerequisite for fluency in a given language.
I have seen that and as others I am a bit puzzled about what this is supposed to mean :
- For me being fluent is about being able to speak the language, which is precisely the one thing Duolingo doesn’t teach at all, so I don’t see how it could possibly measure it
- What does being 50% fluent means? It is that you understand 50% of the words in everyday conversations? It depends a lot on how you are planning to use the language.
- It’s completely wrong, it doesn’t measure at all how much you are fluent but rather just which proportion of the tree you know. For instance my native language is French, yet if I start the EN -> FR tree I doubt that it will guess that I am 100% fluent in French. Saying something like "x % of the course completed" would be less misleading
- Not everyone is interested in fluency, so for those who only want to be able to passively understand the language or just be able to read it, there isn’t much point in trying to guess how fluent they are.
For me being fluent is about being able to speak the language, which is precisely the one thing Duolingo doesn’t teach at all, so I don’t see how it could possibly measure it
This seems to suggest a person who doesn't have the ability to speak will never be fluent in any language.
I'm happy to say that we're making this available to everyone. This has had a huge impact on how much people are learning with Duolingo as well as how many people are learning and how often they come back.
Some of you have expressed concern over the fact that this does not reflect their true fluency. We realize that this is an estimate and that it is not perfect. However, we've decided to label it as fluency because our ambition is to provide the best estimate of fluency possible, not just some measure of "course progress". We'll continue to improve the score over time and include more modes of assessment into the calculation. For now, I hope you enjoy what we have so far.
There already is one that allows you to hide it, "Duolingo: Toggle Fluency Percentage" listed at the wiki: http://duolingo.wikia.com/wiki/Duolingo_Userscripts
Edit: There is also another one here that removes it without adding a toggle: https://gist.github.com/anonymous/c872427f9fb2fcfd35d0
Since this has been deployed, maybe taking an extra step and showing it alongside our profile/flags. In the sentence discussion this would help identify experienced learners, and be a perfect way of implementing the "native" speaker feature. At least this is more valuable than a number denoting "level"/effort. So rather than native speaker there could be a "estimated fluency".
Maybe the progress quiz test could also be extended to offer a fluency tag. Perhaps this means that there any plans to show test center score in the duolingo?
Not necessarily, if the progress quiz score is used to get a more permanent fluency rating for "experts". For example, if some (not all) elements of the test center are incorporated into the progress quiz, so it becomes more difficult, it would then provide a perfect rating for those who want to show their fluency.
True, but my suggestion would be improving the progress quiz so it goes beyond simply what's taught in a course. No educational institution teaches everything a learner needs but it will always evaluate something more than they teach.
In fact proper students consult dictionaries, libraries, read books, and have many other forms of autonomous learning.
For my own personal language-learning needs, I find this unnecessary; however I can understand why others might want something like this.
I would much prefer that it was less-obtrusive, smaller perhaps, possibly further down the page/ sidebar. I find the Friends Leaderboard and Daily XP Count with Weekly Graph to be more useful, and would prefer they were closer to the top of the page than a shield with a number written on it.
Interesting - two thoughts in the light of comments so far
there is clearly some appetite for something that maps onto a wider language proficiency measurement framework - CEFR or whatever - I'd like to see some estimates of what ranges of fluency percentages map to A1, A2, B1 myself
people are expecting reverse trees to be able to measure their fluency in their native language. This is impossible from a testing point of view - the result of a test administered in another language is unlikely to exceed by much what you would score for fluency in that other language. Perhaps this should be made clearer to users?
Immersion helps to strengthen the static words learned list, so I believe it will to some extent.
However, it would be a lot more effective as a metric of fluency, if additional immersion words translated and upvoted were added to the list. I guess since not all languages and interfaces support Immersion, Duolingo can't do that for everyone, so they don't.
Hmm, this makes 3 developer updates in less than 2 months. I'm guessing the policy in Duolingo changed, perhaps each developer is now free to showcase their features in the forums :).
Anyway, I think people in these forums are overestimating their knowledge of linguistics. First of all, all Duolingo courses probably teach more than 3000 words in fact they have at least one probably exceeds 5000, as defined by :
A single distinct meaningful element of speech or writing, used with others (or sometimes alone) to form a sentence and typically shown with a space on either side when written or printed.
What people here and what Duolingo generally uses is to describe the concepts or elements they teach users are lexemes :
A basic lexical unit of a language consisting of
or several words, the elements of which do not separately convey the meaning of the whole.
technically, a lexeme is an abstract unit of morphological analysis in linguistics, that roughly corresponds to a set of forms taken by a single word.
The developers should probably make an effort to emphasize this fact in a tool-tip or help-text somewhere in the interface itself, to avoid more people reporting bugs with their alleged "fluency". Perhaps also make it clear what fluency means to Duolingo (in a tool-tip or something).
Duolingo courses probably teach more than 3000 words per tree?
I cannot confirm that after finishing 6 trees. Spanish: 1586 words German: 1710 words Italian: 1800 words French: 1935 words The Dutch to English and English to Dutch trees do not specify the number of words.
Yes, that's my point. Most people criticizing the word count and relating it to fluency in this thread seem oblivious to the fact that any given article or text of sufficient size will include a relatively small number of lexemes. Also that although native speakers may "know" a word they may not necessarily know what it means or even how to use it in a sentence or phrase.
Books may however include some names or "fictional" things that would be pretty useless to teach in a language learning course, and are better left to be learnt in the real world.
I didn't actually know that new vocabulary is added to it, but it makes my original statement even more true. If we account for the new words added and the synonyms (passive learning) shown in drop down lists (hints) but never really used in the course, then the word count increases by a lot.
Sorry, that is false advertising as far as I'm concerned. If you called it something other than 'fluency', such as "course mastery", then it might be a useful addition. But telling, for example, middle school students that with Duolingo they have become "x% fluent" in a language they are trying to learn can only be seen as intentionally deceptive.
I certainly appreciate the attempt to measure something around fluency, but think there is room for improvement. So far I have only seen a single comment that suggests to do something with the timed practice, and that is exactly where I measure my fluency when practicing with Duolingo. If you cannot complete the timed practice with at least 18 points or so out of 20, then you are either not typing fast enough or not translating fast or accurate enough. Duolingo could even measure the difference between these possibilities I guess. To be honest I had to improve on my typing speed first to be able to complete any timed practice, but at the moment I can finished timed practices in Dutch, German and English and these are exactly the languages that I am fluent in. I finished the Spanish, French and Italian trees as well, but still struggle to complete timed practices in time, therefore I am not fluent in those languages yet. The fact that I am fluent in Dutch, German and English is not because of Duolingo, but the fact that I have enough conversations in these languages. I don't expect to arrive at fluency for the other languages via Duolingo. I am however very thankful for the lessons Duolingo provided and with the levels I can achieve here. What I miss (or have missed) in Duolingo is the option to connect and have conversations with native speakers in an "intercambio" where participants change roles from student to teacher and vice versa. I am aware of the fact that there are platforms for that, but feel it should be facilitated by Duolingo on their own platform.
The fluency metric should be removed. If you take into account what fluency is, how quickly language evolves, and what you can do outside of Duolingo, you can easily conclude that it is unnecessary.
I think that the word "fluency" is misleading. Consider this situation: a scholar is going to a convention to talk about an scholarly text in a different language, but his luggage gets lost in the airport. He doesn't know enough slang to ask for help. But, a 13 year old who has focused only on what her peers might be saying would probably know plenty of slang and would most likely be able to ask for help. Who is do you think is more fluent now?
In addition, language evolves relatively quickly. Just think about English... Even though it is a Germanic language, many of its words are Latinate or of Greek origin. Even the word "language" is Latinate! Its common slang changes with every generation or new major technological advance. Then, compare British English to American English. Saying something is "a brave idea" in American English means the idea is good, while in British English means that it is idiotic. I think that this attempt to measure fluency numerically is a brave idea (British context) due to language evolution.
Also, you can do a lot of learning outside of Duolingo. After I finished the skill tree I took a break from Duolingo– this is my second time back in just under two months– and I started to read French news, listen to French music, watch French videos on YouTube, and I even wrote a few emails in French to my French teacher. All of these activities have ameliorated my French, but when I came back to Duolingo it says I am 6% fluent. Merely six percent! Yet I can speak, read, write, think, and dream in French far better and/or more often then I ever did before. This fact contradicts and possibly even subverts Duolingo's statistic.
Considering fluency, language evolution, and outside resources and activities, I believe that this feature should be removed or have a toggle to disable it.
P.S. I appreciate everyone's efforts to make Duolingo better and I hope you continue to add new features.
I don't understand all the negativity towards this. Can't you ignore it if you don't like it? I think if anyone can make a reasonable estimate of language proficiency from using that data of millions of people trying to learn that language, it would be a Luis von Ahn company. I'm really curious to see how good this "fluency meter" gets. I don't really expect it to be perfect ever, but I don't mind being a guinea pig for them to try and figure it out the best they can. I've taken a number of DoD proficiency tests and I have no doubt that Duolingo's measurements will be better than the government's before long.
I haven't really thought too much about this whole feature, but I'd anyway like to take a stab at guessing why it generates negativity:
Duolingo's spending time and resources on something that has the look of being scientific and measuring something, whereas it really doesn't. Now, to take these one at a time: Many Duolingo users are passionate about it, and would like nothing else than that those many ickle technical problems it tends to have would be worked out, and this seems like a distraction from that. Some of us are also obsessed with getting Duolingo to build a particular course (cough...Finnish...cough), and thus also possibly view this as tinkering. Finally, many here seem to be crazy about data and all things scientific. Something that looks all scientific, while -- quelle horreur -- seemingly spitting out random numbers, feels like an insult to those very people. Combine this with love for and loyalty towards the machine churning out this nonsense, and the cognitive dissonance becomes to much to ignore the wretched thing. And it's a pretty big shield, too.
I do not necessarily support any of the views theorized about in my tongue in cheek explanation above.
I do not seem to have a problem with this fluency badge that duolingo has devised, but it is kind of important to know how the score is derived and if it really isn't measuring "fluency," then, yes, it should probably be called something else. I think a good number of duolingo users will take the time to find out what, exactly, it is measuring, but as many or more will not and my concern for this type of duolingo user is that they then have an overly elevated notion of their true abilities.
Be that as it may, I found your tongue in cheek explanation above quite humorous, especially the subtle editorializing in parentheses and the inclusion of "quelle horreur." You have a way with words and I think I'll start following you on that basis alone.
Why the negativity? I think the answer to that question lies in the name "Fluency". Duolingo cannot make a sensible attempt to calculate our fluency in a language. It simply does not have sufficient information about our overall competence in the language to do so. If that had called this percentage "Duo Rating" or something original then it could have been welcomed as a new item of feedback based on recent performance in the course.
The question really should be - Why does Duolingo call this new measure "Fluency" when it must know that it has nothing to do with fluency? Of course they already call the Translation section "Immersion" when it has nothing at all to do with immersion, so this is not the first time it has deliberately given a false name to something on the site.
I think that's the crux of it.
"Fluency" is a loaded word when it comes to language learning, it is what most are striving for. Duolingo's current formula is giving a false estimate of fluency and that is misleading. particularly as it seems to currently be overestimating.
Now I think the options for Duolingo would be to describe it as something else (like your "duo rating" - simply a percentage of content learned would suffice). Alternatively Duolingo could have a fluency test function similar to those on other sites to determine CEFR levels. Users could take the test periodically and be given an fluency rating then.
The test sounds good, I look forward to it being added for the other languages.
With the fluency rating I think the problem is that on one hand it is giving the impression of monitoring real-world fluency - due to the labelling/name plus the fact it cannot be maxed at 100% through completing the course alone.
On the other it is pretty inaccurate and relates more to course content learned than actual fluency.
Now I'm sure Duolingo's thoughts when introducing this would be that the high looking percentages would be motivating, but I think this has backfired spectacularly. People are just seeing straight through the overestimates and this has the effect of them being quite DEmotivating. I think people want realism, not optimism.
Like Lrtward mentioned, you probably gained 1% (just an example) by learning 9 new words. But, in that interval, probably 20 words you had learned a long time back now lost strength to fall below the threshold and were enough to reduce the percentage by 2%. In such a case, you will see a net 1% drop.
keishacat- That hasn't happened to me. So, I don't know if it's a bug or if there are some other factors involved. Mine has gone up after a) walking my dog (probably a bug) and b) doing strengthening exercises (maybe another factor). My French rating hasn't gone up after doing new lessons, but it hasn't gone down as a result either unless there is some kind of delay on the rating, which is possible. My Spanish rating did go down 1%, but I assume it's because I haven't been doing review lessons as much.
I like it. And I know it doesn't really measure how good or bad my french is. It motivates me to keep working on my tree
It's like a level, but more based in reality and always changing and thus prodding the learner to do the homework. In that, this might prove even more successful than daily streaks and Lingots.
As well as a ceiling bellow the 100%, there should also be a certain percentage below which an experienced learner shouldn't fall, since there will always be some basic words that will stay in the memory.
PS: Fluency levels also lay a nice foundation for future features, especially social ones. There have been some suggestions for a pen-pal or live conversation feature, it's easy to see how a system could do pairings with high percantage/naturals + low percantage or same percantage for pen-palling.
Duo shouldn't claim to provide metrics that it clearly cannot measure. The "Fluency" measurement has NOTHING to do with fluency. Call it something else, please.
If it were measuring fluency in the target language, it wouldn't give me different numbers for the same language on different courses (my fluency in Spanish is either 31% or 5% depending which course I check). If it were measuring fluency in the target language, it would give me a better score for the languages I WASN'T practicing on Duo because I already know all the words in the course, instead of a worse score. Instead they end up in exactly reverse order (I speak English [21%] natively, Spanish [31%] at C1, and German [49%] hardly at all). And honestly, if it were measuring fluency in the target language it wouldn't give me nearly 50% in German, when I can still barely understand a native speaker and still struggle to assemble sentences to express my thoughts.
Look, I'm not saying it's a useless metric...apparently it's an indication of course strength, which might have some value (I'm still undecided, especially since there's no way to know what the maximum is). But CALL IT THAT. Don't call it something that it clearly isn't and clearly can't ever be.
Mine keeps bobbling up and down between 49% and 50% as I finish new sections and/or strengthen old sections. Sometimes when it hits 50% it asks of I want to share that on Facebook ... and sometimes it doesn't. Overall, I have to admit I am extremely puzzled by the behavior of this new fluency estimate.
I am by no means a fan is this new feature. It's basically your skill strength translated into a single figure, and the correlation between it and my own perceived mastery of the language is close to zero.
Also, the very word "fluency" is a bit off, because it only measures your progress on DuoLingo, and only using a website will never make you fluent, you need real-life immersion for that.
For a metric to work, it has to be clear in its definition, reliable and meaningful to the user. This one is not. No metric that bobs up and down during every single session is meaningful in any way. No metric that declines overnight is useful, particularly when a student has perhaps done more work in the previous day than ever before.
One minute I'm 55%, the next 56% then back to 55 and back to 56 then on and on and on. It's silly. It either needs to work or it needs be made to work. Or it needs to be dropped.
I think it's very misleading for people who are new to language-learning and not of much use to anyone else. Conversation is the best gauge of fluency. In Duolingo, the estimate of text you can read was more realistic and also a clearer indication of what you've actually learned.
i have now been made part of the a/b test, or it is allready a done deal. either way it is the stupidest thing i ever saw. The numbers don't mean anything. Apperently they are based on your tree, meaning that it states you as being more fluent in a language in which you have finished less of the tree (but have more golden bages) than in a language in which you have finished way more of the tree and are at a much higher level, but have practised every individual word less. That you don't use a word very often doesn't necessarily mean that you have forgotten its meaning. The count of how well you know a word is in itself quite shaky, but this can be forgiven because repeating a lesson every now and then is not a bad thing. The worst thin of all is that the useless fluency rate now replaces your level rating, which is much more useful because it reflects the amount of time and practise you spent on the language. Hence experience points is quite an adequate term. I really hope that duolingo comes to its senses and quickly reverses this move.
I started Duolingo about 2 years ago, first Spanish, then French, then English from each language. I completed all the trees in about 6 months, then continued intermittently at that point, keeping the French and Spanish trees gold. I also subscribed to Babbel, and supplemented their material with other online resources, books, and movies (unfortunately, no travel yet).
About six months ago, I abandoned Duolingo, because I no longer saw it as a rewarding use of my time. Three days ago, I returned, and engaged in a marathon refreshment of all the lessons (78 in French, 64 in Spanish), using the iPad app.
When I first reopened Duolingo, my "fluency" was rated at 38% in Spanish and 40% in French. After completing 122 Spanish and 140 French exercises, my "fluency" measures rose to 54% and 53%, respectively. I see no value in this unexplained "fluency" metric at all, since I don't believe I raised my true fluency very much beyond where I started, and there is no assessment of my fluency beyond the limited material encountered in Duolingo. I didn't keep an accurate record of the number of errors that I made, but they ranged between 0-3 per exercise, and probably averaged less than 1. At a minimum, the fluency shield needs to be given a title describing what it truly represents. Additionally, Duolingo could benefit by using some kind of test to people first entering or returning to the site, so they could assess their "fluency" by whatever algorithm Duolingo uses, and also so they wouldn't need to wade through unnecessary repetitive material to refresh their trees. There doesn't appear to be a way to test out of each level when doing revisions/refreshments, as there is when you first begin a course.
Still loving Duolingo a crazy amount, but my first impressions are that this is not a particularly motivating feature. I've completed my German tree and am working to regoldify it. I'm apparently at 46% if that helps - perhaps I would be more drawn to it more if I had a longer way to go towards maxing out and progress was quicker. I'd much, much rather that more vocabulary be injected into the tree, rather than having this rather meaningless metric (for me at least) up there, although I recognise that this isn't necessarily easy to do.
Mine seems to be stuck on 46%. I've taken my B2 exam in German and passed with 92%, and am now working towards my C1. I seem to have trouble with the same few words, and though my French and Spanish percentages keep going up, German has been stuck on 46% pretty much since I finished the tree about a week ago, even though I've been practicing every day to make sure everything stays at 100%.
I've been doing German daily on Duolingo for two and a half years (broke my streak on only three days), following years of study, but my "fluency" number for German won't go above 39% no matter what I do -- even multiple daily timed tests with 18-20 answers correct. (My other five completed trees all much higher "fluency.")
I guess it's good to know that someone is getting decent numbers in German. :-)
Granted it's not perfect but I love the feature. I've asked the fluency question quite often and there's really no standard metric (besides tests I don't want to pay for) to go by (that I've found at least). One of the most motivating things about Duolingo for me is the gamafication aspect. I'm very competitive by nature and this is one more thing for me to "win" at (which in turn helps improve my Spanish!). I think it will only get better in the future and could be an optional feature for people who like it and turned off for others. Keep up the great work Duolingo!!
I think the Fluency Shield should be clickable so that the user could get some more information. Why not let the user dig down about that percentage, and find out how to improve said fluency estimate? At least Duolingo could have provided a link to this thread explaining it. Only now (after quite a few weeks seeing it and not understanding why it went up or down) I searched the forums and got here. At least a link to this thread please, those who are still puzzled about it will appreciate it (I would have.) But ultimately, if the feature goes past test mode, a little digging down into how to improve that estimate might be even more useful.
One thing I can say is that I don't have any plans to share this Fluency % on LinkedIn, unless it was something above 75% at least. Anything less looks like a rather low score anywhere on the planet (it's a percentage after all.) I'm sure Doulingo might come to terms with the idea when realizing that maybe quite too few users are sharing said Fluency %.
I put mine on LinkedIn and it show "advanced" - for Spanish - I rock a 55% so I was loath to post that - for German I just started so it says "beginning" or something like that - it didn't give percentages - it does let me go right to my duolingo account though from the link , not sure if that's because I am the account holder - or everyone gets that -
So it shows a % here but a level-describing word and no % in LinkedIn? Talk about inconsistency. Gee, it seems this feature has quite the collection of user interface issues. I wonder why not showing said level-describing word here, or if not showing said strange % there, why choose to show it here alone, particularly with no level-estimate description? Just poor and absurd UI choices, plain and clear.
What is the maximum % you can achieve in spanish?
I think a more useful percentage (e.g., for posting to linked in) would be % course completed, or some notion of "level". As proud as i would be of finishing a full course, I am not going to put on my CV that i am "50% fluent" since it just looks bad! I would rather say I completed "100% of Course X" or completed "Level 2 of 3 in Y".
I finished my French tree awhile ago, I have not been updating it, it is far from golden, I do a lot of immersion. My French shield is at 56%. I am a native English speaker, but have not finished the English tree, and my English shield is only at 44%. Obviously the shields have limited value. However, I find them interesting and fun. Let's keep them!
I find it interesting that: my German is at level 4 and at 11%, my Spanish at level 3 is at 15%, my Italian (from English) level 12 - 48%, (from French) level 7 - 21%
... and I find this all makes sense somehow. Spanish is easier for me than German, I just have not been working on either of them, preferring to concentrate on French and Italian. Obviously, I am better at translating from English to Italian and back the other way, than from Italian and French (especially since sometimes I forget which languages I am doing an put in English translations instead of French ones!)
This fluency score does NOT encourage me. I do DL Spanish 5 or 6 times a week (strengthen skills only, finished tree long ago) and when I first saw my Fluency score it was about 65, which seemed about right. Now it has been 53 for weeks! Not going up at all. How can several more weeks of practice produce a much lower score? DIScouraging!
It might be better to call it something else, like a practice rate. Also, it's somewhat annoying if you're finished with your tree and focused on immersion to see a shield with a low number on your best language. If it were tied in with immersion, that might be helpful, but as it is, the fluency shield is a misnomer.
As much as I love Duolingo I have to say that I have doubts about the addition of this fluency measure. I do not see any logic in it at all so far and it is not very motivating. Out of curiosity I just took the English course for my native tongue (Dutch). I skipped the first checkpoints and this brought me to a 56% fluency score. I couldn't test out of the latter parts at once and tested out of them individually. Even though this brought me to the end of the three with all the elements golden, even though I had to do way more tests than I had to do to skip the first checkpoints and even though I had now finished the most difficult part of the course... the fluency rating remained at 56%. Now I might get that as I took shortcuts I did not use all available words, but then why did I get to 56% from the shortcuts that only took two tests? And why no change at all thereafter?
The same goes for reaching levels. Finishing the whole tree did not get me past level 10. For Spanish, I have used DuoLingo over the past years and have recently completed the tree. I am now at level 13 for Spanish and I am trying to boost my word rating by doing the exercises again. I get new XP points for this. To me it thus looks that if someone whose native language is Spanish would perfectly complete the tree would only get level 10 and 56% fluency rating, while someone who has completed the tree but keeps on redoing exercises can get up to level 25.
Are the levels just a representation of the amount of XP you gather? Is there an end to it? And please give at least some more details about the workings of the fluency rating, as now it just comes across as a silly measure that might even demotivate when given value.
It looks like this percentage is the "fluency" of the duolingo engine behind practice exercises. I don't understand why I am shown this percentage. This strange non-varying pure technical internal number only annoys and demotivates me. It only shows (me) how incredibly, disappointingly slow duolingo learning is.
For example, I am at the 17th level now. All my tree leaves are open and show 100% strength. However, the fluency percentage has changed only 1% (from 55 to 56%) after a whole month of daily practice. This means duolingo has taught me nothing new in a whole month! This is very disappointing!
Lot of discussion about how many words are required for fluency etc. Here's a random google-generated link that addresses the idea. It gives several different sources, and they almost all agree on fewer than 10,000 words for native fluency. There seems to be a common misunderstanding in this discussion between "fluent" and "educated". People with 8th grade educations (in America) are perfectly fluent in American English, despite the huge gap between the number of words they know and the number known by High School graduates, college graduates etc.
There appears to be inconsistency across languages. In German, I've completed the language tree and am at level 15, and have also taken the test and gotten 5 out of 5 (I studied in Germany for a year and am fairly proficient. Yet I only have 45% fluency. Meanwhile, in Spanish, I'm only half-way through the tree, at level 8, and show up at 47% fluency. I have almost six times as much "XP" for German as Spanish, but register as more fluent in Spanish. I get that Spanish is easier than German, but that's not enough to explain it. It just appears to be an inconsistent approach to doing fluency numbers.
Same problem here: German is my highest-level language in terms of XP (level 21), and I've done it daily for two and a half years with few errors. Fluency estimate is 39%. Spanish, however, is level 19 (I've also done that daily over the same time period, but with a few less timed tests), and I'm at 51%. The other four languages I practice here regularly range from 44% to 50%, but are at lower XP levels. I don't understand the logic at all.
So... I had been learning Italian for years before even starting with duolingo and I am up over half way in the tree on level 10 or something yetthe thingo still says I am only 1% fluent lol. Not that it reallmatters it is just anoying that whenever I use it it says congratulations. 1% every time and never updates.
That is interesting :-) I have also noticed that these percentages can be rather stable, but your case sounds extreme. Here is a test you could do: try what happens if you only strengthen your skills for a while without taking new lessons. My theory behind it is that the percentage increases when more words have been practiced often enough to be counted as "moved to long term memory". Maybe you are currently progressing too fast and everything is still regarded as "fresh". I would be interested to learn if it makes a difference or that the thing is simply broken.
Hey jero, you could be right, when I first started Spanish I was very careful to proceed slowly, but have everything answered correctly. I had an amazing 95% then I started working to get the tree done. My percentage went down to about 56%. BUT since then I have on occasion strengthened all the words. I mean when I opened my word list there was only 3 words that were not full strength, but I only increased to 58%. I am sure I am not 58% fluent, but I am certainly way more fluent than when I started. In addition I am a native speaker of English and can only reach 57% in English, that's what my English teacher told me, hahaha
After all this time, still wondering not only why in heaves this rather useless and frustrating shield is still being shown, also why I still have the absurd option to share it of all places exclusively on LinkedIn and nowhere else, and also why I actually still don't have the option to actually remove this absurd shield from my Duolingo experience. The length of this thread and the volume of negative feedback should have given some clues about all this. Is anyone in Duolingo paying any attention?
The idea of tracking fluency is good, but the implementation needs work.
So the Oxford dictionary defines fluency as
The ability to speak or write a particular foreign language easily and accurately
If someone says "fluency in German is essential" I know what that means, and I have a rough idea where I am in relation to that. I want to see that score accurately reflect the progress I am making as I learn the language.
That is what I expected you to be measuring. If you are measuring something that does not align with that definition, I can't trust it, and therefore it doesn't motivate me. When I see my score reduce as I learn more vocabulary, it de motivates me and is very annoying. I know I am more fluent now than I was a month ago when I started learning! I am now able to read simple books in German without having to look up every word , and I now understand the weather report and time checks, and get the gist of news reports on German radio whereas a month ago I could not at all. But my Duolingo fluency score has gone down!
I don't want you to remove the fluency score, I want you to fix it, please.
The fluency rating is not useful at all and is actually demotivating. I was in the 20s when I stopped doing the tree. I came back a few months later, completed the tree and dropped to 14%. Lately I have been keeping it golden every day and steadily dropping. I'm now down to 8%! At this rate I will be at 0 in another month.
It would be helpful for Duolingo to explain how they arrive at this number.
I found it demotivating too, so I did a bit of research and have since hidden it using a Greasemonkey script, http://blog.alexstew.com/original/scripts/userscripts/duolingo-toggle-fluency-percentage/. However, I wasn't able to follow the instructions given, thanks to an error in the URL for the script. After installing Greasemonkey you need to copy the URL given into your browser address bar, and then delete the trailing "?duo" before hitting the Enter button. Greasemonkey will then install the script, and add a button to your Duolingo home page that allows you to hide the fluency shield.
For what its worth, the issue may be related to the platform you are using for Duolingo. My wife was doing Spanish on her iPad for about a year and was disappointed in her lack of fluency progress, being stuck on 18%. I told her to switch to the computer and within a week, she was up to 33%. So the problem may be in the algorithm for pads and phones.
I'm not exactly sure what this is measuring. For me at 27%, it appears to be almost exactly the percentage of French words learned that are "Still Strong" and "Pretty Good". In which case, it is doesn't have much correlation with fluency.
Sorry, but I just don't need to strengthen words like French "neige" - last encountered a year ago - to know it means "snow". Come to that, I haven't even seen snow for a year, but I haven't forgotten how to use the English word either. Presumably my English fluency wouldn't rate much higher in the eyes of the Owl which, frankly, would be an insult.
Good luck with your studies if you think fluency is keeping a list of 1,000 Duolingo words golden, rather than widening your vocabulary towards the 10,000-20,000 words the average native speaker knows and uses.
Edit: I've just had a look at my reverse (English) tree - 23%. Joke!
Edit 2: Since I posted this 7 hours ago, my "fluency" in French has dropped to 26%. I was once told by a client that I have a brain like a sponge. I took that as a compliment at the time but, based on Duolingo's metric, perhaps he was comparing me to the life form!
This sounds definitely off for me too. My mother tongue is at 44%, The language I've been learning on Duolingo for 2 years is at 46% (finished tree, trying to keep it golden) And the one I started 2 months ago and where I haven't reached the second checkpoint is at 48%..
This is the exact opposite of my own impression of fluency in those languages
So several of us have the same problem. I think that the issue is technical- it would be nice to have a meaningful measurement that combines all activity (immersion, lessons, etc) and perhaps our background information as well? So that if a person can pass a speed proficiency test, perhaps they shouldn't be forced to perform meaningless drills if they are also participating in immersion as well. Otherwise we get bombarded with "encouragements" that are either meaningless or discouraging because they don't take all of our duolingo activities into account.
Similar for me. I've been learning/speaking German for over 20 years, but my Duo fluency won't go above 51%. In fact, it has gone DOWN to 50% a few times, then a day or two later, Duo gives me a big reward announcement when I climb "back up" to 51%. Meanwhile, Duo says my Italian fluency (which I just started on Duo a few months ago) is 49%.
yeah it's weird for me as well! My deutsch tree is only half complete but my portugês one is done but both show 46% fluency, although I speak german much better from things outside Duo so it's only not done yet because i havent sat down to bang em all out yet.
But yeah it's odd that they seem have no rhyme or reason to the % it shows
I've been puzzling over that German fluency number myself -- I first studied German forty years ago, took it in high school and college, and have yet to see DL's estimate hit 40%, despite dogged daily practice here for two years. But I passed that in Italian after only three months of study; my numbers for Swedish (where a fluency number was just introduced), Portuguese, French, and Spanish are all (like my Italian now), between 44% and 51%. Go figure.
I agree. I stopped keeping my skills gold after I realized there were many more ways to learn more vocabulary. For example, reading a book in a foreign language or watching a foreign TV show. This website called Babadum is also really fun for learning vocabulary. Why, even Immersion is good for expanding vocabulary. But, like you said, I don't need to practice a word everyday to know it. It's like how, even though I've never used the word "okapi" in months, I still know it. It's not like I'll randomly forget. Even simple exposure to words can increase memory. Strengthening skills is useful, but not needed. And Duolingo's "fluency" doesn't even measure listening or speaking.
Yeah. I really wish Duolingo would do a more anki/mnemosyne approach to spaced repetition - that is keying in on a scale from 1 to 5 how strongly you knew the previous question. There are some times where I want to practice a question again in a couple minutes to reinforce it - and those words don't show back up because I "strengthened it" already - and there are other times where it's an easy word (or concept) that I really don't ever need to see again but the program still insists on giving it to me. Over and over again.
Frankly, I think this fluency rating could be actively detrimental towards ever gaining actual fluency. I will not touch the Spanish for English speakers course because training yourself to immediately translate Spanish words into English instead of functioning in Spanish is the worst thing you can do if you want to be able to use the language one day. These courses work fine at a beginner level, but encouraging people to repeat the same lessons ad nauseum and telling them that this will help them achieve fluency, when what they should be doing is looking for ways to use the language independently, does them a disservice.
Multiple languages. =) I've been doing Spanish off the site, Portuguese on the site, and everything else has been fighting for scraps. Don't get me wrong; I think the program is great. I'm desperately waiting for Catalan because I know from Portuguese and Italian that I can go from zero to functional with it, but that's still a long way from fluent.
I agree. In the (hopefully soon) future where Duolingo has expanded at least some trees so that they could plausibly make you at least sort of fluent, this could be a useful feature. But we live in the present reality where Duolingo, for all its many fine qualities, doesn't even pretend to get anyone to true fluency. Against that backdrop, and without knowing what the top possible percentage in any given language course at any given time might be, I just don't see how this could be motivating.
I agree. I can't say I find the idea helpful at all. I've finished my French tree, and golded it up to 100% a couple of times but I'm not going to find doing that again that much fun any more. I am enjoying immersion and I have no further use for the tree (in fact I'm rather hoping that when I hit Tier 25 it will go away for ever). If DL doesn't know what I am doing in immersion I don't see how it can helpfully comment on my 'fluency' or otherwise.
the 80-20 rule... 20% of words crop up 80% of the time. There are a very small number of words needed to talk fluently. In French I am very aware my vocabulary is limited, but I am able to hold conversations without any trouble because I know those 20% of words super well. the first 1500 words are (if correctly picked) the most powerful. the next 1500 after that improve your fluency less, because they are words that are used less often. As an example, say you learn the word "big". It's not a spectacular word, there are better ones meaning the same thing, but if I learn "huge" "large" or "gigantic" I'm not bumping up my ability anywhere near as much as I did when I learnt that weak word "big". My point is duolingo shouldn't be able to say we are 100% fluent on just 1500 words, but equally I would imagine we can get above 50% fluent on that many.
I read a very useful site that estimated fluency % and linked to what you would likely be able to do and roughly how many words that indicated you may have in your vocabulary. I believe 80% was functionally going to allow you to read most newspaper articles with just a couple of words on a page being unknowns that you can guess, so if I get an 80% from duo I'll be sure to check! (and yes there were writing/listening/speaking things too, but they didn't stick in my head so well)
The french translation of this feature (I'm learning english) is better, I think: "Maîtrise d'Anglais".
At first, it looks wrong (a better translation would be: "Maîtrise de la langue Anglaise" or "Maîtrise de l'Anglais"), but you can understand "Anglais" as "cours d'Anglais de Duolingo" (Duolingo's english course). So, this feature tells you how good you are in Duolingo, rather than your true fluency in real life.
If time is money, I have spent an awful amount of money on Duolingo. The point is that Duolingo is good, and could be great, but by adding features like this, they're taking one step further away from being great. Do we not have the right to voice our opinions? Many companies spend good money to get feedback on their products, and we're offering it for free.
No, even if it was a paid service it would not be immune to making steps in the wrong direction. The point is not about the direction, it is about the overwhelming negativity and outrage about a shield on a free service. If that is worst thing that is going to happen to you, then you are in pretty good shape.
Duo aint free folks, that is unless your time is worth nothing. so here's the model according to the ted ex video. They provide an opportunity to learn in the form of this easy peasy technology. It's good, its fun and it kinda works for awhile. You learn a language. In exchange you translate documents (eventually) for free, which they would otherwise have to pay a translator to do. So we get something and they get something. Win win. The only problem is that eventually, the primary learning vehicle (the tree) eventually gets boring and stops becoming of any value. (Im not there yet but from this long list of concerns it seems that some are.) So the obvious solution is just to keep adding material for advanced users and let users phase out the easy parts of the tree as they please. That way the exchange remains balanced. Users learn more of the language and continue to stick around to do the translations. Again , win win, but at a higher level. Bag the fluency shield and offer psychological rewards with course levels. Don't ever think that duo is free.
@ airelibre - yes, we do have a right to voice our opinion. That is exactly what I did :). It's not just cirtics that are allowed to voice their opinion... In my opinion, I'm happy with what DL offers, and it is free. I don't find the shield terrible and if I didn't like it I would ignore it and focus on the face that I am learning a new language.
Ever since I completed the French course tree (this happened on August 18th, 2015), the fluency indicator of the language has remained stuck at the same level: 59%. Absolutely no change has occurred with this number for months now. Please, note that I have been continously using Duolingo, and strengthening my bars, for more than a hundred days now (I'm currently in an offensive of 105 days). Even so, nothing happens. My fluency indicator is stuck at 59%.
You are roughly at what I expect to be the maximum percentage that DuoLingo will give you for your fluency. In a way it makes sense, because via Duolingo you can achieve a B1 level, whereas to be fluent you should be at the C1 or C2 level. You might be at a level higher than B1, but this is not something you can prove by using DuoLingo. I have never seen a fluency percentage above 60% in any of the languages I study here, despite the fact that I make an effort to keep my tress golden.
Hum, this one is a very nice insight. So, it's likely that the fluency indicator algorithm will not permit percentage levels higher than this number: 60%. Very interesting. Before reading your comment, I thought the highest possible percentage would have to be 100%. But your theory makes perfect sense. Maybe the developpers designed the fluency indicator in a way that the 100% level can't be reached now. This could be due to the fact that they simply do not have a platform mature enough to measure higher levels of proficiency/fluency in the various languages they have here. Very interesting. Thanks, Jero.
I don't think it is true that the maximum fluency you can reach in Duolingo is 60%. I hit 62% Fluency in French just shortly before completing level 17, and the first time I noticed the Fkuency indicator, before I completed the tree, it was at 58% -maybe back at level 12 or 13? The difference, I think, is that I was using the Duolingo iOS app to review my French which was formally learned in a US public junior high school and high school decades ago. My guess is that fluency is based on familiarity and correct usage of a basic vocabularity for an educated speaker of the language. But the only way the app can assess this is to test you on progressively more complex language usage, and have you continue to get this right. It can't assess what it doesn't test. And the only way I've found to increase fluency is to always let the app choose the items to strengthen and test. Progress is slow, but you do advance. And yes, there are a lot of new colloquial expressions that it comes up with to learn in the process.
My prevoius comment is based on my personal experience. By the way, I have also noticed that a fluency percentage can go down despite the fact you are levelling up. Did you complete your French tree at 62%? I have competed seven trees so far, of which five have a fluency percentage (it is not available for English to Dutch and vice versa). I am for example fluent in German, but my fluency percentage has never been 50% or higher for a finished and golden tree. Recently a lot of new lessons were added to the German tree, so maybe when I finish those the percentage will go towards 60% (which probably is not a hard limit, but my guestimate)
I 100% agree. I have been constantly having to start from the top again learning things like "Un garçon" over and over, while I obviously am able to do this while practicing the bottom portion of the lessons over again.
Would anyone be able to recommend a way to start expanding out into those other words and additional fluency, as I sort of feel limited going through the same content day in and day out. (Not implying I'm perfect at it yet, but some variety would be nice.)
My suggestion is to do it every day, for very short bursts. One lesson in the morning while you are waiting for something (anything), then any time you are waiting in line. It really makes a difference, and the more you do it, the more fun it becomes. I know that if I don't practice, the practice becomes annoying and painful. When I practice every day, as I have been for the past 1163 days, it's not annoying at all. Bonne chance!
What would be better: an estimate of how long it will take to become fluent if the student continues at their current pace. This would obviously not be based on a flat number of hours, but rather a ratio of average time spent learning each day and average words learned each day, and the accuracy of the student to determine retention rate. Of course, if a student can only obtain 50% fluency through Duolingo (assuming they are using no other means of linguistic education) then Duolingo can only award 50% fluency in any language, including the student's mother tongue. However, an incredibly simple method for improving fluency past 50% would be to connect users with different mother tongues to practice speaking the language, and give them "challenge words" to expand their vocabulary each time that they connect with an outside user. The 'mother tongue' user can then rate their fluency.
Bonjour de Sinuapa et merci de votre question. Je suis disconsolate de vous informer que mon niveau de " fluency " est en ce moment à 13% !
I haven't been actively using Duolingo since I lost my streak in November, not out of pique, but just because it seemed a good point to give it a rest.
As you can see, the Duolingo learning algorithms are so poor, that according to Duo's own calculations, I have forgotten 50% of everything I learnt in just three months. Still, at least I am now as fluent in French as in my native English. Result!
Thanks Owl - are you sure you are not a Great Tit for promoting such demotivating clap-trap? Those that might be unsure about their abilities in a foreign language - that they may not have had the opportunity to speak yet - are not being sent a good message.
SussexSoleil, Merci pour ta réponse. Your information is quite important, especially if Duolingo updates Facebook/LinkedIn automatically.
Apparently, the algorithm employs some kind of decay counter (like in radioactivity), with the half-life of the fluency level unknown to us.
But, really, it is not that important as long as we know ourselves where we stand with the language. Personally, I'm finding that I can understand a good number of phrases and the main topic when I watch France24, for example, or listen to songs in my French channel in Pandora.
Thanks for your input. Much appreciated.
Surely so long as people understand that this is a quick and dirty metric then it's fine. It will clearly miss the nuance of how close your ability is to native language. Obviously a native speaker or competent translator will be able to assess you nearly immediately to see whether your language was appropriate for buying a sandwich, or as a white collar professional amongst native speakers. But on the other hand, with the metric I could quickly tell if you know some or a lot of french from the % fluency.
I don't have the feature so I don't have anything to comment on regarding how Duo rates my fluency. I DO want to say THANK YOU for announcing this feature and explaining it to us, instead of just springing it on us out of the blue and leaving us to wonder where the number comes from. You get an A+ and a gold star in communications for this one, Duolingo!
Haha I just noticed these! Awesome! My Spanish is at 51% and my French is at 60%. This seems fairly accurate to me (I consider my French a good deal stronger than my Spanish, but I've barely done any French immersion so I'm not surprised by the lower score.)
I think it's neat :D I can't wait to see how it shapes up.
Positive: Hey David, thanks a LOT for announcing it. You guys do a tremendous job and I really appreciate it! I hope that you will continue writing such posts when you introduce new features.
Negative: I dislike this A/B test for a couple of reasons:
- You say one can never reach 100%. This is extremely demotivating and a mere no-go.
- It shows the Duolingo fluency. If somebody wants to know how fluent he really is, this measurement shows nothing concrete.
- To combine these two statements, you basicly say "Hey, you can learn a certain amount of words that are obviously not sufficient enough to reach any fluency in any language - and you will even never be able to fully learn all of them, but we still call this "measured fluency"." That's, again, an absolute no-go and simply not fair.
My solution: I know what Duolingos long term goal is (from the summit). If you really want to go with this fluency batch, then
- first agree on what one word is (for instance "I" is one word, "am" is a second word. Are "You are" two new words or one new word since "are" is only another form of "to be"?). If you agreed on that, you have a basis for your fluency measure.
- Depending on what basis you choose, there are estimations that it takes roughly 6,000 - 20,000 word in order to reach C2 fluency (if you count ALL forms of a word, meaning all tenses, numerous, genders etc as merely one word combined, approx. 6,000 words may be enough, if not, 20,000 would be a good measure if you count all different forms as one new word each a.s.o.)
- Then you agree on quotient: Learnt words on Duolingo/Word to reach a certain level. Example: You agreed on that one needs 6,000 words to a C2 level, then your level borders could be 1,000 words for reaching A1, 2,000 words for reaching A2 etc. And if a course teaches 3,100 words in total, your profile page shows "A1" if you reached 1,322 words, A2 for 2,655 words and B1 for 3,050 words. That would be far more accurate and learners would realistically know how good their comprehension skills really are.
Thanks for reading that, hope you'll discuss it with your group.
Just a thought: Wouldn't the number of words and therefore the limits vary a lot between different languages? I am thinking mainly of agglutinative languages such as Hungarian and Finnish, in which a gazillion different endings mean that there are probably fewer (base) words in total when these endings are accounted for. Like in the myth about the number of Inuit words for snow...
This would mean that the limits shouldn't be the same for every language...
PS. I wholeheartedly join you in thanking a_david for posting this information in the first place!!!
Wouldn't the number of words and therefore the limits vary a lot between different languages? Exactly, this is individual for each language due to different grammar systems, the number of synonyms (the English is among the top 3 languages with the most words; the largest English dictionnary shows more than 1,000,000 entries), the number of tenses (Ancient Greek has more tenses than German) etc. - So they had to individually agree on a specific word count for each language =)
>> Depending on what basis you choose, there are estimations that it takes roughly 6,000 - 20,000 word in order to reach C2 fluency
Fluency is not only breadth of vocabulary, but also the ability to understand and generate speech "on the fly" without translating back and forth in your head, to use verb tenses correctly, and handle idioms and colloquial speech.
A simple word count solution is, I think, too simplistic. I suspect that Duo is tracking not only whether or not you know the word for "hair" when given a set of flashcards to pick from, but also thing slike whether or not you know to put the adjective before or after the noun, and whether or not to make the adjective plural, and whether or not you need to adjust gender or case when you use the word in a sentence.
Sure, but since Duolingo does not provide services to improve your speaking or text writing abilities, I thought it was sure the I talk about text comprehension and translating sentences (because that's what you do on Duolingo, right?). If Duolingo had more services to improve all other language abilities such as speaking, the whole starting point of this discussion would be different.
If we concentrate on text comprehension and the ability to translate sentences (again, that's what you do on Duolingo), a word count does correlate to your language skill level - if you agree on a solid way how to calculate that (unless you know 10 synonyms for a world like "unequivocal" but not words like "newspaper" or "bus", but nobody learns vocabulary that way I assume...). That's what I'm talking about.
It's actually simplistic. As much as Duolingo. I am not sure if I am achieving the skills I would need to speak those languages and to communicate with a foreign person in their own language, but I am learning a lot of words and some sort of idioms and common sentences and stuff, and I am having real fun in the meanwhile. I believe in a certain point, with only vocabulary, and a bit of grammar, you can forge great ideas in other than your modern language. I speak Spanish myself, and I learn English by myself (hence the errors you noticed already), before I met Duolingo. It's a tremendous tool.
Don't forget that it's not just the number of words that one should know, but also how important they are (i.e. how often you will encounter them in day-to-day use). This is the main reason why I use Memrise because there are courses there where you are taught the x number of the most common words (based on various sources like opensubtitles' word frequency database or wikitionary's frequency lists) in a given language.
This is an interesting thread as, until this morning, I was not aware of the fluency shield. Duo said many moons ago that there was no real estate available on the home page for the progress bar which told us how far we had to go to the next level but now they find room for this fluency shield which IMO is a total waste of space, or it could at least have been tagged onto the end of the progress data in the profile.
While it could be useful to know the level of fluency we have achieved if it was based on a known or accepted measure such as CEFR but to just give a percentage of a very vague measure seems a bit pointless for an academically respected site.
We are told that we are unlikely to reach 100% so if I assume that this is equivalent to a C2 level then 50% or 60% could be approximately a B1 level, assuming that the CEFR is a linear scale. Now that would be worth bragging about.
I think what is (was?) said after the lesson is/was "you can now read X% of an article in language Y", which is a whole different thing that being able to speak that language. While this figure has also been generally thought of as inflated, it's at least easier to measure, and part of the "inflation" is explainable by the fact that Duolingo knows you understand every single "es", "y", "no" and so forth in a given article, although you wouldn't necessarily understand any of the key words that actually make you understand the content of it.
Yes. The last messages I got close to finishing trees (and also after it, I think) was that I would be able to read 50 +/- 3 percent of texts. I guess, it's pretty much the same this badge shows. Otherwise I have no idea how to measure "fluency". My guess: the A/B test will show that the badge has zero effect (shocker!). Although it might be eye-candy for new users and a way to keep them, until they realize it means nothing. If this was the intention, I wouldn't be surprised.
They should bring back the bars and stuff — perhaps as options — people add on their own via scripts anyway. But this doesn't look (and presumably) sound as fancy as a "fluency badge".
Personally, I love having the ability to quantify my learning. Language can be so subjective at times, to the point of sapping some enthusiasm. Having a number to look at, and to see it increase, and to strive to make it increase, will surely be another motivational tool for me. I trust that its accuracy will steadily and consistently improve over time. Great idea!
If this fluency shield is to succeed, way more skills in ALL language trees need to be added. In addition, grammar explanations need to be expanded and added. For example, in the French skill tree, there is no explanation of the pattern/paradigm for constructing future and past tenses, for combining past tenses together, for combining conditional and future, etc. Adding those would really help learners. Having said that, I think the fluency shield has a lot of potential, but the current "finished" skill trees might need to be revamped to keep up.
My two cents on the issue:
I found it to be very a very nice surprise to get the value. Which leads met the thought that it should perhaps only be displayed to the user who has once passed say 30% progress, N experience or K active days. To tone down the negative input. Also some kind of feature for measuring returning users, similar to what blizzard does with StarCraft.
I'd prefer a number, picture or animation instead of a percentage. I'd guess ( confirmed by the discussion above), that there are too many users who'd get stuck on the percentage itself. Perhaps a scale towards fluency. Kind a like the classical picture from the old Egyptian painting of the judgement. (Heart vs a feather). Or perhaps a see-saw that adds a lot of objects to the opposing side you're on. You got good artists, anything is possible.
Thank you for a awesome site.
While i realise this feature is at the start and might change significantly i do have some dilemmas: 1. My german is at 39% which is sensible cause i am struggling with german even tho i finished the tree but my french is at 52% even though i havent finished the tree and many of my badges are way further from gold than on the german tree... so that seems like a calculation contradiction to me 2. How exactly am i suppose to use this information to better myself!?
I've been floating between 56% and 57% in French. I finish brushing up on one category and get 19/20 on the timed portion, and my percentage jumps up to 57%. Get a 20/20 on the next one, and it's back down to 56%. It hasn't changed beyond that, and doesn't do much motivating. I'm not a fan.
I have the same question that you KlassyShelly. I have finished the German / Spanish tree twice (all gold now) and I am 250 points short of level 23 but I only yesterday moved from 48 to 49% (there is no immersion in German to Spanish). I found this forum trying to understand the correlations. We stay in the dark.
For me it is the same, 47% for months with a golden tree. I continuously try to get all the accents and capitals correct where needed, but it does not seem to make a difference. For Spanish, Italian, French and Portuguese the golden trees are bouncing around the 56-57% mark.
For what it's worth, the accents aren't relevant. I'm careful to use correct diacritics in Spanish and German, and usually in Italian, but rarely in French or Portuguese, and only intermittently in Swedish (sometimes I can't type fast enough to get all the umlauts in!). Doesn't affect the fluency scores, judging by where mine fall for those six languages.
So happy to be apart of the half that can see percentage. Also I am quite pleased with the transparency here, especially in being honest that just completing your tree won't get you 100% fluency, and you're probably working to churn out enough material to be able to claim someone has 100% fluency, or at least maybe 95%. Keep up the good work :)
I find it incredibly motivating to know that the hours I'm putting in are actually amounting to something quantifiable (Albeit maybe not completely accurate). Since moving to Spain last week, I have increased my language by 7% which helps me feel much more confident and comfortable. And anyway, if they're bringing this into schools to teach children how to learn languages, it obviously has a great deal of merit. Yes, you might not use the word 'bear' on your average summer holiday but Duolingo is offering a much more wellrounded language learning opportunity. For FREE, by the way. People who are complaining about it, whatever, get a guidebook. I think this is great.
When will they add fluency levels for other languages such as Ukrainian? I'm currently learning 6 languages from this app but I'd like to see this. Also, it should explain how it measures the fluecny as these are not all everyday words I would use if I went to the country, will you be adding new words continuously to the languages?
I pretty much think it's pointless as when it came out it said I was 27% and I'd only completed half my tree. Now that it's completed and fully gold it's worked itself down to 17% it makes no sense whatsoever. Not a useful addition, I'd prefer more vocabulary or more bonus lessons.
I am surprisingly motivated by this metric. Even though I know it can't be perfectly accurate and there are many other factors and measurements of progress to care about, I love seeing this number go up! Further, I absolutely hate the idea of it slipping further back, so I work hard to keep it steady. I honestly didn't thinknow this would matter to me very much, but I love it!
I spent 650 + days in a row practicing French on duolingo's iOS and never got above 38% fluency (and sometimes dropped down to 32%). Three weeks ago I switched to only using the website - even on my iOS devices - and moved the fluency rating up to 60%. On the website, lessons seem both more challenging and immersive - plus explanations on grammar, tests, etc.. The only aspect missing from the OS version is repeating words and sentences out loud.
Does the duolingo team have plans to expand the iOS app to match the robust quality of the OS version?
I finished the Spanish Tree several months ago at 57% and have been doing 100 points per day to stay refreshed. The 57% has never changed and I suspect that is as high as it will go unless Doulingo should add more vocabulary to the Spanish tree. Just curious as to whether or not anyone else who has finished the Spanish tree has a different percentage.
Similar experiences here. I have a few golden trees of which I can share the percentages:
- Spanish from English 56%;
- English from Spanish 57%;
- French from English 59%;
- German from English 49%;
- Italian from English 54% (not even close to golden now, but I can recall it being 60% or higher)
A completed tree can drop down in fluence to as low as 30%, again a few examples:
- English from German 32%;
- English from French 30%;
- Portuguese from English 44%.
I'm in the same situation; finished my Spanish tree a while back, 57% fluency. I've been doing a few exercises every day for months without the fluency changing. I've also been doing Wlingua, finishing A2 soon. And then today, after finishing three timed exercises on duolingo with only one miss, it went to 58%. This value seems accurate for my reading comprehension and writing, but my speaking abilities are weak. Which makes sense, since duolingo can't converse with me. Yet.
You can't always improve, though. The limit is somewhere in the high 60s, depending on the course.
Even that figure is blatant fantasy, though, for a course teaching you about 1,500 words -- that's nowhere near more than half the way to being a fluent user.
Duolingo is excellent at teaching you the basics of a language, and that's what it does: teaches you the basics.
The thing is, the algorithm is not accurate. I started Spanish 2 years ago and got halfway to the tree in a couple of months, reaching about 45% fluency. Then I stopped until this year in April, when I completed the tree and reinforced every skill until they got all gold. While I was doing this, my fluency kept dropping, until it reached 11%. (?) The annoying part is that Duolingo keeps asking you to share your fluency on LinkedIn. Let's share something else more meaningful on LinkedIn.
What really boosted me was starting from the beginning and redoing all the lessons. After a while it will go up almost every day. Seems like the algorithms noticed how much you have achieved over time and they are including the 2 year absence. Maybe you need to start a new Spanish account :(
What happen when you cancel? Do you lose your lingots? What happens to your XP's? I am up to level 16 in Spanish so I am not sure that I want to start over. I do not care about fluency estimates that much anyways - it's just slightly depressing.
I took a placement test from Instituto Cervantes and I got B1 - which is pretty good for having only studied it on Duolingo :)
I think I might have scored a new record today with my fluency in English from Romanian. I got an incredible 75%!!!! For years I was stuck at 58-60, but today I noticed this high percentage and I'm so thrilled! Even though I know for sure my fluency in English is around 90 something percent for real, as I have been speaking it for more than 20 years now :))
My French fluency was at 7%...... and it has been steadily going DOWN every day since I started practicing again. My level of proficiency has absolutely no effect on it. Today, for example, I scored quite a few perfects on timed tests and compared to yesterday, I was much less reliant on hints. My fluency, however went down from 3 to 2%. I am sorry, but something is a bit off here....
Congratulations "dercosta"! By the way, I just do not believe that Duolingo can legally promise or guarantee that someone is 100% percent fluent. Such a representation also does not seem wise as a matter of pedagogical psychology. You should feel appropriately gratified - do not feel bad. You have "succeeded" in English in my book!
While I agree that this metric is probably not going to be able to judge "real" fluency with 100% accuracy, I do think it's a useful bit of encouragement, if only to be taken with a grain of salt. The real test of fluency (in my opinion, and I am not a linguist) is making it through a conversation with a high amount of comprehension.
"Native speaker" fluency is not possible with any class, program, or book--- again, my opinion.
I kind of like it, because anything that motivates me to push myself harder is a good thing.
I guess I would prefer to see French in the testing centre because I'd like to know where I am on the DEFL/DALF scale. My best guess would be DALF C1, but I'm not sure what percentage that would add up to.
The only thing that seems a little off is that I'm rated 62% in French, which is a language I've been using every day for basic communications, vs. 40% in German which I can't converse in and only know through Duolingo.
I know 802 words and alla my lessons are stengthen (don't know how to right it sorry) and right now I am almost really close to finish determiners... And my fluency level is 16%.... And my friend which does Spanish (I am doing Italian) knows 532 words and again she has everything strengthen (but I do not remember the lesson she is right now only how words she knows) and her fluency level is 30%... How is that even possible?
The calculation of the fluency percentage is still a mystery to me, but it might be the case that the number of errors you make during practice has an influence on the calculated percentage. A thing to watch is how many questions you need to strengthen a skill. Without mistakes it should take 17 questions. With a lot of mistakes it might take you 30 to 40 questions before you finish an exercise. In the end the skill will be golden, but I expect that there should be a difference in the calculated fluence if you made many mistakes. Probably aslo the warnings for incorrect spelling or missing accents may influence the calculation. My German is much better than my Italian, but my completely finished and golden tree in German is at 46% whereas the Italian, French and Spanish trees are around 55% to 57%. One of the differences is that I do timed pratices on the German tree, because I am fluent, but I don't use the special characters like ß ä ö and ü during the timed practice. This means I get warnings all the time, but the answers are still correct. In Italian I don't do timed practices and take the time to get the spelling correct including the accents like à è and ò. Maybe you can figure out if it has an effect on your fluency when you try to avoid errors and warning.
80% of what we say only requires the first 1000 most common words. Once you learn 200, you are looking at 90%... Of course, the percentage increase DOES slow down as you learn more words. Correct me if I'm wrong...
I wonder how the A/B test is performed, though. I mean, I don't like that fluency shield at all. It doesn't seem accurate nor very useful to me. Nevertheless I won't stop using the website. How is the satisfaction on this new functionality measured ? I'm just curious.
Something else I have thought about: If finishing the tree doesn't get you to 100%, well, how far does it get you? Or does it get you up to a fixed max % at all, let's say 70 or 80 or what? What's the biggest % anyone could get? Why not explain that?
By the way #1: is there an upper bound to what this % can be? Why isn't Duolingo more precise about it? Why the lack of transparency? I wonder if there are didactic reasons, or copyright/patent issues involved in all the lack of transparency related to this %.
By the way #2: I would also appreciate if user comments here clearly showed which comments come from Duolingo employees/developers. I've seen Moderators clearly highlighted, but not sure if employees are clearly identified. (I mean, moderators might not necessarily be employees, and employees might not necessarily be all clearly identified.) Maybe it's just me but I find an issue not knowing how biased a post might be because of...
Obvious lack of transparency is a sure way to foster mistrust, you know?
Will our points decrease the less we practise? My French teacher monitors our points for grades, I have 1040 points, if I practise less will it decrease just because we don't use the app when we have enough points for an A? My email is firstname.lastname@example.org
I think this is discriminating, because not everybody knows English as their first language. I simply had to put English because that's the language I know the best from all the given ones. So, my opinion is that you should give up a bit, because a lot of people who are in my situation lose points because there are some strict rules when translating etc. I don't think of (?) some rough mistakes, but of spelling errors and similar. It's not fair that native speakers of the given languages could get better results just based on their ,,advantage" to be born in those language spoken areas, while the others are constantly annoyed by some silly mistakes and taken back by them.
I am wondering just what I need to do to get the fluency percentage to rise? How does it measure things? Does a typo count against it, even if it's in the translation? Does an accent missing count? Does lack of punctuation count? All of these? How much for each "mistake"? Does it know if I'm looking up the word that's defined in the sentence? Does that count against me?
I am not sure what algorithm is used to measure fluency, but the other day mine went from 53% to 51% right after I strengthened my very first Basics 1 skill which has been non-golden for some time. Before I thought that fluency was directly proportional to how much of your skills you keep golden, but apparently the algorithm is more sophisticated and sometimes strange results like the one above happen.
What a thread! Just to leave my 2 cents. I very recently joined Duolingo, more out of curiosity than anything else. Having moved to Guatemala 3 years ago, I decided to start with Spanish as a "test" and to see if it could help me improve my Ser/Estar problem. I took the initial test and it brought me to level 10, leaving only a few courses in the tree to work on. Most of these courses were also cleared by testing it out. I work primarily in Spanish and only seldomly have to ask my colleagues, family in law or clients to repeat or translate anything.
Nonetheless, the badge shows a 57% fluency. To be honest, it's not a motivational tool for me, on the contrary!
Most of my 'errors' are either because I don't translate word by word but rather translate the overall meaning (ie. saying 'met' instead of the expected 'had a meeting') or because Guatemalan Spanish has some slight variations from the traditional Castellano.
Bottom line, I think it should be optional for those who want it as it could be a counter-productive for many other users, driving them away from the site instead.
I really believe 55 to 60 % is about all you can expect from duolingo! I believe if you download Synergy Spanish you will be amazed how much that takes from what Duo has taught you and increases your fluency! I continue to do 4 or 5 lessons daily of duo/Spanish and about 20 to 30 minutes of Synergy and it is really helping! While the app is not free there are a lot of free add ons they constantly send me. There comes a time when one needs to graduate to the next grade level. Eighth grade was fun but now it is time for high school
What is the highest fluency level that can be achieved for Duolingo Spanish? Currently, I have a 55%. Is it possible to make my fluency level in Spanish higher than 55%? Can the fluency level change after the tree is completed? For example, my fluency level hasn't changed since I completed the tree; I continue to practice on a daily basis.
Hip hip hooray. Duolingo have done something to ensure fluency percentage reflects ones fluency percentage. I was getting deeply depressed watching it fall from 60% to 50% over 5 levels (as I was sure I became more fluent) ..then today, over twenty lessons I went from 50% to 59%. At last!
It's garbage. I finished my tree and tested as a B2 and my fluency was stuck at 1%. So I made a new account, tested in at 45% and then with each new gold I went down, I'm now at 18% on that account. It does nothing. It measures nothing. I like to do timed practice and see how many right answers I can get in 30 seconds, I get 20 most times, if I make a typo my fluency goes down. I can get three golds with no mistakes at all and my fluency goes down.
I have a fluency of
56% in English from Spanish.
56% in Portuguese from Spanish. 44% in Spanish from English. 49% in German from Spanish. 35% in German from English. 43% in Italian.
For Russian, Catalan and Dutch, there is no fluency % displayed
I Find this fluency estimation useful. For instance, i am in level 21 of German from Spanish, however, when i watch tv i feel like understanding almost 1/2, and that's very congruent with my 49%. I am in a high level because is study German harder than any other language, not because i understand it. I don't study English and Portuguese as hard as German, because i already speak those languages, so i'm in a lower level because i have lesse EXP points, but my fluency in those language is higher, 56%, and i do feel that i understand more than 1/2... German from English is easy for me, after studying German from Spanish, but i started later so my exp or fluency% is not yet as high, even though it is the same language, German, in both cases. Spanish is my native language, and i am 44% in my reverse tree Spanish from English. My fluency% raised quickly, and it took me little EXP to finish my tree. I should have 90%, spanish is my tongue, but i find it very boring to study spanish. Italian is OK. For Russian, Catalan and Dutch fluency is not yet meassured, I don't know if they're in Beta... I hope they release the fleuncy% function soon.
I think this is a great idea for a feature, but it appears to be totally broken. My French 'fluency' started at 25% and after several days of learning, went up to 26%. However, one hour later, without having done any more activities and it has dropped to 18%!!! Why is this happening?
It motivates me!~ I like it and think it should be kept. If some people don't like it, you don't have to pay attention to it....or it could be made into an option that is either visible or not when a box is checked or something like this. But I personally like it and feel that the algorithms used for it are quite accurate.
Completed my tree, golden tree I might add.. Fluency 53% This thing seems extremely erratic when calculating fluency. It should really just be removed entirely, and simply have a duolingo progress percentage instead.
I've watched my fluency fluctuate from the low 60's, to the low 40's, all while having my entire tree completed, and keeping on top of the courses that are no longer golden.
I ensure it stays completed, and I continue to do other translations in order to improve my competency within the language, but do not seem to show any reflections towards the percentage.
The fluency estimate is an obvious attention-getter. Its practical value in providing feedback and motivation varies with each individual. I'm currently on a a 526 day streak in German (Level 22) with a fluency rating of 48% that has held steady for several months. Am I 48% fluent? Who knows? What I know is that after having spent the past two weeks in Germany, I was pleasantly surprised by my ability to understand signage, advertising and basic conversation beyond the typical bar and restaurant ordering scenarios. I was able to initiate and respond in simple conversations. By no means can I claim fluency, but there is no question that Duolingo armed me with basic grammar and enough vocabulary to function. My biggest limitation was vocabulary. If you don't have the words on instant recall, it's next to impossible to follow a running conversation. In all fairness to DUO, I have never invested much effort in memorizing vocabulary, but I was surprised at the number of words I knew, not to mention those that I recognized but couldn't translate! My take: estimated fluency has limited value but doesn't correspond to actual fluency; i.e., the ability to function in the language.
I have a question about the "Words" that impact "Fluency," particularly those that are considered "Overdue." There are a lot of "Overdue" words in my account and almost all of them are basic words I translate on my app every day, but under the "Words" tab it says I haven't worked on them in over 2 years. I've completed my tree in French (with 2/3 lessons in gold) and I've been working to improve "Fluency," but my fluency is more likely to go down than up because of these words that never seem to go beyond "Overdue." I'm at a loss as to what I can do to improve "Fluency" at this point since so many "Words" will never improve. Any suggestions?
I have been using Duolingo for over two years and am very consistent with doing 30 points every day. And at one point I was at 47% fluency and now with no break in my consistency of study and completing my daily quota and moving forward completing more lessons I'm down to 33% fluency ? Can you explain this ?
These estimates are so confusing! I have completed all of the branches in my Spanish tree, and it still says I am only 8% fluent. My French three is about halfway done meanwhile and says I am at 54% fluency, and my Italian tree, which I started today and have completed 8 lessons on, says that I am 15% fluent. I would be interested to know how the computer comes to this number.
I've found that the fluency seems really high when I first begin the language and way too low when I'm advanced.
For instance, I still haven't completed the second section of German and it says that I'm 28% fluent, which I don't believe one bit: When I listen to "beginner" exercises outside of duolingo I can understand only a few words out of hundreds, which makes me think my fluency is far lower than 28%.
On the other hand, I'm a native English speaker, have been studying Spanish for 8+ years (I actually have a bachelor's degree in both languages) and I live in South America, yet when I do an English/Spanish combo my fluency is only at 78%.
I like the idea of an estimation of fluency: I think it's useful, and I know that "fluency" means something different than "accuracy", so this number gives another dimension to duolingo's evaluation of my language. But I think the system needs more work before the % fluency is estimated accurately.
Fluency rating fluctuates between 59% (currently) and 60% for me in learning French from English. It seems accurate in my case. However, when I get a duolingo question wrong, it's often in my English translation ... typos. So I'm confident that fluency in my native language is also less than 100%! lol
Side note: As mentioned previously, the best way to increase the fluency rating is to practice duolingo on a computer. And conversely, the best way to decrease the fluency rating is practice duolingo on a mobile device. Maybe a quirk of the algorithm ...
I don't really understand the negativity in this discussion. Duo is quite awesome, and I like all the "bells and whistles". Every day I do it like a game... get this point up, get that bar full... it's tough keeping up. And why are so many people hung up on semantics of "fluency". I suggest everybody just accept a more fluent definition of fluency :).
why are so many people hung up on semantics of "fluency"
Er...because people interested in learning languages care about the meanings of words? Otherwise, you can just make things up as you go along, people may or may not understand you, you might have a great time, but you won't actually learn a language.
I suggest everybody just accept a more fluent definition of fluency :).
The word you are looking for is "fluid". ;-)
People who are interested in learning a language should understand that no number can be pinned on their comprehension of a language, not with a decent accuracy anyway. Any scale is fairly arbitrary. Some goal is set, and we try to achieve it. As long as we are improving, this is a good way to do it.
By the way fluent also means: "able to flow freely" It also comes from latin "Fluent" which means "flowing".
People who are interested in learning a language should understand that no number can be pinned on their comprehension of a language, not with a decent accuracy anyway.
From looking at all the questions posted on this discussion board about this topic, and the mere fact that Duolingo has chosen to use such a figure, people should understand this, but they don't.
Sure, fluent has the more general definition you mention, but that one doesn't quite fit in the context you used it in English, in my opinion. A definition doesn't flow freely, or move with ease and grace, but it can be something that can change easily.
I was reading the comments here about fluency, the idea of 20,000 or 30,000 words would be a measure of fluency. Maybe, but. I think there are many other things to fluency, I have a gold tree in both english and spanish and I still have problems often enough formulating sentences. Even though I know the meanings of the words I want to use, it sometimes still is very difficult to put the idea into a sentence. I especially find the reverse tree a challenge. When I first started doing the spanish tree I was at 95% then slowly and realistically I dropped down. Now I am at 58% , I am sure that is overly optimistic and in my english tree I am at 54% (they didn't hear me swear) but that is because I make mistakes with the spanish part. All in all I can definitely get by in Spanish when travelling. So for me I achieved 100% of what I set out to do.
I love the percentage. This is my 30th day and I'm a level 10, fluency 39%. Personally, my goal is to get to 100% by the end of the next 30 days. I don't know if I will make my exact goal, but this is a major motivator for me to try harder and study more. Right now I'm trying to get all of my skills strengthened to gold, hopefully by the end of the day.
This is also a distinct percentage which I can show off on my linkedin, which is an advantage against my competitors for the same position, who don't speak a second language. Even if they know it's not exactly equal to perfect proficiency, a person reading my linkedin profile will know that I am considered to have sufficient proficiency by duolingo to pass what you might call their course. It's extremely useful to get a point across, especially to back up a claim of "I speak two languages."
I take responsibility for how much I study once I reach my goal of 100%. Knowing every word isn't a measure of proficiency, just vocabulary. Understanding how the language works will ultimately be more useful, with the admitted necessity of a certain level of vocabulary to get through daily life and communicating needs. In my opinion, this is both useful and fun, and I get bored walking into Walmart. Hehe.
You might want to note that the fluency percentage never goes as far as 100% on here (native speakers who know the base language well plateau somewhere in the high 60s, I think).
Which is the one thing that makes sense about this measure, because the Duolingo courses don't teach you enough to make you even remotely fluent.
I agree also, I am rated at 57% in Spanish and 55% in English. Well I am a native English speaker and I can only get by in Spanish. But the one thing the fluency score does well is it compares your progress against yourself. This also includes regular/daily exposure to the language. Even though I had my Spanish tree completely gold, with only 3 words to strengthen on the word list I only had 58%.
Wish you the best for getting that 100%!
I finished the French tree (and of course did the reverse tree) more than two years ago, visit France twice yearly, also practice daily with Memrise and Babble+, and haven't missed a day of studying for over 800 days ... and my level has never gone above 60%. Given how difficult French grammar is and how robust the language is, the estimate of 60% seems more than fair.
I still don't really understand how this fluency thing works. Somewhere at the start I managed to up my Italian fluency to 19%, which remained there for nearly a year, until I recently discovered it dropped to 18%. Of course I've learned much more words since then and am I still maintaining the strength of every single word I learned, so I don't get why it dropped a percent, instead that it increased.
Allow me to translate: I now have a fluency shield in Duolingo. It show that I am 0% fluent. Damn! My tree is completely gold, I know many thousands of words, I have read books and watch videos, founded an Esperanto group in Meetup.com, but all was for nothing! Good grief!
If I could put it to you more plainly, I would say that there is something seriously wrong with your fluency estimates.
Honestly, I think this tool on the site is just a way to keep you motivated, to reach just one percent is great and gives you a confidence boost. Fluency isn't measured in that way though, I think fluency is being able to understand basic words, to have a basic understanding using context clues, and being able to communicate or get your point across. If you can hold a conversation, it doesn't mean you are fluent though, you are just well practiced, you're not fluent until you can talk for hours.
I don't know, maybe fluent is just an idea.
If you truly want to become fluent in a language, it's best to live in a place in which most people speak that language for 6 months to a year minimum. One would learn a lot more with that approach than by sitting at a computer clicking buttons. While I am no where near being able to visit France to become fluent, I hope to someday. The fluency number given on Duolingo is inaccurate because the words taught are nowhere near all of the words offered in those languages, but Duolingo is a great tool to help you get started.
I agree! I believe the crux of this discussion stems from the very term, "fluency." It's nice to have a barometer to measure progress, but calling the measurement standard "fluency," really muddies the waters. One definition of fluency is "the ability to speak or write a foreign language easily and accurately." Given the limitations of a computer-based programmed learning course, it's all but impossible to assess an individual's fluency in a language with any credibility. i don't know what the term should be, but it's not fluency. The percentage of whatever it's really measuring is still motivational, but nothing to stress over. Duolingo is a useful tool nonetheless.. See my earlier comment regarding my experience with my Duolingo German in Germany.
I had the exact same problem in German, but eventually it unstuck itself from 39% and started increasing again (I'm now at 43%). I doubled my daily XP in German most days for a couple of weeks -- I don't know whether or not that made a difference, but you might want to try it.
(All my other language counters have progressed steadily, so go figure.)
From my experience I have concluded that the fluency mainly depends on two things; how far you have progressed in the tree and how "golden" the tree is. If your tree is completely golden the fluency percentage will no longer go up unless you can complete new lessons. Once you have finished your tree and have it completely strengthened (i.e. at least golden), you can reach around 55-60% depending on the language you are learning.
Had to comment, just started with duolingo. Tested where i placed in italian got 60% fluency (lol)... In italian i can communicate basic needs and interests but struggle with anything more complex.
I then tried Danish, i learnt Danish as an infant and regularly communicate with family in Danish, My spelling and grammar are relatively weak, since i never attended school in DK, i and speak a grammatically loose (compared to official danish) and accented danish from rural jutland), but my vocabulary and comprehension are on point. I scored 1% fluent.
really don't rate the fluency test. soz.
I am frustrated by this feature. I want to learn several languages. I find this number to be a poor representation on my language skills. Please explain to me why I can correctly answer 37 questions in a row without a change in my fluency score. But when I get a single question wrong, my score dropped 3%! This is silly. Can I change test groups? I find myself pursuing a fluency score instead of trying to learn. I am not a fan.
You really don't want it. I'm a native english speaker in the middle of a PhD program with a french language requirement. According to the fluency estimator, I have 1% fluency in english and 3% fluency in french. I've just passed a french fluency test that will allow me to do advanced undergraduate french classes.
Dear Duolingo Moderator -- It would be great to have full transparency into the fluency estimate algorithm. That way, we would know what actions bring about the most positive results. After being at an estimated fluency level of 57 pct (or higher) for quite a while, my fluency estimate dropped by 6 points in one day for no particular reason. I am at lever 22, and this drop is very demotivating for me after spending so much time on your platform. I had been working hard to maintain the 57 pct. Please consider providing complete transparency into your algorithm. This would only help users use your platform in the way that you desire. Thank you for your consideration.
Dear A/B product test group -- I like the idea of a fluency measure -- it is helpful for tracking progress, BUT, by process of elimination, I can tell that the fluency badge measure you are using has a PARTICIPATION measure as part of the algorithm. This means that the badge is not accurately measuring fluency. It is also measuring participation. We already have a tracking devices available to use on the site (streaks, daily goals, etc.) to measure fluency. I would like the fluency badge to accurately measure FLUENCY. This is the only goal that I am concerned with achieving. I am using lots of resources to practice my Spanish, so I am not that interested in the participation metrics. Thank you for considering my feedback!!!
It only shows up on German, not on French and Italian. I also notice my skills on German that I originally tested out of don't go down but the newer ones I learn do.
This is the opposite of motivation. Completely useless and a lie. If I finish my skill tree with everything golden and I don't get 100% then what's the point? Just remove the stupid thing from the game! I don't want to finish a lot of lessons just to be told that "congratulations, you are STILL on 60%" ...
Thinking that you're 100% fluent just after completing the tree is kind of a positive way of thinking, because you probably forgot a lot of things, but the estimates are bs. First completion should be around 80%, then allowing you to go up to 100. I finished my Italian tree like 2 months ago, do around 1000 exp a week and am stuck at 52%.
You can go 100% if you keep your strength up but first ask your French teacher if this can make you fluent
I didn't even know any of the courses would give you a "fluency" number that high -- that's amazing. I've been working steadily on six of my languages for three or four years and never got any above (I think) 61%.
(The number is supposed to reflect how much of a real-world conversation you'd be comfortable following. No Duolingo course takes you beyond "beginning intermediate" stage, so an estimate of 60% is actually quite generous.)
Yes, the recommendation to use Strengthen Skills to raise fluency works. When I started using Duolingo to review French skills, I found that using this option was the fastest way to both increase fluency, and also to get tested with examples of more complicated usage and vocabulary. More than a year ago, before they recalibrated the fluency estimates to go higher, I would simply spend 5 or 6 days using this feature at about 100 - 150 XP per day to move up in level. By level 23 I was at 65% fluency.
If your looking for the "Fluency Estimate," it's no longer an item.The "Crown" points on the header replaced it. Probably just as well as "fluency" is an amalgam of vocabulary, conversational acumen, local knowledge, etc. Pretty impossible for DUO to have the data necessary to make such a call with any accuracy.