"Popular" Meanings and false translations

Dear community,

I’m not sure if this has been discussed - I could not find anything searching the forum. I see a problem with the translations of single words that multiply especially with new learners (like myself) that have no way of knowing better and just choosing the first translation, which is in many cases wrong.

Whenever you click a word, the “meanings” (translations - don’t know what the english client says, in German it’s “Beliebte Bedeutungen” … “popular meanings”) are displayed and ranked according to their “popularity”.

Users can add their own translations, so there is a variety of translations available to you, but many of them are simply false. For that purpose, the “report” button has been included. But I have a feeling there are ten times more false translations added than flagged as false with the report function.

Here is one simple example from my italian learning - sorry for theh long explanation: The italian preposition “nelle” is composed of “in + le” (article for feminine plural) and its meaning is "in the … " but in plural feminine (hard to translate that in english).

The most popular suggestion (74) is basically wrong because especially when learning the vocabulary there is no hint of composition or plural. “in” directly translated to italian without m/f or sg/pl is “in”, not “nelle”.

The second one (23) is basically correct, but doesn’t tell you it’s feminin (or plural). Might as well be the translation from “nel” or “nello” (masculine)

The third one (8) is the best, clearly states in brackets (f pl), feminine plural - unmistakable.

The fourth one (1) again is wrong, yes - “nelle” is composed of “in” + “le” but that’s not the meaning, it’s just the grammatical explanation.

In many other cases, users even add the italian word to the translation, that’s not useful in repetitions of the vocabulary, when the solution is already in the question…

Long story short:
My example is one of MANY! I would guess that’s the case for 10-20% of the nouns, around 30% of the verbs in precise coniugations and maybe even 50+% in the more difficult translations like composed prepositions and so on.

How should a new learner know which of the 4 in my examples is really the correct one?

How could we improve that in the future?

I do not want to learn “popular” translations (that are 100% false), I want to learn correct ones.

I have stopped to blindly add the proposed translations and look up everything in solid sources like PONS.


The only solution as you said, at the moment, if you don’t trust suggested users hints, is to search for your own translation by using dictionaries. Hopefully with time, with more and more wrong hints reporting, the correct one will became most popular one and be at the top.


This is a problem that I have as well. Sadly one that probably can’t be solved. The quality of the translations highly depends on the language pair. In your case Italian → German seems to not be very popular. With something like spanish → english you probably have better translations. I am on Ukrainian → German and it is even worse. Very often the only existing translation is the automatic translation from google translate, which for sinlge words is very unrialabe. I added english to my languages so that I can check the english translations as those are sometimes better. But even then you often don’t have the correct translation available.
This problem is amplified by the lack of actually good dictionaries. Anything that is declinated or conjugated is often even wrong translated by the dictionaries that lingq has included. They
often can’t even tell nouns and verbs apart.
So yes in most cases you have to look up the correct translation yourself and add it manually which is very time inefficient.


Yes, the quality of the ‘user hints’ is a known problem. New users may not realize that they’re not consulting a professionally curated dictionary but just good old Google Translate.
The hints are of course inherently subjective, some people include grammar information or transliterations, so the only way to guarantee happiness is to take full control and not rely on other people’s hints.
Also, anyone can ‘report’ a hint, which results in it disappearing from public view. So, even if someone spends time to find a high quality translation it could disappear at any time, so there may not be much incentive to improve the state of affairs.
I remember we had a discussion about this topic not long ago, and one suggestion was to implement a way for librarians to create a ‘featured hint’. Regular users wouldn’t be able to delete (report) it and it would show up toward the top of the list. But that would only benefit one language combination at a time (e.g. Italian → German) and to make an actual impact you would have to add something like 100k meanings, e.g. by importing a dictionary, but that may run into problems with copyright.
Seeing all the issues LingQ is dealing with, I am not confident they’ll get around to improving this situation.

Anecdotally, I don’t think bad translations necessarily to lead to catastrophe. For example, I never found a good dictionary in Romanian and I’ve never seen any user hints over there, so I just kept pressing the right arrow key selecting whatever Google Translate came up with and nothing terrible happened (imho), things got clearer after some time. (Although I believe the same wouldn’t work in Chinese)


Also, anyone can ‘report’ a hint, which results in it disappearing from public view. So, even if someone spends time to find a high quality translation it could disappear at any time, so there may not be much incentive to improve the state of affairs.

How many reports are sufficient for a hint to disappear? 1, 5, 10?
If it’s just one as you’re making it sound, then what’s stopping assholes from reporting high quality translations? What about missclicks, can you undo your own report?

1 Like

My understanding is that one click, from anyone, is sufficient:

I forgot to mention that LingQ does try to incentivize good user hints by awarding coins when users select a hint you created, you can see the result for example on the monthly ‘contributor rankings’, in the weekly report or on your profile (it’s called ‘Translations Shared’).


Thanks for clarifying. I generally write translations just for myself, but that’s a reason enough to never bother adjusting them for other people. This is bound to backfire sooner or later, especially for the beginners in any language.


I have no evidence, but it probably doesn’t happen all that often. Having said that, it seems ridiculous for a single person to be able to remove any translation just with one click, particularly a popular one, or especially a correct one… Something like this, at best, should be reserved for “special” users.


As others have said, this seems to be a problem primarily with language pairs with a smaller user base.

I first encountered something similar with Ukrainian → English when I started in early 2022. I don’t know where the wrong translations came from (possibly an earlier, inaccurate iteration of Google Translate for Ukrainian?), but for many words there was only one suggestion which was plain wrong. Not even wrong in the sense of grammatically inadequate, but simply a completely different word. This seems to have improved a LOT over the past year, presumably as more people picked up Ukrainian!

In the meantime, I simply had to learn to recognize when something didn’t seem right, and check with Google Translate or another dictionary before saving a translation. It made the process slightly more tedious, but it was worth it to me. Admittedly, part of it was also simply turning more tolerant of ambiguity and “noise” in my learning.

As a “quick fix”, you would probably get more accurate translations if you switched the dictionary to English, as Italian is a fairly popular language and surely has a bigger user base there! Of course this is suboptimal if you really prefer translating to German.


It is time-consuming to put in your own definitions. It’s also a time suck to review and delete translations that are completely wrong.

But . . . thirty years ago I’d be making my own flashcards for each and every new word, so I’m just accepting the situation for what it is and making the best of it. I’m not going to ask LingQ for translation help when there are so many other tech issues they have to deal with.

Also, the fact that they put in romanizations for Korean gives me no confidence in their judgment on how Korean should be learned.


Yeah this is what I hope for in the future. The main problem was that I initially (first few days) didn’t really realize that many or most of them were user generated, until I discovered some very bad typos and also totally bad translations. By that time with my initial enthusiasm I had already marked 2-3k words as known because I learned Italian for some years in school.

On the other hand… double checking translations and carefully adding new ones that include more precise meaning and so on is a tremendous boost to my own learning because I’m really careful with that and sometimes look it up in multiple sources, check coniugations and so on.

Maybe, in the future, the simple point system (how often a translation is chosen by users) could be refined to a system where contributions of “higher rated” users (because they add more or correct a lot of things) are considered more and their translations rank higher.

1 Like

Thanks for that hint, I would not change the target language (at least I wan’t to speak one of the pairs perfectly wenn :wink: but I didn’t think about that and it gives me a good explanation for that.

I fully agree, there will be no catastrophe because of that but I really have to “unlearn” quite a few expressions - especially prepositions and binding words. On the other hand, correcting as many errors as I can with careful research also is a strategy for learning :slight_smile:

1 Like

One thing you can do is select the phrase where the word appears. That may give you a better idea of what the word means in context, and may be more accurate than the individual definitions.

As a rule of thumb, I always look up definitions by myself when given ones that do not fit in the sentence. I prefer more concise definitive dictionary entries, and we can put any common grammatical explanations and morphological variances in the note section.

I usually resort to the monolingual dictionary for reference when a direct imprecise translation does not suffice. By the way, it’s understandable to see the pronunciation guide for the definition for the overlapping purpose, but entering a sample sentence in T1 is utterly weird.

That is good to know and somewhat unfortunate. I used to create quite a number of translations (for everyone) - however if those can indeed get lost - I need to reconsider whether I should rather put those in my personal comments instead.

It can be a problem. I use LingQ primarily to read news and magazine articles that I import. My strategy is to accept a word translation if it fits in the context of the sentence and the story. If it seems off, I look up the word in a dictionary. If it appears to be an idiom or common phrase, I’ll google it to see what comes up. If I just can’t resolve it, I pick anything to turn it yellow and just move on. Eventually,I’ll run into those things again in another article, and I might then know the solution. Learning a language is a highly iterative process and bounces all over the place. It’s not linear at all. Even in your native language, you hear new things and have to decide if they are correct, slang, proper nouns, acronyms, etc.



The reason Community Definitions exist is because of copyright. The company cannot buy dictionary licences for every user for every language. This would mean your subscription would cost much much more than it currently does. This is their solution to the issue - allow users to write their own definitions (and give the copyright of them to LingQ).

Personally I’ve written 30,000 definitions for Italian (from English). Even though on LingQ, Italian has been studied by thousands of people before me, I still prefer to write my own definitions because I had (and still have) problems with Italian grammar. Now it’s become a habit that’s hard to break. As you get a higher level, writing definitions becomes less important for several reasons: (1) more words go straight to known; (2) the words have only one definition and the Google Translate is the correct one; (3) you understand the grammar so don’t need to correct the current ones; (4) you know all the other words in the sentence, so can better guess if the community defintions are correct; (5) you know the word is rare and you may never see it again so why waste the time writing a complex definition?

I agree there’s an issue with a lot of the high-frequency words. There are wrong definitions, such as specifying the wrong pronoun. When I was a beginner too, I know I wrote some wrong definitions too, which were saved. What really needs to be done is someone to go through all the Mini Stories and other beginner material (i.e. with high-frequency words) and clean up the definitions and write some good ones. For the mid-frequency words, my definitions are decent (because I improved in how I wrote definitions over time). And for low-frequency words, it’s not worth your time and you are skilled enough to just use Google Translate definitions. As @bamboolzed said, due to the priorities within the company, for cleaning up of the high-frequency definitions to be done, you either need to wait a long time or alternatively someone needs to take it on as a community project.

That being said, I’ve learnt a lot of high-frequency vocabulary, despite using wrong definitions initially. When you advance, you just realise that the definitions are wrong and you either choose another one of the Community Definitions or end up editing it yourself. There’s none of this long and painful ‘unlearning’ because you never knew the word well in the first place (it’ll stay yellow because you feel the word doesn’t make sense to you). You will just eventually realise that ‘nelle’ means ‘in the (pl. f.)’. Or, in these cases, if you learn a bit of grammar, you won’t choose the wrong definition in the first place. It is rare to find definitions that are completely off (with the exception of Google Translate, because the word could have multiple definitions). It is much better to get 80% there (such as ‘in’ instead of ‘in the (pl. f.)’) than to just be watching TV series without subtitles and having not the faintest idea what it is. Usually with the definition, the sentence, and the context of the whole lesson, you get the general meaning and that’s all you really need. Over time, the word becomes clearer and clearer.

All in all, nothing to worry about.


Thank you for your reassuring post!

I have noticed that user-submitted (mis)translations often contain obscene language or other nonsensical translations. Who knew that so many Polish words meant either the F-word or “strawberry flan” or a string of random numbers?

I try to report these user-submitted (mis)translations as much as I can. It seems like a losing battle, though. I guess I’ll have to learn to live with them.

It would be nice, though, if LingQ could employ robotic process automation or something to at least flag obscene language for the librarians or a volunteer committee to review to make sure the translation is correct in both meaning and register.

@ed_shin That said, the mere act of all the clicking and selecting or writing definitions takes a lot of time.

The comparison for me would more be like reading while listening on LingQ at ~55 wpm, reading while listening on YouTube with Language Reactor at ~150 wpm (I just tested it with the Harry Potter audiobook on YouTube with auto-subs at 1.3x, but having to repeat sentences every now and again) or true extensive reading while listening at 1.65x/~240 wpm. In this case, using my reading while listening speeds, we would be talking about 10M words read as extensive reading while listening, ~6M with Language Reactor and ~2.5M with LingQ. As mentioned, the LingQ reading speed is so slow because of the amount of time it requires to get a decent definition (which often includes pausing the audio to open up a dictionary and write a definition). With the translation under the subtitle on Language Reactor (aka bilingual text), you can merely glance down often to get a good definition, without ever having to click anything, hence the increased reading speed.
Is extensive or intensive reading faster for vocabulary acquisition? - #43 by nfera