Duolingo German review: A Vivid Discussion

Thank you for your responses to my Duolingo German review – I will gladly point out what I had in mind in my previous post. For those who are new to this discussion, I have cited Filip’s arguments in italic before my replies. You can find Filip’s initial comment at the very bottom of this article.

Exercises? What am I Exercising Exactly?

You have a strong point here – I’m not sure what the purpose of that exercise is. It may or may not work on subconscious level, as Rosetta Stone people often claim, but one thing you neglected to mention here is how rare these exercises are. I haven’t seen a single one for a very long time and I’m fairly certain at this point that they are only present in the first couple of lessons, where the software assumes you don’t know a single word of German (again, whether learning the first words this way is valid or not, I can’t say). Instead, this is how these exercises look in most of the lessons: //postimg.org/image/lh6wfgog9/ The point of these is much more clear – it’s great to learn both gender and spelling of a word.

There are quite a few of these in the first lessons and they do not serve any purpose but maybe very gently leading the learner towards harder exercises by letting him or her experience a few little successes. While some might like that, I still consider this a waste of time and an insult to any intellect.

The kind of exercises you show here is a simple translation exercise. While those have their charm when used in limits, translating random sentences is highly inefficient as it opposes our mind’s desire for meaning. But I like the muscle-show below your screenshot 😉

That particular exercise that you show does not help you to „learn“ the gender and the spelling because it does not explain anything nor does it help you to connect it to your actual knowledge. It helps practicing it. But still out of any significant context.

Lost in Translation

This one is also mostly incorrect – Duolingo does not assume “is” is not an error. There are (to my knowledge) three types of “minor” errors that pass as “almost correct” – nouns not beginning with capital letter, lack of umlaut and typos. Typing “ist” as “is” counts as a typo and Duolingo reports it as such, with “almost correct”, as seen in the screenshot. //postimg.org/image/jv9o0gc9b/

Well, just because you have found an example where DL is working fine, that does not mean that there are no more mistakes in other places. And when I can not be sure that it is right in all cases, I will get confused or worse learn wrong structures. The problem I point out in my example is that DL is simply overcharged with pointing out two mistakes at the same time and ignores the more grave one as „is“ is wrong. But of course no software is perfect and I should give DL more credit here. Point taken. I also praised it for these little annotations.

Is this OK? Are these errors truly “minor”? Not really – and Duolingo treats them as regular errors, it marks them the same way it marks errors, it plays the error sound and the only real difference is that you don’t lose a heart (losing three forces you to start over). This is OK in my opinion – on many occasions I really did make a typo and Duolingo let it slide. If it didn’t, it would be frustrating to start over since I knew the correct answer, just mistyped it. Students shouldn’t be scolded for every error they make, but every error needs to be pointed out – which is exactly what Duolingo does.

No disagreement here and not my point.

That said, what you displayed in that screenshot appears to be a bug, since it failed to report all the minor errors (also, should three minor errors really pass as “almost correct”? I think not). As seen in the screenshot I supplied, it does not consider “is” instead of “ist” to be correct, but it lets it slide since it’s only off by one letter (BTW typo detection is way more strict than “off by one letter” – on many occasions I supplied the wrong answer which was only off by one letter, yet Duolingo did not report it as a typo).

The fact that this is obviously a bug does not refute my point. As a learner I don’t care if it is a bug or a concentration mistake or simply bad knowledge of the teacher. But you are right, if this is not too frequent as you say from your experience, there’s not much to argue about from my side. Yet it is important to be aware of the fact that there might be significant errors.

Is Duolingo a worthy German learning app
die Eule – the owl / Image via Pixabay

On Randomness

I agree with this one – there is very little context and I understand it would be easier to learn if there were a “story” of sorts, as language learning books usually make it. Still, many of the sentences make sense and Duolingo is highly tolerant if your translation is correct, but in completely different context than expected (using “woman” instead of “wife” as translation for Frau, for example) – simply because it’s aware no context is provided. I cannot say the same for many teachers I had (unpleasant) experience to learn from – they would often scold you if you missed the context, despite the fact the textbook wasn’t clear enough on what the context is.

Be careful: Just because there is worse, that does not mean „bad“ turns automatically into „good“. You actually fully prove my point. That a sentence has some context by nature is indisputable. But just rowing up non-connected sentences after another does not make them any more useful. Actually I would rather say that it is confusing as your brain will always try to make sense of what you do. Even if you do not experience this clearly, this approach is far more inferior than if you had a solid context, e.g. in form of a story.

And why would you want to get away with „woman“ when you should have learned „wife“? There is a significant difference here and „my woman“ might be perceived differently than „my wife“ by many. I assume that you are referring to translations into your mother tongue. In this case, yes, the difference does not matter as you get the idea. But when that error is not corrected when translating into the target language, then this is a crucial failure of the system.

Reden ist Silber…

I agree with this one as well. I had to disable voice recognition as it simply didn’t work for me.

Listen Well

I’m not a native speaker so I can’t actually comment on this one. However, I feel the need to point out you neglected to mention that every exercise like this also has the “turtle” option which reads it to you slowly which makes it quite easy to get in 99% cases. Therefore, I actually like the “fast” computer generated words – they make it more difficult and listening to native language speakers often feels like this to someone who learned the language from teachers who purposely pronounce words very clearly to make it easier for students.

You seem to have a „better than nothing“ attitude which I do not share, yet also do not object per se as that’s your choice and you most likely do fine seeing things like this. I simply like the approach: Do it right or don’t do it. And by „right“ I do not mean 100% perfectly. I am a strong proponent of the so called Pareto Principle without claiming scientific validity for it. 80% of the success derive from 20% of your effort.

While I love the turtle button myself and consider it very useful, I would like to compare the computer generated voices in DL with the experience of sitting in a classical concert while wearing a huge metal helmet if that makes any sense. Language strongly lives from the nuances in pronunciation and sentence melody. Computers can’t generate these yet and therefore rob the language of a very important emotional factor.

If you have watched HER, you might have felt that Scarlett Johansson’s voice was crucial for the feel of that movie. With a voice DL uses I doubt it would have gotten that much attention.

Sure, it might get you through the game here and also give you a solid enough idea of the sound but you also need to consider that we are highly imitative learners and might pick up some of these artificial sounds. I personally would prefer to sound as sexy as in my mother tongue in any language I speak.

Discuss with the Right People

This is a good point – very difficult to know whose advice to follow. Would be nice if Duolingo made some way of distinguishing between native speakers who can offer good advice and people who only *think* they can.

Grammar: The Ugly Stepchild

“No Instruction on how to Learn Anything Whatsoever” This is just plain incorrect – almost all lessons have explanations, and yes, grammar rules are explained and can easily be revised even when you complete the lessons. See screenshot. //postimg.org/image/qrgx3vf0p/

The title of this paragraph is misleading, I agree and understand your objection. Yet read the subtitle more carefully: No instruction on HOW TO LEARN whatsoever. As usual grammar is simply shown and explained in a lingo that is simply difficult to grasp for many. But they do not show you how to learn the Accusative. How do I in fact learn the Accusative prepositions? How do I learn the articles? In this example I can not even see when I need to use it. It simply says that „when the subject turns into an object“. That is surely a correct but at the same time an awful instruction. So I am not wrong at all. If you showed me a single instruction that actually showed me A WAY to get all those informations into my brain, I’d withdraw this point immediately. I haven’t come across one so far.

I was having trouble to find those grammar pieces before but I took another look and can now see and access them easily when I am working on a lesson. Yet I can not find a way to access them independently of the chapter that I am currently working nor did I find a way to search for certain topics at least with help of an index.

How to (not) proof Efficiency

I don’t agree with you completely, but I feel you made a strong point here (see below).

Finally, my opinion is not that Duolingo is perfect – it can never replace real life teachers and classes, not should it attempt to. It’s, however, a good product that actually can teach people the foundations of a new language they want to learn. From that point onwards, it will be easier for them to continue learning or, at the very least, have basic understanding of the language. It’s also very good at this – it learned me a lot and it’s, in my opinion, the best of its kind – so far. Looking forward to Yippiy though – and wishing you the best of luck. Even if I don’t end up using your product, having an alternative to Duolingo is great.

Thank you and please do never take any of my points personally, especially the next one: The problem is in general that our personal experience often seems to differ with the facts. I can strongly recommend this video as an introduction to skeptical thinking.

So, while what you have experienced might be absolutely true, I would always doubt it until I had examined your claim and also compared it with someone who has used a different approach and/or no approach at all. If then your results differ significantly from those others, I would take a deeper look into how you did that. As I do not have the time to evaluate DL to that extent, my assumptions are based on over 20 years of experience with efficient (language) learning and teaching and professional as personal research in exactly this field plus a deep interest and very analytical understanding of the human nature. That does not mean that I am never wrong yet I see a lot that others don’t. I am not selling any opinion here nor my courses unless I clearly state so. I share my experience and observations and knowledge which might be faulty. That’s why I invite anyone to challenge my observations and conclusions. This is how science works, or at least should do. And as DL itself claims to be more efficient than a semester at a high school, the burden of the proof lies with them. And as I have pointed out, they simply fail to do so.

One last thought: When we invest time or money into something we tend to look for the benefit of our doing. Sentences like: „At least I got this and that from it“ or „It wasn’t all bad“ serve mainly one function: to justify that we have most probably just wasted our time. We do not like to admit failure. And I do contribute a lot of DL success experiences with this phenomenon. And this is a provocative yet not unlikely assumption.

Plus of course you are right: there does not seem to be anything better out there. Although thinking about it, I would indeed consider Busuu or Rosetta Stone way better than DL because
they have a broader variety of exercises and also a more solid programming. But better than bad does not mean good.

That being sa(i)d and despite all differences in our perception and interpretation of our findings, I would like to thank you once again for sharing your points with me and your willingness to dispute on a very constructive level.

I wish you a lovely time and a lot of joy with everything you do.

Herzlichst Michael

Filip’s first comment that initiated this discussion

While reading your review, I saw some strong points, some questionable points and some points that were just plain wrong. But the ending explained everything – this review (“critical” in name only) is merely an attempt to market your own (expensive!) course and indiegogo campaign. While I think this is a cool marketing move, I strongly disagree with the idea of criticizing a free product to attract attention.

My initial reaction

Dear Filip, I understand how you come to your assumption. While of course I sell my services -for by the way half the price the local Goethe-Institute  would charge you- that does not mean that I am insincere in my blogging. If as you say things I have found criticizable in Duolingo are wrong I would highly appreciate if you could point them out so that I can review them and change them in case you are right. I am not sharing an opinion here. Everything I do is backed by science and practical experience of over 10.000hrs of teaching. Yet that does not mean I can not err and am happy to learn where I made a false claim or interpretation. Have a good evening.

5 responses to “Duolingo German review: A Vivid Discussion”

  1. OK, once again, your arguments sound good. But re-reading both blog posts again, I realized that we are, in a way, defending two different points.

    My original response to your blog post, and also the one after that, was written because I got the idea that in reviewing duolingo, you simply found several bad things about it and then wrote a post that *only* covers these bad things (while briefly mentioning in the beginning that you like the idea behind it).

    For me (and I must emphasize this, it’s just the way I personally saw it back then and still see it), your original blog post implied or stated outright that:

    -most exercises can simply be guessed or answered without any German knowledge based on pictures (“Exercises? What am I Exercising Exactly?”)

    -translation exercises are completely broken and often gives false positives (“Lost in Translation”)

    -voice recognition doesn’t work (“Reden ist Silber…”)

    -there is very little or no explanation at all regarding grammar (“Grammar: The Ugly Stepchild”)


    All of these things are valid points to a certain degree and I even agreed with some of them in my previous post, but are also exaggerated and give the wrong idea about Duolingo’s effectiveness.

    I initially (foolishly, I admit) assumed that you’re simply trying to demonize Duolingo in order to promote your own product(s), which is why I wrote my initial post. After that, all I was trying to say was that Duolingo is a tool that can (and it actively is, as I’m typing this) learn people to understand German to a certain degree (I’m not competent to judge on how great this degree is, but just the fact that it exists, it’s free and there are people actively improving it is, in my opinion, an amazing thing for anyone who wants to learn German). The title of your original post implied not only your disagreement with this claim, but also that Duolingo can somehow “hurt” your German with which I can in no way agree and that’s the reason why I had put effort into writing these comments.

    • Hi Filip,
      thanks for engaging once again. I have not much to add to what I already wrote. Maybe let me state once again my point of view, which admittedly is kind of strict but I see no way of lowering my expectations without major improvements of the softwares.

      If I compared Duolingo or any other (!) software/web-based language course to a mobile phone, it would make dialing numbers very hard. Dial per voice would completely fail and although the calls are for free, I wouldn’t reach whom I’d like to talk to. At times it even gives me three numbers to choose from. It seems to have all the features of a phone and people fancy it for its looks and funny sounds, so they play with it just for fun and it’s entertaining them.

      When I say “harm” I mean that there are two likely -in my eyes negative hence harmful- consequences:
      1) The learner spends time with inefficient and insufficient practice material like e.g. simple multiple choice or monotonous translation exercises of sentences that are completely out of any context. In that time they could have learned something more useful and also more helpful for further lessons like e.g. grammar.
      2) The German they learn is mainly unsupervised. If there’s anyone believing that the voice recognition works and takes that as valid input and training for speaking German, she or he is likely to acquire a very messy German pronunciation that is ridiculously hard to fix. While that doesn’t seem likely there is always those who might. Why offer something so completely disfunctional? Rosetta Stone has the best working voice recognition but all they do is letting you repeat sentences in parrot manner. That has absolutely nothing to do with communication and does simply not work. A computer needs to ask you questions and then check your answer. But we are still 20 years away from that kind of AI.

      Regarding grammar: In DL it is at least written down and accessible at times but they do not really bother to explain how to learn it. They simply present it. My guess is that most DL-learners do not even bother to read these grammar texts or even if they did, might not get it exactly, so they simply try to derive it from the exercises they do. In Rosetta Stone you have got no choice at all but to figure things out on your own. And that might and often will lead to a German that is based upon unclear, vague guesses as no one can be sure the rules they have derived from the samples they have worked with are 100% valid. Uncertainty is a huge communication killer.

      You and a few others might be talented language learners. You would learn from any source or even in a group of hundred people as you have the ability to figure things out even if there’s a lot of “noise” around you. From my experience the average language learner is not that lucky as the school systems are not really famous for their language teaching approaches. And that is where most of us have got their language learning skill and methods from. And those I sincerely warn from ANY software/app that claims to teach them German on its own, like the Rosetta Stone, Busuu, Duolingo and Babbel do. Rosetta Stone at least now offers supplementary tutoring but that comes with a price.

      I can not and will not argue against personal preferences and subjective impressions. From a admittedly highly rational point of view, I can not recommend any mentioned or non mentioned language learning course so far, only tools that help you to master certain skills, like vocab (memrise) or reading comprehension + vocab (bliubliu). Just because something is free, it doesn’t automatically mean it has any significant benefit. No matter what the individual experiences.

      Those apps claim to teach you a language. They do not hold up to that claim. If someone claims to have learned any language I am able to understand mostly with any of those apps to a degree where she or he can have a basic conversation on level B1 CEFR, I am more than happy to talk to her and see for myself. Should that stand my expectations -and I promise to be generous and willing to let her pass- I will toll my utmost respect. And still would expect her to be the exception.

      And if you use those apps in addition to whatever you do, how can you be sure that the benefit you experience is caused by the app and not fully from the other source you are making use of? DL has ordered that shady study they have published on their page but that one is rather nourishing my doubts than appeasing them as I have written in my original post.

      And once again. If you just want to spend your free time playing around with DL and do not really care about the outcome but rather about having some seemingly useful fun, then, well, DL it is. I personally would and will stay away from all those courses even though I used to be quite a good language learner as they would simply waste my time.

      Having said that, I really appreciate your interest and engagement and understand that my point of view seems strict and makes you want to oppose as it obviously contradicts your subjective (!) experience. I do not want you or anyone to stop working with any app they use but simply to be able to make an educated choice by also seeing the downsides of the path they have chosen.

      I wish you success with your language learning and will soon start to learn Greek in 30 days and see how far that will get me. My aim is B1 with 2-3hrs of daily learning.

      Have a good evening.

      • Hi Michael,

        Thank you for being patient with me and explaining your arguments. I now understand your point of view better (but still have some of the original doubts). I wish you the best of luck with learning Greek and am looking forward to a progress report after 30 days.

        Best regards,

  2. Hi Michael,

    I’ve had a chance to re-read today, and wanted to leave a few comments.

    Firstly, I would love to see an article by you, in which you outline in full, your core principles for language learning, and the reasoning behind them. I keep seeing little teasers (for example, in this dialog), about the necessity of *narrative* to real learning, the link between *understanding* and retention, and the emotional nature of language. But I’m always left wanting more! These are fantastic insights, that could even serve as the grounds for a whole book, if you ponder them for a while….

    Secondly, I didn’t quite understand the reference to the Pareto Principle, in relation to your approach to language learning. This was the thing I was going to mention in lesson, but couldn’t really find a way to express the objection. Mostly because I didn’t understand the claim being made. So, I was hoping you could expand on that.

    Lastly, I have to say, the end of the article shocked me a bit. I could be wrong, but it looks to me as though Filip is just flatly accusing you of some sort of consciously deceptive opportunism in his initial comments. If someone is not even willing to grant you benefit of the doubt in the opening salvo (i.e. that maybe you’re just passionate about what you do), then why even engage him at all?

    Anyway, that was all.

    Schönes Wochenende!

    • Hi Greg, thank you for such a lovely inspiration. I will write my answer into another post as this addresses some fundamental principles that I would like to share more visibly than hidden in a long row of comments. Einen schönen Start in die Woche wünsche ich Dir.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.