Review of Lingvist: Fun, but Incomplete

Share this:

This post contains a referral link for Lingvist. Feel free to use it. Then make your own link before sharing it with your friends!

People are always looking for shortcuts and hacks when learning a language.

  • Can you learn a language in 30 hours doing nothing but studying with an app?
  • How fluent can you get with an app alone?
  • What app is going to get you fluent as fast as possible?

These are just a few of the questions I expect you’ve been asking yourself as you fiddle with your app of choice on the way to work — either solving Duolingo puzzles, listening to native speakers on Memrise or maybe even patiently keying words into Lingvist.

There’s always an app du jour promising that with a simple commitment of 30 minutes of study a day, their AI-based language teaching technique, based on millions of hours of research across multiple languages by data scientists, PhDs and alien superheroes, can teach you a language.

Lingvist is another one of those.

Is this possible? How?

Note: Links to Lingvist are my personal referral code. I don’t earn anything from it, but if you hit 250 words, we each get a free month of ‘unlimited’.

Me speaking German after a month with Lingvist

After using Lingvist for a month, I could have basic conversations like

  • “Where is your wife? She was here.” “I don’t know”. “Why don’t you know?”
  • “I don’t want to eat this!” “Why? You are strange”
  • “Why are you standing here?” “Because I want to eat a sandwich.”

These are the “conversations” I had with my brother, who’s fairly fluent in German (at least compared to me). Let’s cut to the chase. This isn’t super advanced.

But it’s only using one app: Lingvist. It was after using it for a total of twenty hours over the course of a month.

Speaking to him also exposed many things I could NOT say

  • “How are you?”
  • “Nice to see you again!”
  • “I’ll be back in 30 minutes.”

And many other basics. Basically, after a month of using Lingvist — and twenty hours of cumulative study , while I felt I could assemble sentences, there were so many basics that I didn’t know that I realised this isn’t for me.

Why even try to learn a language using only an app?

Any sensible language student would tell you that learning only using an app is a bad idea. In fact, this is part of what we recommend in general. At minimum, any app should be combined with a good teacher like one found through italki (see our full review and guide). German isn’t even one of my highest priorities.

The reason I decided to try Lingvist is that it’s an Estonian company that made it and we happened to be in Estonia for a little while. We tried to meet them (unfortunately, never crossing paths), gave it a try before meeting them and then decided to keep going to see how far we could take it.

The idea of studying for a short amount every day using only an app is alluring on multiple levels.

Firstly, AI seems like a much more intelligent way to teach. Atutor understands what words you know and what level you’re at, and drills you on your weaknesses as well as not letting you coast along on your strengths. A teacher normally does this well, but it seems like something you can entrust to a machine much more precisely. 

Secondly, an app can democratise education. It can make something very inaccessible — language study, typically provided through private tutors or university courses — accessible to a broader range of the population. Anyone reading this is likely fluent in English. Think of everyone out there in the world not fluent in English and not raised in an area blessed with a high level of English language teaching. Making learning English available to those populations is one of the things that could most transform their lives.

Finally, it appeals to the lazy person in all of us. Just thirty minutes a day! From anywhere! You probably barely have to pay attention! 

Overall, I think the results are decent, but incomplete. I didn’t understand a lot of what my brother said, but I managed to hold a basic conversation, and after 20 hours of only clicking and typing my way through an app, over the course of a month.

How does Lingvist teach?

Lingvist has a really interesting contextual algorithm that I haven’t seen anywhere except in the way we naturally study ourselves. 

It never teaches you grammar. You’ll never have to review conjugation tables, understand how sentences are formed or learn the peculiarity of pronouns.

Instead, it tries to teach you all of it contextually. For example, it’ll pose a question. In the above example, it put a question: “Die _________ hat einen neuen Sprecher” and told me that the blank meant “federal government” in the sentence “The federal government has a new speaker”.. The first time I come across this, if I don’t know the word (which I never do, as I’ve never studied German), I’m going to have to just hit enter for it to tell me. The next time, I have to type in “Bundesregierung“. If I do, it’ll show me this question less frequently. If I don’t know, or mistype or or spell it wrong, it’ll bring this card right back to the top of the pile.

How do you measure how efficiently you’re learning?

For many of us, the best way to assess how well you’re learning is speaking to people and seeing how far we can get.

For everyone else, there’s data. Or more precisely, charts. Man, these are great!

This is the one area where Lingvist absolutely excels. For the nerds among us, it’s very satisfying to see.

Lingvist tells you your

  • Active study time. How much time you’re spending per day reviewing your 100-200 flashcards (I typically did between that to continuously add 20-30 cards a day). The surprising thing for me is how difficult it is to keep focus. Half an hour of active study time takes me longer than half an hour.
  • How many words I have learned. I can also pull up a list (not shown) of words I’ve had to study the most, learned most recently, etc. I also find it interesting that it tells me how much German I know as a percentage of the language, based on the words I know. (This depends heavily on the definition of “word” – see below.)
  • Study time per day. This keeps me in check. If you’re the kind of person who likes to see big continuous lines, it’ll motivate you to keep it up there.

I like this level of data much more than I like gamification (like Duolingo, with things like hearts, lingots or whatever) or arbitrary metrics like how sprouted the trees are in Memrise. Those are fine. They’re symbolic, but this is quantitative. 

One note is that this heavily depends on the Lingvist definition of a ‘word’. A ‘word’ exists as one conjugation or one plural form. For example, the word Land (country) is listed as a separate word to Länder (countries).

This is beneficial in one way: you’re not obliged to learn conjugations that are never used, or plurals of words that don’t often need a plural. If you’re a language nerd who likes to geek out on obscure applications of words, you’re not going to learn them here. Also, it will never teach you everything, and will force your brain to improvise. If you’ve learned three conjugations and they look the same, you can look for a pattern and learn the fourth without ever being taught it.

On the flipside, it means that you get penalized for confusing similar conjugations or plural forms, and since the penalty is always the same (back to the top of the deck!) this could slow down your learning.

What I’d try to change if I were Lingvist

Firstly, I know Lingvist is aware of these and is working on them. This is a language teaching app built by data scientists, after all.

The hardest part of any algorithmic teaching system is to make it feel “human”. One principle an early teacher told me was that he’d overload me every day on purpose, teaching me 100 words and expecting me to retain 60% even if I tried really hard. He said this is fine; the brain is not a database, and it’s unfair to expect it to perform as such. Similarly, he recognized that I had certain weaknesses and he’d work on them with me occasionally, but he’d also let me move on and study other things while working on those weaknesses on the side. Finally, he recognized my interests, and didn’t teach me things that are irrelevant (while at the same time forcing me to learn things that are objectively important even if I didn’t know it).

These are the human aspects of coaching that Lingvist doesn’t capture. I think they can be, by implementing a few features, so I’d suggest the following.

Firstly, add topics. The teaching system used by Glossika, for example, lets students choose from a range of topics. I de-selected ‘business’, because I don’t want to discuss business if I’m not living somewhere. This would de-prioritise sentences in the Lingvist database like “The economy is growing”. The challenge would be that Lingvist tends to sprinkle in words you’ve already learned (or will learn later) as ancillary, contextual words.

Secondly, let us make typos. Sometimes, we mistype a word. It could be a fat finger error (because often we’re using the app on the phone) or hitting enter by accident. Maybe just an ‘undo’ button. Trust us! (Oh, I know sometimes people will cheat.)

Thirdly, let us remove words from decks. I don’t know if I’ll ever say “federal government” in German, and I’m sure I’ll get it wrong again. I mostly want to harass my brother and understand what people yell at me when I speed past them at 200km/h on my BMW S1000RR superbike down the famous Nürburgring public race track.

The challenge here is that there may indeed be core words that are important for understanding German, and it would be tedious to explain what those core words are.

Fourthly, tell us when words are related to other words. Lingvist currently considers as separate “words”: a) different conjugations of the same verb, b) different plural forms, and c) different case forms (e.g. when a subject, indirect or direct object, etc.). While it’s nice to see the word count go up, it’ snot realistic. And it’s also hard for us to understand how one word relates to another (unless we have a great memory).

Finally, obviously, Lingvist needs more languages. While I do want to learn Russian and German, they’re far down on my priorities list. There are many languages that are more important and globally relevant, like Arabic and Chinese

The Lingvist sentence database is created from many texts of the most commonly-occurring words, and I suppose what I’m asking is to let us learn using our own database of texts. This may be an impossible ask.

I know the Lingvist team is stacked with scientists and great developers, but I haven’t seen much change happen in the last six months since I started using it (It’s April 2019, vs October 2018 when we first started it). Perhaps a major release is in the works.

How we suggest using Lingvist

The way we’d start using Lingvist is a bit like a sentence bank, like Glossika (a website/web app we far prefer).

  1. Make sure you know the basics of the language first. Even though Lingvist is supposed to teach you passively, it’ll all make a lot more sense if you already know how sentences are built, how to read the characters (assuming they add non-Latin alphabet languages) and so on.
  2. Learn each sentence. Don’t just learn the word you’re supposed to learn: study how the sentence is constructed and memorise it (if it’s useful).
  3. Say each sentence out loud. Make sure you really know how to pronounce it!
  4. Play with substitutions. What other nouns/verbs would fit into the sentence structure? Make your own, and learn those too
  5. Take notes. I ended up taking a list of notes on words that were quite similar but I always got confused (like mond for “mouth” and Mund for “moon”, then Veröffentlichung for “publication” and Veranstaltung for “event”). Make your own word list.

The main caution I’d have against using Lingvist at all is that it’s easy to get stuck into the language learning trap where you do something because it’s fun and not because it’s teaching you anything. Think to yourself: “is this really more productive than time with a teacher?” It’s not likely that it is.

Is “Lingvist Unlimited” worth the price?

With respect to the Lingvist team: for language learners, as much as I enjoyed the app’s interface and the analytics, I don’t think it is worth the price or the time invested.

In free form, Lingvist is limited in two major ways. Firstly, there are only a certain number of new words you can add every day (I think it was 20). Secondly, you can’t use audio input.

In practise, these are not really limitations.

Adding 20 cards a day is pretty hard. If that’s your natural limit, you’ll be fine with the free version. If you’re going to go at a hundred miles an hour and adding many more words a day… well, good for you, but personally I’d spend more time elsewhere. Thirty minutes on any app is enough.

Secondly, we rarely used the audio input. I tried it a few times, and I found it a little tedious. I use it around 10% of the time, just to encourage me to open my mouth. Normally I’d be happy just mouthing the words.

If you use my referral link, you can use the free version and then at 250 words you’ll get a free trial to the unlimited version. Bear in mind that current pricing is extremely geared towards getting a yearly membership.

Give it a try and see what you think and let know if you like it!

Share this: