At Linguineo we like statistics and anecdotes about language learning and we will try to share as many of those as possible with our community. In this post we will take a first look at the statistics that a bit more than 100.000 exercise reports offer us.
Linguineo currently offers 9 language classes: Spanish, French, German, Italian, English, Japanese, Portuguese, Russian and Dutch.
As the table below shows most exercises are being completed on Spanish. The runner-up is different per platform: for Android this is Italian, while for iOS both English and French come in second. On both platforms, the share of these 4 most popular languages is about 85-90%, dwarfing the share of the other 5.
We also want to look at who is learning which language and in which country each language is being learned. Unfortunately, our data with regard to this is polluted due to the fact that some of our apps are only available in English, most are available in Dutch and English, and only some are available in 5+ languages.
Our Dutch Class app for example is mostly used by English speakers, but since this app is only available in English, it is unclear that this is because there are more English speakers learning Dutch than say French ones, or that the fact the app is English only poses a significant barrier to entry for French speaking people. Maybe the latter?
Let’s take our English Class app as another example. This app is available in 8 languages and is mostly used in the United States, which might seem a bit odd, since English is the main language there anyway. But a closer look shows that many of these downloads are done by users whose native tongue is Spanish. This is an example of one of the patterns we often see emerge: by looking at the usage of a language app in the country where its subject language is spoken, you can predict which foreign speakers are learning that language there. Of course, again, entry barriers of the language that the user is being tutored in not withstanding.
Just as interesting is which nationalities are learning the language of another country while not moving there. English speakers in the US for example clearly prefer learning Spanish, with Italian and French coming in second. In Europe, a variety of languages are learned, with French, German, English or Italian being the most popular depending on the country. In countries like Brazil and India, English clearly wins: these countries all have higher download numbers for the English Class app than any European country while almost not having any downloads for any of our other language learning apps.
We noticed some other patterns and interesting things, but decided to keep them for a future blogpost. The new app that we are developing will be available in 15+ languages and tutor more than 20 languages, which should provide for a less polluted and less incomplete source of data.
We also tried to get an indication on “What is the most difficult language?”. The averages of the sum of all exercise report results for a language all end up somewhere between 70% and 90%, with Japanese, Russian, Dutch and German at the lower end of the spectrum(<80%), English in the middle, and the Romance languages French, Spanish, Italian and Portuguese at the higher end of the spectrum(80%>). It is not clear that this measurement has anything to do with the difficulty of the language though. It is possible that if we dig deeper into the data, which we plan to do later on, we might find completely other reasons for this difference. Other things we noticed with regard to difficulty are that users of the demo apps make a lot more mistakes(average result of 70%) than the users of the full course apps(average result of 85%) – dedication anyone? 🙂 – and that some exercise modes – for example writing versus multiple choice – are much more difficult than others.
Now we are talking about exercise report statistics: most exercises are completed in the weekend – even a bit more on Sunday than on Saturday – and the exercises with 10 and 25 exercises are most popular: for each language we land somewhere between 12 and 24 for the average number of exercise items. With the demo users – dedication, again, anyone? – preferring shorter exercise series than the users of the full apps.
Again, we save the most interesting data for last: which exercises are our users doing?
The table below shows that word exercises are clearly the most popular, accounting for more than 50% of all the exercises being done.
|Speaking exercise (Android only)||2%|
The next table shows which word exercises are done most(we left out some rare ones, such as “Pick the correct letter” and “Phonetic -> Tutor language” which are only available in very specific cases), with the easiest default category of “Language being learned -> Tutor language” having a share of more than 80%!
|Word exercise on gender||1%|
|Word exercise – recognize the image||7%|
|Word exercise – language learned to tutor language||83%|
|Word exercise – tutor language to language learned||9%|
This means that almost 50% of all exercises being done are word exercises in which the user answers what a foreign word means by translating it to his own language. Although – no more jokes anymore about dedication 🙂 – this pattern is much more visible among our demo users and full app users complete a much wider variety of exercises and are clearly taking advantage of the grammar sections of our apps, it made us view our current course apps from a new angle. We treated this information as valuable input for our upcoming new app.
We conclude this post with some trivia: the most popular word – the word that users answered the most is the Spanish word “llevar” – “to take, carry” which was answered 1235 times in the sample. The most difficult words are the Spanish word “el corzo” – “roe” – which is the word that was most often answered incorrectly (but also 544 times correct)- and the Japanese word “～つづける” – “to continue doing”, which is the word that was never answered correctly with the highest number of incorrect answers: 4.
ASK YOUR QUESTION
This post is our first post about our language data. If there is enough interest we will be posting more of these. If you have an interesting question or viewpoint you would like us to take a look at, do not hesitate to ask us here or on our Facebook page.