Rakuten, Japan’s 21st Century Online Powerhouse, to Launch 19th Century Language School

One of the leading online retailers in Japan, Rakuten, is getting into the English-teaching business. According to their press release, their “Super English” lessons will use software developed by a new startup in the language teaching “space,” Lingvist.io.

Here’s how Lingvist summarizes its system:

Lingvist applies mathematical optimization and statistics to make the language-learning process as fast as it can theoretically be. Learners are introduced to vocabulary according to statistical relevance, ensuring that they start by learning the most widely used words. The program keeps track of a user’s mistakes and progress, and it adapts the course to each and every user with machine learning algorithms.

A quick look at their lessons reveals that what Lingvist really does is spaced repetition to help you memorize vocabulary words, an idea taken from the latest in cognitive science research, Ebbinghaus (1885).

In most variations of spaced repetition, you are presented with a series of “paired-associate learning” tasks: here’s a word in English, here’s that word in Japanese. You are shown (or guess) the meaning first, then the words are shown to you again at certain intervals that are supposed to help you remember them.

There are at least four problems with this “scientific” system:

  1. No one actually knows what the optimal “algorithm” for the spacing should be, nor for what kinds of words – and it’s been studied continuously by scores of psychologists for more than 100 years
  2. Teaching a word by giving its definition (or word and a one or two-word translation), a form of what is called “shallow instruction,” isn’t nearly as effective as other methods when it comes to helping you comprehend real-world language use –  say, when you read something with those memorized words in it. The results of shallow instruction, no matter how “mathematically optimized” you make it, will always be shallow.
  3. Acquiring new words using spaced repetition has not in fact proven to be any faster than other, less complicated and painful ways of acquiring new words, like reading and listening to interesting language that is comprehensible to you.
  4. There is no published evidence that I’m aware of that the solution can “scale;” that is, there’s no evidence that anyone has actually acquired a language to a high level of fluency primarily through memorizing a large number of words via spaced repetition (remembering that to read fluently in English requires upwards of 8,000 word families).

Studies of spaced repetition usually report on their subjects’ success in learning X number of words, not on acquiring fluency in a language, and even this encompasses a very short range of time (typically hours or a few days at most). Memorizing isolated vocabulary words is far from actually acquiring the full range of abilities needed to become fluent, and embedding the words in few sentences, as Lingvist does, is probably only slightly more useful.

There is, on the other hand, a massive amount of evidence you can acquire a language through listening to and reading comprehensible input. Getting good comprehensible input gives you not only a broader and deeper knowledge of vocabulary, but also grammatical fluency, writing, pronunciation, reading speed, spelling competence, and knowledge of the world (Krashen, 2004). It is also much more enjoyable.

Despite the slick website interface, fancy charts and graphs, and promise of a “scientific” approach, Lingvist’s method (and those like it) is in fact a highly inefficient use of time when it comes to language acquisition.

Full disclosure: I help run a website that teaches English to intermediate and advanced students based on giving them comprehensible input.



Print Friendly, PDF & Email