The Lingua.ly app is available for free on Android and will be rolled out on IOS later this year

Virtual language teacher on your mobile

A virtual language teacher on your mobile phone that can recommend real-world reading material tailored to your ability has launched today.

Lingua.ly is a new cloud based translation tool that helps users to read articles from foreign language publications on their mobile by allowing them to click on words they do not understand to get an instant translation.

But unlike most online dictionaries the app then saves those words the user has clicked on in a series of flash cards, which can be augmented with images and audio, and uses a space repetition system to periodically tests the user and see if they know the word, with intervals increasing each time the user gets it right and decreasing when they do not.

The app also collects statistical data on words the user has clicked on, but also words they have seen and not clicked on, to build a map of their vocabulary and then rank real-world articles by suitability for the user’s level of comprehension – the optimum level of word recognition for learning is between 90 and 95 per cent.

“Most online dictionaries, once you’ve got the meaning the interaction is over,” said CEO and co-founder Jan Ihmels. “Google Translate doesn’t map knowledge, it doesn’t take learning criteria into account so basically it’s impossible to learn using Google Translate.

“It needs to be able to look across all the thousands of web pages and change the parameters for each user. That’s the main technological challenge.”

The firm launched a web browser extension with no marketing to test their concept last August, which now has 30,000 regular users, and the firm is also carrying out an in silico experiment, the results of which they plan to publish in summer, to investigate how virtual users’ vocabulary changes over time.

So far the research has found that one third of virtual users become regular users, while 15 per cent become power users using the app four to five times a week or even daily. The research also found there is no upper limit to vocabulary expansion.

“The system is unbounded. You can even reach very advanced or technical levels of vocab. The system keeps on propelling you forward,” said Ihmels.

The app is currently only available on Android but an IOS version will be available by the summer, and the company is not only targeting individuals but also teachers following a pilot with a several college classes.

“The teachers absolutely loved it because it takes away one of the most frustrating aspects of their daily work – choosing content for their class,” said Ihmels. “And for students it was also great because it allowed them to choose something they’re interested in rather than something centrally dictated.”

According to Ihmels, this ability to allow users to study a language using real-world resources, rather than the artificial content found in text books, is one of the key advantages of the app.

“Almost no-one learns from a class or a text book. They give you the initial basis but then you have to go and expand that basis on your own,” he said.

“I think this is exactly the strength of our system. It is exposing you to real world articles that show how a particular word you look up is used in real life. I think this is how children learn, basically discovering how these words are used in a real context.”

The app combines an API for an online dictionary called Babylon with an algorithm designed by Ihmels, who has a background in computational biology and physics. “The algorithm we developed for this is inspired by something I used in the context of gene expression data for my graduate work,” he said.

Both the app and the browser extension are currently free, but the company aims to add paid content later this year – the firm is already working with a synthetic speech company to provide custom audio lessons for premium users.

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them

Close