Richard Howeson has founded and run language learning company, uTalk, for more than 30 years and seen first-hand the pros and cons of using technology in education. The company’s app, also called uTalk, gives beginners the choice of more than 150 languages to learn. Howeson discusses the use of automatic speech recognition and where he thinks EdTech works best.
As the boss of a language learning company, I’m often asked about automatic speech recognition (ASR) – one of the standout developments of interactive tech – which we have absolutely no plans to use.
This is despite the fact that many other language learning apps use ASR to give learners feedback on their pronunciation.
The problem is that ASR is not accurate 100% of the time and sometimes tells you that you’ve got something right when you’ve got it wrong and vice versa.
It can also struggle to understand accents and if, like us, you have 150+ different languages including, for instance, five different varieties of Arabic, that’s a big deal.
So we looked carefully at ASR when we were designing our apps (and we still closely monitor it) but what we found worked best was the same high-tech piece of kit that helped us learn our first language: the human ear. So, in the uTalk app, learners listen to native speaker audio in their target language, record themselves copying it and then compare the two and mark their efforts with a tick or a cross. Perhaps counter-intuitively, we’ve also found that learners are more likely to be hard on themselves rather than overly generous when it comes to self-marking.
I mention all this because, when it comes to EdTech, it’s crucial to evaluate what works best for the people choosing to learn with you rather than jump on the latest bandwagon.
That’s not to say that language learning hasn’t massively benefitted from developments in EdTech – in particular the way in which educational content can be tailored to people’s learning strengths.
For instance, research shows we’re generally better at remembering visual rather than written information and, in the same way, we also tend to be better at remembering spoken words and phrases when they’re linked to a picture. So our app, for example, uses a picture of someone pouring a cup of coffee next to audio of ‘I’d like a coffee, please’ in whatever language you’re learning. It’s a learning technique called ‘dual-coding’.
It also seems likely that, in the future, this technique will be further developed so that learners can use Augmented Reality (AR) headsets or an AR device experience to ‘see’ or personally interact with a visual representation of the word or phrase they’re learning – alongside audio of the word or phrase. It’s likely to make for lasting memories.
Another way that technology has impacted on language learning is the ease with which developers can now use interactive apps to help people learn by playing games.
You can play educational games interactively against yourself, against your peers or against Artificial Intelligence. You can also play online or offline, remotely or in the same room.
The best language learning apps also use fun, progressively challenging games, track a user’s progress, set attainable motivational goals and reward achievement. (In the uTalk app we even test learners more on words or phrases they get wrong.)
But there is another side to educational apps. Some of them, like ours, give individual students the chance to work at their own pace during the different learning exercises, letting them repeat those exercises as often as they need and, importantly, giving them the choice of learning content that interests them. (As many of us can vouch from our school days, learning is much harder when you’re not interested in the subject matter).
Looking into the future, I think it’s this personalisation of EdTech that will be the one of the next big things. Educators are increasingly aware that we all absorb and respond to new information at different speeds, it’s called processing time. Importantly slow processing speed doesn’t equate to lower intelligence because you can be a deep thinker without being a fast thinker. But it does mean some of us need to be given new information at a different rate than others. So it’s likely that apps will soon be measuring the rate at which individual users learn and then automatically either slowing down or speeding up information delivery depending on a person’s needs. The ‘one-size-fits-all’ method of learning is definitely on its way out.
So, what are some of the dangers of the use of tech in education? In terms of language learning apps, I’d say it’s expecting people to learn a new language by copying computer generated speech rather copying audio from real people. This is happening with at least one of the big players in the industry with the aim of increasing their range of languages and cutting costs. Call me old-fashioned but I don’t think AI can ever replace the emotional timbre of a real voice and nor would I want it to. At uTalk we’re committed to supporting languages, particularly minority and endangered ones, and that means supporting the people who speak them as well. So we always work closely with our native speakers and the days we spend alongside them in the recording studio are some of the best. That’s when our speakers share their stories with us, not just their language, so we get a real sense of what it means to be part of their culture.
As well as connecting with our native speakers, we also go out of our way to talk to our learners: our language staff go to all the big language learning conferences, our customer service team respond to requests for advice, a staff member replies to online customer reviews and our languages manager keeps a wish-list of all the languages people have asked us to add to our app. I also chat to customers every week because we believe in keeping things real.
So my take-home message to you is that EdTech is a very useful tool but you have to use it rather than let it use you. By that I mean you have to prioritise the human side of language learning over the AI side. For us, that means recording authentic speakers, avoiding ASR and putting our customers at the heart of everything we do.
