Living in a speech bubble

Living in a speech bubble
Notes from learning languages later in life – is there a digital solution?

I grew up in rural oxfordshire in the 1970s. I was taught languages in the 1980s. It didn’t go well.  In the 90s I studied AI and was happy learning programming languages. But natural languages remained a mystery.  This century I wanted to change that. I started to travel more and the more I travelled the more obvious it was that I couldn’t communicate outside of english and my own ‘language’ of hand waving… which I think we can all agree is all rather embarrassing. Hands up if you’ve been embarrassed by me…

But then smartphones and apps arrived and Duolingo was born.

I had something that would help me learn languages anytime anywhere. The key to my multilingual future had arrived.  And I’ve gotten pretty good at it.  The only downside to my amazing DuoLingo french and german is that while I’m amazing in DuoLingo… I’m not quite so good in the real world. It’s a little bit like how well homeopathy and vitamin supplements work incredibly well.. Right up until the point that you become sick… It’s easy to convince yourself that Duolingo is teaching you to understand and speak a new language. The Koolaid feedback of “you’re 20% fluent” can very easily lead you down the path of false hope. Don’t think for a minute that you will survive a taxi queue at Tegel airport if you have learnt German from that friendly multi-wardrobed owl…

Something needs to change. I’ve become incredibly adept at making that DuoLingo ‘sentence’ sound that gets you through the voice recognition task.  I have, like many other ‘games’ learnt to play the app more than I’ve learn to learn a language.  Don’t get me wrong, I think DuoLingo is a very useful tool and a powerful learning aid – but it’s speech recognition is currently terrible.

So I’ve started to play with other things.

In October I turned Siri to German.  It’s kind of weird and as in your native language is mostly totally useless. However, It’s useful for spelling out the words as you say them. It’s useful, but always seems to resort to “I’ve found this on the web for you” or “Ich habe dies für Sie im Internet gefunden”.   But I like it  and I’m not sure why. Perhaps I feel someone is speaking to me in German not judging me like the man in my local wine shop does…

I’ve tried a range of free translation apps – and the UI on them is generally so incredibly bad you never really get to evaluate their translation qualities. Popups when you’re learning is a terrible feature.

Smartphone apps have  amplified the power of speech recognition. They’ve taken it from a lab or a desktop activity into real life.  An example of this is with our polish builders in Berlin who use Google Translate to tell us the things they want to tell us

PB: “Kein wasser heute”
Me: “Wann ist  das”
PB: “Jetz”.
Me: “OK”.

What was interesting is that they started using the audio on google translate as a natural thing they do. I then started to play with google translate audio functions. And it’s pretty good.  I tend to use it to support the german texts I’m learning from. I find it a powerful way to test pronunciation – and also to hear the correct way to pronounce things when I’m not sure.

google_translate_1Some of my attempts at google translate. Interesting how Wie Geht’s was translated as Gate Inn and this prompted a Tripadvisor link…  

What’s great is that if you think the technology isn’t working – let’s say you don’t think the speech engine isn’t up to scratch and it’s not your german, it’s google’s tech…. Then you can switch back to your native language and check.  Simple things work well. Like learning the German alphabet.   I was convinced google translate wasn’t working when it wouldn’t recognise my version of the alphabet. So I checked in english… and it worked fine. So I went back. Listened to google’s translation of the english alphabet and then retried it. And it worked.  I’m now more confident when I ask for me “ah-bee tageskarte” from the shop outside Nordbahnhof.  

What I think is interesting is just how much difference putting speech recognition on a mobile device made and how much it is opening the door not just to translation but to learning.  
With the potential for the IoT to further amplify  (think in-ear live translation) language translation and its effect on the world, I think we’re heading into incredibly exciting opportunities. However, it will also amplify the ability for surveillance, for large-scale data harvesting and algorithmic decision making. Things that in the current turbulent political world, I think we should all be thinking a lot about.  

Leave a Reply

Your email address will not be published. Required fields are marked *