Multilingualism and technology
This is the English version of a post originally in French, accessible here.
Ceci est la traduction anglaise d’un article écrit en français, disponible ici.
As a computer engineering student, and somebody who lives in the information age, it never ceases to amaze me how prevalent technology is in every industry of modern society. You’d have to be in a prison or living on a desert island with no civilisation to be unaware of its spread! And yet, in spite of how it’s everywhere, it still has growing pains when being used by people who make use of multiple languages every day.
First, something that we use every day: a keyboard.
The list of keyboards that I use on every computer is four or five items long: English, French, Spanish, Simplified Chinese, and Traditional Chinese. On the phone, I use Swiftkey, which lets me scribble in English and French; unfortunately, it doesn’t support adding a third language. And in addition to those five, I have two ways of entering each version of Chinese: with pinyin, and manual handwriting.
On each platform that I use (iOS, Android, Windows, Mac, Ubuntu), the software configuration for keyboards is ever so slightly different. For example, I’m typing now with the Canadian French arrangment - on Windows, it looks like this; on the Mac, it’s more like this. Switching between keyboards is a whole other beast too: the list of keyboards is presented by a user-configured order on Windows, iOS, and Android; on the Mac and Ubuntu, the list is organized by the last time you used that keyboard.
The world of keyboards is very chaotic, and there really isn’t an optimal solution.
Next, let’s take for example, virtual assistants like Google Assistant or Alexa. I love home automation and smart speakers, but there are some serious issues with their multilingual capabilities.
Alexa, created by the ubiquitous everything store Amazon, didn’t speak French until last week, four years after its launch in English in the United States. She still doesn’t speak Canadian French, which, we could say, has some pretty major differences with its European cousin. She also does not speak Spanish nor Chinese, leaving out a substantial portion of the world.
The competition: Google Assistant, which does a bit better in terms of languages. It can, at time of writing, speak eight languages, with English (localized for six different regions), French (both Canadian and French French), Spanish (American, Mexican, and Spanish Spanish), German, Italian, Japanese, (and apparently Korean and Portuguese, but I couldn’t find those options in the Google Home settings). Google says that there will eventually be more than thirty languages supported in the future, et also promises multilingual comprehension for English, French, and German.
Siri trounces the other two, with so many languages as to even distinguish between mainland Cantonese and Hong Kong Cantonese. What brings it down is the price of the Homepod, its absence in the majority of marketplaces in the world, and its atrocious understanding of user intent. In the grand scheme of things, she’s pretty irrelevant.
All three have problems when they’re exposed to a multilingual environment:
- Alexa (in France French mode) can’t turn on nor turn off my lights because they’ve got English names that she apparently can’t pronounce. She also doesn’t respond to “Turn off all my lights” because she thinks I have a lightbulb called “all my lights”. (???)
- When Google Assistant’s Spanish became a [thing]((https://9to5google.com/2018/06/11/google-home-spanish-dialect-us-mexico-spain/), I tried it on the first day… but it couldn’t control my lights for some reason.
- Google Assistant couldn’t play my playlists on Spotify unless I put on a reeeeeeeeaally exaggerated Québecois accent (and nobody wants an Asian anglophone attempting to sound like somebody from Saguenay.) “OK Google, jouez ma liste de lecture deeeescauveeerrr weeeeekleeeeeeee!” In fact, it’s generally just difficult to get Google Home to play any specific track with a title or artist in a different language.
- Google Assistant, for some reason, couldn’t pronounce “Waterloo” in French for several months, butchering it to make “Watam” - sorta like “what time”. So every time I would ask for weather, it would respond “Today, in Watam…” (but in French, of course).
- Siri, set to Cantonese, can’t understand my friends’ names in English, like “Edward”, “Megan”, or “Ellen”.
- Siri, in English, can’t understand my family’s Cantonese/Mandarin names.
I get that these problems are some of the most “first world problem” problems that one could have. “Oh, boo hoo, your smart speakers don’t work. What will you do?” But my attitude towards technology is this: it should be so easy to use that we no longer recognize to what point we depend on it. And every time my smart speaker tells me “Sorry, I didn’t get that!”, the illusion shatters a little bit more.
At the same time, I do have to be very thankful to all the benefits that technology has brought to people who speak multiple languages or that want to learn multiple languages. I don’t think I would have learned to speak French without podcasts, like the great Le rendez-vous tech and Les années lumière. My Kobo e-reader has given me the ability to access a vast library of foreign literature, without needing to have a dictionary beside me at all times. Forums and websites in other languages are plentiful, and are great places to be exposes to a more more informal register of the language. And finally, of course, the telephone, which I’ve used to connect with my family that lives an ocean away.
I’m optimistic. I’m immensely interested by human-computer interaction, and I think that an improved understanding, and in particular, a recognition of the existence of a multiple-language use case, will make technology even more seamless in our everyday lives.
aaron at 20:04