I got to record the video thanks to the Chartered Institute of Linguists. Here’s my full take on the issue.
This app is helpful when comes to basic expressions and moving around a city you don’t know, to a similar extent a phrasebook with pronunciation would work, but it saves you the trouble of pronouncing the text yourself. A significant improvement lies in the fact that the app can also capture what the other person says in a foreign language and translate it for you. However, all of this only works if we’re talking about the literal meaning of words, at a dictionary level.
It’s important to take into account that meaning as such doesn’t reside just in single words. Let’s take ‘a dog’ as an example. On a basic, literal level, I’m sure Google Translate would handle questions such as ‘Is it your dog?’ or ‘Where’s my dog?’ without any problems. In here we’re referring to the literal meaning of ‘a dog’, i.e. a furry animal with four paws and a wagging tail. Things get a bit more complicated if we enter the figurative level of meaning, used more often than you realise. Imagine you’re enjoying small talk with a tourist and you say ‘it’s raining cats and dogs’. In here, you obviously don’t mean that there are actual cats and dogs falling down the sky. It’s a metaphor for something else. This is an easy metaphor, but even this one already causes Google Translate problems. The reason behind it is that figurative language refers to the meaning that doesn’t come from just the sum of words, but a combination of specific words and context.
The biggest challenge, and at the same time the beauty of communication, lies in the pragmatic level of meaning, that is the meaning that is constructed based on knowledge held by the participants of a conversation. Take ‘it’s a dog’s life’ as an example. It may mean two totally opposite things, depending on who, when and where you’re talking to. It’s a dog’s life will mean a bad or difficult life if your interlocutor knows that when first domesticated, dogs used to guard houses outside and lived in poor conditions. or it will mean a great and care-free life if your interlocutor happens to live in Chelsea and give her dog frequent bubble baths. How will a machine ever know who’s talking to who without actually being a participant in the conversation? How will a machine ever be able to analyse language not only through its dictionary meaning, but figurative and pragmatic levels, too? And we’re only talking about dogs here, imagine the complexity of human expression in arts or sciences, law or medicine, or the entire world of business (e.g. ‘go to the dogs’, ‘dog-eat-dog industry’, or even ‘cats and dogs’ in finance).
There’s more to language than just the literal meaning, and it’s often believed that the pragmatic level of understanding is reserved only to humans. At least for a few more years to come.