If I were really doing what I wanted to be doing most of the time, what I would be doing (aside from spending time with my wife, my son, my brothers, my mates, enjoying good things to drink and eat, listening to music, sometimes dancing to it, occasionally making it myself, wandering around London, wandering around the world, running, cycling, or playing tennis, watching other people play football, being in the water, being on boats in the water, fiddling with gadgets, and doing lots of reading, at home, in the pub, in the park) is writing. This website is a place for me to share some of what I write (and build), presented in a way that I determine.
Two topics that I spend a lot of time thinking about are language and technology, which I guess tells part of the story of how I ended up calling myself, professionally, a computational linguist. It’s my good fortune that I get to spend a lot of time using machines to do things with words. If you’d told my ten-year-old self I’d be making a living this way some day, I would have been pretty happy.
But there is a trap that people like me who work with language technology can fall into: we start to think of language itself as a technology, and this is a problem. It happens when we start to see language as a tool the function of which is to specify an informational process, outside the language itself, outside any effort to directly communicate. Part of the problem with thinking of language as this kind of tool is that it assumes the main functionality of language is to organise the world into objects with types and attributes. The problem compounds when language-as-a-tool is applied to language itself, risking a reduction of language to merely syntax waiting to be dressed up with semantics. This is not to say that thinking of language in formal syntactic terms is not productive – programming languages allow programmers to build fantastic things. But things go wrong when we start thinking of all linguistic communication as a way of just porting around information; we begin to lose site of each other, when something about what it’s like to be ourselves is really the most important thing that language can communicate. I think this must be part of what the philosopher Ludwig Wittgenstein meant when he said “philosophical problems arise when language goes on holiday”.
There is a lovely story to tell about how, when our ancestors first started talking to one another, it was done as a method for them to align with one another in an unpredictable and perhaps often hostile environment. According to this story, the earliest use of language was not about acts of naming, describing, and classifying; it was actually about empathy and the virtue of humans being able to exchange perspectives on the world in order to accomplish things together that could never have been accomplished alone. In this version of how language started, communicating with words is wrapped up with the emergence of a species – our species – that was able to travel great distances and work collaboratively to bend the environment to its will, thanks to an ability for strangers to come together and, through words, achieve meetings of minds. This origin story of language is recapitulated by almost every individual human when they go through the process of developing their own specific linguistic ability. It’s significant that we are all born with a capacity for learning some language, but none of us are born with a particular language already encoded. This aspect of being human means that we each need to go through the process of coming to grips with the world we find ourselves in, and, at the same time, in coming to grips with the world, we have a chance to update the language that our specific communities and cultures use to more accurately reflect what we find and what we make. The fact that language can adapt to a changing world much more rapidly and much more directionally than a genetic code is the main reason that humans have been able to supersede the glacial pace of evolution by natural selection. It is these two features of language, its telepathy-like ability to transmit mental experience and its developmentally enforced adaptability, that has made humans the way we are, and in turn has allowed us to make our world the way it is.
I’m trying to clear some space here to talk about AI. I’ve always been interested in the idea of machines that could communicate and collaborate with people, and I’ve been lucky to study and work directly in this field since 2012 – I’ve seen some stuff. And language and linguistic data have become incorporated into AI in some ways that have surprised me (and I think probably anyone who’s being honest about it), but AI does not learn or use language like we do, at all. In particular, AI is not an entity that is equipped to take a stance on something, to promise, to avow, to threaten, to doubt: these are speech acts, things done with language as in a way that makes language not technical but instrumental. The thing that AI models is not having these intentional stances; the thing AI models is merely semantic, acts of summarisation, translation, encoding, or converting.
So, about me: I’m a human, who, like everyone else, learned a language from scratch, on the go, the hard way. I only speak one language with any degree of authority, but I really like to speak it, and one of the things I like doing with language is talking about language and technology for processing language. Language-about-language is a special category of language, and the people who indulge in it tend to end up writing fairly long sentences. I also like to use language-about-language-technology, which is in part language-about-language but one more degree removed, because the technology itself doesn’t use language, doesn’t “do” language, is not “languaging”.
But I personally do not use technology to write language for me. This is all me, for better or worse.