Experimenting With Eye-Tracking

It seems an age ago now that I was sitting in a Cuban restaurant in downtown Stockholm with my supervisors discussing a workshop I'd attended on eye-tracking with the NorPhLex group. Over a delicious glass of Chardonnay and the most colourful coconut risotto I threw out a suggestion from the very top of my head: "I mean, it'd be great to test how infants respond to cross-linguistic onomatopoeia, I reckon". I still remember the look on their faces: delight mixed with regret that it couldn't be done back in York. Five months later I'm back in Stockholm, using that very same eye-tracker that I was using on that day, and carrying out a metamorphosed version of that very experiment that I suggested over Chardonnay and risotto that evening in Stockholm.

Eye-tracking basically involves displaying words on a screen (with accompanying audio, in my case) and using infra red to trace participants' eye movements and fixations throughout. These are then stored as data in a nice cushy analysis software, which will generate reaction times, p values, averages, the lot. The technique can be used in myriad fields: marketing, (neuro)psychology, medicine, media, cognition, development and linguistics, and is easy to manipulate to different audiences as you can display pretty much any kind of media on there. I believe you can even track the eye movements of animals, but seeing how hard it is to get one-year-olds to focus for 3 minutes and 49 seconds, I imagine that calibrating a dog's eyes is only for those with the most solid determination.

If only all my participants would behave like this.

Conjuring up my own experiment has been a long and pernickety process, very much in the vein of 'one step forwards, two steps back'. Still, having nurtured this project from the tiny seedling of an idea that it was has been one of the most exciting things I've ever done, and seeing it out there in real life, with real participants, is a little bit of a geeky dream come true. Until now I've always worked with databases, using information gathered by others to come up with my own analyses and ideas. This works fine, and it makes me tick in ways that nothing else ever has (I kind of think that I was made for data analysis), but I have felt like a bit of a fraud since starting my PhD, having had the initial intention to ride on other people's hard graft to generate my own results. I work with babies in theory, but in practise I hadn't seen a single baby until Wednesday, when my first participant rolled up in his pushchair, enthusiastic mother in tow.

Running an experiment that is my own has given me the taste of project management that you don't often get in the workplace: the investment, wholeheartedness and obsessive pernicketyness is on a whole new level, and the sense of ownership is both scary and magical. Luckily for me, I am working with a team of PhD students who are astoundingly excellent, and they have shown me the ropes while also respecting the fact that the experiment is completely mine. I have learnt a whole world of new skills and knowledge in only 2 weeks, from input statistics to basic coding to rudimentary Swedish and essential childcare methods. Without them I would have (and just about did) flounder at the first hurdle, and even as I lead the project in its reality, I am still learning from them with each participant that comes my way.

Testing, testing...

The biggest challenge in using eye-tracking with one-year-olds is the one-year-olds. They are all so different, and all so excited by every small thing that happens within the realms of their five senses. It hasn't helped that I am a complete novice with infants; an awkward and rather hesitant Englishwoman whose Swedish is questionable even to those who have not yet learnt to speak it. At first I wasn't convinced that they would even look at my experiment, never mind give me enough data to use in the paper that I hope to write once I've finished here. But they do, they really do look, and they really do respond in the way that one would hope they would respond. It's not consistent or focused, but it's natural and it allows a sight into the developing mind of an infant that can't be found in even the most magnificently detailed database (even Deb Roy's amazing approach doesn't provide details about pre-speech understanding on the input level). Plenty of people out there snub any sort of generated experimental data, believing it to take away from the natural process of language production/perception, but there are always going to be limits in research, and if we want understanding I think we have to be prepared to reach those limits, and considering the potential impurity of the results is part of our jobs as researchers.

So I am excited to be using eye-tracking. I'm excited to see my idea come to life, to have results as a result of my idea, and to lead something that will eventually (I hope) lead me. It's kind of fun working with babies too, as some of them are delightful, but I think it will take a lot of time for me to come round to being comfortable with them. To me, infants are exclusively interesting, and I'm not really bothered about studying adult language at all, so career-wise it has come as a rite of passage to actually work with real-life language learners. Whether or not this will translate into a full-blown turn from databases towards lab work I do not know; seven participants in and I can see that there is nothing less predictable than a one-year-old in an unfamiliar setting. I guess that if I liked predictable then I wouldn't have ever considered a return to academia. It has certainly been a challenging couple of weeks, with organising, setting up, arranging appointments (with thanks to my very helpful Swedish research assistant!), greeting, explaining, playing, distracting, analysing and endless processes of learning and re-thinking, but I feel a little bit stronger and a little bit smarter and oh so in need of a glass of wine!