For best experience please turn on javascript and use a modern browser!
You are using a browser that is no longer supported by Microsoft. Please upgrade your browser. The site may not present itself correctly if you continue browsing.

Once a linguist, then a phonetician, soon a cognitive scientist

An interview with Kateřina Chládková
K. Chládková

“My current position here at the UvA is for just one year, so I have only two months left to finish what I am doing! The project is about speaker and accent normalization in speech perception. It is a joint project of the University of Western Sydney, where Paola Escudero (the project leader) is based, the University of Amsterdam and Leiden University. Here in Amsterdam, we test babies and adults, in Leiden they mostly do experiments with zebra finches, and in Australia they also test babies and adults. Our experiments are aimed at finding out whether listeners handle speaker differences between sounds and dialect differences in the same way. It seems that for dialects you obviously have to have the lexical knowledge, or linguistic experience to be able to normalize for those differences. But it might be the case that the speaker or gender normalization are something more fundamental in our auditory system, maybe not only for humans but for other species as well. That is why we use so many different groups: little humans, big humans, and birds.”

“I did my BA at Palacký University in Olomouc, Czech Republic. I actually started doing mathematics and English philology there, but after 1,5 years I dropped mathematics. I do not know whether it was the subject that was so difficult, or the approach in which it was taught. You had to memorize a lot of things and I did not like that. I stopped when I had my exam for geometry, together with my friend who was studying English and mathematics as well, and was also struggling with this subject. At the exam I just told her: ‘I am done with this, I am leaving, are you coming with me?’ And she did.”

“I continued with English and Dutch philology, but I was very much interested in linguistics and phonetics. Because I was studying Dutch, I did an Erasmus at Utrecht University, for half a year. That is how I got to know people at Amsterdam, and I decided to do my masters here, because it was not possible to do phonetics and linguistics in depth at my old university back then. I wanted to do the two-year research master originally, but I did not get a scholarship, so I chose the one-year master, to be sure that I would have enough money to survive.”

“My PhD project started really suddenly. After the MA, I was planning on going back to Olomouc, but then Paul Boersma got the VICI grant for the project ‘Emergent Categories and Connections’. I really liked the project, and applied for a PhD position. My project was supposed to be about how we learn categories in speech. But it turned out that we do not really know what these categories are, so I wanted to find out what the phonological categories are that we actually learn and use when communicating with people.”

“My project was about phonological features in speech perception. I did experiments with adults, behavioural perception and EEG experiments. Also, I simulated speech learning with neural networks. I found that real people listen to speech sounds in terms of phonological feature categories (and not only in terms of phoneme categories). A phoneme, like /p/ or /a/, is the linguistic representation that corresponds to what we experience as sound segments in speech. A feature corresponds to (most often) a smaller entity than a segment. For instance, the segment ‘a’ has several important frequencies, a particular duration and intensity, and some of these composite properties of the sound segment are, in the language user’s grammar, encoded as features. My simulations also show that the artificial brain is able to create feature categories on the basis of the input that it gets during language acquisition. In short, in my PhD project I showed that we have phonological feature representations, and that we use them when listening to speech and when learning it.”

“The idea that people have “feature detectors” already came up in the 1950s. But then quite suddenly researchers seem to have abandoned this idea, and started to argue that we listen to phonemes. So nowadays, when you say that language users have feature detectors, people not always believe you immediately. For example, I presented my findings at the University of California, at a phonetics and phonology seminar, and they seemed to be perceived as quite controversial. But that is actually a very good thing, because it heats up the discussion and can lead others to design their own experiments to challenge the findings.”

“I think it is quite important to include modelling in my research. If you design your experiments with humans very carefully, you can make quite robust conclusions about what kind of representations they have. But what you are actually looking for is what language in our brain looks like. And you cannot really look into the brain; normally, you cannot cut open your participants’ brains and see what happens in their neural networks. It is therefore important to include computational models in linguistic or psychology research. If you mimic the performance of real people with a model, then you have the model and you can look inside it.”

“Earlier in my research, I used Optimality Theory, which is one of the models for phonology, and in the PhD research I did modelling with neural networks. You never know which model resembles the human brain most closely, but it seems that neural networks are more likely to be like the human brain than Optimality Theory is. For instance, Optimality Theory assumes an infinite number of possible candidates for every single word that you are planning to produce or will perceive, and such an infinite database seems to be very implausible to be contained in the brain of a real person.”

“My decision to do PhD here was because of the excellent research group and their researcher approach. I do like the Netherlands, and I love the city of Amsterdam, but it still does not feel entirely like home. It feels like a second home; I am still rooted very much in the Czech lands, where I have close friends and family. Luckily, my boyfriend came with me when I started my PhD. I would not have started if he had not came along. He is a software developer, so he can find a job anywhere.”

“There have been a few moments when I was doubting whether I should stop doing research. During my PhD, sometimes I really did not see the end. When your experiment does not work, or does not turn out as you expected, and you have trouble interpreting the results. Or even when you are not able to find the right next question that you want to answer, because maybe your previous experiment gave a null result, you can get demotivated. Obviously, it is not so exciting when you are not getting nice results. There are many people who get null results from their experiments, and they get demotivated partly because these results are impossible to publish and then it looks as if they have done nothing. I think that it is important that these results are publishable as well: after all, they do tell us something about the question that they were designed for and should be made known to others (who can then try to replicate your experiment or try to include different variables than you did, for instance).”

“Before I started my PhD I would have called myself a linguist, during my PhD I would have said I was a phonologist and a phonetician, and now I think that I have mostly been a psycholinguist – well, actually a “psycho-phone-ist” if that exists. But what I would like to be is a cognitive scientist. I really want to get more grip on other domains of cognition than speech or phonetics, like attention or visual cognition. That is why I want to do my next project at a department of cognitive psychology. I like linguistics and phonetics very much, but I also want to learn new things and explore other fields.”