An interview with Giulio Severijnen.
What was the main question in your dissertation?
In my dissertation, I examined how we can understand other people while listening to them, despite differences in how they speak. People can speak differently because of where they grew up (based on their language dialect), biological differences (like having a lower or higher voice or people who have a lisp tend to pronounce ‘s’ differently), or just the way they are used to speaking (some people speak faster, others more slowly). Despite all these differences, we usually understand one another with little effort. The main question of my dissertation was: how do we accomplish this?
Can you explain the (theoretical) background a bit more?
Understanding spoken language includes various steps. First, we have to understand the correct sounds in a sentence. For example, we have to make sure that when someone says “Can you give me that pear?”, we correctly understand that they meant “pear” and not “bear”. Second, we have to correctly interpret the prosody of their sentence, meaning their tone, the rhythm, and the emphasis of their speech, to understand, for instance, whether they are asking a question or making a statement, or which words or syllables they want to highlight.
However, understanding each other is more complicated than it seems because of all the unique ways people speak. This uniqueness leads to countless differences in how people pronounce sounds, use rhythm, and place emphasis in their speech. Yet, we are still able to understand the same words, sentences, and meaning. For us as researchers who want to understand how people perceive speech, this is a fundamental question. And gaining knowledge of how this works reveals many interesting aspects of speech perception.
Why is it important to answer this question?
Apart from the scientific reasons I described above, this question could have a huge impact on artificial intelligence. For example, think about when you are driving and using the microphone on your cellphone to enter a new destination into your navigation system. How many times has your cellphone not understood you?
This happens because there are so many different ways that people can speak! For AI to correctly perceive speech, it has to overcome the same challenge humans do. If we can discover how humans deal with these challenges, we can then apply them to AI to improve speech recognition.
Can you tell us a bit more about one particular project?
In one project, I tested whether listeners can learn how different talkers speak. Specifically, how they stress some syllables or words in their sentence to make them more noticeable. To do this, people came to the laboratory and did an experiment on the computer. During this experiment, they heard two talkers who each had their own unique way of putting stress on words. We saw that, even though participants were initially confused about how each talker spoke, they improved at interpreting what each said, showing they adapted to each talker's style.
This shows that understanding spoken language from different talkers is a flexible process and that people adapt to how each person speaks.
Can you share a moment of significant challenge or failure during your PhD journey and how you overcame it?
A very difficult moment during my PhD was when one of my experiments did not turn out the way I expected. There was a period when I was trying to find the best way to design an experiment, and I ended up running many studies because we were not getting the results we expected.
This was quite challenging because I kept putting in a lot of effort without really getting the results I hoped for, which felt hopeless. So it felt like I was not making any progress, while I was doing a lot of work.
In the end, I overcame this by changing the experiment in such a way that meant somewhat abandoning what I had done before and trying something completely different. This finally gave the result I was hoping for. Yet, this did not mean that all the work I did was worthless. I learned a lot about what worked (and what did not), which led to the final experiment.
This showed me that while it sometimes seems as if you are not making any progress, it is never lost work. You are always learning something. Working in academia is not ‘easy’ in that sense. Things do not always work out exactly as you plan. But this showed me how important it was to be flexible, look for alternatives, and keep believing that things will work out in the end.
What was the most rewarding or memorable moment during your PhD journey?
It is going to be cliché, but the defense was definitely the best part. There were many smaller rewarding moments (publishing a paper, presenting at conferences), but the defense was the absolute pinnacle. It was beautiful to see how interested the professors and colleagues were in my research and to feel recognized for years of work. But the most rewarding part was being able to do this in front of my family and friends. It was truly unique to see all the proud faces, and it made me realize how special doing a PhD is.
What do you want to do next?
My goal is to continue in academia, doing research on topics I am excited about and combining this with teaching. I am now working as a postdoctoral researcher in Italy, and I’m very excited to see where a career in research takes me. Eventually, I would like to try to start my own research group, but let’s take things one step at a time.
Share this page