Graduate Researcher Series: an interview with Rena Gao
Juerong Qiu
Rena Gao is a first-year doctoral student at the School of Languages and Linguistics. Her research aims to develop a language prediction model to map and predict how an overseas educational experience may shape and change the way speakers of English as a second language communicate.
Juerong interviewed Rena about her doctoral research, her PhD journey, and the challenges she has encountered so far.
***
To start us off, could you please tell us about your doctoral project?
My doctoral research project aims to design a language prediction model and to use this model to predict different target features in human conversations. My target population is speakers of English as a second language who are international students studying in Australian universities. I want to use this language model to see how they gradually adapt or modify the language features that they use in communication during their study abroad experiments. We plan to have different focus groups. The study may focus on different groups of students and the length of their studies in Australia. For example, we have the short-term study abroad group, and the long-term study abroad group. We may also include Australian native language speakers as a comparison-based language group. If we have enough budget and time, we may expand the data size, but it’s hard to say for now.
***
What’s the significance of your research?
My research is important in terms of second language interactional competence because it’s quite hard for researchers in interaction competence to investigate quite a large population to see how the research subjects actually adapt or change the language features they use. This will involve a lot of conversational data which is quite hard to transcribe and analyse. With the language model that I intend to develop, it could be possible to analyse and predict these kinds of features.
Secondly, I focus on the population who speaks English as a second language, which is quite a large population worldwide that has been long ignored by computer science and natural language processing fields. So the data can actually provide a lot of insights on how to better design a language model for different language groups and how to use this kind of model in different fields of study.
***
How did you embark on this topic?
I speak English as a second language and have also been studying in Australia for quite a long time. Because of my experience, I think that it will be quite interesting to investigate changes in language use for English second language speakers in Australia. Overseas studying experience can shape or change international students’ identities and ideologies, and I want to look at how they actually develop their verbal communication abilities during this experience.
***
I learn that your research also involves a lot of coding. What’s the rationale? How do you include machine learning related content?
My project will include a relatively large sample size. The more participants we have, the more data sets we need to deal with. This is the reason why we want to develop such a language model from the perspective of natural language processing to investigate and predict the language features. Also, I want to express my thanks to my two supervisors: my primary supervisor, Professor Carsten Roever from SOLL who provides a lot of suggestions from the perspective of linguistics, and my co-supervisor, Professor Trevor Cohn, from the School of Computing and Information Systems. He gives me a lot of guidance in machine learning and statistical methodologies, which helped me a lot in language model building.
***
What advice would you give to someone who has no programming background to start coding?
Coding is quite useful still for a lot of humanities researchers if we want to deal with quantitative data more efficiently. We can use some software such as R, Python and Java. There are a lot of online resources and some tutorials at the University of Melbourne for students with zero background in computer science. Also, coding requires a lot of background knowledge in mathematics and statistics. Hence, if some researchers in humanities want to improve their coding skills, one of the best ways to do that would be to improve their mathematical abilities. In that case, you can understand the logics in the code and how to connect different code functions to your application.
***
You have had your pre-confirmation and the next big milestone for you is your confirmation. How do you prepare for your confirmation?
Currently, I’m drafting the literature review of my thesis. This will take up most of the 10,000-word report for confirmation. At the same time, I’m working on the language model. The results seem quite promising now. We can achieve accuracy of around 80%, but we haven’t recruited human participants yet. We’ll see how the application of the model goes.
Also, I divided the whole candidature into different stages. The first stage, which is my current stage, is to expand the literature and design a language prediction model. I will focus on data collection in the second year and then move one to connect the language model with the collective data and to see how we can apply the model into data analysis.
***
As a first-year PhD student, what’s the biggest challenge for you so far?
The biggest challenge is to adjust my learning style from being a coursework student to an independent researcher. I’m not a coursework student now, which means there are no assignments and tutorials to guide me at different stages. As a coursework student, what I had to do was to finish those assignments by the due dates. But as a graduate researcher, it’s quite different, because I don’t have to complete assignments anymore and instead, I have a big project to finish within a couple of years. I need to motivate myself and give myself deadlines in order to finish the PhD thesis. In that case, how to plan my time and how to motivate myself can be extremely important.
The second thing is loneliness. Even in the same group, graduate researchers are working on different topics. Basically, we are doing research by ourselves together with our supervisors. For fresh PhD students, we need to learn how to deal with the feelings of loneliness and how to share our research topics and progress with our colleagues.
The last thing is to maintain work-life balance. As a PhD student, we are faced with the pressure to graduate within a stipulated timeframe. Comparing to undertaking coursework subjects and completing assignments for coursework, I could study for a period of time and have a rest in the midterm break. Now I need to manage my time well, dedicating my time to my project and taking a rest when the need arises.
***
Talking about work-life balance, could you tell us how you achieve a healthy balance between your work and personal life?
I’m a part time photographer. I love taking pictures: especially scenic landscapes and people’s portraits. Dabbling in photography work actually helps me to meet new people, visit more beautiful places and gain new perspectives. I think it’s quite important for PhD students to have some hobbies, like sports, board games, or video games, as long as they provide some form of relaxation or respite. We may feel quite frustrated especially when we don’t have very good results sometimes. Other activities can help to fulfil our life goals.
***
Thank you for taking the time to talk with me, Rena.
Thank you, Juerong.