INTERNSHIP REPORT
by
Anastasia Shaboltas
1. Introduction
My internship took place at the Arctic University of Norway (UiT) in Tromsø at the Department of Language and Culture in March-April 2020. I chose this place for the internship due to several reasons. First of all, I have been interested in Scandinavian languages since obtaining my first degree in Swedish language and literature. Knowledge of Swedish gave me the ability to understand spoken and written Norwegian to some extent, so I was interested in conducting a research in a native environment with one of the Scandinavian languages.
During Academic writing course in 2019 my teacher Irina Sekerina suggested UiT as a potential place for the internship. She also helped me to establish initial contact with Björn Lundquist who later became my supervisor. In addition to that, in September of 2019, I attended AMLaP conference in Moscow where I met Björn Lundquist in person. At the conference, I also had an opportunity to listen to the presentation by Natalia Mitrofanova who spoke about her research project at UiT. This project was about second language acquisition and involved visual world paradigm and eye-tracking – the technique I later used in my internship and thesis project.
1.1 Arctic University of Norway (UiT)
The Department of Language and Culture at UiT has many active research groups. During the internship I had the opportunity to join the activities of Polar lab and Language Acquisition lab. These labs are part of AcqVA (Acquisition, Variation & Attrition) research group. The group is involved in various research projects about changes in linguistic knowledge, how people acquire language and how the language changes in case of heritage speakers and bilinguals. Polar lab research is evolving around bilingual and multilingual language acquisition and processing and cognitive and structural changes in the brain during life span. For example, one of the guest lectures organized by the Polar lab was dedicated to investigating the differences in hippocampus of monolingual and bilingual speakers, the role of proficiency compared to the role of immersion in bilingual processing. Hippocampus was found to be bigger in bilingual speakers than monolingual that can refer to bilingual advantage.
My internship research activities were directly connected to the Language Acquisition lab which provided the eye-tracking hardware for my research project and was the place where data collection took place. Some of the projects conducted at the lab involve visual world paradigm
experiments about gender change in Norwegian and case processing in bilingual and monolingual speakers.
2. Internship activities
The goal of the internship was to get hands-on experience in collecting eye-tracking data, to learn how to analyze eye-tracking data and later use the analyzed data in the thesis. These activities were a follow-up of the eye-tracking study in part carried through at the Arctic University of Norway. This study focused on how speakers from three different L1 languages produce and comprehend grammatical gender in L2 Norwegian and investigated the role of L1 gender systems in mastering grammatical gender in L2. One of my goals was to implement a change in the experimental design of the previous study, presenting a screen with three objects to the participants instead of two objects as in the above-mentioned research, and to test how it could affect looking pattern of the participants and time resolution of their predictive looks.
Unfortunately, due to corona virus situation, my presence at the university was limited. I could be present at the university only for the first one and a half weeks of the internship. During that time I was designing the experiment, setting it up using special software and ran one pilot session with several participants.
The initial plan of my activities for the first month included two weeks dedicated to setting up the experiment and two weeks for testing 20 Norwegian speaking participants. The rest of the internship was supposed to be dedicated to the work with data analysis.
2.1 Designing the experiment
In the beginning of the internship, I was introduced to eye-tracking software called SMI Experiment center and also saw the Language laboratory with eye-tracking hardware.
First step of designing a follow-up experiment included creating a table with the list of stimuli, controlled for phonological onset of the nouns inside each item, position of the target noun on the screen and other characteristics. First half of the list with stimuli was created together with the supervisor and the other half independently.
Second step involved putting new experimental items into the SMI Experiment center program, installed on the laptop. This software contained several older experiments that allowed me
to use them as an example for my experiment. While learning how to create an experiment, I used visual and audio materials from the previous experiment provided by the supervisor to create new experimental items for the later presentation on the experimental computer. Then, created items were divided into groups for randomization purposes. I also ran the experiment several times on the laptop to see if the visual stimuli are presented correctly on the screen and checked if all audio files work.
2.2 Running the experiment
When the experiment was ready, I was able to test it in the Language lab. The supervisor told me about the eye-tracker itself and I learned practical aspects of the experimental procedure such as where a participant should be seated, how calibration and validation should look like. Final experimental set-up consisted of the laptop with the experiment and a computer with the eye-tracker and the screen where the experiment was presented. I had an opportunity to check how the experiment goes from a participant’s perspective and then also test myself in the role of the experimenter. This testing demonstrated that we needed to make some changes in the experimental presentation such as adding some additional text in the beginning and also adjusting the validation procedure.
Later I ran the experiment with the participants invited to the lab by the supervisor. I could test the participants under the supervision and ask the supervisor questions between the sessions. I was able to train my communication skills and learn how to explain the task to the participants, how explicit I should be in my instructions. There are different views on the way the task should be expressed so I tried different strategies during the test trials. We also received some feedback from the participants if they felt the experiment was long or if they got tired or not.
2.3 Working with the data
2.3.1 Data extraction and preprocessing
After the test trials were complete, I was shown how to download data from the eye-tracking software. This was done by my supervisor and he explained to me what parameters we need to download for the purpose of our experiment. In addition, we needed to choose a specific set of variables so we do not exceed processing abilities of the computer during data analysis.
Then, I learned how to preprocess raw data output to get the data ready for the statistical analysis. Here I received large assistance from my supervisor who explained me stepwise what kind of manipulations we should do. He showed me some examples of the older scripts so I could follow each step. We loaded the data set into R program and made it more readable by adding an extra time-column with time slots, 50 ms each, and a time-column with areas of interest. I also learned how to create additional columns with proportion of looks and add them to the initial data table. These activities were the last steps I participated in at the university in person before the university closed due to the pandemic. After that the majority of my work took place at home.
In the end, this preprocessed data was received only from a handful of participants and, as it was only first testing sessions of the experimental procedure, we did not aim to control the participants for their first language and collect all the demographics. Therefore, I was advised by the supervisor to perform data analysis on the previously collected data and use this procedure in the later analyses if we will be able to collect more data.
2.3.2 Analysis of the old data
The majority of my internship time was dedicated to the data analysis in R program. I had regular Skype calls (once a week) with my supervisor when he explained the steps of the analysis to me. My first goal was just to get used to working with the eye-tracking data. I performed various manipulations with the data in R, formatting the data set in different ways with the help of various R functions.
In the old study the researchers were not specifically interested in, where the participants were looking at early in the trials, so they did not include any data about early looks as separate columns in the data set. I was supposed to create and include additional variable of “early looks” into the analysis. At the next step I learned to create graphs with proportion of looks and visually inspect the data. The final step of the analysis included fitting mixed linear models to test the significance of interaction between early looks and proportion of looks.
To my mind, linear mixed modeling was the main part of the analysis where I could develop my previously gained knowledge during my EMCL+ program. This was the first time when I was able to employ these models with the real data set. I and the supervisor spoke a lot about various strategies that could be used in the analysis. We discussed fitting the models to the whole data set or to the separate subsets of the data, differentiated by the early looks of the participants. Finally, we also discussed the results of the data analysis and how to interpret the output of linear models.
2.4 Other activities
Besides the tasks directly associated with the experiment and data processing, I participated in offline and later online events organized by the Polar lab. I attended guest lectures such as the one about the changes in the brain mentioned earlier in this report. I also attended one online seminar dedicated to online collection of the data using OpenSesame program. All these talks were very useful to me.
3. Conclusion
Despite the unforeseen events which prevented me from fulfilling my internship obligations as planned, I find my internship experience to be positive. I regularly received help and support from my supervisor and had informative discussions with him.
I got the opportunity to get acquainted with different stages of research such as creating the experiment, testing it on the participants and analyzing the real data. This practical experience complements theoretical knowledge provided by EMCL+ program. I am especially glad that my internship activities included some novel component when I could test slightly different experimental design than was used in visual paradigm studies before.
I find my internship activities to be potentially useful for future research. New experiment created during the internship is suitable for collecting more data for other studies. Besides that, the additional analysis of the old data can help to adjust design of the future studies and give new insights.
