Internship Report
26.03.2020
Internship providing partner:
Name: Department of Linguistics, Cognitive Science and Semiotics, Aarhus University Address: Jens Chr. Skous vej 2
City + postal code: 8000 Aarhus C Country: Denmark
Telephone/email: +4587150000/ au@au.dk
Daily supervisor of the intern, on behalf of the above-mentioned organization: Byurakn Ishkhanyan
Internal supervisor of the intern on behalf of EMCL+: Vânia Correia de Aguiar Intern:
Student number: S3857077 Name: Sona Zohrabyan
Address: Van Goghstraat 119, City: 9718MX, Groningen Country: The Netherlands
Telephone/email: s.zohrabyan@student.rug.nl
The School of Communication and Culture if Aarhus University focuses on academic training of researchers to engage them in multidisciplinary projects to promote academic progress both within the school and in collaboration with international colleagues. The Department of Linguistics, Cognitive Science and Semiotics is one of the units of the School. In this department researchers tackle problems related to the science of language, human mind and brain, as well as, meaning, perception. This research institution was my first choice internship host for a variety of reasons including and not limited to the expertise of researchers in the chosen project, the availability of research laboratory equipment, the mind-provoking topics investigated in the project, the readiness of the internal supervisor to engage the intern in the day-to-day research related activities, as well as, her rich background in language processing in neurotypical and brain-damaged populations, which provided me with ample opportunity to get answers to my many questions to tackle problems in my Master’s thesis project.
I conducted my internship in the project The Puzzle of Danish at the Department of Linguistics, Cognitive Science and Semiotics at the University of Aarhus under the direct supervision of postdoctoral researcher Byurakn Ishkhanyan. This project studies multiple aspects of the Danish language, including both language acquisition among children and language
processing among adults. Previous research has investigated questions about the late acquisition of the Danish language, particularly vocabulary, among Danish children, as compared to children with other native languages. A potential cause identified is the large number of vocalic sounds in Danish (Bleses et al., 2008). For adults, the opaque phonetic system of spoken Danish has raised questions about how contextual information helps or hinders adult Danish speakers to process information (Ishkhanyan, et al, 2019). One of the topics within The Puzzle of Danish project focuses on investigating why and how the Danish language is difficult to acquire or process among children and adults due to its non-transparent phonetics while drawing comparisons to Norwegian. Some of the main results so far is the finding that vocoids in word boundaries hinder young children from acquiring Danish fast (Trecca, et al., 2020) and how only acoustic-phonetic information alone is not quite enough to recognize a word and that context facilitates word processing among adults (Ishkhanyan, et al, 2019). The research I was involved in was built on and further developed the aforementioned findings, particularly the acoustic-phonetic information and context integration in speech processing among adults.
My internship duties included the following:
Cleaning previously collected sensitive background information using the R Studio program, which resulted in a dataset ready for analysis. During this assignment an appropriate R code was used to find necessary data points in a dataset and fill them in another file. The main item learned from this was the proper handling of sensitive information.
Cleaning audio data using Praat (Boersma & Weenink, 2018) by editing, separating and segmenting text grids to prepare the data for acoustic analysis. This assignment involved working with two speakers’ phonetic and orthographic transcriptions. The initial data consisted of both types of transcriptions of both speakers in one file. Each type of transcription for each speaker was segmented separately which resulted in a large number of separate files with text grids for each speaker. Later, these text grids were segmented into smaller intervals.
Using ELAN linguistic annotator (Sloetjes & Wittenburg, 2008) to prepare audio data for acoustic analysis within the context of child language acquisition in Danish. This task involved acoustic data of mothers and their toddlers speaking in Danish in a naturalistic setting for around one hour per dyad, where the mother addressed the child in child-directed speech, as well as, the orthographic transcription of the speech. The audio and the transcription, however, were out of synchronization. To be able to analyse the data, e.g. to identify pitch, vowel hyper- or hypo-articulation those two had to be synchronized. I synchronized some of the data manually, however, as this method turned out to be quite time-consuming, another method was developed which included synchronizing at least 6 time-points in each file to later automatize the synchronization process.
Conducting data collection for a pilot study investigating categorical perception in speech. The pilot study was conducted with Danish speakers in the Aarhus university campus with the PsychoPy tool. I recruited six participants from the campus and conducted the experiment in the study room of the campus. The participants were instructed to listen to words and make a decision about the first sound of the word by clicking the appropriate key on the computer keyboard.
Setting up the lab and collecting data for a study investigating categorical perception in speech. After the successful completion of the pilot study, I participated in setting up an experiment in the Cognition and Behaviour lab and collecting data for the study with PsychoPy programming tool. The setup of the experiment included preparing a lab room for the participants, printing necessary forms for the participants to sign, receiving participants, informing them on the study, handling sensitive information (consent forms, compensation receipts) about participants, giving them instruction on how to complete the experiment, supervising participants and taking notes in case of deviations from the standard behavioural criteria for a successful completion of the experiment, recruiting more participants by preparing and handing out flyers when needed. Due to unpredictable circumstances, data collection had to be postponed until further notice. Handing over of the collected data, sensitive information about participants and other necessary documents has been done.
During my internship I was also involved in other academic activities such as attending lectures of Clinical Neurolinguistics conducted once a week by Byurakn Ishkhanyan, attending seminars and meeting on recent research outcomes such as Patti Adank’s talks on speech production and perception, a meeting on current state of a number of projects in the fields of linguistics, management, economics and such. The advantage of the project was that it was very hands-on, in that I could immediately get involved in the handling, collection, and cleaning of data. This allowed me to get a direct experience in the middle of a project. Unfortunately, the data collection could not be finalized due to the Corona virus, and thus not made within the time frame of the internship. No data analysis was performed on the collected data so far.
References
Boersma, P., & Weenink, D. (2018). Praat. Doing phonetics by computer (Version 6.0.43). Bleses, D., Vach, W., Slott, M., Wehberg, S., Thomsen, P., Madsen, T., & Basbøll, H. (2008). Early vocabulary development in Danish and other languages: A CDI-based comparison. Journal
of Child Language, 35, 619–650.
Ishkhanyan, B., Højen, A., Fusaroli, . ohansson . Tyl n . hristiansen M. . . Wait for it! Stronger influence of context on categorical perception in Danish than Norwegian. In A. Goel, C. Seifert & C. Freksa (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society, 1949–1955. Presented at the 41st Annual Conference of the Cognitive Science Society, Montreal, Quebec. Austin, TX: Cognitive Science Society.
Trecca, F., Bleses, D., Højen, A., Madsen, T.O. & Christiansen, M.H. (2020). When too many vowels impede language processing. An eye-tracking study of Danish-learning
children. Language and Speech.
R Core Team (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Sloetjes, H., & Wittenburg, P. (2008). Annotation by category - ELAN and ISO DCR. In: Proceedings of the 6th International Conference on Language Resources and Evaluation (LREC 2008)
