**Note** - The code was not submitted for peer-review because I did not know how to anonymise the repository back then. But now I do, so I will try to share codes and data in future work as much as I can! **Update 03 August 2023**: - I found that a line was missing, ending up yielding slightly different results. Although it did not affect any interpretations of the study results, I have updated the code here so the results on the paper should be fully replicated. - Also, I have added sections "Check the data" to make sure of the number of tokens. - Accordingly, I tidied up some notations and the folder organisation of the supplementary materials. As far as I understand, if you download the folder and open the R Markdown document from there, the code should run without issues, including loading data. If not working properly, however, please set your own directory and/or use the absolute path to each file. If you have any questions about the data, please let me know at: t.nagamine@lancaster.ac.uk