Welcome to my research page. Below you will find my primary research themes. I also include a list further below of my publications, both published and in the process of publication. Feel free to contact me via email at shemelstrand@link.cuhk.edu.hk for queries regarding my research. I am also happy to provide copies of my articles if you are unable to access them.
Most of my work so far focuses on the idea of graphic complexity, or the complexity of characters in a writing system. Graphic complexity is of core interest to those who study visually complex writing systems like Chinese or Japanese, particularly at an early age. I discuss projects related to this idea below.
Cognitive skills (e.g. implicit awareness of sound rules) have been shown across several studies to predict reading ability. Contrarily, graphic complexity has been shown to have a negative effect on reading. What if there was a relationship between the two? This is the general concept of the graphic complexity hypothesis, which states that cognitive skills, particularly those related to visual-orthographic awareness, should buffer the effects of complexity.
To investigate this idea, I analyzed data from primary school L1 Cantonese students in Hong Kong. This data showed that two particular skills, morphological awareness and rapid automatized naming (RAN), seemed to exhibit this buffering effect. The plots below show this relationship. When complexity is low and these two cognitive skills are high, the probability of reading success increases (shown in yellow). When the opposite is true (complexity is high and cognitive skills are low), the probability of reading success is extremely low (shown in red).
This all entails intra-script complexity, or the complexity of graphs within a script). However, does inter-script complexity (the complexity between scripts) play a role? I looked at this idea in an indirect way by posing a more flexible version of the orthographic depth hypothesis and testing it on primary-age Japanese readers of hiragana and kanji. The regression lines show poor, average, and excellent readers in ascending order. One can see that the association between morphological awareness and reading for each group of readers differs more dynamically in kanji, suggesting potential specialization in skills.
Furthermore, I tested some of the boundary constraints of this idea due to the perceived relationship between complexity and frequency of characters in visually complex scripts, which may pose a confounding relationship. We achieved this by testing the principle of least complexity, which states that the relationship between the frequency of a character and it's respective complexity should be negative due to the need for humans to use words more economically to achieve goals efficiently (the core idea of this was originally proposed by Alexei Koshevoy and his colleagues in an analysis on other languages).
We used three lexical lexical databases to test this idea. Across all our models, we found the association to be universally negative, but nonetheless incredibly weak. This seemed to suggest that this influence may not be as fundamental as we originally believed, which was a surprising but important finding.
I have recently begun writing on methodological issues as well. My most recent analysis looked at the dangers of dichotomizing continuous variables in language research. Utilizing a handful of Monte Carlo simulations, it was demonstrated that dichotomization often decreases effect sizes like R2 and increases the standard errors, producing less precision in statistical testing. This was shown to be the case in simple regressions, multiple regressions, path analyses, and different predictor distributions.
Published Scientific Articles
Hemelstrand, S. & Inoue, T (2024). The extrema of Japanese literacy: Beyond the bounds of average reading. Reading Research Quarterly 0(0). 1-14. https://doi.org/10.1002/rrq.536
Hemelstrand, S., Wong, B.W.L., McBride, C., Maurer, U., & Inoue, T (2024). The impact of character complexity on Chinese literacy: A generalized additive modeling approach. Scientific Studies of Reading, 28(1). 1-20. https://doi.org/10.1080/10888438.2023.2217967
Wong, B.W.L., Hemelstrand, S., & Inoue, T (2024). Revisiting the influence of phonological similarity on cognate processing: Evidence from Cantonese-Japanese bilinguals. Quarterly Journal of Experimental Psychology. 1-19. https://doi.org/10.1177/17470218241242631
Book Chapters
Hemelstrand, S., Wong, B.W.L., & Inoue, T (2024). The orthographic orthodoxy: How Japanese flies in the face of typical literacy development (Accepted).
In-Progress Scientific Articles
Hemelstrand, S. & Inoue, T (2024). A tale of two scripts: Applying the principle of least complexity to simplified and traditional Chinese (In revision).
Hemelstrand, S. & Inoue, T (2025). Stop splitting hairs: The problems with dichotomizing continuous data in language research (Submitted for peer review).
Presentations and Published Editorials
Hemelstrand, S. & Inoue, T (2025, February 27-February 28). A tale of two scripts: Applying the principle of least complexity to simplified and traditional Chinese [Paper presentation]. Association for Reading and Writing in Asia. The Philippines.
Hemelstrand, S. & Inoue, T (2024, February 29-March 1). The extrema of Japanese literacy: Beyond the bounds of average reading [Paper presentation]. Association for Reading and Writing in Asia. South Korea.
Hemelstrand, S. (2023, August 2). What Hong Kong educators can learn from the pandemic. South China Morning Post [Editorial]. https://www.scmp.com/comment/letters/article/3229637/what-hong-kong-educators-can-learn-pandemic
Hemelstrand, S., Wong, B.W.L., McBride, C., Maurer, U., & Inoue, T (2023, February 23-24). The impact of character complexity on Chinese literacy [Paper presentation]. Association for Reading and Writing in Asia. Virtual.