Functional writing units of Chinese: evidence from handwriting data

Introduction. Previous studies have proposed that orthographic units of different grain sizes are organized in orthographic representations at the same level in the mental lexicon (Law, Yeung, Wong, & Chiu, 2005). With the advancement of tablet technology, the search for the effects of orthographic units of different grain sizes on Chinese writing can be accomplished using handwriting measures. In the current study, the significance of character frequency, radical frequency and logographeme frequency in explaining handwriting data obtained in Chinese character copying was investigated. Method. The dataset used was extracted from the Database of Radicals in Written Chinese with Reliable Logographeme Boundaries (Lau, 2019; Lau, accepted). In this database, there are 856 radicals identified from 6480 most frequently found traditional characters in newspapers in Hong Kong (Leung & Lau, 2010). For each radical, a corresponding character with the highest character frequency was selected to serve as stimuli. One hundred undergraduates (age ranged from 19 to 22 years old, gender balanced, with no reported prior linguistic training or literacy problems) were recruited for a direct copying task, each consisting of 172 of the total stimuli. This guarantees that handwriting responses of at least 20 participants were collected for each stimulus. Data collection was conducted via 7-inch tablets and capacitive stylus pens with a homebrew Android application that recorded the elapsed time and the coordinates each time the stylus pen left / touched the tablet screen. The inter-stroke intervals (ISI) and inter-stroke distance (ISD) were calculated accordingly. In this study, the handwriting data of 384 phonetic compounds were analyzed because (1) they contain only unambiguous logographeme and radical boundaries, and (2) less than 10% of the participants made errors when copying these characters. Results. Figure 1 shows the results from the linear mixed effect model. The results showed that ISI increased with ISD (0.46 ± 0.01), and increased with stroke number (5.17 ± 1.85). ISI also decreased with Character Frequency (-6.78 ± 0.79, average count) particularly at the Radical Boundary (interaction of Radical Boundary/Character Frequency: -5.57 ± 0.38), decreased with Radical Frequency (-7.33 ± 0.83, average count) at the Logographeme Boundary (interaction of Logographeme Boundary/Radical Frequency: -8.63 ± 0.86), and decreased with Logographeme Frequency (-5.28 ± 0.38). Discussion. The significant Character Frequency effect at the radical boundaries and the significant Radical Frequency at the logographeme boundaries observed in this study are consistent with previous suggestions that more time is needed for the retrieval and/or planning of low frequency radicals and/or logographemes in the writing process (Lau, 2019; Lau, accepted). Finally, the significant Logographeme Frequency effect in predicting ISIs (particularly within logographemes) supported the notion that logographemes are the functional processing units of Chinese writing (e.g. Law & Leung, 2000; Lau, 2019). Theoretical implications of whether each of the frequency effects reflects central and/or peripheral processing of writing (Ellis & Young, 2013) will be discussed.