Reading does not depend on writing, even in Chinese

Abstract

A recent hypothesis proposes that reading depends on writing in a logographic language – Chinese. We present a Chinese individual (HLD) with brain damage whose profile challenges this hypothesis. HLD was severely impaired in the whole process of writing. He could not access orthographic knowledge, had poor orthographic awareness, and was poor at delayed- and direct-copying tasks. Nevertheless, he was perfect at visual word-picture matching and read aloud tasks, indicating his intact ability to access both the semantics and phonology in reading. He was also able to distinguish between fine visual features of characters. We conclude that reading does not depend on writing, even in Chinese.

Introduction

Reading is a learnt yet central cognitive process literate people employ everyday. Given that oral language is acquired far earlier than written language both on the species level and individual level, one of the foci of the past research on language processing mechanisms has been the role of phonology in processing written language. A new notion has been put forward recently that promotes the contribution of writing to reading in a logographic language, Chinese. Motivation and support for this hypothesis come from two perspectives, including reading development research and neuroimaging studies of normal adult subjects. First, Tan, Spinks, Eden, Perfetti, and Siok (2005) studied a group of dyslexic Chinese children and observed that handwriting skills were the most significant predictors of reading ability, and were more significant than phonological awareness, which is traditionally assumed to be the major explanatory factor for dyslexia in various languages (e.g., Ho, Law, & Ng, 2000; Vellutino, Fletcher, Snowling, & Scanlon, 2004). Two potential mechanisms underlying the involvement of writing in reading were identified by the authors: orthographic awareness (assessed by copying ability of pseudo-characters) and motor programming/motor memory skills (measured by picture copying ability). Both variables showed significant explanatory power of reading skills in beginning and intermediate Chinese readers. Such writing skill measurements were not found to explain the reading disabilities in their English-speaking counterparts (Schatschneider, Fletcher, Francis, Carlson, & Foorman, 2004; Vellutino, Steger, & Kandel, 1972; Vellutino, Smith, Steger, & Kaman, 1975).

Furthermore, a series of functional brain-imaging studies (e.g., Siok, Perfetti, Jin, & Tan, 2004; Tan et al., 2003) consistently found that the posterior portion of the left middle frontal gyrus (BA9) was activated when subjects made phonological judgment about Chinese written words. This same region was not observed to be responsive when native English speakers carried out similar tasks on English words (see He et al., 2003). Also significantly, structural and functional abnormalities (reduced grey matter volumes and reduced activation in language tasks) in this region was found in dyslexic readers of Chinese (Siok et al., 2004; Siok, Niu, Jin, Perfetti, & Tan, 2008). Given that this region is close to the premotor cortex, Siok et al. suggested that its activation in judging visual words reflects the potential involvement of writing in reading Chinese characters. Taken together, the contribution of writing experience to reading was indicated both by behavioural, anatomical and functional imaging measures in developmental study and by functional imaging of the mature brain.

Tan et al. (2005) offered a theoretical speculation about the role of writing in the reading process that is specific to speakers of Chinese. They argued that (1) learning strategies modulate the cognitive system and (2) for English children the learning-to-read strategy emphasizes the sound structure of speech, whereas Chinese children learn to read by repeatedly copying Chinese characters. This strategy is driven by the language-specific characteristics of Chinese words. As a logographic language, its basic writing units are characters, which correspond to a syllable and usually a morpheme. There is no visual-sound correspondence at segmental level. The visual-spatial configurations of characters are rather complex, involving the spatial arrangement of strokes into a two-dimensional square in complicated ways. The degree of homophony is also high in Chinese, with each syllable corresponding to multiple characters. Indeed, how these differences from alphabetic languages influence cognition has been the subject of most cognitive neuropsychological work with acquired reading and writing disorders (e.g., Bi, Han, Shu, & Caramazza, 2007; Han, Zhang, Shu, & Bi, 2007; Law & Leung, 2000; Law & Or, 2001; Law, Yeung, Wong, & Chiu, 2005). Within the current context, Tan et al. (2005) assumed that the linguistic differences between Chinese and alphabetic languages led to learning differences, and therefore result in different processing mechanisms of Chinese and English words – reading in Chinese depends on writing while reading in English does not. Corroborating the effects of learning strategy, Longcamp et al. (2008) recently showed that the learning modality of novel characters shapes the visual recognition efficiency as well as brain activation patterns. In their study speakers of alphabetic languages learnt novel characters by either writing or typing and were tested on mirror image judgment tasks. It was found that characters learnt by writing were not only judged better behaviourally, comparing to the characters learnt by-typing they also elicited more activations in areas that are involved in normal letter identification (left Broca's area, bilateral parietal inferior lobes, left dorsal premotor, and left postcentral regions). Such results confirmed the potential participation of motor processing in visual word perception, and further indicated that learning strategy plays an important role in inducing this participation.

While seemingly counterintuitive, the proposal that writing (an output process) mediates reading (an input process), echoes an important notion developed in the “simulationist framework”, which roots in the motor theory of speech perception (e.g., Liberman, Cooper, Shankweiler, & Studdert-Kennedy, 1967). The core assumption of this school of simulationist or “embodied cognition” theories is that in order to understand a physical stimulus one must internally “run” or “simulate” the corresponding production process. Recent discovery of mirror neurons in both monkeys and humans has revived and further highlighted this notion. Researches using both single-cell recording and functional neuroimaging studies on macaques and humans have observed a specific population of neurons in the premotor cortex that are activated both by the performing of an action and the viewing an action performed by others (e.g., Gallese, Fadiga, Fogassi, & Rizzolatti, 1996; Kohler et al., 2002). The existence of these mirror neurons advocates the hypothesis that the recognition of action (and even objects) might entail the same production processes that are associated with this action (and corresponding object) (e.g., Gallese et al., 1996; Martin, Ungerleider, & Haxby, 2000; see Caramazza & Mahon, 2006; Mahon & Caramazza, 2005; Mahon and Caramazza, 2008, for alternative arguments and review of counter evidence). This notion has been adopted in theories about a wide range of cognitive processes (e.g., Federmeier, 2007; Pickering & Garrod, 2007).

Along the same vein, it is not unreasonable to imagine that the visual recognition of word stimuli involves the simulation of producing such visual stimuli – writing. This proposal of reading-by-writing simulation can be taken with various degree of strength. The extreme hypothesis would be that reading depends on writing, i.e., successful reading has to go through some kind of writing process. A weaker hypothesis could be that the writing process influences or modulates the reading process in some manner but is not a necessary step in reading. And finally there is a theoretical possibility that the writing process is activated but does not influence reading at all. Moreover, there might be a distinction between the developing system and the mature system. For instance, being able to write might be necessary in learning-to-read but not in the adult system. Although Tan et al. (2005) opted for a strong version of the hypothesis – reading depends on writing in both developmental and adult systems – to explain the developmental data and the adult imaging data, these empirical findings by themselves do not necessitate a strong version of the reading-through-writing hypothesis.

This strong hypothesis should predict that an individual who fails at developing writing should also fail at developing reading.¹ Also if a literate Chinese speaker loses writing ability due to brain damage, his or her reading ability would also be impaired. In the current article, we present a case that is contradictory to this prediction. Our patient, HLD, is severely impaired throughout the whole writing process, including both accessing abstract orthographic knowledge for output and on more peripheral motor programming levels. Yet he was intact at reading character aloud and reading comprehension.

Note that such neuropsychological profile is not novel in Western language speakers. While dysgraphia (writing difficulty) frequently associated with dyslexia (reading difficulty) (e.g., Nickels, 1992, Roeltgen, 1985, Saffran, 1985), dysgraphic individuals have been reported for having relatively spared reading ability or having different impairment patterns for reading and writing (e.g., Hamilton & Coslett, 2007). However, since Tan et al. proposed that reading depends on writing only in Chinese and not in alphabetic languages, the study of Chinese brain-damaged individuals is crucial.