BRAIN
AND
LANGUAGE
30, 381-386 (1987)
REVIEW Neurolinguistic Aspects of the Japanese Writing System. By Michel Paradis, Hiroko Hagiwara, and Nancy Hildebrandt. New York: Academic Press, 222 pp.
Reviewed by Department
of Oriental
JEROME
L.
PACKARD
Studies, University
of Pennsylvania
This volume, the latest in the series on Perspectives in Neurolinguistics, and Psycholinguistics, provides the reader with a detailed summary and analysis of processing and alexia factors specific to the Japanese written language. Coauthors Paradis, Hagiwara, and Hildebrandt have contributed a valuable theoretical and reference work which will be of use to students and researchers in the fields of reading psychology, acquired reading disorders and the experimental psycholinguistics of orthographic systems. The first chapter of this work presents a wonderfully clear and precise linguistic analysis of the Japanese writing system (preceded by a glossary of terms relevant to Japanese orthography). In this chapter, Paradis et al. examine the relationship between the printed form (grapheme) on the one hand, and both sound and meaning on the other. The authors emphasize that while the Japanese writing system may be cleanly divided into two primary subsystems of phonetically based alphabetic kana and the nonphonetic or ideographic kanji, these two subsystems nonetheless each have unique properties, and interact in complex fashion with a number of other factors. These other factors (such as frequency or part of speech) are singled out as serious potential sources of experimental artifact in previous kanalkanji processing studies. The authors demonstrate how, once these factors are controlled, the unique characteristics of the Japanese writing system make it a valuable research tool. Their treatment will undoubtedly help guide experimental design in future research on orthography, especially Japanese orthography. Chapter 2 provides a valuable synopsis of psycholinguistic processing studies involving kana and kanji (including Chinese characters, from
Neuropsychology,
381 0093-934X/87 $3.00 Copyright Q 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.
REVIEW
which the Japanese graphemes are historically derived). This chapter describes and evaluates the studies according to factors such as task and stimulus type, and then outlines the empirical findings in terms of the analysis presented in the first chapter, The chapter is divided into two parts, the first dealing with general psycholinguistic issues such as Stroop effects, naming latencies, etc. The second part of the chapter provides a breakdown of kanalkanji laterality studies. The laterality section includes a useful evaluation of factors involved in the use of the tachistoscope and also a table which conveniently summarizes all the relevant experimental variables and results. In evaluating the processing studies, Paradis et al. argue against the traditional interpretation, i.e., that kana and kanji achieve lexical access exclusively through phonological and visual routes, respectively. The authors demonstrate how the experimental results which have suggested a kana/phonological kanji/visual relationship can be accounted for by the potential sources of artifact detailed in Chap. 1. In suggesting that ideographs may access the lexicon phonologically, the authors take the somewhat controversial position that prelexical phonological processing may occur with ideographic script (cf. Coltheart, 1980, pp. 208, 222). A corollary of the authors’ position is that prelexical phonological processing of ideographs must be a direct association between grapheme and wholeword sound, rather than phonological decoding, since the sound/grapheme relation in an ideographic script is to a large extent arbitrary. The authors state that “lack of phonological regularity need not preclude phonological processing, since it is possible that whole-word pronunciations could be accessed directly and automatically” (p. 25). Their idea that kana words may access the lexicon visually (as well as phonologically) is also a nontraditional view, but it is less controversial than their position regarding ideographs . In their treatment of laterality studies, Paradis et al. not only carefully outline experimental variables and results, they also briefly review models of asymmetry of function which have been proposed to account for the laterality results. After a careful examination of these studies, the authors conclude that due to conflicting results and uncontrolled variables, “there can be no empirical claims based on kanjilkana laterality data for any particular model of cerebral asymmetry of function” (p. 45). The authors criticize as overly simplistic the popular notion that ideographs are processed in the right hemisphere, and note that the only pattern that has consistently emerged regarding ideographs is that “two-character word stimuli . . . will produce a RVF advantage” (p. 33). The authors conclude that given proper controls, kanji and kana laterality patterns may in fact be similar to those found for alphabetic scripts. The third chapter is a summary of all clinical studies and reports since 1900 which involve alexia and agraphia in Japanese subjects. Such a
REVIEW
summary (including an exhaustive table) is a valuable reference work in itself, since most of the studies are translated from the original Japanese. But in addition, the first part of the following chapter (Chap. 4) provides a clear breakdown of the findings according to patterns of dissociation (e.g., better reading comprehension with kanji than kana), correlated with lesion site and type of aphasic disturbance. It is clear from this summary that virtually every conceivable dissociation is found in the reading and writing of kana and kanji (including the variant On- and Kun-readings). Regarding localization, the authors find no support in the clinical reports for the common view that kanji are processed in the right hemisphere, but note that a clear neuroanatomical correlation does nonetheless exist, with “a greater involvement of the left temporal area for kana processing, and of the left parieto-occipital area for kanji processing” (p. 165). The authors also point out a correlation of dissociation with type of aphasic disturbance, noting that “performance with kanji is better than with kana in the context of Broca’s aphasia . . . [and] . . . performance with kana is better than with kanji in the context of transcortical sensory aphasia” (p. 157). The authors analyze these findings according to the processing and orthographic factors discussed in the first two chapters, and present various interpretations previously given in the literature. In the latter portion of Chap. 4, Paradis et al. present the implications that the empirical findings have for theories of reading. In this section, the authors summarize the various dissociation patterns discussed in the previous sections and apply them to a theoretical reading model proposed by Marshall (1982). They suggestthat since the samepatterns of dissociation are observed in English- and Japanese-speaking alexic subjects, and since proper experimental controls may result in similar laterality patterns for alphabetic and ideographic scripts, it must be true that all the pathways that are available to readers of languages such as English are also available for both types of Japanese script. Continuing along these lines, the authors argue that the different reading strategies generally assumed for the two types of Japanese script (i.e., [shape + sound + meaning] for kana, and [shape ---f meaning + sound] for kanji, p. 192) are mere tendencies, and should not be considered to be totally determined by the type of script. The authors make this point throughout the text, suggesting that “prelexical processing for any word in any script can be both visual and phonological as long as the word is known and the script is familiar to the fluent reader” (p. 20), and further arguing against “elevat[ing] to the status of assumption” (p. 20) the notion that kanji are processed visually and kana are processed phonologically. Finally, the authors make the somewhat stronger claim that “Routes (or reading strategies) are independent of representations, since . . . the route available for reading
384
REVIEW
kanji may or may not be available for reading kuna script, and vice
versa.” (p. 194). Paradis et al. are on firm ground in calling into question the putatively inviolate nature of the kuna/phonological kanji/visual relationship. However, they may be making too strong a statement when they claim that routes are independent of representations. The authors themselves state that “assuredly the use of one strategy is likely to predominate for each type of script in the normal course of reading” (p. 191). They further state that “Though both routes can be used for either type of script, it appears from the various patterns of dissociation that the visual route is more common for reading kanji and the phonological route for reading kunu” (p. 194). This leads us to the reasonable conclusion that routes and representations are in fact not totally independent, for if they were, it would not be true that the visual and phonological routes are more common for reading kanji and kunu, respectively.’ To help clarify matters, it may be useful to make a distinction between the terms route and strategy. We may consider strategy to be a more or less conscious decision by the reader to enter or retrieve a word in a certain way, for example, by sound association or directly by visual word form, This decision could very well be a function of education, training, or a number of other factors. Route, on the other hand, must refer to a neurofunctional pathway that is activated during the reading process. Given this distinction, it is dear how strategy, but not route, might be independent of representation. It seems reasonable to assume, with Paradis et al., that readers may use either a sound-association or word-form strategy regardless of the type of script. Thus, readers of totally phonetic orthographies often read by whole-word recognition, and readers of nonphonetic orthographies often read by first recalling the sound value of a printed form. Given this fact, strategy and representation would appear to be truly independent. To be sure, other factors, such as the emphasis on one strategy over another in a given educational system, are certain to come into play, and undoubtedly result in significant, statistically demonstrable tendencies toward the use of one strategy over the other for a given orthography. The point, however, is that these tendencies are not automatic, i.e., there is no direct cuusul relationship between type of orthography and reading strategy. On the other hand, there is reason to believe that a causal relationship exists between orthography and which neurofunctional reading pathways ’ It is also, of course, possible that the observed effect is primarily due to the choice of strategy being dependent upon orthography. While it is true that the choice of reading strategy must in many cases be determined by the type of orthography, I argue in the following paragraphs that orthography and strategy are functionally independent.
REVIEW
385
are activated, with a phonetic orthography causing greater prelexical activation of phonetic pathways, ceteris paribus. Given an orthography which contains no phonetic information (i.e., one in which the association of sound and grapheme is completely arbitrary), either a visual word form or sound-association reading strategy may be used, and whether the visual pathway or the phonetic pathway is prelexically activated depends on strategy. In this situation, route is independent of orthography and is determined by strategy. However, as the orthographic system becomes increasingly phonetic (i.e., as the relationship between the graphemes and the sounds becomes less arbitrary), the normal reader is automatically (perhaps even reflexively) cognizant of the fact that a principled relationship exists between the sounds of the language and its written form. If a reader is aware of the nonarbitrary relationship between sound and orthography, surely this knowledge is manifested to a certain extent in the activation of the phonetic pathways. Evidence that readers do possess this type of knowledge comes from a number of experimental studies which show that such knowledge affects lexical access. Katz and Feldman (1983) found that the extent to which a reader relies on articulatory recoding for lexical access is a function of the degree of correspondence between the orthography and sounds of a given language. They found that while semantic priming facilitates lexical decision in both English and Serbo-Croatian, priming does not facilitate word naming in Serbo-Croatian (which has a very close correspondence between orthography and speech sounds) but it does in English (in which the correspondence is less direct). The investigators interpret their findings as evidence that lexical access involves more prelexical recoding in SerboCroatian than in English, and suggest that the reason for this is the “shallow” (i.e., nonarbitrary) nature of the orthography. Parkin (1982) found that lexical decision was significantly slower on words with irregular spelling-to-sound correspondences than on words which had regular (i.e., less arbitrary) correspondences in English. These results suggest that the less arbitrary the relationship between the sounds and the written forms, the greater the likelihood that prelexical (Parkin’s “nonlexical”) phonological recoding will occur. In addition to replicating the above results, Parkin and Ellingham (1983) provide further evidence that the recoding process referred to is an obligatory component of lexical access, rather than being under the strategic control of the reader. Treiman et al. (1981) found that for subjects whose task was to determine the truth value of written sentences, readers of Chinese were significantly less impaired by homophone sentences relative to control sentences than were readers of English. The investigators suggest that the greater phonological interference for the English speakers is due to the relative strength of the correspondence between sound and grapheme in English. As these experiments suggest, the more principled the relationship
386
REVIEW
between sound and symbol, the more likely it is that readers will be aware of the principled nature of the relationship, and accordingly, the more likely it is that phonological information will play a part in lexical access. Given the likelihood that the relationship between orthography and neurofunctional pathway is deterministic, it seems reasonable to suggest that route and representation are not totally independent as Paradis et al. claim, but rather that they are independent only to the extent that the representation contains arbitrary phonetic information. My suggestion that route is dependent on representation does not deny the authors’ conclusion that the Japanese dissociation data support the neurofunctional modularity of cognitive skills. As the authors state, all the various neurofunctional modules that underlie each cognitive skill are integrated through their numerous interconnections. Each module, however, appears to be neurofunctionally independent. Not only can each module be selectively impaired, but it is also able to continue to function in isolation, that is, in the absence of its usual connections with other modules. (p. 199).
The point to be observed here is that while modules may be neurofunctionally independent to a significant degree, we should not deny the importance of the interconnections, especially those interconnections whose activation appears to be an automatic consequence of the activation of a given module. Nor should these comments detract from what is clearly an excellent and valuable piece of scholarship. Michel Paradis, Hiroko Hagiwara, and Nancy Hildebrandt have produced a work of remarkable quality which shall be considered an important and useful contribution to the field. REFERENCES Coltheart, M. 1980. Reading, phonological recoding, and deep dyslexia. In M. Coltheart, K. Patterson, & J. Marshall (Eds.), Deep dyslexia. London: Routledge & Kegan Paul. Katz, L., & Feldman, L. 1983. Relation between pronunciation and recognition of printed words in deep and shallow orthographies. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, 157-166. Marshall, J. 1982,March. Tavonomies of dyslexia. Lecture presented at Montreal Neurological Institute, McGill University. Parkin, A. 1982. Phonological recoding in lexical decision: Effects of spelling-to-sound regularity depend on how regularity is defined. Memory & Cognition, 10, 43-53. Parkin, A., & Ellingham, R. 1983. Phonological recoding in lexical decision: The influence of pseudohomophones. Language and Speech, 26, 81-90. Treiman, R., Baron, J., & Luk, K. 1981. Speech recoding in silent reading: A comparison of Chinese and English. Journal of Chinese Linguistics, 9, 116-125.