LINGUIST List 2.322

Tuesday, 25 June 1991

Disc: Orthography

Editor for this issue: <>


  • Mark Seidenberg, orthography and reading
  • Scott Horne, Writing systems

    Message 1: orthography and reading

    Date: Fri, 21 Jun 91 02:32:31 PDT
    From: Mark Seidenberg <>
    Subject: orthography and reading
    John Phillips has raised the important point that it can't simply be assumed that alphabetic writing systems are easier to learn to read and to process. However, I don't think that psycholinguistic research on reading supports several of his other assertions. He writes, >There is a considerable body of evidence >suggesting that, in the initial stages, learning a logographic script is >very much easier than learning an alphabetic script. The difficulty >comes later because of the number of logographic signs which must >eventually be learnt. I'd be interested to know what evidence he has in mind, because my own reading of the literature is quite different. It has been asserted at various times that one or another writing system is easier to master. Phillips thinks that Chinese is easier in the initial stages of learning to read. The writing systems for Serbo-Croatian (there are two, Cyrillic and Roman) are said by others to confer an advantage because of the simple and consistent correspondences between graphemes and phonemes. There are Japanese authors who have argued the merits of their orthography. One way to turn this into an empirical question would be to ask whether there are differences in the incidence of dyslexia associated with different types of writing systems. In about 1968 there was a paper (rather notorious among reading researchers) asserting that there is no dyslexia in Japanese because of the ideal organization of the writing system. However, a very impressive, large scale study of children learning Japanse, Chinese or English did not support the claim that type of orthography has a big impact on reading achievement or incidence of dyslexia. (I don't have the reference handy, but could find it on request. The principal investigator was Stevenson from the University of Michigan and the study was published in the journal Child Development. There _were_ important differences between the groups in terms of learning arithmetic, but that is another story.) There was one study by Rozin and colleagues, published in Science some years ago, suggesting that some children who were dyslexic readers of English did somewhat better when taught to read some Chinese. This was an interesting and surprising finding (that's what psychology articles in Science always are) but it would not be valid to conclude from the study that Chinese is easier to learn to read. (Born without a leg, I might find it easier to learn to ski than to walk, but ....) It should also be noted that some reading researchers (Uta Frith in the UK, Linnea Ehri in the US) think that children initially approach written English as a logographic system, and only later figure out that it is alphabetic. This would also tend to suggest that there are greater similarities in the early acquisition process than the differences between the writing systems might otherwise imply. In general, I know of no compelling evidence indicating that there are substantive differences between writing systems in terms of ease of acquisition. There may be very local differences in learning rate (e.g., Japanese children seem to get going pretty quickly with Kana) but these quickly wash out over the first couple of years of reading instruction. John Phillips continued: >Once learnt though, logographic writing is easier to read than >alphabetic writing - I believe this has been demonstrated for both Chinese >and Japanese. This is because fluent readers read logographically >anyway - English readers do not read letter by letter, they recognise >and interpret whole words at once. Since the shapes of Chinese >characters are more compact and distinctive than the shapes of English >words, they are easier to read. There is an anecdote about >the Chinese (a couple of centuries ago) comparing printed English text >to pictures of rows of worms and wondering how it could be read. Well, all that Chinese stuff just looks like squiggles to me, too. Again, there has been a lot of empirical research on how orthography affects the reading process, and it does not support Phillips' assertions. There is no basis for concluding that Chinese is "easier to read," and it is a gross oversimplification to assert that "English readers... recognize and interpret whole words at once." It is quite difficult to draw broad conclusions about the processing, of, say, alphabetic and logographic scripts because differences between the writing systems are confounded with differences between the languages they represent. For example, it is a fact that latencies to recognize and pronounce words in Chinese are considerably longer than for comparable words in English (see, e.g., my paper in the journal Cognition, 1985). It would be an obvious mistake to conclude that Chinese is necessarily "harder to read" because of other differences between the languages (e.g., there are also fewer words in the average Chinese sentence). In terms of who reads "logographically" ("whole words at once"), the "whole word" process Phillips attributes to skilled readers of English seems to apply rather better to the behavior of very young, unskilled readers (cf. the Ehri and Frith research mentioned above). They are the ones who seem to use pattern recognition processes like those used in recognizing non-linguistic stimuli, such as objects or faces. It's the skilled readers who are sensitive to the structure of words (and of the orthography), as indicated in numerous experimental studies of actual reading performance (see,e.g., the Rayner and Pollatsek textbook on the Psychology of Reading for review). These studies also suggest that there are very striking similarities in the basic processes used in reading different scripts. For example, the characterization of the Chinese script as "logographic" tends to obscure the fact that many words contain systematic cues to pronunciation. Empirical studies suggest that these cues are exploited by skilled readers, but they are more helpful in reading words that occur relatively infrequently in texts. A similar effect occurs in the skilled reading of English, the alphabetic orthography. There are systematic correspondences between the written and spoken forms of the language; this is supposed to confer an advantage (in terms of both ease of acquisition and ease of processing) over the logographic script. These regular correspondences are violated in words such as HAVE, SAID, GIVE, DONE, etc. What numerous studies show, however, is that for skilled readers, the regularity or irregularity of these correspondences only has an impact on the processing of relatively infrequent words. Thus, in both Chinese and English, common words are recognized on a visual basis. For less common words, readers exploit the cues to pronunciation each writing system provides. Note that, to the extent that words are recognized on a "visual" basis, this would tend to minimize the effects of differences among orthographies in terms of how simply or directly they encode phonological information. Thus, one of the principal dimensions along which writing systems differ seems to be IRRELEVANT to much of skilled reading. Bottom line: it's hard to sustain the claim that a particular type of orthography is either easier to learn to read or to process. The writing systems that have survived represent various solutions to the problem of representing spoken language in a cipher. They all seem to be learnable and processible (sic) at roughly the same rate. I would say that this is because the writing systems that exist reflect some pretty obvious tradeoffs among a variety of constraints (number of symbols, complexity of symbols, bandwidth, ease of production vs. ease of perception, ease of acquisition vs. ease of processing, etc.). There are universal aspects of reading and learning to read, dictated by facts about our common cognitive, linguistic, and perceptual capacities. There are language- and orthography-specific factors, but they don't seem to have resulted in big differences in terms of acquisition or processing. A lot like grammar, I would say. Finally, I would stress that the issues about acquisition and processing can't be resolved by intuition. They require careful and systematic research on peoples' actual performance. John Phillips thinks that Chinese words are more distinctive and therefore easier to recognize than English words in their respective written forms). Well, one could try to quantify distinctiveness in some way and maybe it would turn out that Chinese characters are more distinctive than English letters or words. The consequences for acquisition or performance are not obvious, however. The characters being more distinctive from one another might facilitate identification because less information would be needed in order to discriminate any given character from all of the others. However, we also know that human perception exploits redundancy, which reflects similarities across exemplars. The fact that a word is structurally similar to other words might facilitate perception (it certainly does in many domains, including word recognition). If that is the case, the putative distinctiveness of Chinese characters would be a disadvantage. People interested in current research on these topics might look at Marilyn Adams' book Beginning to Read (MIT Press, 1990). Lots of interesting things about the controversies over how to teach people to read, too.

    Message 2: Writing systems

    Date: Fri, 21 Jun 91 01:38:47 EDT
    From: Scott Horne <horne-scottCS.YALE.EDU>
    Subject: Writing systems
    I agree with John Phillips's claim that learning a logographic writing system is much easier at first than learning an alphabetic one. The biggest difficulty with learning alphabetic writing systems is learning to form a word from its constituent phonemes (and, when writing, to determine which phonemes make up a word). As a volunteer literacy tutor, I've found that this is what most troubles adults learning to read English. A colleague who used to teach English as a second language (ESL) to Chinese immigrants once claimed that teaching adults to read English by showing them how to use a dictionary to look up the phonetic respellings of words is a trivial matter. In other words, he regards the irregular English orthography as the biggest hindrance to acquiring literacy: according to him, substituting the (alphabetic) writing system which dictionaries use in respellings would trivialise the task of learning to read. I disagreed, of course, on the grounds given above, and suggested that a Chinese-speaker could learn to read his language much more quickly with the help of a dictionary. (Both this colleague and I speak Chinese, but I'm the only one of us who reads and writes Chinese.) A Chinese-speaking adult can, after his very first literacy class, read and write entire sentences and paragraphs (granted, they're simple and of limited subject matter); the English-speaking adults I teach don't start doing this until after a month or two of lessons. My colleague responded that his ESL students learn to read and write English words very quickly; to this I said that they are already familiar with one writing system (that of Chinese) and thus have acquired most of the skills needed for learning another. (Indeed, many Chinese have even been exposed to the IPA.) --Scott