LINGUIST List 7.400

Sat Mar 16 1996

Review: Daniels & Bright The World's Writing Systems

Editor for this issue: T. Daniel Seely <>

What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Daniel Seely at


  • Richard Sproat, Review of Daniels & Bright The World's Writing Systems

    Message 1: Review of Daniels & Bright The World's Writing Systems

    Date: Fri, 15 Mar 1996 15:09:52 EST
    From: Richard Sproat <>
    Subject: Review of Daniels & Bright The World's Writing Systems

    Peter T. Daniels and William Bright (Eds.) The World's Writing Systems. New York, New York. Oxford, England: Oxford University Press. 1996. Pp. XLV + 920. ISBN 0-19-507993-0

    Reviewed by Richard Sproat (Bell Laboratories)

    Previous books on writing systems that can claim to be comprehensive have invariably been single-authored volumes, where the author is a specialist on a particular writing system, and has only second-hand knowledge of others. For example DeFrancis (DeFrancis 1989) is a specialist in Chinese, and his discussion of writing systems as a whole stems ultimately from his irritation at common misconceptions about the nature of Chinese writing. Gelb, the father of modern studies of writing (Gelb 1963), specialized in ancient Near Eastern writing systems. Coulmas (Coulmas 1989) is evidently a specialist in Japanese.

    Daniels & Bright is different, in that it is an edited volume consisting of contributions from a large number of specialists (too many to mention by name here). Discussions of particular writing systems are typically written by experts on those systems, and one is fairly confident that one is getting reliable information.

    The book is divided into thirteen main parts. In the first part, Daniels introduces the field of grammatology, giving a brief history, and a broad typology of writing systems. The typology includes some familiar terms: "logosyllabary" for writing systems like Chinese where the individual characters of the script simultaneously represent both morphemes and syllables; "syllabary", where the individual characters represent syllables (or at least CV morae, following on Poser's 1993 LSA presentation, which is lamentably unavailable in written form); and "alphabet", where the individual characters represent phonemes. Less familiar terms are: "abjad" ("formed from the first letters of the most widespread example, the Arabic script, in their historic order"), where the basic characters of the script denote consonants (and only peripherally, vowels), typical examples being Semitic scripts as used for Semitic languages (adaptations of Semitic scripts to other languages such as Hebrew for Yiddish, Arabic for Uyghur or indeed Phoenician for Greek, often end up functioning as alphabets); and "abugida" (derived from the first four vowels and first four consonants of the traditional order of the Ethiopic script) where the characters denote consonants with a designated inherent vowel, and other vowels are denoted by diacritics on the basic consonant symbols. Examples of abugidas are Indian and Indian-derived scripts; and of course Ethiopic. Finally there is a "featural" script, namely Korean Hankul, where "the shapes of the characters correlate with distinctive features of the segments of the language" (cf. Sampson 1985). It is actually not clear from the above definition in what sense Daniels means that Hankul is featural, and thus constitutes a separate type of script. There is no question that that some of the basic Hankul symbols are derived graphically from aspects of vocal tract shapes, and that they are combined to form the basic letters of the script in a somewhat compositional way. On the other hand it is certainly not the case that Hankul represents features in the same sense as a phonologist represents them in a distinctive feature matrix (DeFrancis 1989). Indeed, Koreans (at least non-linguistically informed Koreans) evidently think of Hankul as being composed of segmental letters (again, DeFrancis 1989), so it may perhaps make the most sense to view Hankul as being an intelligently constructed alphabet. Note that Daniels' terms "abjad" and "abugida" are only somewhat consistently used in contributions to this volume: the term "alphasyllabary" (which is Bright's preferred term) is often used instead of "abugida".

    The second part deals with Ancient Near Eastern writing systems: Mesopotamian cuneiform in all its various forms and functions; Egyptian writing; Anatolian scripts; Aegean scripts such as Linear B; epigraphic Semitic scripts; and others such as Berber and Iberian scripts.

    Part three turns to decipherment, giving short accounts of some of the better-known as well as less well-known accomplishments in deciphering ancient scripts. Also presented in this section are scripts which are only partly deciphered (Mayan) and scripts which are largely or completely undeciphered (Proto-Elamite, Indus Valley, Rongorongo).

    Part four deals with East Asian writing systems. In addition to Chinese, Japanese and Korean, various "Siniform" scripts -- Tangut, Kitan and Jurchin are discussed, along with the Yi script which was originally a logographic system and then developed in modern times, with extrinsic planning, into a syllabary. There is also a chapter devoted to Asian calligraphy. A particularly useful addendum to this part is a comparative table of Chinese characters, giving traditional, simplified, and Japanese forms; pronunciations in Mandarin, Korean, and various On (Sino-Japanese) and Kun (native Japanese) pronunciations; and glosses. In addition, some examples of wholly Japanese creations are given.

    Part five concerns European writing systems. Various theories of the transmission of Phoenician script to the Greeks are discussed. Discussion of the Greek and Anatolian alphabets is followed by discussion of the Coptic and Gothic alphabets; Italian scripts, and the Roman alphabet itself; Runic and Ogham; Glagolitic and Cyrillic; and Armenian and Georgian. The history of the Roman alphabet is given in somewhat more detail than the history of other writing systems, with particular attention being given to the various forms that the letters took on in different scribal traditions.

    South Asian writing systems are discussed in the sixth part. The discussion starts with Kharoshthi and Brahmi, the latter being the forebear of most of the modern scripts of India, and many of those of Southeast Asia. The remainder of the part includes detailed discussions of Devanagari and its relatives -- Gujarati, Gurmukhi and Bengali; and South Indian scripts Oriya, Sinhala, Kannada, Telugu, Malayalam and Tamil; and the Tibetan scripts. Two points that are striking about Indian writing that are brought out well in this part are the following. First, unlike in many cultures, in particular Muslim cultures, the written word in India is not considered to be particularly sacred: sacred texts in (non-Islamic) India were typically intended to be memorized, and the written form was considered to be secondary. The second point, which is also peculiar, is the notion that each language should have its own distinctive script (cf. also Section 65 in part 11). This view of the intimate relationship between language and script has carried over even into Southeast Asia: one of the forces helping to promote the Pahawh Hmong script (Smalley et al., 1990, and Section 57 of Bright & Daniels), was the notion that Hmong, should have its own distinct script. I was somewhat disappointed in the discussion of Tibetan. Tibetan script is extremely unphonemic: the written form of a word seems only abstractly related to its pronunciation. What was not made clear is whether the written form actually corresponds to an earlier stage of the language.

    Part seven continues the story of Indian writing systems into Southeast Asia. Scripts covered in detail are Burmese, Thai and Lao, and Khmer. An interesting overview of various insular Southeast Asian scripts is also given. Mention is made of the curious practice in South Sulawesi of marking the first glyph of personal names in red ink, the only instance I've seen reported of a distinctive conventional use of color in writing systems.

    The eighth part discusses Middle Eastern Writing systems --- Aramaic, Jewish, Arabic and Ethiopic; there is also a short discussion of the curious Dhivehi (Maldives) writing system, whose most interesting feature is that its symbols are derived from Arabic and local numeral symbols. (A curious point in the discussion of Arabic is the following: while the form of Arabic numerals is given, no mention of the fact that the order of writing numerals is left-to-right, rather than right-to-left, contrary to the rest of the script.) The most extensive discussion in this part involves the adaptations of Aramaic writing to languages of various groups, including Altaic, and Iranian Indo-European languages. The latter display one of the more interesting phenomena among writing systems, namely "heterograms", words or morphemes which are spelled as in Aramaic, but are pronounced as in the Iranian language being written. Furthermore, these heterograms may combine with "phonetic complements", which spell grammatical endings. For example, the Parthian word "bawaand" `they shall become' would be written "YHYEnt", where the "-nt" is a phonological spelling of the third person plural ending, and the "YHYE" is the phonological spelling of the Aramaic verb "become", to be pronounced as "bawaa". The situation is, of course, highly reminiscent of Japanese adaptation of Chinese script, or Assyrian adaptation of Sumerian script, except that here we are dealing not logographic or morphosyllabic symbols, but with segmental ones. This, to my mind, is one of the phenomena that is most telling about the nature of writing. Much of the literature on writing, both popular and professional, has been devoted to the question of whether writing systems like Chinese represent language in a fundamentally different way from obviously phonologically-based systems. Lost in this debate is the fact that the primary purpose of writing is not phonetic transcription, but the representation of words and morphemes; purely phonologically-based systems are arguably merely the most efficient way of achieving this goal. Iranian scribes of the late pre-Christian era knew how to write in Aramaic, since it was the chancery language of the empire. The use of heterograms was not due to any lack of understanding of the phonological basis of Aramaic script; the phonetic complements discussed above, as well as the many Iranian words which were spelled phonologically show that the principle was well understood. Rather, in adapting Aramaic script to writing their own languages, the scribes apparently simply saw no reason to change the spelling of some common and familiar morphemes.

    The ninth part discusses the invention of writing in modern times, both by experienced linguists, and by previously illiterate individuals. In the latter case, the possibility of writing was either imparted to the individual by direct contact with a culture which possessed writing; such was the case with Sequoyah, the inventor of the Cherokee script. Or else, the inventor of writing was supposedly inspired by a dream, as in the case of Momolu Duwalu Bukele, the inventor of the Vai (West African) script, or by a vision as in the case of Shong Lue Yang, the inventor of the Pahawh Hmong script. Discounting the scripts (such as Cree) invented by linguists, most of these writing systems are not particularly remarkable, qua writing system; this, of course is not to say anything about the remarkableness of their invention. Most are syllabaries representing simple core (largely CV) syllables; Ol Cemet' is alphabetic (but note that the inventor was familiar with Roman writing). Pahawh Hmong, on the other hand, is typologically unique. The system is often termed demisyllabic (Smalley et al. 1990), though this is somewhat of a misnomer, at least given the way the term "demisyllable" is typically used in work on phonetics and phonology: each syllable is represented by maximally two symbols, the first [sic!] representing the rime plus tone, and the second [sic!] representing the onset consonants. The second feature has to do with the ordering of the two subsyllabic units noted above. While many Indian-derived scripts have a few vowels which are written before (or surrounding) the consonant clusters they follow, no writing system besides Pahawh Hmong is consistent in this reversal of the "natural" order (cf. Poser 1994)

    Part ten deals with the adaptation of scripts, in particular Roman, Cyrillic, Hebrew and Arabic. Needless to say, a good many writing systems discussed in other sections --- Akkadian, Japanese, Iranian languages written in Aramaic script --- could easily have been incorporated into this section. Adaptations of Hebrew and Arabic are of particular interest since they often involve a change in the script type over and above any changes there may be to the inventory of symbols. Hebrew script used for Judeo-Arabic still functioned as an abjad; Hebrew script used for Yiddish or Sephardic Spanish functions as an alphabet. Most adaptations of Arabic are basically abjads, except for Uyghur, Kurdish and Kashmiri, which are alphabets.

    The eleventh part is a very short section dealing with a few sociolinguistic issues surrounding scripts. Issues discussed include the rise and fall of Fraktur script in Germany; a very brief sketch of some psycholinguistic results from Serbo-Croatian a language which was (until very recently) taught in two scripts; the coexistence of scripts in India; Christian missionary activities; and script reform in and after the Soviet Union. Interesting in the latter discussion is the number of languages that have undergone at least one change of script, such as Roman to Cyrillic, Arabic to Roman to Cyrillic, or even Cyrillic to Roman to Cyrillic.

    Part twelve gives a useful overview of notational systems for purposes other than the normal conventional encoding of language. These include systems, such as phonetic notation or shorthand, whose purpose is certainly to encode spoken language, but for special purposes (i.e., speed or scientific accuracy); and they include numerical notation, music notation, and the notation of movement. The latter topic includes some discussion of the notation of signed languages, and a fairly detailed analysis of a passage of Plains Sign Talk transcribed into Labanotation is given. Mention is made of systems for ASL, but unfortunately no details are given, except for a table of signs, which is not very interpretable to readers who don't know ASL.

    The book ends with a part entitled "Imprinting and Printing", which comprises a single, largely speculative section on imprinting technology from Sumer to the present digital age.

    One particularly nice feature of this work, which is adopted fairly uniformly throughout, is short sample texts in each of the scripts and languages discussed. For modern languages, one is given the original text, a fairly abstract Roman transliteration (where necessary), a phonetic transcription (though it is not always clear what dialect this represents), a word-by-word gloss, and a translation. For ancient languages the phonetic transcription is generally skipped, for obvious reasons, and in general deviations from the ideal sketched above are found where there is insufficient information about the script or language in question. Left arrows and up arrows are used to draw one's attention, respectively, to a script's right-to-left or (very rarely) bottom-to-top direction.

    While this book is clearly the most comprehensive treatment of writing systems available, there are of course still questions that it does not really address. One question that is not entirely answered relates to a posting of mine to LINGUIST on the "Phonemicity of writing" (6.1094). In that posting I asked the question of how many languages had writing systems where one could accurately predict the pronunciation of a written word, presuming only a general set of phonological rules, and minimal or no lexical or morphological information. In a posted response to my query (6.1096), Daniels stated that "the number of languages with a `phonemic' orthography approaches zero." There is no question, given the detailed descriptions in the book, that this statement seems to be true, in that writing systems tend to represent phonological structure that is certainly not "surfacey". On the other hand, without a more detailed phonological description of all of the languages sampled in the book -- something that is certainly outside the scope of the collection -- it is hard to judge exactly to what extent writing tends towards the "morphophonemic" rather than strictly "phonemic". It is also hard to judge exactly what this all entails for my original posted query: it is perfectly possible for a writing system to represent a fairly abstract level of phonological representation, but at the same time be quite "phonemic" in the sense I meant, in that the mapping between this abstract phonemic representation and a more surface phonemic representation is accomplished by regular phonological alternations that are not dependent on morphological structure. Put another way, having established that a script is morphophonemic, what that entails for how easy it is to predict pronunciation from spelling depends in large measure on how the phonology of the language behaves. On top of this problem, there is the additional problem that in some sections of the text, one is not given information which is crucial to knowing how "phonemic" a script for a particular language is. So in the discussion of the adaptation of the Roman script to African languages, mention is made of fact that the designers of those scripts attempted fairly phonemic systems, and usually succeeded, modulo some problems with the representation of vowels in vowel-rich languages. But no mention is made of the fact that while a large number of African languages have phonemic tone, tone is essentially never indicated in the orthography.

    I also missed a discussion of Daniels' objection to the use of the term "grapheme", something he also raised in his posted Linguist response. In that posting, he cited some articles in LACUS, and these articles are again cited in Daniels & Bright. Lamentably, these papers have proved inaccessible to me, but in any case room could have been made in a 920 page volume for a short summary of the arguments against the use of this rather common term. Note that some of the contributors to the volume use the term "allograph", which seems to suggest that not everybody is uncomfortable with the notion of "grapheme".

    Finally, an unfortunate omission from the book is any description of electronic coding schemes, or of electronic fonts. There is discussion in various places on the technology of writing, but nothing of significance on the topic of the electronic representation of text. This is particularly ironic, since the beautiful electronic typesetting of this book would have been impossible without this technology. Multiscriptal computing owes a great deal to early work at Xerox PARC by Joseph Becker (cf. Becker 1984), and to Xerox "Star", the first truly multilingual desktop publishing system. Certain aspects of this work have continued in the development of the UNICODE multilingual character-coding standard (Unicode Consortium, 1991, 1992). Just as important as the coding of multilingual text are methods for the creation of fonts, and some space could have been devoted to this matter. Needless to say, the computational representation of writing is of immense importance in today's world, not only for obvious technologies such as word processing, but also for computational systems --- machine translation systems, text-to-speech systems --- that interact with people using human language and speech. It would certainly have been appropriate to include a discussion of the electronic representation of text in this book.


    Becker, Joseph. 1984. Multilingual Word Processing. Scientific American. July, 96-107.

    Coulmas, Florian. 1989. The Writing Systems of the World. Oxford: Blackwell.

    DeFrancis, John. 1989. Visible Speech: The Diverse Oneness of Writing Systems. Honolulu: University of Hawaii Press.

    Gelb, I. J. 1963. A Study of Writing. Chicago: University of Chicago Press.

    Poser, William. 1994. Review of Smalley et al. Phonology, 11, 365-369.

    Sampson, Geoffrey. 1985. Writing Systems: A Linguistic Introduction. Stanford, CA: Stanford University Press.

    Smalley, William; Vang, Chia Koua; Yang, Gnia Yee. 1990. Mother of Writing: The Origin and Development of a Hmong Messianic Script. Chicago: University of Chicago Press.

    Unicode Consortium, The. 1991. The Unicode Standard, Version 1.0. Volume 1. Reading MA, Addison-Wesley.

    Unicode Consortium, The. 1992. The Unicode Standard, Version 1.0. Volume 2. Reading MA, Addison-Wesley.

    - --------

    Richard Sproat is a Member of the Technical Staff in the Speech Synthesis Research Department at Bell Laboratories. His research includes multilingual text processing for text-to-speech synthesis; his interest in writing systems is thus guided to some extent by the practical problem of designing working computational models that map between written language and linguistic representation.