This book "asserts that the origin and spread of languages must be examined primarily through the time-tested techniques of linguistic analysis, rather than those of evolutionary biology" and "defends traditional practices in historical linguistics while remaining open to new techniques, including computational methods" and "will appeal to readers interested in world history and world geography."
AUTHOR: Coulmas, Florian TITLE: Writing Systems SUBTITLE: An Introduction to Their Linguistic Analysis SERIES: Cambridge Textbooks in Linguistics. PUBLISHER: Cambridge University Press YEAR: 2003
Reviewer: Peter T. Daniels, Jersey City, NJ
''That such a book as this should have been permitted to go forth to the world with the _imprimatur_ of the University of Cambridge, affords matter for very grave reflection.'' --John Churton Collins, 1886
This is the forty-fourth contribution to the distinguished series of Cambridge Textbooks in Linguistics (the fifty-first, counting second editions), but it might have inspired Mr. Collins to repeat his remark (on a work by Edmund Gosse) some sixscore years on (Cohen 2003). Florian Coulmas is a sociolinguist who from time to time addresses writing systems, and his work on the place of writing in society is always insightful and rewarding. He has also published three books on writing systems _per se_ -- *The Writing Systems of the World* (1989), *The Blackwell Encyclopedia of Writing Systems* (1996), and the volume under review. Unfortunately, however, his attention to detail in this volume has not met Cambridge's wonted standard.
The book begins with a (misguided) Note on Fonts. ''A font for a script that was used 3,500 years ago,'' the author claims, ''is anachronistic. Using it in a scholarly book amounts to a distortion and to underestimating the importance of media'' (xviii). While this is true to an extent for his example of Linear B (where no one uses such fonts, because the transliteration is straightforward), it is most emphatically not true for Egyptian hieroglyphs -- any philological treatment of the language uses typeset hieroglyphs to a very great extent, because it is virtually impossible to transliterate the signs with sufficient clarity to support the discussion, and because transcriptions of ancient Egyptian remain speculative as to the vowels (and as to the values of some consonants). As it happens, even for contemporary scripts, Coulmas uses very, very few script examples within the text (and some of those few are quite mistaken; see below).
Chapter 1: What Is Writing? (1-17)
In the first chapter, Coulmas surveys various definitions of writing, beginning with Aristotle's, and contrasting Saussure's well-known banishment of written language from the purview of descriptive linguistics with the Eastern (scil. Chinese) tradition that still accords the written sign pride of place; he finds in I. J. Gelb (1952/1963) the notion ''that writing _became_ a device for expressing language rather than having been such a device from its inception'' (15). But this is a distortion of Gelb's position; Gelb, unlike Coulmas, carefully distinguished synchronic description from diachronic explanation. Coulmas also slides, apparently unwittingly, from Gelb's statement that writing expresses _language_ to a claim that writing expresses _speech_ (16) --a straw position against which he proceeds to argue.
Chapter 2: The Basic Options: Meaning and Sound (18-37)
Coulmas's second chapter concerns visible representations of meaning that are not writing (what Gelb called ''forerunners'' of writing, although none of them actually ''foreran'' writing; DeFrancis 1989), and also iconic representations of speech such as Korean hangul and Bell's Visible Speech. It culminates with lists of four ''assumptions'' and three ''principles'' (33):
Writing and speech are distinct systems. They are related in a variety of complex ways. Speech and writing have both shared and distinct functions. The bio-mechanics of the production and reception of speech and writing are different.
the principle of autonomy of the graphic system the principle of interpretation the principle of historicity
As in most treatments of writing systems that enunciate ''principles'' in their opening chapters (Daniels 2002: 94), these principles are subsequently pretty much ignored. The chapter concludes with a ''Note on terminology and notation.''
Chapter 3: Signs of Words (38-61)
Seven chapters follow that treat the various ways units of writing are paired with units of language. Each begins with an attempt to define the unit in question linguistically and follows with a description of one or two writing systems operating at that level -- in chapter 3, Sumerian and Chinese exemplify logographic writing.
Chapter 4: Signs of Syllables (62-88)
A number of writing systems, modern and ancient, are exhibited to represent syllabographic writing, but none is presented in detail.
Chapter 5: Signs of Segments (89-108)
The roman alphabet, primarily as used for English but also as supplemented with diacritics and with additional forms (as in the International Phonetic Alphabet), is the topic of this chapter.
Chapter 6: Consonants and Vowels (109-30)
This chapter attempts to discuss Hebrew, Arabic, and Greek.
Chapter 7: Vowel Incorporation (131-50)
Several scripts of South and Southeast Asia, in which vowels are represented by alterations of basic consonant symbols, are presented.
Chapter 8: Analysis and Interpretation (151-67)
The principal example here is Korean hangul, in which letters of the alphabet were originally intended to iconically represent the contours of the vocal tract used in producing their sounds, additional strokes on letters represent phonetic features, and the letters are combined into syllable units.
Chapter 9: Mixed Systems (168-89)
Egyptian, Akkadian, Japanese, and English are discussed here.
Chapter 10: History of Writing (190-207)
A highly compressed history is presented.
Chapter 11: Psycholinguistics of Writing (210-22) Chapter 12: Sociolinguistics of Writing (223-41)
Brief surveys of reading and writing, and of literacy, standardization, and spelling reform round out the volume.
Appendix: Universal Declaration of Human Rights, article 1 (242-46)
Although this appendix is referred to nowhere in the book, it offers its passage in 21 languages, 10 in roman alphabets, 5 of the others without transliteration, and none with analysis to suggest how they represent the meaning of the text.
A bibliography and indexes of names and subjects follow.
In evaluating Coulmas's *Writing Systems*, we can invoke Chomsky's old criteria of observational adequacy, descriptive adequacy, and explanatory adequacy (1964: 924). The first might assess factual accuracy, the second the organization of information, and the third the analytic understanding of writing. The book fails on all three counts.
Readers uninterested in a catalog of errors might skip to the section titled ''DESCRIPTIVE ADEQUACY.''
''The lowest level of success is achieved if the grammar presents the observed primary data correctly'' (Chomsky 1964: 923f.).
This book is wrong, as Humpty Dumpty once said, from beginning to end. The very first indisputable factual mistake is both inexcusable and prototypical: the noted linguist Fred Householder is called ''Frank'' (12, cf. 251). This cavalier attitude toward proper names is pervasive. In the bibliography alone we find Laroch for Laroche, Algeria for Alegria, Empleton for Embleton, Givon for Givón, Rölling for Röllig, and a ''John'' M. Unger listed separately from J[ames] Marshall Unger. This treatment of data that are so easily checked does not bode well for treatment of less familiar materials, such as details of writing systems and their history.
I will begin my catalog of errors in the areas with which I am most familiar, writing systems for Semitic languages, following these with examples that are so egregious as to be evident to the non-specialist.
General. To baldly assert that when using ''Semitic consonantal alphabets'' vowels ''are indicated optionally'' (113) is a vast oversimplification. In Phoenician writing, vowels are never indicated. In Ancient Hebrew, some long vowels are indicated by means of some of the consonant letters (the label for them is _matres lectionis_ 'mothers of reading'), and in Arabic all long vowels are always indicated (with a handful of lexically determined exceptions). The vowel-indication systems that were introduced during the first millennium CE for Syriac, Hebrew, and Arabic are of course optional, but they are not the only device for notating vowels. C seems unaware that he echoes the consensus of scholars in rejecting Gelb's claim that the Semitic consonantaries represent ''syllables with indeterminate vowel,'' since he puts it as ''Following O'Connor (1996) [i.e. in Daniels & Bright 1996, henceforth WWS] I, therefore consider Semitic writing as encoding the consonants of the West Semitic languages'' (114). In discussing the absence of V-letters he adduces the non-existence of vowel-initial words in West Semitic languages, but fails to take into account the probable inspiration for the West Semitic signary from the consonant-only Egyptian writing system; but when he goes on to inquire, ''What if V-initial words ... need to be written?'' (127), he appears to be unaware of the existence of Ugaritic writing, a form of West Semitic script used in a city-state where over half a dozen languages were spoken, among them ones with initial vowels, which includes extra letters that (when writing Semitic) denote glottal stop followed by /i/ or /u/ (the original Alep-equivalent being used for glottal stop followed by /a/). Where C does allude to an Egyptian-Semitic connection, in discussing ''Proto-Sinaitic'' (194), he offers as a reference a 1948 article rather than the 1966 revised edition of the work of W. F. Albright, which perhaps ought not to be cited at all in an elementary textbook in view of its refutation in the next work cited in the same footnote, Sass 1988.
Hebrew. The term ''Paleo-Hebrew'' does not refer to the ''script which became extinct in antiquity when the Hebrews adopted a cursive variety of the Aramaic alphabet from which eventually the 'Square' Hebrew/Jewish script evolved,'' or to the Phoenician alphabet (116bis); it is the label for an archaizing revival of the earlier Hebrew script, at the time the Dead Sea Scrolls were being written around the turn of the Era, which was used within documents otherwise in Square Hebrew letters for the Tetragrammaton (the four-letter sacred name of God), for a scroll of Leviticus, and on coins of the Bar Kochba Revolt against Roman rule. C commits his share of the common confusions of Hebrew letters: final Kap for Dalet (116) and Chet for He (121), and also Pe with Rafe instead of final Pe without Dagesh (119). He chose to transliterate the subphonemic lenition of stops, but failed to indicate several of its occurrences (118). He refers to the epenthetic [a] inserted between a non-low vowel and a final laryngeal (''furtive Patach'') as a ''glide'' (120). His explanations of the realization of the vowel points known as Shwa and Metheg (121) may represent attempts to condense accounts in some grammar-book, but they bear little relation to reality. C alternates between ''gimmel'' (117, 119) and ''giimel'' (126) for the name of the third letter, which is in fact ''gimel'' (or ''giml'').
Arabic. To the practiced eye, Arabic writing is not ''quite different'' (116) from the Nabataean script from which it emerged in the two or three pre-Islamic centuries (Gruendler 1993). The transliteration of the Arabic consonantary omits the underdot from the second (123). C adopts the shorthand claim/pedagogic device that each Arabic letter has ''four different shapes'' depending whether it is connected or not to the preceding and following letters (though the variation is merely a matter of interactions in the creation of ligatures), leading him to coin the phrase ''writing group'' for a sequence of letters that happen to all be traced without lifting the pen (123). The diacritic _shadda_ is printed in the text as a raised, slanted omega rather than with an actual Shadda. In the example of its use (where the correct form appears), the letters of the example words are bizarrely spaced apart (123). Diphthongs are wrongly claimed to be written by combining Waaw, Yaa', and 'Alif (123). C states both that endings (case inflections, specifically) ''are only pronounced in connected speech'' and that they are ''omitted in connected speech'' (125, in adjacent sentences); in fact they are omitted phrase- (or breath-group-)finally.
Akkadian. The East Semitic branch of Semitic is not ''sometimes called North Semitic'' (111); that term has been used by a minority of scholars who wish to avoid the notion of a stark divide between East and West Semitic languages. C has badly misinterpreted the table he reprints (76) from WWS: 57. It is not, as C says, the ''basic grid of cuneiform 'Syllabary A'.'' What it is, as identified in his source, is the ''inventory of basic signs used in the pronunciation column of Syllabary A.'' (A couple of the signs in that inventory aren't even included in Syllabary A.) Syllabary A was not, as C says, ''widely used as a basic standard'' (67). What it is, as explained by Cooper (WWS: 47), is ''a cuneiform syllabary of the first millennium [BCE, which] would have been the first list of signs and their values for a student scribe to master in Nineveh during the age of Ashurbanipal.'' One wonders where C got the notion that the 211 (actually 212; he overlooked no. 56A) signs of Syllabary A ''as applied to Babylonian ... decreased further to a set of about 110 signs'' (67). Cooper clearly states (47) that ''scribes writing Akkadian in most periods had a working repertoire of between two and three hundred signs,'' though scholars would be familiar with two to three times as many.
C has badly misunderstood the relation between the Sumerian and Akkadian writing systems. He believes (177, 208) that Sumerograms represent loanwords in the Akkadian language. All they are, though, is logograms, which happen to be transliterated by modern scholars with the Sumerian readings of the signs, just as logograms in Linear B or Luvian happen to be transliterated with Latin words. Sumerograms were read in Akkadian only. (His main example, LUGAL [177f.] is actually not even a sign in Sumerian -- it is a sequence of LÚ 'man' and GAL 'big' -- and there is no trace of a loanword *lugallu 'king' in Akkadian.) We don't, in fact, know the pronunciation of a few very common words in Hittite (cf. 208), because they happen to be written only with Sumerograms or even Akkadograms. Misled by the Neo-Assyrian typographic form of the sign DINGIR 'god' / AN 'sky' (despite the ''Note on Fonts''!), C does not realize that 'sky' is not a ''semantic extension'' of 'god' (177), but that both senses come from the original 'star' (and the sign's original shape is rather like an asterisk).
On p. 77, C offers an incorrect explanation for ''broken writing'' (a term he renders as ''broken graphics,'' perhaps as a calque on the French), saying ''sequences of -(C)VC-VC signs must be interpreted as containing geminated consonants, as in _li-in-ik-ta_ or _li-in-kat-ta_ for /linkta/ 'he vowed'.'' Not only is the sequence between slants not a possible word of Akkadian, it represents neither an exemplification of the rule as he states it, nor a transcription of either of the sequences of signs offered! If those sequences were to occur, they would be read _lin'ikta_ (with glottal stop; Reiner 1964: 169f.; Cooper, WWS: 48) and _linkatta_ respectively, though neither of these could relate to _naqû_ 'vow'. In Akkadian, a CV sign can be followed by an echo V sign. C claims (77) this was to disambiguate an ''indeterminate'' vowel of a CV sign -- by which he refers to the fact that a sign containing /i/ can also be read /e/ if the word it's in historically contained one of the three laryngeals -- but the signs for /i/ and /e/ are not so used. Rather, CV-V optionally indicates a long vowel, as he also says.
Ethiopic. Since the middle of the reign of King Ezana of Axum, ca. 350 CE (coincident with his conversion to Christianity), vowels have been notated in the script of the Ge`ez language (which is still used in Ethiopic churches) and of its modern relatives including Amharic in Ethiopia and Tigrinya (not ''Tigrinia'') in Eritrea, though not of Somali (154). C makes quite a hash of his description of it. The vowels are written, Indian-fashion, by appendages to the basic consonant letters, except that /a/ is assumed as the vowel following the consonant in the unaltered letter-shape. Vowellessness is not indicated using the same letters as (i.e. the plain consonant letters), as C states (154), but by the same letters as are used for the short high (central) vowel (often misleadingly transliterated with Shwa, although it is not a reduced vowel; I will use <+> below for the nonce). C seems to be hung up on a notion that there is something special about a left-to-right sequence, even within a graphic unit, since he draws attention to the fact that some of the vowel-attachments are on the left of the letters and some are on the right. But he chooses a particularly unapt example to discuss, beginning with the syllable /l+/. The letter is (sort of) an inverted Y, and the appendage for /+/ is a loop attached to the left leg _inside_ the tent of the legs. Yet C describes this as to the right! His eyes seem to betray him further, since he describes the circle-to-the-right that marks /ee/ as a ''curly hook,'' and claims that ''/a/ is marked on the left'' (154). But /a/ is not marked at all, and /aa/ is either an extension or addition of a right leg or a bending of a single vertical to the left. A few only of the marks for /+/ and a few of those for /oo/ are on the left of their letters (as can be seen in the chart on p. 155).
While the counts (173) of 26 monoconsonantal and ca. 80 biconsonantal hieroglyphs are correct, there are not 40, but 87 triconsonantal signs (Ritner, WWS: 78). Perhaps C interpreted Ritner's statement ''Exclusively logographic writing is relatively rare in Egyptian'' (74) to mean that ''only few [sic] Egyptian signs are logograms'' (170), but of the 700 or so signs used in Middle Egyptian, fewer than 300 are accounted for by the phonetics and the nearly 100 semantic determinatives. I can find no warrant (e.g. Gardiner 1957, Loprieno 1995) for C's assertion that ''more than 400 signs were rarely needed at any one time'' (170).
Greek alphabet. C claims that ''The relative shortage of V letters, characteristic of the alphabets of many other European languages as well, testifies to the descent of the Graeco-Latin alphabet [the what?] from its Semitic precursor where vowel indication was even more sparing'' (111). One wonders how this testimony was elicited, given that at least one vowel letter, Omega, was added to the Greek alphabet quite independent of the Phoenician forebear. It is not at all clear why C offers Hebrew letter names and words as if they underlie the Greek letter names (126); the Phoenician originals are either known or easily reconstructed (Noeldeke 1904). C is quite mistaken in stating (122, 127) that the Greek innovation of vowel letters was an extension of the _matres lectionis_ principle -- because the Greek alphabet was developed from the Phoenician consonantary, and Phoenician uses no _matres_ at all. (In the footnote to this discussion, C fails to mention the standard reference on the development and varieties of the Greek alphabet, Jeffery 1990. Familiarity with that work might have kept him from the bare statement ''The Greeks also invented the three letters , , for the consonants /ph/, /kh/, and /ps/''; they were the outcome of a very complicated process.) While it is certainly true that ''For some time both horizontal directions of writing [Greek] were possible, the direction being reversed with every line,'' one must wonder what C has in mind when he says ''This way of writing frequently manifests itself in _scriptura continua_ or writing without word separation'' (128); it is not common at all around the world, and is perhaps most noteworthy in the South Arabian sphere, where inscriptions on very long walls were written in such a fashion presumably so that one could continue reading while walking back to the start of the text.
Cypriote syllabary. On p. 82, C gives a Greek word written with the Cypriote syllabary, _ptolin_, transliterated _po-to-li-ne_, but according to the chart on the facing page, the characters he shows are _pa-ta-li-ne_.
In discussing what to call the writing system in which, for instance, this review is composed, C mentions that the term ''Latin alphabet'' is ambiguous between ''the writing system of the Latin language'' and ''a set of 26 letters serving the writing systems of a great number of languages, ... also referred to as 'Roman' or 'roman''' (32). But whence this number 26? That's the number of letters in the English writing system, not the Latin or roman: Latin has 23, roman an indefinite number of letters, since different writing systems have made additions and subtractions over the ages.
English. In what phonemic (scil. phonetic; slants rather than brackets are simply wrong) scheme are ''she,'' ''he,'' and ''me'' transcribed as /SI: hi: mi/ respectively (63)? In what dialect of English is there a word [ju:fImiz@m] (98)? The charts of English consonants and vowels contain mistakes: the affricates are notated as sequences of stop+fricative rather than with the ligatured characters; and is printed instead of the phonetic symbol for the high back lax vowel (185). In the discussion of spelling reform, unconscionable space and attention (238f. and in the appendix) are devoted to ''Kånådån,'' a development by a Canadian, which is not simply a proposal for reformed spelling but also (as C seems not to notice) festoons English with a variety of inflectional suffixes (so that ''All human beings are born free and equal'' becomes ''Al humanes ar bornized friis and ekwallik'' [appendix]).
Danish. If ''_Stød_ marks a phonological distinction, as in _mor_ 'mother' vs. _mord_ 'murder', homophone words only distinguished by the presence of _stød_ in the latter'' (106), how can they be ''homophone''?
Africa. The chart (102) of the Africa Alphabet (which was promulgated in 1927, not 1930) is missing its letter for the bilabial continuant (International Institute 1930).
Vietnamese. The additional letters for vowels -- consonant letters are disregarded -- are mistakenly said (106) to be plain letters with diacritics (that, of course, is what they look like, but they are by any analysis separate letters), and they are misprinted as such (properly printed Vietnamese can be seen in the appendix), leading to the incorrect statement that ''there are several graphemes consisting of a letter base and double accents'' (107), with more examples of typographic improvisation. The claim that ''The romanization of Vietnamese was a reform of the writing system, as Chinese characters, a completely different system, were abolished and a new system was specially designed for Vietnamese'' (234) reveals ignorance of history. The romanization was devised in the mid 17th century, but the Chinese-based script continued to be used for over two more centuries, and the romanization was not officially adopted until 1910 (Nguyen Dinh-Hoa, WWS: 691).
Chinese. The pinyin diacritic for the Third Tone is correctly given on p. 106 as hachek, but throughout the discussion of Chinese (51-58), it is wrongly printed as breve.
Peter T. Daniels, co-editor and principal author of *The World's Writing Systems*, holds degrees in linguistics from Cornell University and the University of Chicago. He is also Production Editor at Gorgias Press.