This page intentionally left blank

Writing SystemsAn Introduction to their Linguistic Analysis

During its long history the problem of reducing language to writing, and con-versely that of interpreting written signs as language, has found a variety ofsolutions, which still exist in the form of different writing systems. Written bya leading expert, this new textbook provides an accessible introduction to themajor writing systems of the world, from cuneiform to English spelling. FlorianCoulmas presents detailed descriptions of the worlds writing systems and ex-plains their structural complexities as well as the intricate relationship betweenwritten and spoken language. The book also provides a clear and engaging ac-count of the history of writing and its consequences for human thought and literatesociety.

This illustrated textbook includes questions for discussion at the end of eachchapter, and an up-to-date explanation of theoretical issues. Clearly organized andengagingly written, it is the ideal textbook for use in courses on writing systems.

FLORIAN COULMAS is Professor of Japanese Studies at Gerhard MercatorUniversity, Duisburg. He has published several works on writing and written lan-guage including The Blackwell Encyclopedia of Writing Systems (1996) andThe Writing Systems of the World (1989).

C A M B R I D G E T E X T B O O K S I N L I N G U I S T I C S

General editors: P . A U S T I N , J . B R E S N A N , B . C O M R I E ,W . D R E S S L E R , C . J . E W E N , R . L A S S ,D . L I G H T F O O T , I . R O B E R T S , S . R O M A I N E ,N . V . S M I T H

In this series

P . H . M A T T H E W S Morphology Second editionB . C O M R I E AspectR . M . K E M P S O N Semantic TheoryT . B Y N O N Historical LinguisticsJ . A L L W O O D , L .- G . A N D E R S O N and O . D A H L Logic in LinguisticsD . B . F R Y The Physics of SpeechR . A . H U D S O N Sociolinguistics Second editionA . J . E L L I O T T Child LanguageP . H . M A T T H E W S SyntaxA . R A D F O R D Transformational SyntaxL . B A U E R English Word-FormationS . C . L E V I N S O N PragmaticsG . B R O W N and G . Y U L E Discourse AnalysisR . H U D D L E S T O N Introduction to the Grammar of EnglishR . L A S S PhonologyB . C O M R I E TenseW . K L E I N Second Language AcquisitionA . J . W O O D S , P . F L E T C H E R and A . H U G H E S Statistics in Language StudiesD . A . C R U S E Lexical SemanticsA . R A D F O R D Transformational GrammarM . G A R M A N PsycholinguisticsG . G . C O R B E T T GenderH . J . G I E G E R I C H English PhonologyR . C A N N Formal SemanticsP . J . H O P P E R and E . C . T R A U G O T T GrammaticalizationJ . L A V E R Principles of PhoneticsF . R . P A L M E R Grammatical Roles and RelationsM . A . J O N E S Foundations of French SyntaxA . R A D F O R D Syntactic Theory and the Structure of English: a Minimalist

ApproachR . D . V A N V A L I N , J R , and R . J . L A P O L L A Syntax: Structure, Meaning

and FunctionA . D U R A N T I Linguistic AnthropologyA . C R U T T E N D E N Intonation Second editionJ . K . C H A M B E R S and P . T R U D G I L L Dialectology Second editionC . L Y O N S DenitenessR . K A G E R Optimality TheoryJ . A . H O L M An Introduction to Pidgins and CreolesC . G . C O R B E T T NumberC . J . E W E N and H . V A N D E R H U L S T The Phonological Structure of WordsF . R . P A L M E R Mood and Modality Second editionB . J . B L A K E Case Second editionE . G U S S M A N PhonologyM . Y I P ToneW . C R O F T Typology and Universals Second editionF . C O U L M A S Writing Systems: an introduction to their linguistic analysis

Writing SystemsAn Introduction to their Linguistic Analysis

FLORIAN COULMAS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, So Paulo

Cambridge University PressThe Edinburgh Building, Cambridge , United Kingdom

First published in print format

- ----

- ----

- ----

Florian Coulmas 2003

2002

Information on this title: www.cambridge.org/9780521782173

This book is in copyright. Subject to statutory exception and to the provision ofrelevant collective licensing agreements, no reproduction of any part may take placewithout the written permission of Cambridge University Press.

- ---

- ---

- ---

Cambridge University Press has no responsibility for the persistence or accuracy ofs for external or third-party internet websites referred to in this book, and does notguarantee that any content on such websites is, or will remain, accurate or appropriate.

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org

hardback

paperbackpaperback

eBook (EBL)eBook (EBL)

hardback

Writing . . . a mooring post for those who travel on mud

Contents

List of illustrations xiList of tables xivAcknowledgements xviiA note on fonts xviiiList of abbreviations and conventions xix

1 What is writing? 1

2 The basic options: meaning and sound 18

3 Signs of words 38

4 Signs of syllables 62

5 Signs of segments 89

6 Consonants and vowels 109

7 Vowel incorporation 131

8 Analysis and interpretation 151

9 Mixed systems 168

10 History of writing 190

11 Psycholinguistics of writing 210

12 Sociolinguistics of writing 223

ix

x Contents

Appendix: Universal Declaration of Human Rights,article 1 242

Bibliography 247Index of names 259Index of subjects 263

Illustrations

1.1 Chinese character wu nothing page 61.2 Rene Magritte 1929, The betrayal of images 71.3 Saussures model of the linguistic sign 122.1 Carved bones, approximately 35,000 years old 192.2 Quipu knots and their numerical values 202.3 Winter count of the Dakota 222.4 Freges Begriffsschrift. Is it writing? 242.5 Example of Otto Neuraths International Picture Language 252.6 Hebrew letter mem, m, as iconic depiction of tongue position

according to van Helmont 272.7 Iconicity of Hangul consonant letters 272.8 John Wilkins physiological alphabet symbols 292.9 Visible Speech, Alexander Melville Bells Universal Alphabet 303.1 Archaic Uruk tablet with pictographs 423.2 Archaic Uruk tablet containing calculations of rations of beer

for a number of persons for consumption on the occasion of afestivity 44

3.3 Sumerian rebus sign of Jemdet Nasr period 463.4 Multifunctional Sumerian cuneiform signs 473.5 Sumerian example sentence 493.6 The Chinese character shu` number and its graphic composition 533.7 Variant forms of Chinese characters 543.8 Chinese example sentence 594.1 The structure of a simple syllable 634.2 The structure of a syllable in a tone language 654.3 Cherokee syllables, phonetic and graphic 724.4 Consonantal onset of Old Akkadian syllables is lost in Akkadian 774.5 Syllable analysis and mora analysis of Japanese hon book

and the words kana notation 804.6 Fanqie, the turn and cut method of showing the syllabic value

of a Chinese character 86

xi

xii List of illustrations

4.7 Fanqie and phonetic analysis of the syllable hong insect 875.1 The ideal model of phonemic writing 935.2 The letter x and some of its phonetic interpretations 955.3 Graphic representation of [] in English 965.4 Complex phonemegrapheme correspondences 1005.5 The Africa alphabet 1025.6 The International Phonetic Alphabet (revised to 1993) 1046.1 West Semitic languages 1126.2 Lineage of ancient Semitic scripts 1146.3 The Hebrew Gezer Calendar, tenth century BCE,

in Phoenician script 1206.4 Ionian Greek inscription of the early sixth century

BCE without word boundaries 1297.1 Scripts descended from Brahm 1508.1 Amharic lbs cloak 1548.2 Hangul calligraphy by Kwon Ji-sam, seventeenth

century 1588.3 A passage from Hunmin Chongum haerye

explaining the vowel letters 1628.4 Graphic composition of the syllable kwon 1639.1 Francois Champollions decipherment of royal names 1719.2 Francois Champollions rst list of Egyptian

phonetic signs 1729.3 Logographic hieroglyphs 1739.4 Polyvalence in Japanese writing 1819.5 English and its phonemes 1869.6 English [] and its graphemes 187

10.0 Egyptian ofce. Mural relief in the tomb of ofcial Ti inSaqqara (fth dynasty) 191

10.1 Sandstone sphinx from the Middle Kingdom temple at Serabitel-Khadim with inscriptions in Egyptian hieroglyphsc Copyright The British Museum 194

10.2 Sign system to writing system: changing semioticrelationships 197

10.3 Schematic derivation of the Mongolian alphabet 20612.1 Standard and dialects: some dialects are closer to the

standard than others 22812.2 Diglossia 23112.3 Digraphia 234

List of illustrations xiii

12.4 Chinese characters, men gate and wu without, in fulland reduced form 237

12.5 The General Rules of Kanandan, promulgated bythe Internasional Union for Kanadan (IUK) Toronto,On., Kanada 239

Tables

3.1 Graphic development of cuneiform signs page 453.2 Graphic development of four Chinese characters 513.3 Major Chinese script styles 513.4 Liu` shu, the six writings 523.5 Chinese compound characters, formed of semantic

determinatives and the phonetic indicator gong 563.6 Chinese words of location 584.1 Some Manyogana, Chinese characters used as phonetic

symbols to write seventh-century Japanese 684.2 The Cherokee syllabary 704.3 The Cree syllabary 714.4 The Vai syllabary 734.5 Parallel development of Sumerian cuneiform sign and Chinese

character adapted to other languages 754.6 Basic grid of cuneiform Syllabary A. With permission from

P. T. Daniels and W. Bright, The Worlds Writing Systems,Oxford University Press, 1996, p. 57. 76

4.7 Basic kana syllabaries 794.8 Middle Chinese nal consonants are dropped in Old

Japanese 804.9 Hiragana and katakana for palatalized onset syllables 81

4.10 The Cypriot syllabary 834.11 The Linear B syllabary 844.12 The standard Yi syllabary of 1980 85

5.1 Latin vowels and diphthongs 925.2 Latin consonants 925.3 Some ways of spelling /u/ in English 995.4 Commonly used diacritics to enlarge the scope of the

Latin alphabet 1036.1 The Semitic root qbr to bury, imperative forms 1116.2 The Aramaic alphabet 115

xiv

List of tables xv

6.3 Scripts for Hebrew. Adapted from Avrin 1991: 126f. 1176.4 The Hebrew alphabet 1196.5 Tiberian pointing in relation to C letter B b and j h

.121

6.6 Hebrew vowels and their graphic indication 1226.7 The Arabic alphabet 1246.8 Greek vocalic reinterpretation of Phoenician consonant

letters 1287.1 Correspondences between Phoenician and Brahm signs 1337.2 Brahm primary vowels 1347.3 Brahm secondary vowels 1347.4 Brahm matras, diacritic vowel signs on consonant

letters k and l 1357.5 Brahm aks

.aras, consonants with inherent a 135

7.6 Devanagari plosives 1377.7 Devanagari sonorants and fricatives 1377.8 Devanagari vowel signs for Hindi 1377.9 Devanagari conjunct consonants 138

7.10 Tamil consonant signs 1417.11 Tamil vowel diacritics 1427.12 Tamil independent vowel signs 1427.13 Tamil Grantha letters for Sanskrit sounds 1437.14 The Tibetan syllabic alphabet 1447.15 Vowels of modern Tibetan 1447.16 The Thai syllabic alphabet 1467.17 Thai vowel diacritics 1477.18 Thai tone diacritics 1477.19 First page of K. F. Holles Tabel van oud en nieuw-Indische

alphabetten 1498.1 The Mangyan syllabic alphabet 1538.2 The Ethiopic syllabic alphabet 1558.3 Hangul basic and derived consonant letters 1608.4 Hangul tense consonant letters 1608.5 Combinations of consonants and vowels 1648.6 Hangul letters and their modern interpretations according to

the South Korean Ministry of Education 1669.1 Some common triconsonantal hieroglyphs 1749.2 Some common biconsonantal hieroglyphs 1749.3 The uniconsonantal hieroglyphs of the Egyptian alphabet 1759.4 Hieroglyphic determinatives 1769.5 Kokuji, Japanese native characters 180

xvi List of tables

9.6 English consonants 1859.7 English vowels and diphthongs 185

10.1 Proto-Sinaitic signs. From Sass 1988, Table 4. 19510.2 The Etruscan and Latin alphabets 20210.3 The Old Hebrew and Mongolian alphabets 20510.4 Phoenician and Greek sibilant letters 207

Acknowledgements

I want to acknowledge my indebtedness to some of the scholars whohave helped me in various ways, sharing their knowledge and friendship withme: Bill Bright, Danny Steinberg, Jean-Pierre Jaffre, Young-Key Kim-Renaud,Jan Assmann, Yuko Sugita, Giri Suzuki, Yoko Tamai, Minglang Zhou, YoshimiShimizu, Mahmoud Al-Khatib, Kim Hakhyon, Lun Du, Makoto Watanabe. In thefall of 2001, I had the good fortune to spend a sabbatical at the Centre of Asian andPacic Studies of Seikei University, Tokyo. The library and administrative staffwere as hospitable as they were helpful, especially Hiroshi Hasebe and SatokoUno assisted me in many ways that helped the book along and made my stay atthe Centre pleasant and gratifying.

xvii

A note on fonts

Technological advances in the electronic media have made it possible toproduce fonts for a great variety of languages and scripts, ranging from decorativetype for your personal correspondence to Egyptian hieroglyphs and other ancientwriting systems. In this book I have by and large avoided using these fonts, atleast for writing systems that ceased to be used long before even the printing presswas invented. A font for a script that was used 3,500 years ago, such as Linear Bin Greece and Crete, is anachronistic. Using it in a scholarly book amounts to adistortion and to underestimating the importance of media. Before print, all writingwas by hand. True, some cuneiform inscriptions, Egyptian papyri, and Greekepitaphs look as sharp as copperplate, but the travail behind these chirographicdocuments cannot be compared with a mouse click. It is different with scriptsthat have been used continuously from antiquity until the present, such as Greekand Chinese. They have gone with the times and been adjusted accordingly. Butthe tradition of Hittite hieroglyphic and the Indus script broke off hundreds ofyears ago. Presenting these languages in the guise of a modern font is like lettingHannibal traverse the Alps in a tourist bus.

xviii

Abbreviations and conventions

BCE Before common eraC consonantCA Classical ArabicCE Common eraF FrenchGk GreekGr GermanI ItalianIPA International Phonetic AlphabetL LatinMC Middle ChineseOE Old EnglishOJ Old JapaneseSkrt SanskritV vowel

enclose graphemes[ ] narrow phonetic transcription/ / broad phonemic transcription

Italics in running text denote quoted forms and are used, rarely, for emphasis.

xix

1What is writing?

The men who invented and perfected writing were great linguists and it wasthey who created linguistics. Antoine Meillet

Writing has been with us for several thousand years, and nowadays is more im-portant than ever. Having spread steadily over the centuries from clay tablets tocomputer chips, it is poised for further dramatic advances. Although hundreds ofmillions of people are still unable to read and write, humanity relies on writingto an unprecedented extent. It is quite possible that, today, more communicationtakes place in the written than in the oral mode. There is no objective measure, butif there were any doubts, the Internet explosion has laid to rest the idea that for thehuman race at large writing is only a minor form of communication. It is not riskyto call writing the single most consequential technology ever invented. The immen-sity of written record and the knowledge conserved in libraries, data banks, andmultilayered information networks make it difcult to imagine an aspect of mod-ern life unaffected by writing. Access, the catchword of the knowledge society,means access to written intelligence. Writing not only offers ways of reclaimingthe past, but is a critical skill for shaping the future. In Stanley Kubricks 1968 mo-tion picture 2001: A Space Odyssey a computer equipped with a perfect speechrecognition programme, which is even able to lipread, threatens to overpower thehuman crew. This is still science ction. In contrast, the ability of computers tooperate in the written mode, to retrieve, process and organize written language inmany ways surpasses unaided human faculties. Mastering the written word in itselectronic guise has become essential.

The commanding relevance of writing for our life notwithstanding, it is anythingbut easy to provide a clear denition of what writing is. Partly this is because ofthe multiple meanings of English words and partly because of the long historyof writing and its great importance. At least six meanings of writing can bedistinguished: (1) a system of recording language by means of visible or tactilemarks; (2) the activity of putting such a system to use; (3) the result of such activity,a text; (4) the particular form of such a result, a script style such as block letterwriting; (5) artistic composition; (6) a professional occupation. While in this book

1

2 What is writing?

my principal concern is with (1), the relationships with the other meanings are notaccidental or unimportant. The various uses of writing reveal the many aspects ofsociety and culture touched upon by what cultural anthropologist Jack Goody hasaptly called the technology of the mind. It can be studied from a great variety ofangles in several different scientic elds. Philologists, historians, educationalists,perceptual and cognitive psychologists, cultural anthropologists, typographers,computer programmers, and linguists all have their own interest in writing basedin their disciplines specic understanding of how writing works, what functionsit serves, and which methods can be applied to its investigation. What is more, of atechnology that has evolved over thousands of years it cannot be taken for grantedthat it has not changed substantially. There is little reason to believe that writingmeans the same in different linguistic and cultural contexts. Rather, the meaningand validity both of past and contemporary theories of writing are contingentupon the historical and, perhaps, cultural circumstances within which they wereconceived. Indeed, properties of writing systems may have an effect on how writingis conceived, and, conversely, conceptions of writing may inuence the way certainsigns are dealt with. Maya writing is a case in point. Anthropologist Michael Coe(1992) has shown how the refusal to recognize the Maya glyphs as writing longstood in the way of their linguistic decipherment, which, once accomplished, addeda new facet to our understanding of the multiformity of writing. Every attempt ata single universal denition of writing runs the risk of being either ad hoc oranachronistic, or informed by cultural bias. To appreciate the difculty it is usefulto review some of the denitions that have been provided by writers who concernedthemselves with the issue.

Aristotle

What is probably the most widely quoted denition of writing was givenby Aristotle. The second part of his propositional logic, Peri Hermeneias, beginswith some basic explanations about things, concepts and signs. Before discussingnouns and verbs as parts of sentences that can be true or false, Aristotle discusseshow these linguistic entities relate to ideas and to things of the material world. Heexplains:

Words spoken are symbols of affections or impressions of the soul; written wordsare symbols of words spoken. And just as letters are not the same for all men,sounds are not the same either, although the affections directly expressed by theseindications are the same for everyone, as are the things of which these impressionsare images. (1938: 115)

Aristotle 3

Aristotles main concern here was not with writing. Rather, his purpose was toalert his readers to the need to clarify the complicated relationships that obtainbetween things, ideas and words, as a prerequisite of developing logical thinking.He only dealt with writing because words manifested themselves in two differentforms: as sounds produced by the human voice and as letters. Explaining the re-lationship between the two was a matter of systematic rigour and terminologicalorderliness, but of little importance for the rest of his treatise on proposition.Yet, this brief statement became hugely inuential in Western thinking aboutwriting.

Much has been written about it. His pronouncement that spoken words are sym-bols of affections or impressions of the soul what we would call concepts orideas while written words are symbols of spoken words allows for interpreta-tion. What is a symbol? Aristotles term is symbolon which is usually translatedas symbol in English. Other translations of the Greek original have preferredthe term sign, which is more general in meaning and thus makes it easier to ac-cept that a relationship between nonperceptible entities (impressions of the soul)and perceptible entities (spoken words) should be of the same order as a relation-ship between perceptible entities of two different sorts (spoken words and writtenwords). A variety of verbs such as depict, designate, signify or stand for havebeen used to give expression to the nature of the relationship between a symbolonand that which it symbolizes. The common element of all of them is the implicitassumption that this relationship is characterized by linearity and directionality,rather than being symmetric:

things affection of the soul spoken word written word

This formula can be given a temporal and an ontological interpretation. Thingsexist. You think about them, then you speak, then you write. The phenomenal worldprecedes cognition which precedes language which in turn precedes literacy.

The central element of Aristotles denition is that it determines the functionof writing as forming signs for other signs as their referents. Writing is not onlypreceded by, but also subordinate to, vocal speech. This assumption reects theliteracy practice of Greek antiquity. The notion that letters stand for sounds wasrmly established, and that both individuals and societies used speech before writ-ing was evident. Literacy had a place in society, but did not embrace large sectionsof society yet. It was not a form of life as it is now. Letters had not yet broken freeof sounds. It followed that writing, at least Greek writing, was a secondary signsystem serving the sole purpose of substituting for or representing the primarysign system, vocal speech. When writing was invented, such a linear representa-tional relationship between speech and writing did not exist, but that was none ofAristotles concern. Nor did he address the question of whether the relationship

4 What is writing?

he had identied might change in the course of time as the consequences of literacymade themselves felt in society. His remark that letters are not the same for allmen, although affections of the soul are, and the fact that it was part of a treatiseon proposition suggest that he had a general statement in mind, and this is how itwas understood by subsequent generations of scholars right to the present time.Writing is secondary to and dependent on speech and, therefore, deserves to beinvestigated only as a means of analysing speech. This is the gist of Aristotlesdenition of writing, which became axiomatic in the Western tradition.

Liu Hsieh

It has been argued that Aristotles denition is a direct result of the natureof the Greek alphabet, which is said to be the rst full-blown phonetic writingsystem humanity developed. Thus, writing systems, rather than being conceptuallyneutral instruments, are thought to act on the way we think. In this connection anexplanation of what writing is and whence it came that emerged within the contextof Chinese literary culture is of some interest. It bears resemblance to Aristotles,but upon closer inspection also differs in important respects. In his celebratedessay Carving of the Literary Dragon writer and philosopher Liu Hsieh (465522) states:

When the mind is at work, speech is uttered. When speech is uttered, writing isproduced.

The Tao inspires writing and writing illuminates the Tao. What in mind is ideawhen expressed in speech is poetry. Isnt this what we are doing when dashingoff writing to record reality?

Writing originated when drawing of bird trace replaced string knitting.(1983: 1317)

This denition shares a number of elements with Aristotles. A mind at workis what Aristotle calls affections of the soul. It produces speech that in turngenerates writing. The Tao corresponds to nature, that is, things about whichideas are formed in the mind. However, Liu Hsiehs statement also contains anelement that lacks a counterpart in Aristotles denition. Writing is credited witha creative analytic potential: it illuminates the Tao. Moreover, the Tao inspireswriting, apparently unmediated by speech. An idea in the mind is expressed inspeech, but also in writing that is employed to record reality. While Aristotleunambiguously places speech between ideas and written words, Liu Hsieh seemsto concede the possibility that ideas are expressed poetically in speech or in writing,where the relationship between the two is not necessarily unidirectional. This does

Zen 5

not imply that, unlike the Greek philosopher, the Chinese denied that writing wasbound up with language, but from his account of the relationship between ideas,speech and writing it cannot be concluded that he conceived of writing as a meresubstitute for speech.

Plato

Liu Hsieh and Aristotle speak of the same four elements: in modernparlance, objects, concepts, vocal signs and graphical signs, but the mappingrelations between them suggested by their denitions are not identical. In theWest, Aristotles surrogationalist denition has been the basis of the bulk of schol-arly dealings with writing ever since, although it was also recognized early on thatwriting does more and less than represent speech and can never replace it. Moreclearly than Aristotle, Plato sensed the unbridgeable chasm between discourse andtext, between speech and speaker that writing brings about. He was concerned withthe communicative function of writing and saw that it was the tool of articial in-telligence as opposed to empathetic dialogue-generated insight, but he was deeplysceptical of the new technology and the form of knowledge it made possible. In thePhaedrus dialogue he lets Socrates say, Written words are unnecessary, except toremind him who knows the matter about which they are written (Phaedrus 275d).Writing, he reasoned, was just a memory aid, but could not substitute for speech,which was always bound to a speaker who could be asked for clarication. In con-trast, written words were silent, they lacked the immediacy of speech, they weredead. In Platos day, knowledge and knower were not separated, as is typically thecase in fully literate societies.

Zen

Platos critique of writing has been an undercurrent of Western thinkingwhich, however, has strongly favoured the Aristotelian notion that writing is arepresentation of oral language. As a tool of enlightenment it has met with similardistrust in the Eastern tradition. For example, consider the common Zen sloganwritten words are useless (Japanese: furyu monji), which protests the distancebetween message and author/reader and the reliance on objectied knowledge.Enlightenment is practice, consciousness in action, the Way; it cannot be capturedin xed signs. Notice, however, that there is no consistent Zen view on writing,just as there is no such thing in Plato. In both cases, scepticism is coupled withveneration. Plato put his misgivings about writing into writing. It was he who

6 What is writing?

Figure 1.1 Chinese character wu,nothing

preserved in his writing Socrates philosophy for posterity. Excluding from hisRepublic poets who at the time were seen as reciters rather than creators of songs,he did more than anyone to usher in a literate culture grounded in analytic thinking.And much as Zen adherents denied the cognitive value of writing, they practisedthe art of writing. Calligraphy is one of the most highly valued and sublime arts in-spired by Buddhism, shodo the Way of writing. Consider, for example, the Chinesecharacter for nothing (Chinese wu, Japanese mu) in gure 1.1 at which many aZen master has tried his hand. The overwhelming presence of what means the ab-sence of everything is striking and at least as amazing as Rene Magrittes paintingThe betrayal of images (gure 1.2). It is hard to imagine that, in the absence ofwriting, the thingness of nothing would have become a philosophical problem. Wuis not nothing, it just means nothing, a relationship much like that between a pipeand a picture of a pipe. The visual nature of the sign does the trick.

It is perhaps not surprising that something that touches the human mind so deeplyas does writing should evoke diverse and countervailing responses. There is some-thing inherently contradictory about writing, the paradox of arresting the transitory.In this book I am not concerned with the philosophical aspects of this paradox orthe artistic expressions it inspired, but we cannot ignore its consequences for lin-guistics. It is common practice in linguistics to ignore the paradoxical character ofwriting down language, of treating as achronic something whose very essence isits existence in real time. At best it is treated lightly as a necessary and legitimateabstraction. However, this proves nothing but the fact that linguistics, notwith-standing its claims to universality, is a Western science thoroughly rooted in theAristotelian tradition. For the scientic study of language is confronted with thisparadox from the very beginning. Before anyone thought of writing them down,words were evanescent, verba volent. Recording the ephemeral, providing theeeting word with a permanent form ready to be inspected and reinspected is therst step of linguistic analysis, a step that, strictly speaking, is as impossible to takeas it is impossible to give a straight answer to a koan, an illogical riddle developed

Zen 7

Figure 1.2 Rene Magritte 1929, The betrayal of images

by Zen masters as a technique to discredit the verbal side of the mind. How doyou see things so clearly, a Zen master was asked. I close my eyes, he answered.This little episode warns of the danger of believing in ones own systems and cat-egories, the categories, that is, that guide the seeing eye. Another koan describesthree monks watching a streamer utter in the breeze. One of them comments,The streamer is moving, while the second objects, The wind is moving. Thethird monk says, You are both wrong. It is your mind that is moving.

To distinguish the categories that are inherent in the object of observation fromthose that are in the mind is a fundamental problem of linguistics, as of all empiricalsciences. Writing suggests xed categories and stability: words, syllables, letters.This would not be a problem if writing systems were the object of inquiry andanalysed in their own right in order to discover the structural relationships betweentheir constitutive elements. However, they are often studied for what they wouldreveal about the nature of language as well as the mental processes underlying it.The very existence of writing is taken as proof that language can be studied as ifit were a stable object consisting of xed parts. Even though it is recognized asonly a representation of speech, its categories are allowed to intrude into linguisticinquiry. In order to avoid confusion, it is of great importance, therefore, clearly todistinguish that which writing represents of language from what it imposes ontoit. This is no easy task, as the following denition, which we nd in an ancientEgyptian text, indicates.

8 What is writing?

Egypt

Egyptian hieroglyphs were understood as models of the totality of allthings. An ancient Egyptian onomasticon, that is, a list of words ordered for sub-jects, is described in the introduction as the beginning of the teaching for clearingthe mind, for instruction of the ignorant and for learning all things that exist: whatPtah created, what Thoth copied down (Gardiner 1947: 1). It was things that wererecorded, not words. In his introduction to the lists he edited, Gardiner (1947: III),therefore, remarks:

Their title to be called Vocabularies could be upheld only if the lists could beshown to refer primarily to words, rather than to things, and that was clearlyagainst the intention of the compilers.

That a direct relationship between things and written signs was assumed by theEgyptians is also suggested by a text about creation in which the hieroglyphs playa crucial role.

And the whole multitude of hieroglyphs were created by what was thought in theheart and dictated by the tongue. And thus Ptah was content when he had createdall things and all hieroglyphs.

All things and all hieroglyphs, Egyptologist Jan Assmann explains, means theforms of nature and their rendition in writing. The heart envisages the forms, thetongue voices them as words, which, by demiurgical powers, attain a physicalexistence as things. Things are modelled as inner writing in the heart subsequentlyto be vocalized by the tongue and transformed into perceptible entities of thephenomenal world. There is a virtual congruency between the corpus of signs andthe corpus of things (Assmann 1991: 91). According to this view the signs precedethe things, they are models rather than images. Creation is an act of articulationin the heart, which nds expression in written signs rst and then in speech.Externalized writing is thus more properly viewed as a discovery than an invention.

This account puts Aristotles linear order of the elements involved in writing onits head and, therefore, from an Aristotelian point of view, strikes us as bizarre. Howis it to be understood? The pictorial clarity of Egyptian hieroglyphs is well knownand offers an explanation. Does not the Egyptian understanding of writing differfrom the Greek because of the iconic relationship between signs and objects sostrikingly evident in Egyptian writing but lacking in Greek? This explanation, onceagain, implicitly assumes that properties of writing systems have repercussions onconceptions of what writing is. On this ground, the Egyptian idea of writing couldbe easily cast aside as irrelevant for a theory of writing proper, which consists in therepresentation of words, rather than things. Disturbingly, however, the Egyptians

Massias 9

are not alone. Similar denitions of writing have been proposed within the Westerntradition and about Western, that is, alphabetic writing.

Massias

After alphabetic literacy had shaped Western ideas of writing for morethan two and a half millennia, in the nineteenth century, Nicolas de Massiaspublished a book in Paris entitled The Inuence of Writing on Thought and onLanguage. At the time, writing attracted much attention among European intellec-tuals because Francois Champollions decipherment of the Rosetta Stone in 1822had demonstrated to the world that Egyptian hieroglyphs could actually be readand thus constituted writing, though of an utterly different kind than alphabeticwriting. Like many of his contemporaries, Massias thought that writing, especiallyphonetic writing, was closely linked with language. He thought of it not just as ameans of representing language or of cultivating it, but as something much moreessential, which permits language to fully develop:

Here then is man, able by means of language, thought, spoken and written, tocommunicate with himself and with his present and absent similars. But theselanguages resolve themselves into a single one, which is limited, written speech.It is this necessity of writing which gives its name to grammar, osteology andframework of discourse. (Massias 1828: 5)

The rst writing, without which man could not speak to himself and which dis-tinguishes him from animals is that which the mind has traced in itself by its ownaction. (p. 7)

Phonographic writing is favorable to speech; it is speech; it makes up and breaksup the smallest elements of sound; and it sustains all movements and operationsof the human spirit. (p. 96, quoted from Aronoff 1992: 72f.)

That writing is equated with speech sounds nebulous, but from the earlier quotesit is obvious that Massias does not speak metaphorically. Writing, for him, isat the heart of every language. Thought and spoken and written language arecollapsed into one, written speech. As an ideal code it actively articulates ratherthan reproduces articulation performed in vocal speech.

In the Egyptian account of writing hieroglyphs are models of things created bywhat is thought in the heart; in Massias account language itself, its categories(grammar), structure (osteology) and framework of discourse are traced as writingin the mind. As we will see, the idea that writing is a blueprint for, rather thana representation of, speech is not as bizarre as it seems, although most linguiststoday would reject it out of hand.

10 What is writing?

Contemporary views

Although there is plenty of evidence that, in literate cultures, writingintrudes into the linguistic behaviour of people and that without writing manylanguages would not be what they are, the notion that writing is an active agent oflanguage is unpalatable to many linguists for a number of reasons. One is that inmodern linguistics languages are stripped of their historical dimension. Althoughthe obvious fact that languages change in the course of time is acknowledged, thepossibility that their nature may be affected by external factors such as writingis strictly denied, allegedly on the grounds that writing could not possibly haveexercised any inuence on the faculty of language because it is too recent an in-vention. The oldest records reach back a bit more than ten thousand years at best,while language must have evolved hundreds of thousands of years ago. Diachroniclinguistics is essentially unhistorical, because, as a dening capacity of the humanrace, language is not supposed to change by virtue of a humanly contrived technol-ogy. There are no highly or less highly developed languages. This is a primitive oflinguistics. Artifacts and technologies, such as writing, for example, are grantedthe potential to change the environment, but not humanity itself. Since languageis conceived as an essential part of human nature, while writing is a mere tech-nology, the effects of writing on language and by implication the complexities oftheir interrelationship remain largely unexplored.

Scholars in the language sciences who do believe that the invention or discoveryof writing does make a difference, both with respect to what language is and howwe think about it, are in a minority. Linguistic orthodoxy happily concurs withFerdinand de Saussures apodictic statement that made Aristotelian surrogation-alism a cornerstone of modern linguistics:

Language and writing are two distinct systems of signs; the second exists forthe sole purpose of representing the rst. The linguistic object is not both thewritten and the spoken forms of words; the spoken forms alone constitute theobject. (Saussure 1959: 23)

Following this prescriptive instruction, most introductory textbooks of linguis-tics simply exclude the problematic of writing or make do with a cursory reviewof a number of writing systems in the nal chapter. Notice in passing that thisis quite different in the Eastern tradition of the scientic study of language. TheEncyclopedic Dictionary of Chinese Linguistics (Zhongguo yuyanxue da`cdian19912), for example, treats writing systems as its rst topic at great length. Anoble and widely accepted reason for ignoring writing or treating it lightly in theWest is that all human languages are thought to be equal in the sense that theyare expressions of the same inborn faculty of language. The concepts and theories

Contemporary views 11

of linguistics, therefore, have a universal appeal and should be applicable to alllanguages. Since the majority of the languages of the world are unwritten, it isonly prudent to ignore writing when studying language. However, this argumentis not as sound as it seems. For, if all languages are of a kind it follows that if somelanguages are writable all languages are, and since writing is undeniably not thesame as language, it is a legitimate and interesting question how the two relate toeach other. Two questions linguists should not sidestep are: What happens whena language is written down, (1) in terms of linguistic description, and (2) in termsof linguistic evolution? As a matter of fact, linguists never study any languagewithout recording speech and writing it down. This alone is a compelling reasonfor studying writing instead of assuming that writing, whose essential propertiesare so radically different from speech, can be ignored in the research process. Someof the differences are the following.

Speech Writingcontinuous discretebound to utterance time timelesscontextual autonomousevanescent permanentaudible visibleproduced by voice produced by hand

Each one of these contrasts warrants careful investigation because it is by nomeans self-evident how an audible sound continuum produced by the human voice,which can only be perceived at the time of utterance, relates to a discrete sequenceof xed visible marks produced by the human hand, which can be perceived atany time. One way out of the difculty is to say that all of the above are externalcontingencies of language, which linguists are not really interested in. Linguistics isconcerned with the abstract system of language, not with its physical manifestationthrough speaking, writing or signing. The unwelcome consequence of this line ofthought is that vocal speech, too, would have to be expelled from the realm oflinguistics and with it what many consider the heart of the science of language,phonetics and phonology.

Both the medium of sound and the physiological apparatus for modulating soundwaves are deemed essential for the evolution of language. The human faculty oflanguage cannot really be divorced from vocal speech. A soundless linguistics,therefore, most mainstream linguists would agree, is not just truncated, but anoxymoron. Relating sound to meaning is the very essence of language. Accord-ingly, a theory of language a grammar must specify rules for mapping semanticstructures onto phonetic structures. Since Saussure, grammatical theory has under-gone revolutionary changes, but the central concept of relating sound to meaning

12 What is writing?

Figure 1.3 Saussures model of the linguistic sign

in a structured way has remained the same. Saussures model of the linguisticsign still captures the main point. Sound in language has three aspects, whichhe distinguishes: physical (sound waves), physiological (audition and phonation)and psychological, that is, sounds as abstract units, which he calls sound images(images acoustiques).

The linguistic sign unites a concept and a sound image. The latter is not the ma-terial sound, a purely physical thing, but the psychological imprint of the sound,the impression that it makes on our senses. (Saussure 1959: 66)

Image, imprint, impression are the terms he uses to clarify what he means,plainly visual terms that he preferred to, or was not able to exchange for, others lessreminiscent of writing. Saussure denounced the tyranny of the letter and degradedwriting as a distortion of speech. He may have fallen victim to this tyranny himselfin unexpected ways. There has been little discussion about what exactly a soundimage is. The cardinal question is what it is an image of, or of what sound it is animage. The sound shape of words varies from one speaker to another, and even oneand the same speaker is unable to produce an exact copy of an earlier utterance.How then do we recognize sounds as the same? Is there some kind of matrix orideal sound that Saussures sound image incorporates? Some scholars think thatthis is so, Frank Householder, for example. He speaks of a proto-written varietyunderlying speech arguing that in a literate speech community speakers intuitivelyfeel that speech is a rendition of writing, not vice versa (Householder 1971: 253).In many cases this is undeniable. An ever increasing part of the vocabulary ofwritten languages come into existence in writing. They can be given a phoneticinterpretation, which, however, is decidedly secondary. What is more, thanks tothe impact of literacy schooling, it is likely that most educated peoples conceptionof language should be inuenced by writing. A number of scholars have suggestedthat linguists are no exception and that sound image and other important termsin linguistics are proof of that. They are derivative of writing.

Writing and linguistics 13

Writing and linguistics

In 1982 Per Linell published a monograph with the telling title The WrittenLanguage Bias in Linguistics in which he presented elaborate arguments to theeffect that

Our conception of language is deeply inuenced by a long tradition of ana-lyzing only written language, and that modern linguistic theory, including psy-cholinguistics and sociolinguistics, approaches the structure and mechanism ofspoken language with a conceptual apparatus, which upon closer scrutiny turns out to be more apt for written language in surprisingly many fundamentalaspects (Linell 1982: 1).

Ever since Saussures above-quoted postulate, the primacy of speech is taken forgranted in linguistics but belied by actual research and theory formation. Aronoff(1992) points out that, like Saussure, Edward Sapir, Noam Chomsky and MorrisHalle appeal to alphabetic writing in developing their phonological theories. Faber(1992: 110) interprets the observation that many speakers cannot divide words intophonological segments unless they have received explicit instruction in such seg-mentation comparable to that involved in teaching an alphabetic writing systemas evidence that historically segmentation ability was a consequence of alphabeticwriting, not a prerequisite. Various sounds such as diphthongs and prenasalizedconsonants, which in alphabetic writing are represented by sequences of letters,cannot realistically be conceived as isolated steady units. Segment-based phonol-ogy, Faber concludes, is an outgrowth of alphabetic writing and may not be suitedto represent language as a mental system.

Other key concepts of linguistics have been linked to writing in a similar manner.Building on literacy and education research, David Olson (1994) stresses the pointthat the concept of the word as a distinct unit is a by-product of literacy acquisition.Morphology, the study of words and their parts, is deeply imbued with notions ofliterate word processing, such as lexical entry, for example. Lexicon itself issuch a term. A lexicon is a list of isolated words, a kind of usage that does notoccur naturally in speech. The word is an articial entity in another sense as well.It is basically the kind of unit that is listed in a dictionary and thus not necessarilythe same in all languages. It stands to reason, therefore, that the lexicon as a part ofgrammar that supposedly we have stored in our heads and that grammarians inves-tigate would not be a research object for grammar if it was not for the written model.The same is probably true of the unit on which syntax is centred, the sentence.

Many researchers who analyse unelicited real-life discourse have observed thatin speech the sentence is unimportant and more often than not hard to iden-tify. There is no cognitive, content, or intonation unit in spontaneous speech that

14 What is writing?

corresponds to a grammatical sentence. At the same time, attempts at an unam-biguous, uniform and universal denition of sentence have been inconclusive.The sentence is a unit of written language, and a sequence of words between twofull stops is as good a denition as any. Further, sentences are said to have literalmeanings. A question we may want to ask, without jest, is whether this also holdsof unwritten languages. Is it just the language of writing from which we borrowa descriptive term suitable for the phenomenon, or is it the phenomenon itselfthat derives from written language? Olson (1994) has argued that the distinctionbetween a speakers meaning and literal meaning is a by-product of literacy. Staticentities like the stock of words, sentences and written texts are alien to the spokenlanguage where meaning is constituted in the act of speaking, bound to situation,speaker, context, the interaction history of speaker and listener, and so on. Takeaway all that and you get the literal meaning, true to the letter, that is. As Olsondemonstrates at length, this terminology is not fortuitous but speaks of the factthat linguistics is grounded in written language. Since linguistics is concernedwith natural language while writing is an artifact, this is difcult to openly inte-grate into linguistic theory, which, as a result, is characterized by scriptism, whichhas been dened as

the tendency of linguists to base their analyses on writing-induced concepts suchas phoneme, word, literal meaning and sentence, while at the same time subscrib-ing to the principle of the primacy of speech for linguistic inquiry.

(Coulmas 1996: 455)

As Olson sees it, linguists are in this respect representative of literate societyat large where writing provides the model for speech, rather than the other wayaround. We pronounce as we spell, we judge our utterances against the yardstick ofwritten sentences and qualify as ellipsis, anacoluthia, reduction, false start and soon those which do not conform to these patterns. The literal meaning of a sentenceis basic. Other meanings are taken to be derived from it. To a scholar who, likeOlson, looks at language as something to be learned, such a conception, perhaps,comes quite naturally because it is the written form of language that is made theobject of instruction, memorization and testing. As an institution, the school instilsinto the collective mind the primacy of writing. In contrast, those who prefer tolook at language as a natural capacity tend to insist on the primacy of speech. Theseseemingly irreconcilable positions reect the two sides of language, the acquiredand the innate. Since no human being exists as a purely natural creature, both canbe dissociated only in theory. This is the deeper meaning of Olsons notion thatwriting is a model of speech. Acknowledging the cultural and historical nature ofhumanity, it takes seriously the possibility that an artifact, writing, may act uponone of its most essential natural endowments, language.

Writing and linguistics 15

In his conception of the relations between speech and writing, Olson has beeninuenced by Roy Harris (1986), one of the most outspoken critics of Aristoteliansurrogationalism. Harris project is to demonstrate that the development of writtensigns is independent of spoken language. Accordingly, his notion of writing isextremely comprehensive, encompassing both glottic, or language-based, notationsystems and non-glottic systems such as musical and mathematical notations. Anyanalysis of glottic writing, he argues, should start from here rather than from theallegedly secondary character of writing as a representation of speech, howeverimperfect. Stressing the continuous nature of speech, he insists that there couldbe no complete isomorphism between any system of visible marks and any systemof sounds (Harris 2000: 189). It is quite unclear, therefore, what it means thatwritten signs represent sounds. The relationship between speech and writing isfundamentally different from, for example, the representation of a city by a citymap where an inch represents a mile in a straightforward and well-dened way.Hence it is necessary to rethink the conceptual model that has guided Westernthinking about writing for so long. One alternative view is to conceptualize therelationship between speech and writing as one of interpretation. Rather thantrying to depict sounds, written signs are given a phonetic interpretation. There isa historical justication for this view in that writing did not evolve as a means torecord speech but as a system of communication.

I. J. Gelb, whose A Study of Writing was long the most widely cited work onwriting, in a rst attempt offered a very wide denition of writing as a system ofhuman intercommunication by means of conventional visible marks (1963: 12).Various kinds of visible marks seem to fall under this denition, because it saysnothing about whether, how and to what extent language is involved. But Gelb,too, held a surrogationalist view of writing. While acknowledging that in historythe representation of speech was not the origin or the initial purpose of writing,he sharpened his denition stating that fully developed writing became a devicefor expressing linguistic elements by means of visible marks (1963: 13). Gelbsexplanation that writing became a device for expressing language rather than hav-ing been such a device from its inception still seems to leave room for recognizingnon-linguistic functions of writing. But since he considered its becoming such adevice to be the rst step of a goal-directed development it is hard, from his pointof view, to see in the non-linguistic functions of writing anything but signs ofimmaturity. He saw writing evolve from a rather loose connection with languagequasi-naturally towards an ever closer relationship, as the units of representationgot smaller and fewer. The evolution, Gelb was convinced, could not but lead to-wards pure sound representation culminating in the Latin alphabet, the most perfectof all writing systems. This quasi-social Darwinist view has been criticized as thecommon Latin alphabet fetishism (Battestini 1997: 285), because it makes writing

16 What is writing?

systems that communicate information by other means not mediated through therepresentation of sounds appear deviant, decient and underdeveloped.

Combining evolutionism with Aristotelian surrogationalism, Gelb tried to ac-count for the multiformity of the worlds writing systems in a uniform and theo-retically founded way. However, by committing himself to the superiority of theLatin alphabet he not only opted for a very one-sided criterion for judging progressand the goodness of writing systems, he also made it difcult to appreciate otherwriting systems for their own merits. Gelbs scholarship was unrivalled in his day,and much of what he contributed to our understanding of scripts, especially ofthe ancient Near East, is still valid. But his theoretical approach should no longersatisfy the study of writing.

In this book, great importance is attached to Gelbs observation that writingbecame a means of expressing language, but his contention that an inevitableteleological evolution was thus initiated is where we part company. Recordinginformation by graphical means is a basic function of writing that is never nar-rowed down entirely to the representation of sounds. Writing cannot and shouldnot be reduced to speech. Saussures above-quoted observation that language andwriting are two distinct systems of signs must always be kept in mind, but thesecond part of his denition, that writing exists for the sole purpose of represent-ing speech, must be rejected, for writing follows its own logic which is not that ofspeech. From the above discussion about scriptism and the written language biasin linguistics it is clear that there are alternatives to the received opinion that writ-ing is but an imperfect, distorted and hence misleading representation of speech,which deserve to be taken seriously. The relationships between speech and writingare undoubtedly highly complex, but if the medial, bio-mechanical and cognitivedifferences between them are acknowledged there is no reason to assume a perfectrendition of the former by the latter. If, as I will try to do in the chapters thatfollow, we free our grasp on writing from the Western preconception that writingshould, really, be a faithful representation of speech, then there is little reason toblame writing for whatever discrepancies we discover in the analysis. No writingduplicates speech. Precisely for this reason a thorough understanding of writingis a necessary prerequisite to doing linguistics, to reect on what we as linguistsare doing when we record speech for the purpose of analysis.

As will be demonstrated in the chapters that follow, a wide gap between spokenand written language is very common in the literate cultures of the world, and thefact that there are many functional and structural characteristics of writing thathave no counterpart in speech is taken by their members as a matter of courserather than a deciency of writing. Both historically and conceptually, writing hasa certain autonomy. At the same time it would be unreasonable to ignore the impor-tance of writing as a means of linguistic communication. With Harris I, therefore,

Writing and linguistics 17

avoid normative surrogationalist assumptions, but unlike Harris I reserve the termwriting to what he calls glottic writing. Any denition of writing reects bothan understanding of, and a particular interest in, the object of inquiry. Since myconcern here is mainly with the linguistic aspects of writing, only systems withan unmistakable linguistic interpretation are considered within the framework ofthis book. Precisely because writing is targeted here as a means of linguistic com-munication, due attention must be paid to the differences between the expressivepotential of spoken and written language, which make it imperative to dispensewith the reductionist assumption that writing does nothing but represent speech.Writing changes the way we think about language and the way we use it. By virtueof the fact that writing is based on an interaction of hand and eye, the writingsystems of the world have many characteristics in common. Yet, they also dif-fer in important respects due to their different histories and the diverse structuralprinciples on which they are based. Before going into detailed examinations ofindividual writing systems, the next chapter gives an overview of the major typesof writing systems and a number of attempts at their classication.

2The basic options: meaning and sound

I remember telling you about a plan for an extraordinary character whichwould be a means of painting not speech but thoughts like algebraic factsin mathematics. Putting ones discourse into this character one would makecalculations and proofs rationally. I believe we could nd a method to combinethis with the ancient characters of the Chinese.

Gottfried Wilhelm Leibniz 1701

The language of this country is different from that of China, so that it isimpossible (for us Koreans) to communicate by means of Chinese characters.[ . . . ] If there are sounds natural to Heaven and Earth, there should certainlybe writing natural to Heaven and Earth. Thus ancient people made lettersaccording to the sounds and through them the feeling of the myriad thingswere communicated. King Sejong of Korea 1443

The diversity of the worlds writing systems is enormous, but they can all be in-terpreted semantically and phonetically. The communication of meaning is theprimary purpose of most writing, and in one way or another conventional rela-tionships between graphic and phonic units are established to accomplish this.Meaning and sound are the two referential dimensions utilized by all writing sys-tems. It has sometimes been assumed that writing could work by relying on oneof them only. However, a graphic system that expresses meaning directly is asunrealistic as pure transcription, or visible speech. As exemplied by the abovequotes, both meaning-based writing, or semiography, and sound-based writing, orphonography, have been envisioned by inuential thinkers who viewed the graphicexpression of pure reason and pure sound, respectively, as ideals to be pursued inthe design of writing systems. Notice that one and the same writing system, theChinese, is adduced in support of their arguments by either side, though for oppo-site reasons. This indicates that real writing systems do not conform to ideals andthat the true nature of a given writing system may not be easy to determine. Yet,the two ideals of semiography and phonography bring the main issues of the studyof writing into focus. They are, therefore, reviewed here together with some otherbasic features of writing.

18

Visual perception 19

Visual perception

Writing is visible. It is a form of communication created by the hand andappealing to the eye. Students of writing are agreed on these two points acrossall theoretical differences. The dexterity of the human hand, visual perceptionand the ability of the central nervous system to maintain a feedback circuit byusing a visual input for controlling delicate manual movements are basic to writ-ing. Another characteristic of great importance is that writing consists of signs,that is, relatively durable marks that have an assigned external referent. Humanbeings produced graphic signs for many millennia before writing was invented.They drew pictures, cut notches into sticks, arranged pebbles in heaps and gures,tied knots in cords, scratched patterns onto rocks. Already the oldest pictures yetdiscovered, the cave paintings of Lascaux in south-western France dating backmore than 30,000 years, exhibit unmistakable evidence of aesthetic sensibility.There is no way of knowing why they were produced and exactly what functionsthey were meant to full. Perhaps the drawings corresponded to narratives or werein some other way connected with linguistic utterances, but nothing suggests thatsuch a connection, if any, was systematic. What can be said, however, is that,being images of animals encountered by the palaeolithic artists, they were clearlysigns.

The functions of other ancient graphic signs are easier to determine. For exam-ple, the main functions of notches carved into sticks or bones, known in Europeat least 35,000 years ago and still in use in medieval times (see gure 2.1), as also

Figure 2.1 Carved bones, approximately 35,000 years old. After Leroi-Gourhan1964: 264

20 The basic options: meaning and sound

10

9

8

7

6

5

4

3

2

1

Figure 2.2 Quipu knots and their numerical values

of the knotted cords, or quipu, of Inca Peru (see gure 2.2), was memory supportand social control. These systems are mnemonic devices, which embodied socialobligations and conventionalized promises such as repayments of debts and war-rants of goods to be delivered. The quipu were a recording system that enumerateddifferent classes of objects and people. It has been reported by early Spanish sourcesthat the cords also held non-quantitative information such as history, mythologyand astronomy, although no conclusive evidence has been established in supportof this assumption. Perhaps they were a means of communication for the initiated.If so, what is it that distinguishes quipu, tally sticks and other mnemonic devicesfrom writing?

Conventionality 21

Auto-indexicality

No writing system is immediately comprehensible without instruction.In this sense writing, too, is a means of communication for the initiated. But thereis a difference. Most scholars with a linguistic interest in the subject, such as, forexample, Cohen (1958), Gelb (1963), Diringer (1968), Nakanishi (1980), Gaur(1985), Catach (1987), Coulmas (1989), DeFrancis (1989), Gunther and Ludwig(1994, 1996), and Daniels and Bright (1996) recognize as writing graphic systemsthat, in addition to being codes learned by instruction, embody the principles oftheir learnability. By virtue of their graphic composition they reveal the procedureson the basis of which they must be used. In this sense writing is auto-indexical.Every written document not only embodies the message I am meant to be read butalso instructions, however indirect, as to how this can be done. In other words, thesystematic make-up of writing contains a key to its own decipherment. Mnemonicdevices such as quipu and tally sticks and many others lack this level of structuralorganization. Like both writing systems and pictures, they may indicate by theiroutward appearance that they are intended to be perceived as signs, but they donot incorporate any information about the procedure of their own interpretation.This must be communicated separately, as is usual also with writing, which isconventionally learned at school in accordance with established procedures. Butwhere such procedures are unknown, for example because a tradition has beenterminated, it is still possible to recover them by inspecting the written documentsalone, as the great decipherers of Egyptian, Akkadian and Maya writing havedemonstrated.

Conventionality

Pictures such as the cave paintings of Lascaux or the pictographs ofNorth American Indians (Mallery 1972) may have had communicative functions,for instance by being associated with storytelling. Perhaps such an association, ifany, was habitual, but it was not conventional. This difference is crucial. Habitsestablish a practice that is recognized by the members of a certain reference groupas right or good or appropriate and that is transmitted from one generation to thenext by demonstration and situated example. In contrast, conventions establish acode that assumes an independent existence. Codes are conventional procedures forusing signs. They can be cracked, pictures cannot. They can be cracked becausethey encode information about themselves. The kind of relationship that holdsbetween a code and its rules is not picturable, but it can be deduced from the

22 The basic options: meaning and sound

Figure 2.3 Winter count of the Dakota. Pictographic signs serve as memory aids(Mallery 1893).

graphic properties of its composite elements and their arrangement. Writing in thesense understood here is such a code. As we shall see, this implies that writing isbound up with language.

Taking auto-indexicality as the main criterion to distinguish writing from othergraphic signs is a theoretical decision. Other criteria are conceivable. Semioticianand Africanist Battestini, for example, considers writing rst and foremost as ameans of conservation of the collective memory for which a great variety of visualmarks, much like those of the Dakota winter count, are available. He argues againsta narrow language-bound notion of writing. His reasons are threefold. One is thefunctional similarity of writing in the restricted sense with other graphic signs alsoserving the function of collective memory aids. Visible marks with an assignedmeaning should all be analysed from the point of view of this functional similar-ity. His second reason refers to the medial difference between speech and writing,arguing that no system of graphic notation has ever been capable of effectivelyreducing speech. What is written is the trace of thought (Battestini 1997: 32).Accordingly, he postulates that the function of writing, in the semiotic sense,is not to represent speech or language as generally assumed in the West but thoughts (1997: 102). The ultimate reason why Battestini favours such awide notion of writing is that he wants to include all graphic means of con-serving and communicating thought, especially language-neutral systems he callsmythographic of which there is an abundance in Africa, the continent whose

The ideal of language neutrality 23

peoples have commonly, and wrongly, been characterized as writingless. His pur-pose is to correct this perception and to expose its Eurocentric and demagogicfoundations.

Battestini is certainly right to debunk the idea that writing dened in a particularand, as he puts it, restrictive sense can be used to evaluate civilizations. His isa powerful and convincing voice in the post-colonial discourse, and his 1997work on Africa and writing is an important contribution towards a reappraisalof Africas rich traditions, literal and nonliteral, and of the manifold semioticmeans of conserving and communicating ideas. But he overshot the mark with hisclaim that a narrow denition of writing necessarily implies a denigrating view ofcivilizations that do not rely on writing systems that fall within its range.

The ideal of language neutrality

A wide semiotic conception of writing as advocated by Battestini, Harrisand others gets support from another direction. That writing should not only beconceptualized but also developed as a mode of communication sui generis is anold dream. Do graphic sign systems hold the potential to overcome the obviouslimitation of phonetic language of being comprehensible only by the members of agiven speech community? Could writing be elaborated to become a universal codebypassing language? Leading European intellectuals of the Enlightenment suchas Francis Bacon and Gottfried Wilhelm Leibniz were fascinated by the idea of auniversal script as a logical instrument, an aid to memory, and a means of interna-tional communication. Bacon believed that Egyptian hieroglyphs, undecipheredat the time, referred to objects and could be used for transmitting knowledge sothat countries and provinces, which understand not one anothers language, cannevertheless read one anothers writings because the characters are accepted moregenerally than the languages do extend (Large 1985: 12).

In search of a means to transcend the limitations of language, Leibniz likewiselooked at an extant writing system, the Chinese. In the seventeenth century, Jesuitmissionaries returning from China described a kind of writing consisting of tensof thousands of characters directly expressing ideas. Based on such reports hetried to discover the principles underlying this script, which he thought might holdthe key to the deepest philosophical problem, the relationship between reality andwhat we know about it. Logic and the theory of signs played a central role inhis philosophy throughout his life. In his Ars characteristica universalis of 1666he combined the two with logical atomism seeking to construct an alphabetumcogitationum humanarum, an alphabet of human thought. He was convinced thatideas could be visualized if only we could discover the fundamental concepts in

24 The basic options: meaning and sound

all possible existence. The basic premise underlying this project is a cornerstoneof Western thought: the world is exhaustively divisible into individuals. AlthoughLeibnizs search for conceptual atoms became ever more elusive the more heworked on it, he held on to his belief that it was possible and that the Chinese writingsystem actually was an imperfect realization of a universal script that could directlyexpress ideas and thus provided a model. Chinese characters he believed referrednot to words but to things. This was a fundamental advantage because the scripturauniversalis he envisioned would be a semantic script whose characters representconcepts independent of a particular language, but that could be pronounced inany language (Widmaier 1983: 33).

Leibniz did not succeed. The information he had about Chinese writing wasfragmentary and misconceived. But his aspiration to construct a universal scriptof scientic thinking in which all true sentences could be formally deduced livedon long after him. Of the many attempts that explicitly followed in his footstepsGottlob Freges Begriffsschrift, or concept script, of 1879 is the most noteworthy.It was intended as a lingua characteristica as Leibniz had envisioned it. Concepts toFrege were unchangeable meanings, but while Leibniz saw the greatest and eventu-ally insurmountable challenge in devising a catalogue of such semantic primitives,Freges main concern was with propositions and their logical connections. His no-tation is completely different from conventional notations of sentential logic in thatit exploits the two-dimensionality of the writing surface. The antecedent and theconsequent of a conditional are written on separate lines connected by a verticalline. In contradistinction to common logical notations the Begriffsschrift has nosymbols for alternation, conjunction and the existential quantier. All of these re-lations are expressed by hierarchies of horizontal propositional lines and verticalconnections as in gure 2.4.

a

c

b

c

a

b

c

Figure 2.4 Freges Begriffsschrift. Is it writing?

The ideal of language neutrality 25

shoe

works

shoe-works

shoes producedby machine

shoes producedby handwork

Figure 2.5 Example of Otto Neuraths International Picture Language

A straight line symbolizes a sentence regardless of any claim to its truth. Avertical bar added at the left end indicates what we do when we assert the truthof a sentence. a, then, is the assertion that a is the case. Using this as thebase and adding just a couple of other graphic signs that need not concern ushere, Frege produced a complete formalization of rst-order logic. His notation isingenious and uniquely adapted to two-dimensional visual expression of complexrelationships between sentences, but is it writing?

After Frege many thinkers have devoted considerable effort to universal writ-ing schemes. Logician and philosopher Otto Neurath designed an internationalpicture language, paving the way for modern pictographic icons (see gure 2.5).The project of a universal character is periodically discussed in philosophical jour-nals (e.g. Cohen 1954), while philosophers of language continue their quest forthe language of thought conceived as the smallest set of vocabulary items interms of which the entire vocabulary can be dened (Fodor 1978: 124). These

26 The basic options: meaning and sound

are thought to be units of an internal code over which cognitive operations aredened (Fodor 1978: 115). At the same time, the Chinese script has not lost itsspell to inspire discussion of a world script (e.g. Nagel 1930), notwithstandingrepeated explanations of its language-bound nature on the part of Sinologists.Anthropologist Jack Goody states that globalization clearly requires some formof international communication

. . . While the Chinese script is a model, and while there would be some advantagein using a script already employed by a fth of the worlds population, a preferablealternative might be to attempt to construct a new, and possibly more logical,script, using existing icons but developing a new system altogether.1

The language in which this proposal is couched is modern, but the project is stillthe same as Leibnizs three hundred years ago.

The ideal of faithful transcription

The drive for a concept script according to what may be called theLeibnizian tradition was internationalist from the outset: one system for all ofhumanity. Internationalism was also the intention of a project that approached theproblem of writing from the other end, sound, and again it was predicated on thesame idea that the phenomenal world can be broken down into elementary parts.On the face of it, this project has been more successful than the search for semanticprimitives. Early on, the description of speech sounds in terms of the physiolog-ical processes of articulation was recognized as the most promising path to asolution.

In 1667, Franciscus Mercurius van Helmont, a Dutch scientist and philosopherwho believed in physical and mental atoms, or monads, and who may actually haveinuenced Leibniz, published a book entitled Alphabeti vere naturalis hebraicibrevissima dilineatio in which he speculates that each Hebrew letter pictures theposition of the articulatory organs as it is pronounced (see gure 2.6). Helmont waswrong about the Hebrew alphabet, but the idea that the physiology of articulationshould be the basis for a scientic description and representation of speech soundswas pursued by many subsequent schemes, and preceded by at least one. Theinventors of the Korean script, which was created because the Chinese script wasill-suited for the Korean language, claimed an iconic relationship between lettersand sounds. Basic letter shapes were designed to imitate tongue positions duringarticulation (see gure 2.7). Such writing would be natural to Heaven and Earth,

1 Jack Goody, A World Script A Modest Proposal, 5 July 2000; unpublished ms.

The ideal of faithful transcription 27

Figure 2.6 Hebrew letter mem, m, as iconic depiction of tongue positionaccording to van Helmont

Figure 2.7 Iconicity of Hangul consonant letters

as the introduction to the promulgation of the new script quoted at the beginningof this chapter put it. More will be said about the Korean script in chapter 8. Here itsufces to note that iconicity between graphical signs and positions of articulationorgans has been regarded, for all we know independently in two parts of the world,

28 The basic options: meaning and sound

as a formula to make speech visible. In the same spirit John Wilkins, an Englishbishop, in his Essay towards a real character and a philosophical language of1668 developed a physiological alphabet whose 34 signs were meant to depictarticulatory processes (Dudley and Tarnoczy 1950: 154), because there shouldbe some kind of suitableness, or correspondency of the gures to the nature andkind of the Letters which they express (see gure 2.8). Notice that Wilkins usedthe term Letter, as was common practice in his time, to refer to speech sounds,whereas gures were graphic signs. A century later, Sir William Jones (1786),one of the founding fathers of modern comparative linguistics, believed that thesigns of all scripts at rst, probably, were only rude outlines of the different organsof speech.

The different organs of speech were thus what had to be studied, and out of theseand similar attempts to understand the physiological basis of speech productiongrew the science of phonetics in the West, which had been highly developedmuch earlier in the East and in South Asia. The development of signs suitablefor the graphic symbolization of sounds was an important part of these attempts.In 1867, Alexander Melville Bell published a system he termed Visible Speechconsisting of the Universal Alphabet, a set of specially designed symbols ofplaces of articulation, manners of articulation and phonation (gure 2.9). Sincethese symbols were not readily available in type, Bell provided a grid of romanletters and numbers to express the mechanism of articulation in common type.His portrayal of vocal action was intended to be so detailed as to include not onlyspeech sounds but also paralinguistic sounds such as whispers, sobs and hisses,and even coughs and grunts. The Visible Speech letters are to be understood asgraphic symbols as indicated in the diagrams, for consonants at the left side andvowels at the right side. For example, the consonant diagram shows three vocalcord positions, a bar for voicing, a circle for wide opening, and an X for closedglottis.

Bells Visible Speech is immediately recognizable as a notation for specic pur-poses. Another approach was pursued by the International Phonetic Association,which was founded in 1886 in Paris. Following a suggestion by Otto Jespersen,the society devoted itself to devising an international phonetic alphabet (IPA) ap-plicable to all languages, a rst version of which was ready by 1888. The IPA wasdesigned to meet practical needs of linguists and language teachers furnishinglearners of foreign languages with phonetic transcriptions to assist them in ac-quiring the pronunciation, and working out romanic orthographies for languageswritten in other systems or for languages hitherto unwritten (International Pho-netic Association 1949: 1). The IPA presupposes that sounds can be counted andtries to approximate a bi-unique mapping relation of one sound, one letter, al-though the Principles of the International Phonetic Association warns that vowel

The ideal of faithful transcription 29

Figure 2.8 John Wilkins physiological alphabet symbols

30 The basic options: meaning and sound

Figure 2.9 Visible Speech, Alexander Melville Bells Universal Alphabet

symbols are necessarily elastic in their values. It denes phonemes as families ofrelated sounds and admits that it is not possible to design letters for the represen-tation of all distinguishable shades of sound (International Phonetic Association1949: 4). As for graphic design, the IPA was deliberately based on the Latin alpha-bet, and additional letters almost twice the number of its classical version werefashioned to harmonize well with it. This has the obvious advantage of facilitatingthe learning and use of the IPA. But its closeness to the Latin alphabet also hasthe incalculable disadvantage of fostering a conceptual confusion of writing andtranscription. Two faulty and misleading conjectures suggested by this confusionare (1) that the Latin alphabet and its Greek precursors were transcription systemsand that whatever discrepancies between sounds and letters found in Greek andLatin texts are simply a result of sound change over time. (2) The other is that theLatin alphabet is a writing system whose structural characteristics are determinedby the set of roman letters and that, therefore, all systems that make use of theseletters belong to the same type.

The ideal of faithful transcription 31

As for (1), it is obvious that the Greek and Latin alphabets were meant to beinterpreted in terms of Greek and Latin speech sounds, but this does not mean thatGreek and Latin writing was transcription. Writing and transcription are function-ally very different. Writing is for readers who have little need for minute phoneticinformation because they know the language that is written and, therefore, do notdepend on such information for identifying meaningful units in the text. Rather,they are better served by a system that lters out unnecessary phonetic informationand even omits phonological information for the sake of morphology and gram-mar. It is in this sense that a writing system is a grammar a description of alanguage (Scholes and Willis 1991: 230). Writing systems are conventionalizedtechniques of segmenting linguistic utterances in such a way that the resulting unitscan be interpreted as linguistic constructs such as words, morphemes, syllables,phonemes, as well as higher-level units such as clauses and sentences. In contrast,transcription, ideally, focusses on sound alone disregarding grammar. Transcrip-tion is a scientic procedure based on the insights of phonetics and phonology,which, in contradistinction to conventional orthographies, does not assume thatthe reader knows the language. While orthographies provide information aboutgrammar and meaning by means of word spacing, capitalization, hyphenation,homophone differentiation and so on, it relies on phonetic information alone. Agood transcription provides a graphic model that can be interpreted phoneticallyfairly accurately even without meaningful understanding. To the foreign languagelearner phonetic transcription is a supplement to written text, which tends to re-quire more than supercial knowledge of the language in question to be given acorrect phonetic interpretation. A written text, then, is functionally and structurallysomething completely different from a piece of transcribed speech. For this reasonalone (and there are several others) it is wrong to regard the IPA as the ultimateperfection of the Latin alphabet. It is letters of the Latin alphabet put to new anddistinctive use.

(2) Turning now to the second misunderstanding invited by the IPA, it mustbe emphasized that a set of symbols does not determine the nature of the writingsystem in which they are put to use. Consider, for an extreme and hence clearexample, the Cherokee script (Table 4.2). The majority of its symbols look likeLatin or Greek letters, the rest like modications thereof. But the graphic similarityis deceptive. The symbols of the Cherokee script are interpreted as syllables quiteunrelated to the range of sound values usually associated with alphabetic letters.Cherokee, for example, is read as /go/ and as /tso/. While the relation-ship between letters and their phonetic interpretations is different in the variouswriting systems that are historically derived from Latin, the main point at issue hereis the same. The set of elementary graphic symbols must be carefully distinguishedboth from the interpretations they are given in isolation and from the operational

32 The basic options: meaning and sound

rules governing their use. To put it differently, the Latin alphabet is not a writingsystem in a generic sense. The rules governing its use in Latin and in Englishand other languages are drastically different, so much so that Latin and Englishwriting should not be treated as writing systems of the same type. Hence, the ex-pression Latin alphabet is ambiguous. The two meanings at issue here can beparaphrased as

the writing system of the Latin language; and a set of 26 letters serving the writing systems of a great number of languages.

In the latter sense it is also referred to as Roman or roman. The spelling witha small initial r is indicative of the general signicance of this script which isno longer associated with a particular language or culture. Cherokee, Englishand the IPA make use of the same symbols, roman letters, but not only are theirinterpretations disparate, the functions they serve are unrelated. This differencetends to be overshadowed by the fact that the IPA looks like an extension of theLatin alphabet and thus like just another writing system, which it is not.

The confusions just discussed, it should be made clear, are not the fault of theIPA per se or the International Phonetic Association. They are simply suggestedby graphic similarity. This is what happens when technical terms are couched ineveryday language. The IPA is a technical instrument, but it resembles our ABC.Moreover, as a technical instrument it is subject to deliberate improvement andunnoticed change. The IPA is the most widely used system of phonetic transcrip-tion, but this should not be taken to mean, as unfortunately it sometimes is, thatit incorporates the ultimate solution to the problem of rendering a continuousstream of sound into a discrete array of graphic signs. There is no ultimate so-lution. Since its inauguration the International Phonetic Association has changedits position on many topics and revised the IPA several times. Pullum and Ladus-aws (1986: xvii) observation that the tacit understandings about transcriptionrepresent not a rm common ground but one that shifts over time like any othercultural system alerts us to the important point that transcription, as long as itmakes use of discrete symbols,2 is not a purely mechanical procedure. The trans-guration of audible into visible signs requires interpretation, which means thatit is susceptible to cultural inuence. It is bound to be imperfect because themedial difference between sight and sound stands in the way of isomorphic cor-respondence. It is not just that succession in time is not the same as extension2 These need not be letters. Notice that a transcription based on an exact description of sounds can be

quite complex. What in IPA is represented as [n] corresponds to the following expression in a notationdeveloped by Otto Jespersen: ,,0f,,23. Clearly, this expression is not to be interpreted as asequence of features to be realized one after the other. For details on phonetic notation systems seeMacMahon 1996.

Real writing 33

in space, but a sequence of graphic symbols with a linear expansion, suggestingas it does temporal duration, may actually correspond to articulatory events thatoccur simultaneously rather than one after another. That transcription is possi-ble at all is thanks, on one hand, to the fact that speech production is groundedin the physiology of articulation organs, and, on the other, to the exploitation ofarticulatory sound modulation and the formation of classes of sounds for com-munication. Transcription systematically maps onto these two levels. Very fewwriting systems do, although many writing systems are in a broad sense phono-graphic. But this is not a deciency. Many important features of speech are ignoredin writing, and vice versa. Many writing conventions are unre