Roy Harris-Rethinking Writing-Athlone (2000)

RETHINKING WRITING
This page intentionally left blank

Rethinking Writing
ROY HARRIS
Ai continuum
• • W L O N D O N • NEWYORK
First published in 2000 by
The Athlone Press
Continuum
The Tower Building,
11 York Road, London SE1 7NX
370 Lexington Avenue, New York, NY10017-6503
Reprinted in paperback 2001 by Continuum
© Roy Harris 2000
All rights reserved. No part of this publication may be reproduced

or transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording or any information storage or retrieval
system, without permission in writing from the publishers.
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available
from the British Library.
ISBN 0-485-11547-6 (hardback)

0-8264-5798-3 (paperback)
Typeset by RefmeCatch Limited, Bungay, Suffolk

Printed in Great Britain by Bookcraft, Midsomer Norton.
Contents
Preface vii
a
Foreword: Writing and Civilization 1
1 Aristotle's Abecedary 17
2 Structuralism in the Scriptorium 39
3 Writing off the Page 64
4 Notes on Notation 91
5 Alphabetical Disorder 121
6 Ideographic Hallucinations 138
7 On the Dotted Line 161
8 Beyond the Linguistic Pale 184
9 Mightier than the Word 215
Bibliography 243
Index 251
Preface
A profitable Invention for continuing the memory of time

past, and the conjunction of mankind, dispersed into so
many and distant regions of the Earth; and with all difficult,
as proceeding from a watchfull observation of the divers
motions of the Tongue, Palat, Lips, and other organs of
Speech; whereby to make as many differences of characters,
to remember them.
Thomas Hobbes
Hobbes's matter-of-fact description summarizes centuries of tradi-

tional wisdom concerning writing. More specifically, the kind of
writing in question was that which has formed the basis of Euro-
pean education since Graeco-Roman antiquity: that is, alphabetic
writing. Hobbes does not mention writing of any other kind. That
alone bears witness to the tyrannical hold that the alphabet has
exercised over Western thinking on the subject. An alternative title
for this book might have been: The Tyranny of the Alphabet and How
to Escape from it.
Escaping from it has never been easy for anyone educated
from childhood onwards in European schools. For the traditional
wisdom about alphabetic writing is deeply embedded in ele-
mentary pedagogic practice. This, being based on teaching the
alphabet in one or other of its several European versions, varied
viii Preface
little over time. It has varied little since Hobbes's day either.
Many of us were taught our 'ABC' by methods that are not far
removed from those described by Quintilian in the first century
AD.1
The traditional wisdom is based on certain assumptions about
the relationship between speech and writing. Throughout the
Western tradition, original contributions to analytic reflection on
those assumptions have been few and far between. Two of the
most important, one ancient and one modern, will be considered
in some detail in this book. They form an essential background to
any attempt at rethinking writing today.
Why should there be any need to rethink writing at all? Anyone
as satisfied as Hobbes was with the traditional wisdom will pre-
sumably see no reason whatever. But the more closely the tradi-
tional account is examined, the clearer it becomes that there are
basic questions that have been left unanswered or dodged. So
trying to answer those questions is one reason for a new attempt to
rethink writing. Another reason is that modern technology makes
available resources for reading and writing that have never been
available before. But perhaps a more important reason than either
of these is that how people think about writing is bound up in
many subtle ways with how they think about their fellow human
beings. There is more to rethinking writing than just looking for a
more satisfactory intellectual analysis of particular communi-
cational practices.
Before developing this point further, a fundamental point about
alphabetic writing itself needs to be made clear. Alphabetic writ-
ing is one form of language-based, or 'glottic' writing. Reading
this sentence aloud would be a trivially easy task for millions of
people; but impossible for anyone - even if literate - who knew no
English. Similarly, anyone who knows no Thai will be hard put to
it to read a Thai newspaper (either aloud or silently), even though
Thai writing is also alphabetic. These disparities are diagnostic
indicators of glottic writing. However, the alphabet as such is not a
1
Institutio Oratoria, I.i.2 Iff.
Preface ix
glottic system. Furthermore, glottic writing is not the only kind of

writing human ingenuity has devised. Musical and mathematical
notation do not require any knowledge of English (or Thai); but
they do require an understanding of the principles on which the
notation is based. These principles, however, are quite different
from those of glottic writing. Which is why you do not read a
musical score or a mathematical table in the same way as you read
this page (or any other page of glottic writing in any other
language). Forms of non-glottic writing will occasionally be
mentioned in this book, but only for purposes of comparison with
glottic writing.2
Those of us who have learnt to read and write on the basis of
using at least one glottic system are usually counted 'literate'. This
is not primarily a book about literacy (nowadays a highly contro-
versial concept) but about the concept of glottic writing under-
lying it. So anyone looking for literacy statistics or discussion of
teaching methods will be disappointed. Unless, like the author,
they think that arguing about literacy is likely to be unproductive
until glottic writing itself is better understood.
Literacy does nevertheless come into consideration, since it
affects ways of thinking about writing.3 In other words, how liter-
ate people view writing is often coloured by their opinions con-
cerning literacy and their own status as literate members of the
human race. So much so that the very suggestion that there might
be any need to rethink writing may seem almost heretical in such
an 'advanced' literate society as ours is usually deemed to be.
That kind of complacency is itself one symptom of the kind of
2
For more on non-glottic writing in general, see R. Harris, Signs of
Writing, London, Routledge, 1995. Like all distinctions in this area, that
between glottic and non-glottic writing can be problematic. The ques-
tion of the limits of glottic writing comes up for discussion in the final
chapter.
0
It would be tedious to keep repeating 'glottic'. So from this point on the
reader should take it that the term writing refers to glottic writing unless
there are contextual indications to the contrary.
x Preface
difficulty involved in trying to reconsider the question of writing

objectively.
In a society which invests so much in literacy it is reasonable to
expect that the topic of writing might be considered of some
importance. So it is. But chiefly because parents are nowadays
concerned that their children are not being trained adequately in
the relevant skills. Teachers are concerned because they are
blamed for this. Governments are concerned because they are
urged to improve matters and worry about their electoral popular-
ity and their education budgets. There, however, public concern
with writing often ends. Wider issues are rarely raised, and it is
rarer still for anyone to ask whether the traditional assumptions
commonly taken for granted about writing are sound. These
assumptions are among those that have to come under scrutiny in
any serious attempt at rethinking writing today.
People commonly measure the level of literacy in a society by
estimating what proportion of the population can read and/or
write. Exactly how that should be measured is itself controversial.
But there is a quite different way of looking at the question. What
people think writing is must be counted a no less important aspect
of literacy than counting how many of them have mastered the
practice. That kind of thinking is the aspect of literacy that this
book addresses.
As regards understanding what writing is, societies may be seen
as going through three stages from the point at which writing is
recognized as something different from drawing pictures and
other ways of marking surfaces. The first of these stages, which
may be called for convenience 'crypto-literacy', is one in which
everything surrounding writing is still regarded as a form of magic
or secret knowledge, reinforced by various superstitions and shib-
boleths. These include myths about the origin of writing (such as
its invention by gods or animals).4 The second stage - with which
this book is mainly concerned — might be called 'utilitarian lit-
eracy'. It begins when writing is no longer regarded with mystical
4
R. Harris, The Origin of Writing, London, Duckworth, 1986, Ch.l.
Prefacea xi
awe, but as a practical tool or technique for doing what would

otherwise have to be done by means of speech, or left undone.
Hobbes's statement about alphabetic writing is a typical product
of a society at the stage of utilitarian literacy. The third stage, 'full
literacy', which arguably no society has yet quite reached, is one in
which writing is no longer regarded just as a 'profitable Invention
for continuing the memory of time past, and the conjunction of
mankind', but as a particular mode of operation of the human
mind and the key to a new concept of language.
Chronologically, these three stages are by no means well sepa-
rated. The first clear indications that writing is regarded as some-
thing more than a utilitarian technology, useful to bureaucrat,
teacher and poet alike, begin to emerge in the nineteenth century.
On the other hand, long after typically utilitarian improvements
like the invention of printing, European scholars were still discuss-
ing whether the alphabet their printers were using had been
designed by God - an idea which clearly belongs to the stage of
crypto-literacy. Nowadays much nonsense about runes is popular
with adherents to some kinds of'New Age' lifestyle. Survivals and
revivals of primitive beliefs, however, do not alter the general
picture. The kind of literacy that has predominated throughout
the Western tradition, and still does, is utilitarian literacy. In the
Western perspective alphabetic writing is regarded as the nee plus
ultra among systems of writing because, as Giovanni Lussu has
recently put it, we have a conceptual model in which, implicitly or
explicitly, writing is regarded as 'nothing more than an ingenious
technical device for representing spoken language, the latter being
the primary vehicle of human communication'.5 It is this con-
ceptual model that needs rethinking.
For as long as that model prevails, writing will be valued chiefly
for the ways it offers of replacing speech in dealing with and
disseminating the kinds of information that are regarded as
important. It is essentially a surrogational model; that is, writing is
viewed as a surrogate or substitute for speech. The particular
5
G. Lussu, La lettera uccide, Viterbo, Nuovi Equilibri, 1999, p. 11.
xii Preface
utility of the substitution is seen as being that writing can transmit

a verbal message in circumstances where vocal communication
would be difficult or impossible. In this sense the written surrogate
provides a convenient extension of the spoken word.
The old notion that writing is 'but a substitute for the art of
speaking'6 goes hand in hand with the idea that the basic function
of the marks used in writing is phonoptic: i.e. serving to make
sound 'visible'.7 For Voltaire, writing was a 'portrayal of the voice'
(lapeinture de la voix). This seemed so obvious in the mid-nineteenth
century to R.C. Trench that he declared it to be an 'unquestion-
able fact that the written word was intended to picture to the eye
what the spoken word sounded in the ear.'8 The phonoptic thesis
is sometime summed up even more succinctly in the phrase 'vis-
ible speech'.9 This curious oxymoron still survives as a reminder
of the assumptions underlying the surrogational perspective on
writing.
The comparatively recent introduction of communication by
telephone, radio and television, together with various electronic
forms of recording, have to some extent compensated for
the 'disadvantages' of speech as compared with its phonoptic
6
Joseph Priestley, A Course of Lectures on the Theory of Language and Universal
Grammar, Warrington, 1762, p.22.
7
Optophonic devices convert light into sound, rendering what was
visible audible: phonoptic devices do the reverse.
8
R.C. Trench, English Past and Present, London, Parker, 1855, Lecture V
(Everyman edition, London, Dent, 1927, pp. 135-6.)
9
The term Visible speech' was applied by Alexander Melville Bell to his
own system of phonetic writing (1867), which he regarded as 'self-
interpreting physiological letters, for the writing of all languages in one
alphabet' - a kind of apotheosis of glottic writing. It was later the title of
a well-known book on sound spectrography (R.K. Potter, G.A. Kopp and
H.C. Green, Visible Speech, New York, Van Nostrand, 1947). More
recently, it has been reapplied to traditional writing systems (John
DeFrancis, Visible Speech. The Diverse Oneness of Writing Systems, Honolulu,
University of Hawaii Press, 1989). But the notion that writing makes
speech visible goes back a long way in the Western tradition.
Preface xiii
counterpart. But that has not as yet undermined the standing or

the indispensability of writing as a utilitarian device, in spite of
alarmist latter-day prophecies of its impending demise.
The kind of thinking which seems characteristic of utilitarian
literacy belongs to a stage in cultural evolution where those who
have learnt to read and write are still congratulating themselves on
not being 'illiterate'. It is not knowing how to substitute writing for
speech and, conversely, substitute speech for writing which, for
them, epitomizes the state of ignorance they have just left behind.
Thus, for all their sophisticated acquaintance with books, docu-
ments and records of various kinds, their reflection on such topics
is still backward-looking. They think of literacy as an unnatural
but beneficial state of affairs which has quite late in human history
(within the last few thousand years) begun to supersede the natural
orality of the human race. In short, writing is viewed as an arti-
ficial superimposition on speech, which still provides its founda-
tion and to which it can still be held accountable. (Hence, for
example, periodic outbursts of zeal for 'spelling reform', in which
the reformers mainly want to persuade us to do things like write
tough as ft^'because that is how it's pronounced'.)
A society can remain in this stage for a very long time. People
can fail to understand how writing has entirely reshaped their
understanding of the world, because they still hark back to the old
assumption that writing is just a useful alternative to or, at best,
improvement on vocal communication. That is the stage we are
still in today, and will remain in for as long as we continue to pat
ourselves on the back for progress in eradicating 'illiteracy', regard
reading the right books as the basis of all education, and postpone
the rethinking of writing that is long overdue.
Anyone who can both read this sentence aloud and copy it out in a
notebook realizes that writing and speech are quite different bio-
mechanical activities. One involves the hand and the other
involves the mouth. Hand and mouth engage quite different sen-
sory motor programmes. These can be independently studied as
xiv Preface
such by physiologists and psychologists. Reading aloud and read-

ing silently are also biomechanically different. It would be possible
to read (and understand) this sentence silently without having any
idea of how to read it aloud. It might even be possible to read it
aloud without understanding what it meant. But if a society is to
progress beyond utilitarian literacy to full literacy, writing and
reading must be viewed not just as biomechanical activities but as
biomechanical activities involving signs. In order to distinguish
this kind of thinking about writing from the various other ways of
regarding it, the term used in this book will be 'semiological'.10 A
theory of writing without a theory of signs is like a theory of
eclipses without a theory of the solar system.
The first important thinkers in the Western tradition to think
about writing semiologically were Plato and Aristotle. Aristotle
was the first to describe writing explicitly as requiring 'symbols'
(symbold). His view of the relationship between speech and writing
- or simplified versions of his view - became part and parcel of
received scholarly opinion on the subject. It was not challenged in
any significant way until the twentieth century. The challenge
came from one of the founders of the intellectual movement sub-
sequently known as 'structuralism': Ferdinand de Saussure. It was
also Saussure who introduced the term semiology (semiologie] into the
vocabulary of modern cultural studies. This explains why there is
so much discussion of Saussure in Rethinking Writing. Saussure tried
to rethink writing at a basic level for the first time since Aristotle
and put it into a wider semiological framework. But at the same
time he tried to provide a better theoretical foundation for essen-
tially the same view of the alphabet that Aristotle had taken.
Whether Saussure succeeded in what he attempted to do is open
10
To those who find the term semiological rebarbative, the author offers no
apology, since he did not invent it. Those who find it pretentious into the
bargain are invited to stay the course and then judge for themselves
whether what was presented as 'semiological' was, in this instance,
pretentious or not. Those who wonder why semiological was chosen in
preference to semiotic will find that question dealt with later.
Prefacea xv
to question. If he did not, it seems important to understand his

failure. His work is in any case the starting point for subsequent
theorists. Whichever way we look at it, Saussure's thinking about
writing is pivotal and commands attention from anyone interested
in the subject.
Those who are content with the traditional wisdom about writing
may well ask what is the point of questioning it, unless so doing
can be shown to lead forward to practical pedagogic or remedial
advances of some kind. These are demands characteristic of utili-
tarian literacy. But this book does not promise any dramatic
improvement in students' spelling or new insights into the treat-
ment of dyslexia. Nevertheless, the Foreword tries to indicate why
we should not be happy just to go along with the traditional utili-
tarian wisdom for ever and a day. Perhaps the most immediate
worries have to do with the fact that the traditional wisdom is
nowadays being revamped as a 'scientific' basis for all kinds of
projects: and that it certainly is not. Unfortunately, it is often
invoked also to justify linguistic policies and attitudes that should
be considered unacceptable in a society of the twenty-first
century.
The reader does not have to share the author's worries on this
score in order to follow the argument. For the argument is based
independently on two strategies. One is pointing out the internal
contradictions in earlier views (as judged by their own criteria).
The other is pointing out that an alternative approach to writing is
available which does not fall into these contradictions. Anyone
prepared both to examine the contradictions and look for avail-
able alternatives is already committed to rethinking the assump-
tions underlying traditional utilitarian literacy. That is exactly the
kind of rethinking the author had in mind.
Most of the chapters began life as lectures given at the Sorbonne

in the academic year 1991—2. An edited version of those lectures
xvi Preface
was published in Paris by the Centre National de la Recherche

Scientifique in 1993 under the title La Semiologie de I'ecriture. Since
then I have frequently been told that it was a pity no English
version had been made available. However, there was no English
version, since the original lectures were in French. Furthermore,
when I came to look at the possibility of translating them, I was
struck by how odd certain aspects of them might seem to an
English readership. I also felt unhappy with the way certain points
had been dealt with. So Rethinking Writing is not a translation of La
Semiologie de I'ecriture, but an updated re-writing in English of some
of the same topics, often illustrated with different examples. It
incorporates both condensations and expansions of the original
material, as well as certain changes of emphasis.
I should like here to reiterate my thanks to Marine Valette,
Colette Sirat and Rita Harris, all of whom gave invaluable assist-
ance at various stages with the preparation of the French publica-
tion, to Sylvain Auroux, to M. Marcassou at CNRS Editions, and
to my former colleagues at the Ecole Pratique des Hautes Etudes,
where I was privileged to spend the year during which these ideas
first took shape. For the development of my ideas on integration-
ism I am indebted, as usual, to Daniel Davis, Hayley Davis, Chris
Hutton, Nigel Love, Talbot Taylor, Michael Toolan and George
Wolf. Rita Harris read the whole of the present book in draft form
and made many excellent suggestions.
R.H.
Oxford, June 1999
FOREWORD
Writing and Civilization
When Boswell objected to Johnson's calling the Chinese 'barbar-

ians', Johnson replied curtly: 'Sir, they have not an alphabet.' Eric
Havelock reminds us of this anecdote in his original and chal-
lenging book on literacy in Classical antiquity.' The reminder is a
timely one, since the study of writing has been beset in the past -
and doubtless still is - by many misconceptions and prejudices.
None of these is more pernicious than the persistent belief that
writing, and particular forms of writing, are somehow to be taken
as diagnostic indicators of levels of civilization or mental progress
among the peoples of the world.
Johnson's reply to Boswell is, on the face of it, puzzling. One
might have been less worried if he had said: 'Sir, they have
no democratic government.' Or even: 'Sir, they have no knives
and forks.' But why should a particular method of writing be
taken as marking a dividing line between civilization and
barbarism?
Johnson was not alone in the eighteenth century in holding
such a view. Rousseau distinguishes three general types of writing
in his Essai sur I'origine des langues. The first of these types involves
the depiction of objects, the second the use of conventional
1
E.A. Havelock, The Literate Revolution in Greece and Its Cultural Consequences,
Princeton NJ., Princeton University Press, 1982, p.2
2 Rethinking Writing
characters to stand for words and sentences, while the third is

alphabetic writing.
These three ways of writing correspond almost exactly to three
different stages according to which one can consider men gath-
ered into a nation. The depicting of objects is appropriate to a
savage people; signs of words and propositions to a barbaric
people, and the alphabet to civilized peoples [peuplespolices].2
Rousseau's example of the writing of a barbaric people is Chinese.

But although he is rather more explicit than Johnson in his corre-
lations between forms of writing and forms of society, it is still far
from clear how these correlations are established. Or why the
order of superiority should be as Rousseau evidently takes it to be.
But it may be worth pointing out that by Rousseau's day the
French term for a person unable to read or write was already
analphabete: which by implication equates literacy with mastery of
alphabetic writing.
Although it is not explicitly invoked either by Johnson or by
Rousseau, there was a religious controversy lurking behind atti-
tudes to the alphabet in the eighteenth century. Many people held
there was Biblical authority for believing that the alphabet had
been invented by God. Joseph Priestley, who rejected this view,
nevertheless acknowledged that:
The transition from speaking to writing is so far from being
thought easy and natural by many persons, that some of the
greatest men this nation ever produced [ . . . ] have had
recourse to supernatural interposition to account for it, and
suppose that the first alphabetical writings were the two tables
of stone, which were written by the finger of God himself. And
it must be acknowledged, that the oldest accounts we have con-
cerning the use of letters in Asia and Greece is so circumstanced,
as by no means to clash with this hypothesis.3
2
J-J. Rousseau, Essay on the Origin of Languages, trans. J.H. Moran,
Chicago, University of Chicago Press, 1966, p. 17.
3
Priestley, op. cit., pp.23-4.
Writing and Civilization 3
Priestley himself came to the conclusion that 'the imperfection

of all alphabets, the Hebrew by no means excepted, seems to
argue them not to have been the product of divine skill . . .'4 But
he also thought that alphabetic writing, for all its imperfections,
was superior to that of the Chinese:
we are told that it is, in fact, the business of half the life of a
Chinese, philosopher to learn barely to read a sufficient variety of
books in their language: and the difficulty of inventing, and
establishing the use of new characters (without which they
could have no way of expressing new ideas) must itself prevent
the growth of arts and sciences in that nation.0
This kind of a priori reasoning about writing systems is by no

means extinct even today.
Underlying the notion that links the invention of writing to the

dawn of civilization we can detect at least two supporting theses.
Both are plausible, widely held and misleading. One thesis is that
the practice of writing requires of the practitioner feats of mental
discipline and agility that are beyond the powers of immature
minds. Thus the observation that children commonly learn to
speak before they learn to write feeds the equation between the
pre-literate community and social immaturity. The other thesis is
that unless individuals can express themselves in writing and
enrich their own minds by reading, their lives and personalities
will never reach their full potential. Learning to read and write is
seen as an achievement which makes it possible in principle to
overcome what would otherwise be inescapable limitations of the
human condition, and even to overcome human mortality itself.
As Samuel Butler put it:
The spoken symbol is formed by means of various organs in
or about the mouth, appeals to the ear, not the eye, perishes
4
Priestley, op. cit., p.29.
5
Priestley, op. cit., pp.35^6.
instantly without material trace, and if it lives at all does so only

in the minds of those who heard it. The range of its action is no
wider than that within which a voice can be heard; and every
time a fresh impression is wanted the type must be set up anew.
The written symbol extends infinitely, as regards time and
space, the range within which one mind can communicate with
another; it gives the writer's mind a life limited by the duration
of ink, paper, and readers, as against that of his flesh and blood
body.6
This in turn leads to the conclusion than when the bonds of

communicational contact are not merely between one living indi-
vidual and another but - through writing - between present, past
and future generations, the result is a superior form of social
entity.
This seems to have been the conclusion generally accepted by
anthropologists of Butler's day, who draw the line between 'primi-
tive' and 'civilized' peoples by using the presence or absence of
writing as a yardstick. Edward Burnett Tylor, who became
Oxford's first Professor of Anthropology in 1895, wrote:
The invention of writing was the great movement by which
mankind rose from barbarism to civilization. How vast its effect
was may be best measured by looking at the low condition of
tribes still living without it, dependent on memory for their
traditions and rules of life, and unable to amass knowledge as
we do by keeping records of events, and storing up new obser-
vations for the use of future generations Thus it is no doubt
right to draw the line between barbarian and civilized where
the art of writing comes in, for this gives permanence to history,
law and science.7
By this criterion, the Chinese are at last upgraded from barbar-
6
S. Butler, Essays on Life and Science., ed. R.A. Streatfield, London, Fifield,
1908, p. 198.
7
E.B. Tylor, Anthropology, London, Macmillan, 1881, Ch.8.
ism to civilization. (Tylor magnanimously attributes their failure to

develop an alphabet to the proliferation of homonyms in spoken
Chinese.) And Tylor at least gives a reason for looking to writing as
a measure of human progress. But it is a reason which puts the
emphasis on the consequences of writing, rather than on writing
itself. And this in a way makes the rationale even more puzzling. It
is rather like taking the invention of the steam engine as a defining
moment in modern history. (A case can be made out for the steam
engine, to be sure: but it was what the steam engine made possible
rather than the invention itself which, retrospectively, makes it
seem of importance. As for actually operating a steam engine for
some routine task, it may well be within the competence of a
nincompoop.)
The notion that human beings did not become civilized until
they became literate was perpetuated by historians of writing for
generations after Tylor's book appeared. In 1949 a respected
authority on the alphabet stated:
In the growth of the spiritual human advance, that is of civiliza-
tion, the origin and the development of writing hold a place of
supreme importance, second only to that of the beginnings of
speech, as an essential means of communication within human
society.8
This claim appeals to a different rationale from Tylor's. But it

raises even more serious problems. For the fact is that this
allegedly 'essential means of communication' was for many cen-
turies in the hands of a fairly small elite. Thus it can be argued
that for a long time writing did not contribute a great deal to
communication 'within human society'. It was 'essential' only for
those to whom it brought power and prestige. The social function
of writing was not like the function of speech: it did not unite the
community, but divided it.
That division, it might be said, is intrinsic to the Western
concept of literacy in at least the following respect:
8
D. Diringer, The Alphabet, 2nd ed., London, Hutchinson, 1949, p. 19.
The key factor in the formation of a specialist literate class in

the West has been the difference that existed since the begin-
ning of the alphabetic revolution in ancient Greece between
the language of the texts used to teach literacy and the lan-
guage of the people who were to learn it. This difference, which
could only increase with time and literary conservatism, created
the need for a class of interpreters of the language preserved in
the texts - the literary language. It is important to realize that
the expertise of this class does not lie simply in being literate.
Nor does it lie simply in its linguistic mastery of the language or
dialect of the texts used to teach literacy. These are, of course,
necessary attributes but not sufficient. The defining feature of
the class is its expertise in constructing justifications on the
basis of the texts alone for the linguistic features that the texts
happen to manifest.9
Put more simply, the point is that it would be naive to construe

literacy as a mere matter of knowing how to perform some
mechanical but complicated task(s) with writing materials and
their products. Literacy involves the creation of a social status for
initiates, and the acceptance of a scale of values over and above
those which the tasks themselves entail. And this has generally
required inter alia a revision of ideas about language. As the same
author observes:
most of those who wanted to become literate had to assimilate
the conceptions of this [sc. literate] class as to what linguistic
competence involved.10
The views we encounter in histories of writing to this day still

reflect the views of the literate class about its own superior status.
Thus one authority assures us: 'Writing exists only in a civilization
and a civilization cannot exist without writing.' This is little more
9
Bennison Gray, 'Language as knowledge: the concept of style', Forum
Linguisticum, Vol.3 No.l, 1978, p.30.
10
Gray, op. cit., p.30.
than a generalization from the same scholar's view of the position

of the illiterate individual in contemporary society. He even
extends this without hesitation to comparative assessment of
backward and more advanced nations in the modern world:
Nowadays an illiterate person cannot expect to participate suc-
cessfully in human progress, and what is true of individuals is
also true of any group of individuals, social strata or ethnic
units. This is most apparent in Europe, where nations without
any noticeable percentage of illiterates, like the Scandinavians,
lead other nations in cultural achievements, while those with a
large proportion of illiterates, like some of the Balkan nations,
lag in many respects behind their more literate neighbours.''
What is questionable here is not simply how 'advanced' the

Scandinavian countries are by comparison with other parts of
Europe. There is a circularity involved in appealing to notions like
'human progress' and 'cultural achievement' in an argument
addressed to readers whose own education has already taught
them to identify literacy as the indispensable basis of progress and
culture.
One could go on adding to the list of dubious reasons that

scholars have advanced in justification of their own treatment of
writing (or particular forms thereof) as landmarks in the history
of humanity. Some have even divided this history into three
great phases: (i) pre-literate, (ii) literate, and - yet to come - (iii)
post-literate.12
Anticipation of phase (iii) is recorded as early as 1895, when
H.G. Wells published one of the original masterpieces of science
fiction. In The Time Machine, Wells's narrator describes a society of
11
I. J. Gelb, A Study of Writing, 2nd ed., Chicago, University of Chicago
Press, 1963, p.222.
12
Ch. Higounet, L'Ecriture, 7th ed., Paris, Presses Universitaires de
France, 1986, pp.3-4.
the future in which writing has long been forgotten. In this society,
orality has regained the monopoly of verbal communication.
Books are museum exhibits, arid even the museums are in ruins.
The passage quoted below tells of how the time-traveller discovers
that he is now in the post-writing age. He is exploring what
remains of a large palace, accompanied by Weena, a young
woman of the aristocratic Eloi race, who speak a language of
which he has picked up a few words.
The material of the Palace proved on examination to be indeed
porcelain, and along the face of it I saw an inscription in some
unknown character. I thought, rather foolishly, that Weena
might help me to interpret this, but I only learned that the bare
idea of writing had never entered her head.13
This is one of the great missed opportunities in English litera-

ture. Wells does not tell us whether what Weena failed to under-
stand was the semiological status of the inscription per se (i.e.
whether it was a sign at all) or whether she misconstrued it as a
sign with some different kind of meaning (e.g. as a drawing, as a
mark of ownership, etc.). The 'bare idea of writing' is not quite as
bare - or innocent - as that casual phrase seems to suggest.
What the 'bare idea of writing' is - or might be - is a question
central to this book. It is not a popular topic, because there are so
many vested interests concerned to promote ideas of writing that
are far from bare. And it might be argued that if there ever was a
'bare idea' of writing people have long since lost sight of it. Never-
theless, the intellectual exercise involved in trying to reconstruct
that idea might well be worth the pursuing
Comment is called for at this point on one of the 'less-than-bare'

notions of writing dear to some scholars in the humanities. It
impinges on our present concerns because what is assumed is that
the worth of a civilization is ultimately to be judged by its 'litera-
13
H.G. Wells, The Time Machine, London, 1895, Ch. 8.
ture'. Hence the curious restriction of the term writing and the
verb write to apply to the production of literary compositions. As
when Roland Barthes, in a seminar at the Ecole Pratique des
Hautes Etudes, delivered the oracular pronouncement:
On the one hand we have what it is possible to write, and on the
other what it is no longer possible to write.
This is one of those truisms of such striking banality that they

sound oddly like the result of profound and original thought. For
Barthes this truism applied to one culturally privileged activity:
that of the writer qua author. It is the sense of the verb write in
which the answer to teacher's question 'What did Balzac write?' is
expected to be: 'Balzac wrote La Comedie Humaine.'' And doubtless
everyone agrees that it is no longer possible to write the novels of
Balzac. They are already written and Balzac is dead. What he
wrote can be republished, imitated, or adapted; but it is no longer
possible to write it — that is, write it in the sense in which writing is a
creative activity and the product is a literary work. It is in this
sense that modern universities offer courses in so-called 'creative
writing': the students would be surprised and perhaps alarmed if
offered instruction in innovative calligraphy. They want to be
poets, novelists, playwrights.
Other interpretations of Barthes's equivocal dictum can cer-
tainly be proposed; for example, the point might be there are
subjects which a writer can tackle today, but others that it is no
longer possible to write about; or there are styles of writing that
are nowadays acceptable, but others that are quite outmoded.
Straight away the proposition become less banal. But in all these
cases the activity of the writer is conceived as that of literary
composition. When Barthes invented the notion of a 'degree
zero' in writing he was thinking not of the illiterate's struggle to
master a script but of a certain neutrality of literary expression.14
His categories of the modes of writing were bourgeois writing,
Marxist writing, poetical writing, etc.
14
R. Barthes, Le degre zero de I'ecriture, Paris, Seuil, 1953.
Similarly when Ezra Pound, many years before Barthes, pro-

claimed that 'writers as such have a definite social function exactly
proportioned to their ability as writers' he did not have in mind the
ability to hold a pen and form legible characters, or to display a
versatile competence in the many writing systems of the world.15
A parallel restriction applies to the term reading. When I.A.
Richards published a book entitled How to Read a Page, he did not
begin with the question of which way up to hold it.16 His advice
never addresses the problem posed by the fact that different scripts
and different formats require different scanning procedures if we
want to make sense of them. Nor is reading a newspaper treated
differently from reading a legal contract. 'Reading', for Richards,
begins with the assumption that the archetypal experience is deal-
ing with a (Western) literary text. What is interesting about
Pound's emphasis on the ability of writers 'as writers' is that it
epitomizes a literate society's reverence for the literary text and
hence its tendency to identify 'writing' as the mode of serious
expression par excellence. The elevation of 'the writer' to a position
of high cultural prestige is characteristic of a society imbued with
the assumptions of utilitarian literacy. 'The writer' is one whose
works deserve to be circulated and preserved for posterity: and only
writing can ensure this. In the Western tradition this attitude goes
back to ancient times: Cicero already uses scriptor interchangeably
with auctor.
But the writing that writers like Balzac and Cicero do consti-
tutes no more than a minute fraction of writing as understood by
the historian or the anthropologist. Writing in this broader sense
too could, at a pinch, be brought within the scope of Barthes's
dictum. For there are writing systems that have disappeared as
well as writing systems no longer in use, just as there are lost
languages and dead languages. In this wider context, Barthes's
truism recovers its reassuring banality. But underneath this banal-
15
E. Pound, ABC of Reading, 1934, Ch. 3.
16
1.A. Richards, How to Read a Page, London, Kegan Paul, Trench,
Trubner, 1943.
ity one senses an attempt at definition. The suggestion seems to be

that if, for any particular society over any given period of time, we
could somehow draw a moving line separating what literate
members of that society could write and what they could not, then
we would at one stroke have sorted out the history of writing.
This, however, would be a jejune idea. For what it is possible to
write (either in the broader or the narrower sense) is secondary
and adventitious. Before anything can be written at all, writing has
to be available. Not so much in the sense that a system of writing
has to be in existence before anyone can use it, but in the sense
that without the concept of writing the question of what to write
does not even arise.
As soon as this conceptual priority is given its due, Barthes's
pronouncement is no longer the truism it seemed to be, and to
take it seriously would be a mistake. The history of writing is not a
steady, gradual progression in which some primitive concept of
writing is replaced by concepts that become successively more
sophisticated or more correct.
Nor should the current popularity of the odd expression oral
literature be allowed to obscure what is at issue here. What the
oxymoron oral literature reflects is Western culture's belated and
somewhat embarrassed attempt to come to terms with the realiz-
ation that the production of poetry, narratives and other verbal
compositions that engage the intelligence and the imagination is
not the exclusive prerogative of literate civilizations. But the
downmarket label 'oral literature' makes it clear what is happen-
ing here: outsiders are being reluctantly granted a cultural status
to which 'literally' they are not entitled. Authors of'oral literature'
are thus automatically branded as second-class citizens in the
Republic of Letters.
There is a certain historical irony in the fact that Wells's prophecy

of a post-literate society was addressed to readers who belonged to
the first generation to benefit from Disraeli's famous education act
of 1876, which laid down that every child in the land should be
taught to read and write. In other words, at the very moment

when universal literacy is recognized as an achievable ideal (at
least, in one country) we find already a forecast of eventual rever-
sion to a state of universal illiteracy. In fact, more than a hundred
years later, the number of countries where more or less everyone is
taught to read and write as a matter of course remains quite small.
It is by no means out of the question that in some parts of the
world literacy will simply be bypassed by the advent of new com-
munications technologies which make it unnecessary to teach the
skills of reading and writing to the whole population.
This thought should provide us with another reason to reflect
on what we mean by a literate society and on the connexion
between literate societies and civilization. According to historians
of writing, we know nothing that allows us to suppose that there
were literate societies before, roughly, the beginning of the third
millennium BC. But this is because most historians of writing
regard the evidence as beginning with the appearance of the
earliest surviving texts (or what are nowadays recognized as such).
Is it reasonable, however, to speak of a literate society if we know
that the production of such texts remained in the hands of a
small, privileged class of professionals? This question reflects back
upon the notion that there is some kind of equation between
writing and civilization. And in turn raises the question of
whether it is only the literate members of a society who are
civilized.
From at least the time of Disraeli onwards, it is indeed possible
to detect in educational reform the assumption that teaching the
illiterate to read and write is a matter of civilizing them, of mak-
ing them fully participating members of society. This idea is often
overlaid and reinforced by the further assumption that literacy
brings with it access to an essential religious text: the Bible in the
case of most Western educators. Carried to its logical conclusion,
this brings us to the notion that literacy not only civilizes but,
more importantly, opens up the prospect of spiritual salvation for
the individual. That conclusion has been overtly embodied in
many programmes of missionary linguistics, and still is. It is no
coincidence that in many parts of the world the first writing sys-
tems were devised by missionaries and the first published texts
were religious texts.
How does the 'bare idea' of writing relate to these far-from-
bare enterprises that are seemingly based upon it? Do they have
anything to do with it at all? Or is writing itself merely a neutral
technology, which can be used for whatever purposes the user
decides? And if the latter, why should one continue to insist on the
advent of writing - rather than, say, the advent of the steam
engine or the advent of electricity - as heralding the dawn of
civilization?
The most ambitious answer that theorists have recently pro-
posed as regards the importance of writing in human evolution
involves the claim that writing is not 'just' a technology, a means
of storing and retrieving information. Writing is said to be some-
thing more, and this something more is psychological. Writing, it
is held, 'restructures consciousness'." The claim is that when
Homo scribens eventually succeeded Homo loquens a new kind of
mentality made its first appearance in the history of the human
race.
A deeper understanding of pristine or primary orality enables
us better to understand the new world of writing, what it truly
is, and what functionally literate human beings really are:
beings whose thought processes do not grow out of simply nat-
ural powers but out of these powers as structured, directly or
indirectly, by the technology of writing. Without writing, the
literate mind would not and could not think as it does, not only
when engaged in writing but normally even when it is compos-
ing its thoughts in oral form. More than any other single inven-
tion, writing has transformed human consciousness.18
More extreme forms of this claim have also been put forward,
some of which are very reminiscent of eighteenth-century views
17
WJ. Ong, Orality and Literacy, London, Methuen, 1982, Ch. 4.
18
Ong, op. cit., p. 79.
concerning the superiority of alphabetic writing. According to

Marshall McLuhan, for example, 'the Chinese are tribal, people
of the ear', their writing is non-alphabetic and only the alphabet
'has the power to translate man from the tribal to the civilized
sphere'.19 R.K. Logan relates 'the lack of theoretical science in
China' to the non-alphabetic nature of Chinese writing.20
If these or similar propositions have any foundation, then there
would seem to be grounds - at least in principle - for saying that
Johnson and Rousseau, together with some nineteenth-century
anthropologists and their successors, may not have been
altogether misguided in attributing to the advent of writing, or
particular forms of writing, a diagnostic status. That would still
leave room for a great deal of argument about the details. But if it
is - or may be - true that there is a huge gulf separating the
mentality of the preliterate individual from the mentality of one
who can read and write (and habitually does so), it is not difficult
to see how writing falls into place as one of the most crucial
developments in human evolution. And if so, the eventual
regression to a Wellsian post-literate society would be a matter of
considerable and legitimate concern, not only for educationists
but for everyone else. A future without writing would be, quite
simply, an impoverished form of living with an impoverished form
of thinking; a life from which the 'bare idea' of writing had
vanished.
Perhaps this recognition of the cognitive role of writing marks
the beginning of the transition from utilitarian literacy to full
literacy. Perhaps it heralds at long last the advent of a mature
literate society. But how are we to assess these possibilities? By
what criteria are such claims and their validity or invalidity to be
evaluated? The contention motivating the present book is that the
only sound basis for such an evaluation depends on developing an
objective and independent semiology of writing. 'Objective' and
19
M. McLuhan, The Gutenberg Galaxy, Toronto, University of Toronto
Press, 1962, p.38.
20
R.K. Logan, The Alphabet Effect, New York, Morrow, 1986, Ch.3.
'independent' in that it will be free from the more or less ethno-

centric assumptions discussed above, which have shaped the study
of writing and of its history hitherto; free from any tendency to
flatter our literate self-esteem by tacitly equating the readers and
writers of the world with its civilized population. And this, for
obvious reasons, is a problematic task. How can scholars distance
themselves from presuppositions that are exemplified by - and
enshrined in - their own publications, their own scholarly
endeavours and their own institutional existence?
What is needed - the goal - is nevertheless clear enough. It
must be a semiology which breaks with the old tradition of treat-
ing writing systems as indices of cultural progress or cognitive
advancement, the tradition which judges writing systems by their
'accuracy' in transcribing the spoken word, the tradition which
invariably treats the alphabet, either tacitly or overtly, as the
ultimate human achievement in the history of forms of
writing. For only then can we feel confident that we have available
a semiology of writing which does not merely recycle the old
prejudices.
One might have hoped that such a semiology would emerge
from the work of the two great founders of modern thinking
about signs, Saussure and C.S. Peirce. Peirce, however, has disap-
pointingly little to say about writing: he is, in fact, a telling - one
might even say shocking — example of the intellectual myopia that
utilitarian literacy can induce about its own procedures and their
semiological implications. Saussure, on the other hand, at least
saw to what extent his own society's views of linguistic matters
were distorted by the accumulated biasses of literacy. He therefore
insisted as a prerequisite for any modern science of language and
languages that a clear separation should be made between the
spoken and the written.
That lesson was scrupulously learnt by his successors and
became one of the maxims of twentieth-century linguistics.
Unfortunately, the lesson was generally interpreted as meaning
that linguists need not bother with writing, because writing was
not really 'language'. Thus, as one observer put it, 'the study of
writing began to suffer at the hands of linguists.' It still does.21 Any

opportunity that linguistics might have had to contribute to the
self-understanding of a literate society was thus lost. For while
insisting on that separation, Saussure at the same time reinstated
(for reasons connected with his own academic politics) some of the
most questionable traditional assumptions about writing.
Today, as a result, we are still lacking a viable semiology of
writing, one which will be sufficiently robust to stand up to the
critical scrutiny of a generation willing to take less for granted -
both about language and about communication in general - than
Saussure's did. The rethinking of writing that was already needed
in Saussure's day still remains to be done.
21
K.H. Basso, 'The ethnography of writing', p.425. (R. Bauman andj.
Sherzer, Explorations in the Ethnography of Speaking, Cambridge, Cambridge
University Press, 1974, pp.425-32.) The passage continues: 'Depicted by
members of the emergent structural school as a pale and impoverished
reflection of language, writing was consigned to a position of decidedly
minor importance. Textbooks continued to include brief chapters on the
subject, but this was to emphasize that writing and language were
entirely distinct and that the former had no place within the domain of
modern linguistics.'
CHAPTER ONE
Aristotle's Abecedary
Plato was the first authority we know of who questioned the think-
ing about writing that was current in his day. His conclusions, if
we can rely on the Phaedrus and the authenticity of Letter 7, were
deeply sceptical.1 He thought that writing compared disadvan-
tageously with speech, because it gave a specious permanence to
words. This is a contention which does not make sense unless
addressed to a society already well into the stage of utilitarian
literacy, where one of the great advantages of this surrogate form
is seen as being, precisely, that it gives words a permanency — or at
least a durability, a verifiability — that the human voice does not
afford. Plato's objection to writing, in short, is not theological or
metaphysical or pragmatic but semiological. He objects that writ-
ing does not do what people commonly assume it can do, i.e. stand
as an alternative to or even a definitive version of an oral dis-
course. This objection implies that what a written text signifies
cannot be equated with what the corresponding oral discourse
signifies. (It is a purely negative objection, i.e. Plato does not
commit himself as to what a written text does signify. Aristotle later
exploited this lacuna.)
Plato had parallel misgivings about painting as a way of
1
Plato, Phaedrus and Letters VII and VII, trans. W. Hamilton, London,
Penguin, 1973.
capturing what the eye can see.2 The eye can see a person; but all
the painter can produce is a lifeless image.
The productions of painting look like living beings, but if you
ask them a question they maintain a solemn silence. The same
holds true of written words; you might suppose that they under-
stand what they are saying, but if you ask them what they mean
by anything they simply return the same answer over and over
again.3
Here we glimpse the beginnings of what was to become a cross-

modal semiological topos in the Western tradition, whereby writing
is commonly assumed to be a kind of 'depiction' of speech. It is
clear from the comparison, however, that what Plato is objecting
to about writing is not the accuracy of the depiction - the question
that was to preoccupy the grammarians of antiquity and their
successors. He is not, in other words, complaining that the Greek
alphabet is deficient in resources to capture all the nuances of
spoken Greek. That would be like complaining that the portrait
does not resemble the sitter. Plato's objection goes much deeper
than that. What he is rejecting is the whole assumption that writ-
ing provides a viable surrogate for speech, or at least denying that
writing can be treated as a reliable substitute where important (i.e.
philosophical) matters are concerned. Adopting a semiological
perspective on writing, as far as he is concerned, is a move in an
argument about language and philosophy. Had he lived to witness
the advent of the tape-recording he would have said exactly the
same thing about that. The problem is not that writing does not let
us hear the voice of the author. The problem is, rather, that writ-
ing (like the tape-recording) is not language in vivo, but merely the
lifeless shell or trace that language leaves behind.4 To suppose
2
For detailed discussion of Plato's reservations about art, see Iris
Murdoch, The Fire and the Sun, Oxford, Oxford University Press, 1977.
3
Phaedrus, 275.
4
A superficially similar but much shallower complaint reappears in the
work of some twentieth-century linguists. (See Chapter 8.)
Aristotle's Abecedary 19
otherwise is to make the mistake of equating language with its

material residuum. And this Plato implicitly compares with
mistaking the portrait for the living person.
Plato's scepticism strikes at the very foundations of utilitarian
literacy. It questions the validity of divorcing the speech from the
speaker, the text from the writer. And insofar as writing encour-
ages us to believe in the viability and usefulness of that divorce it
is, for Plato, a deception. It masquerades as something it cannot
possibly be. From this conclusion it is not a million stadia to St
Paul's famous warning that 'the letter killeth'. Utilitarian literacy
needs its detractors. They play an essential role in maintaining
the perception that writing and the interpretation of writing are
forms of expertise, not to be entrusted to fools. The lawyer, the
theologian, the historian, the grammarian and the literary critic
all have a vested professional interest in perpetuating that
perception.
Part of the background to Plato's complaint concerns the per-

vasive practice in his time of discussing speech in terms of writ-
ing.3 There were two terms stoikheion and gramma that could be
used either to refer to letters of the alphabet or to the sounds of
speech. As we see from Plato's Cratylus, Plato himself tends to use
them interchangeably. To add to the problem, both letters and
sounds were commonly identified by their letter-forms or letter-
names (alpha, beta, etc.). The earliest surviving Greek grammatical
treatise, that of Dionysius Thrax, bears witness to the fact that this
potentially confusing state of affairs survived long after Plato's
' Saussure, more than two millennia later, castigates his own immediate
predecessors for doing exactly the same. But they did not have the excuse
that might be offered on behalf of Plato's generation; namely, that no
adequate metalinguistic terminology was available. The longevity of this
conflation points to the endemic scriptism of Western thinking about
language.
day.6 There we find a section devoted to the stoikheion which starts

by announcing that there are twenty-four letters (grammatd) in the
alphabet and that they are called grammata because etymologically
that means 'scratches' or 'traces'. But then we are told that they
are also called stoikheia because they are arranged in a row
(stoikhori). But the subclassification which is then given into vowels,
diphthongs, consonants, etc. seems to be based on phonetic cri-
teria. The letter-shapes themselves are never discussed, although
they are used to identify the stoikheia.1
A similar state of affairs, as Franôise Desbordes points out, is
found in Latin with the terms littera and elementum. As she goes on
to observe, however, although authorities of this period are
apparently quite capable of declaring that 'the human voice
consists of letters' and so on, it would be absurd to suppose that
they imagined armies of little black marks issuing forth from the
speaker's mouth.8 But what exactly did they imagine? And what
did they mean? These are difficult questions. But perhaps no
more difficult than the corresponding questions concerning
people who nowadays claim that some words, and even some
languages, are 'pronounced as they are written'. It seems
unlikely that they believe the movement of the vocal organs
imitates the movement of the hand holding the pen. Or perhaps
they do?
These are questions that bring us face to face with the difficulty
of conceptualizing and explaining, even in a quite sophisticated
literate society, exactly what the relationship between speech and
writing is. And especially difficult if one wishes to interpret the
6
Dionysius himself seems to have lived in the 2nd century BC, but the
surviving treatise that bears his name is probably a much later
composition.
' A good modern edition is that of Jean Lallot, La Grammaire de Denys le
Thrace, Paris, CNRS, 1989. The section on the stoikheion is Chapter 6 of
the text.
8
F. Desbordes, Idees romaines sur I'ecriture, Lille, Presses Universitaires de
Lille, 1990, p. 113.
relationship in semiological terms, as Aristotle, for one, certainly

wished to do.
Aristotle's response to Plato (if that is the right way to look at it) is
to counter Plato's scepticism by conceding, in effect, that one must
not confuse the functions of speech and writing and then proceed-
ing to draw a clear semiological distinction between the two. He is
the first to try to explain the semiological gap between speech and
writing; i.e. exactly why the written forms do not capture the
thought expressed by the corresponding spoken forms.
Sounds produced by the voice are symbols of affections of the
soul,9 and writing is a symbol of vocal sounds. And just as
letters are not the same for all men, sounds are not the same
either, although the affections directly expressed by these indi-
cations are the same for everyone, as are the things of which
these impressions are images.10
This turns out to have been one of the most cryptic but also one
of the most influential statements about writing ever made in the
Western tradition. Because it is so cryptic, certain details are less
than clear, but it nevertheless leaves us in no doubt that for Aris-
totle the essential function of writing is to provide signs for other
signs, i.e. visual metasigns which 'symbolize' the sounds of speech.
How are we to intepret this? The first requirement, often neg-
lected by commentators, is to determine the level of abstraction at
9
'Affections of the soul' is the traditional translation, which I retain here
in preference to replacing it by 'concepts' or 'mental impressions', partly
because either of the latter would stand as much in need of interpre-
tation as 'affections of the soul'. Keeping this somewhat quaint term at
least draws attention to the obscurity of what Aristotle meant by it. Even
in the latter part of the nineteenth century there were still linguists who
referred to language as an external revelation of 'acts of the soul' (e.g.
W.D. Whitney, The Life and Growth of Language, New York, Appleton,
1875, p.303).
1(1
De Interpretation, 16 A.
which Aristotle's account is situated. For it is probably only the

synopsis of a more elaborate discussion by Aristotle that has not
survived in the extant texts.
Aristotle evidently recognizes four basic elements, including
two types of sign: oral and written. The former belongs to spoken
discourse and the latter to visual communication. He further dis-
tinguishes an internal element, the 'affection of the soul', which is
related to an external element (the object or thing in question). So,
according to Aristotle's schema, there are two channels of com-
munication which link certain elements occupying, so to speak,
two different spaces: an internal, psychological space and an
external, physical space.
EXTERNAL INTERNAL
objects affections of the soul
vocal sounds
letters
Let us take, for example, talking about the city of Athens.
According to Aristotle's semiology, the starting point is the object,
the city itself as an existing complex. Then there is the affection of
the soul, a psychological item, the idea one forms of this city.
Third, there is the name of the city, the name we utter when we
talk about Athens. Finally, there is the written form of this name,
that we find in edicts, inscriptions, etc. So there are four elements,
each perfectly distinguishable from the other three.
The passage from De Interpretations is concerned with setting out
the general relations obtaining between these four types of ele-
ment; i.e. between what we say, what we write, what we think
and the thing about which we speak, write and think. Straight
away it is evident that the discussion is situated at a higher level of
abstraction than that which relates to the study of particular lan-
guages, or even that of logic. And there is nothing surprising in
that. For we are dealing here in De Interpretatiom with the preface to
a treatise on the proposition.
But what is remarkable in Aristotle's presentation of this very
general and preliminary discussion is that the relation between the
vocal sign and the written sign is assumed to be of the same order as
the relation obtaining between the vocal sign and the affection of
the soul, as indicated by the use of the term symbolon. This is
Aristotle's contribution to rethinking writing. In both cases we are
dealing with a 'symbolic' relation. Written signs are symbola of
vocal sounds, and vocal sounds in turn are symbola of affections of
the soul.
What does Aristotle mean by this? What is a symbolon? Accord-
ing to Liddell and Scott, it is one half of an astragalos or similar
object that two xenoi or parties to a contract broke in two, each
keeping half as proof of the contractual relationship.!' Modern
translations commonly render it as 'symbol' although, as Whitaker
observes, this is 'not the most informative word to choose'.
The normal use of the Greek word was for a tally or token. A
contract or other agreement might be marked by breaking a
knucklebone or other object in two, one portion being taken by
each of the parties to the agreement. Each person kept his
piece, and could identify the person who presented the other
piece by matching it with his own. The word hence comes to
denote any token, for instance, for admission to the theatre.12
As a communicational metaphor, Aristotle's use of the term is

both striking and profound. What is involved is both an agreement
(between the two parties) and a physical relationship between the
items (since the two halves were one before being sundered).
Furthermore, they were broken apart deliberately (by the parties
in question) in order to provide proof of that relationship.
Here we have a basic type of connexion which is of great
semiological interest. For the two disjoint parts of the symbolon
have no value at all individually. Each is significant only as a counter-
part of the other. More importantly still, this relationship is not
11
H.G. Liddell and R. Scott, Greek-English Lexicon, rev. ed., Oxford,
Clarendon, 1996, p. 1676.
12
C.W.A. Whitaker, Aristotle's De Interpretatione, Oxford, Clarendon, 1996,
p. 10.
representational. Unlike the relationship between the affection of the

soul and the object: that is representational.
Aristotle draws this distinction quite clearly, albeit succinctly. It
is all the more dispiriting to read modern commentators who
rabbit on about 'representation' without apparently noticing that
right from the beginning of the Western tradition the more alert
minds were perfectly capable of distinguishing representation
from symbolism.13 Semiology today might perhaps take the
rehabilitation of that distinction as ranking among its urgent pri-
orities. Without it any attempt to think clearly about writing — and
many other practices involving signs - is on a hiding to nothing.
With the two halves of the symbolon there is no question of one
representing the other or being a substitute for the other. They are not
identical, nor equivalent. Substituting one for the other is a mean-
ingless operation. (It would leave things exactly as they were.) Nor
is there any question of one half being a copy of the other. (On
the contrary, it is important that they should differ.) The whole
point is that they are both different and unique. Both their difference
and their uniqueness are united in their complementarity. Anyone
who does not grasp that has not understood what a symbolon is, at
least as Aristotle is using the term.
Aristotle invites us to see certain relations as 'symbolic'.
(Whether we accept that invitation is another matter.) He invites
us in effect to see a Greek writing down the name of the city of
Athens as someone inscribing grammata which are symbola of the
sounds uttered when this name is spoken. At the same time - and
one might say almost in the same breath - he invites us to see the
sounds uttered as symbola of something in the mind of the speaker
(the idea of the city thus named, i.e. the city of Athens). The point
to note is that according to this account the grammata as such have
nothing to do with Athens. It would not matter if Athens were a
13
Oddly, perhaps as the legacy of nineteenth-century poetry, symbolism is
now the vaguer term. Almost anything can be a symbol of something
else, provided the mind makes the requisite connexion. That was not the
case in Aristotle's day.
country village or a foreign land or another planet. All that mat-

ters (in this relationship) is that the grammata 'symbolize' the right
sounds, the right sounds being those of the name 'Athens'. How-
ever, there is an indirect relationship between these grammata and
the city in question, which hinges on the name 'Athens' being - as
a matter of fact - the name of that very city.
But what is it to be 'the name of that city? The name, according
to Aristotle, is the sound uttered which symbolizes that 'affection
of the soul' relating to the city of Athens. (Presumably someone
who has never seen or heard of Athens will not have any corre-
sponding affection of the soul: so for such a person the name will
be meaningless, i.e. the sound in question will not be a name of
anything, or else a name of something unknown.) The affection of
the soul, it should be noted, is not a symbolon of the city of Athens,
but its image (homoioma). In other words, symbolic relations do not,
for Aristotle, obtain generally between external and internal
elements. Vocal sounds are symbols of affections of the soul, but
letters are not. The city of Athens is not a symbol of anything.
According to Aristotle the affections of the soul (pathemata tes
psyches) are homoiomata of objects or events (pragmatd). So it is the
external world which supplies the originals for those images that
dwell in the soul. This world ofpragmata a outside, which is the same
for all of us, thus has only an indirect connexion with speech and
an even more indirect connexion with writing.14
14
At the beginning of De Sophisticis Elenchis Aristotle seems to suggest that
names (onomatd) are symbols of things (pragmatd) and we employ them
because we could not use the things themselves for purposes of dis-
course. But this is not to be taken too seriously, as Swift saw when he
turned it into a joke in Gulliver's Travels about the sages who save the wear
and tear on their vocal cords by carrying around large sacks of things on
their backs in order to obviate the need for words. Aristotle's purpose
here becomes clear when the remark is put in context. He is warning
against the dangers of supposing that the world of words provides a
faithful image of the world of things. His point is, precisely, that words
are not to be taken as standing in a one-to-one relationship with particu-
lar things. A single name signifies more than one thing.
In the same passage in De Interpretatione Aristotle distinguishes

between a symbolon and a semeion. The latter is speech as a sign of
what is happening in the speaker's soul (psyche) at a given moment.
In other words, when I pronounce the name of the city of Athens,
that is for my listeners a semeion of what I am thinking, i.e. an
indication of the fact that I have that city in mind. My listeners,
however, will not know that unless they know what the name in
question is a symbolon of. So understanding the semeion presupposes
acquaintance with the symbolon. Aristotle also distinguishes
between what we write (gmphomena) and the letters (grammata) we
use for that purpose.
Aristotle, in short, is quite careful to distinguish terminologic-
ally between the units of the communication system (spoken or
written) and their use in given circumstances. But unfortunately
these distinctions were not always preserved in the exegetical
tradition. Already in the Latin translation of Aristotle by
Boethius the two terms symbolon and semeion are rendered by a
single Latin word, nota, which in any case has no technical mean-
ing corresponding to that of symbolon in Greek. This example
illustrates how easy it is to lose in translation the nuances of
Aristotle's quite subtle formulation. Boethius has already blurred
the schematic picture that Aristotle took some trouble to delineate
quite precisely, and many subsequent commentators have done no
better.
If the above interpretation of this key passage is on the right
lines, Aristotle's semiological framework provides for a series of
relations between quite separate theoretical items, which can be
arranged by order of origin as follows:
object > affection of the soul > vocal sign > written sign.
The series begins with objects in the external world, which
generate affections of the soul (psyche}. These in turn give rise to
vocal signs, through the social institution of language. Vocal signs
in turn give rise to written signs, through another social institu-
tion, which is writing. Thus the affection of the soul presupposes
the object, the vocal sign the affection of the soul and the written
sign the vocal sign. Relations between written signs and vocal
signs, as between vocal signs and affections of the soul, are sym-
bolic. Relations between affections of the soul and external
objects are based on likeness.
Thus Aristotle's solution to the problem of defining the rela-
tionship between writing and speech is to incorporate both into a
model of language which displays their respective roles with
respect to (i) thought, (ii) the expression of thought and (iii) the
external world. In this model, writing does not directly express
thought. (To that extent, Plato would have been pleased.)
Aristotle's solution became incorporated into the body of received

truths about language that were handed down in the Western
tradition. Thus, for example, we find the grammarians of Port-
Royal in the seventeenth century initiating their young pupils
into the mysteries of grammar by explaining the relation between
writing and speech as follows:
Grammar is the Art of speaking. Speaking is explaining one's
thoughts by signs, that men have invented for this purpose.
It was found that the most convenient of these signs were vocal
sounds.
But because these sounds do not last, other signs were invented
to make them durable and visible, which are the written charac-
ters that the Greeks call grdmmata, from which came the word
Grammar.1^
This is unadulterated (albeit simplified) Aristotle, with the topos

of'visible speech' added for good measure.16
Where does Saussure stand in relation to the Aristotelian
lj
Grammaire generate et raisonnee, Paris, 1660, p.5.
16
The Platonic parallel between painter and poet had long since been
epigrammatized in Horace's famous dictum utpicturapoesis. The concept
of writing as 'visible speech' extends this. If poetry is depiction, the
written poem is a depiction of a depiction.
legacy? When Saussure says that writing exists only to represent

spoken language he at first sight appears to align himself
unambiguously with this tradition. Indeed it might seem that what
he says on the subject adds nothing new.
A language and its written form constitute two separate sys-
tems of signs. The sole reason for the existence of the latter is to
represent the former.] 7
However, Saussure does not follow Aristotle all the way. When
Saussure refers to the 'sounds' of a word, it is important to be
clear what he means.
A linguistic sign is not a link between a thing and a name, but
between a concept and a sound pattern (image acoustique). The
sound pattern is not actually a sound; for a sound is some-
thing physical. A sound pattern is the hearer's psychological
impression of a sound, as given to him by the evidence of his
senses. [. . . ]
We grasp the words of a language as sound patterns. That is
why it is best to avoid referring to them as composed of 'speech
sounds'. Such a term, implying the activity of the vocal appa-
ratus, is appropriate to the spoken word, to the actualisation
of the sound pattern in discourse. Speaking of the sounds and
syllables of a word need not give rise to any misunderstanding,
provided one always bears in mind that this refers to the sound
pattern.18
No such interpretation can reasonably be put on what the Port-

Royal grammarians say about writing; for they make it quite clear
that its basic purpose is to make ephemeral sounds 'durable and
17
F. de Saussure, Cours de linguistique generate, 2nd ed., Paris, Payot, 1922,
p.45. All page references are to this edition. English translations are
from F. de Saussure, Course in General Linguistics, trans. R. Harris, London,
Duckworth, 1983.
18
Saussure, op. cit., p.98.
visible'.19 And these ephemeral sounds can only be what Saussure

refers to as the actualisation in discourse {parole}. Aristotle likewise
is talking explicitly about the sounds produced by the human
voice, not anything in the 'soul' of the speaker. On this point,
therefore, Saussure takes very great care to distance himself
from Port-Royal and Aristotle. He is insistent that if there are
'sounds' signified by alphabetic letters, these sounds must already
be envisaged as psychological entities.
The result of Saussure's refusal to commit himself fully to that
line of thinking raises a problem about what exactly the relation is
between 'what we hear' in one sense and 'what we hear' in
another. For an auditory impression, mediated by acquaintance
with the language (langue in Saussurean terminology), may quite
possibly not be an accurate reflection of the sounds, i.e. the
external acoustic facts. So how do the latter stand in relation to the
former?
In the Aristotelian tradition this is not a problem because the
question of auditory impressions does not arise. So the 'accuracy'
of a written transcription can only depend on correspondence
between the sounds uttered and the letters inscribed. In fact, in
this perspective 'words' as linguistic units do not strictly come into
the picture. In theory, at least, the Aristotelian secretary does not
need to understand what is being dictated: all that is necessary is a
good ear for speech sounds and a knowledge of the grapho-
phonic correspondences.
For the Saussurean secretary, however, that would not be
enough. Only an adequate knowledge of the language (langue)
19
They also concede, however, that a legitimate function of spelling is to
'help us to grasp what the sound signifies.' Thus they think that it is
useful to spell champ ('field') with a final -p (even though it is not pro-
nounced) because it reminds us of the derivation from Latin campus and
thus avoids confusion with the homophone chant ('sing'). For a discussion
of the view of writing in the Port-Royal grammar, seeJ-C. Pellat, 'La
conception de 1'ecriture a Port-Royal', inJ-G. Lapacherie (ed.), Proprietes
de 1'ecriture, Pau, Publications de 1'Universite de Pau, 1998, pp. 153-60.
would yield the information necessary to identify the internal

sound patterns (images acoustiques] correctly. Otherwise, as far as
Saussure is concerned, writing would have no connexion with
language at all. It would be an external instrument quite detached
from the signs involved in discourse, rather as a camera enables
the spectator to take photographs of what is going on in the
stadium, but does not help at all in understanding the rules of
play.
Now Saussure qua linguist needed a theory of writing which
would justify, in principle, the use of written texts as linguistic
evidence (because otherwise the past history of languages would
be inaccessible to 'scientific' study). And that is exactly what the
Aristotelian account could not provide. For a linguist operating on
strictly Aristotelian lines writing would be of dubious value. In
order to establish phonetic correspondences for letters, the testi-
mony of written texts would be worthless unless confirmed by
speakers of the language. But if the direct evidence of speakers is
available, the linguist does not need the texts anyway.
There is also a more profound difference between Saussure's view

of language and that taken for granted by Aristotle and his succes-
sors. The Aristotelian tradition is a nomenclaturist tradition. It is
natural within that tradition to view the relationship between let-
ter and sound on the model of the relationship between name and
item named. This explains how and why, within that tradition, the
question of the phonetic adequacy of alphabetic writing is con-
stantly being raised as a problem. Quintilian, for example, has this
to say about it:
It is for the grammarian (grammaticus) [. . . ] to consider whether
our alphabet lacks certain letters, not for writing Greek [ . . . ]
but for writing Latin words. For instance, in writing the words
serum and uulgus we lack the Aeolian digamma. There is also a
sound intermediate between u and z, for we do not pronounce
optimum and opimum in the same way. In the word which is
written here, neither the vowel e nor the vowel i is clearly
Aristotle }s Abecedary 31
heard. And then there is the question of superfluous letters

[...r
In this connexion, Quintilian asks whether one might not dis-
card the letter x.
Now Quintilian did not have either the mind or the tempera-
ment to try rethinking writing from scratch on his own account;
but he did confront certain questions arising within the semio-
logical model of his day, which is Aristotelian. The way Quintilian
discusses the Latin alphabet would be incomprehensible except
within a theoretical framework where two things are taken for
granted:
(a) the number of sounds to be heard in speech should determine
the number of letters in an alphabet, and
(b) that should result in a neat set of one-to-one correspondences
between sounds and letters.
Without these twin assumptions Quintilian's examples make
little sense. In the words seruus and uulgus, the two letters u are not
identically pronounced: therefore, according to Quintilian's rea-
soning, they should not be written identically either. But the Latin
alphabet has no letter to indicate the sound of the first u - i.e. no
letter which corresponds to the Greek digamma. Likewise, in the
two words optimum and opimum, despite what the orthography
seems to indicate, the vowels of the middle syllable are not the
same: therefore, they should not both be written i. In the word
optimum, the vowel is half-way between i and u. (In Latin texts the
spelling is sometimes optimum and sometimes optumum.} But in this
case the Greek alphabet offers no letter to fill the gap. A similar
case of an 'intermediate' sound is illustrated by here. This is a
vowel which is not quite i and not quite e.
According to Quintilian, therefore, there are Latin sounds
which cannot be indicated by the Latin alphabet (as it existed in
the Classical period). On the other hand, this alphabet included at
20
Institutio Oratoria, I, iv, 7-9.
least one unnecessary letter: x. This letter is superfluous because it

is simply a means of indicating the combination of two sounds
that can be separately represented as c followed by s.
Quintilian's verdict, then, is that the Latin alphabet is doubly
defective. On the one hand, it does not have enough letters to
register unambiguously the sounds of spoken Latin. On the other
hand, it has letters that serve no useful purpose, because the
sounds they indicate are already catered for.
Quintilian never asks why it should be assumed that all the
minutiae of pronunciation have to be indicated alphabetically;
nor whether that scrupulousness might not complicate Latin
orthography unnecessarily. He proceeds as if the recognition of a
phonetic difference is eo ipso a sufficient reason for trying to find
some way of indicating it alphabetically. Given this perspective,
the alphabet becomes a basic vocabulary for constructing descrip-
tions of any speech event. Such a vocabulary will be adequate
insofar as its inventory of terms matches the number of items to
be distinguished in the description required; and a nomenclature
is automatically inadequate if it emerges that there are nameless
items to be taken account of. These assumptions about an
adequate writing system conform to a more general concept of
'representation' in the Graeco-Roman world, where representa-
tion is envisaged ideally as a matter of one-to-one relations
between the thing represented and that which represents it.
The whole of Aristotle's schema as outlined in De Interpretatione is

based on bold (or naive?) pragmatic and psychological assump-
tions that have particularly struck modern commentators; namely,
that not only the external world but the internal world (the affec-
tions of the soul) are the same for everyone. It is on precisely
this issue that Saussure rejects the foundations of Aristotelian
semiology lock, stock and barrel. For according to Saussure our
internal mental world, or at least that part of it which includes the
conceptual domain, is language-dependent; and since languages
vary from one society to another there can be no question of
conceptual uniformity for the whole human race. Saussure is quite

categorical on this point:
In itself, thought is like a swirling cloud, where no shape is
intrinsically determinate. No ideas are established in advance,
and nothing is distinct, before the introduction of la langue.21
For Saussure, this is a basic problem and he keeps coming back

to it. We cannot take it for granted that the world out there is full
of things waiting to be given names, as Adam named the animals
in the Garden of Eden. Saussurean semiology sets itself the task
of explaining what Aristotelian nomenclaturism simply takes for
granted.
Saussure's need for a theory of writing that legitimizes the use of

written texts as linguistic evidence is not the only relevant factor to
bear in mind in assessing his account of writing. An even more
serious requirement now needs to be considered. Saussure could
not afford to admit that writing played a role, however small, in
the articulation of thought; that is to say, in establishing a con-
ceptual system which finds its mode of expression in language. For
that would have necessitated setting up two separate branches of
general linguistics, each with a different theoretical basis: one for
pre-literate communities and another for literate communities. In
order to avoid this embarrassing schism, it was necessary to limit
the function of the written sign to the representation of speech.
Aristotle, with quite different preoccupations from Saussure's,
faced no such restriction. What the modern reader finds difficult
to understand in Aristotle's case is why he placed the relationship
between sounds and letters on the same footing as that between
sounds and affections of the soul. Treating the relationship
between word and object as an indirect one, with a 'concept'
"' Saussure, op. cit., p. 155. The students' notes suggest that what Saus-
sure may have said was not la langue but le signe linguistique (the linguistic
sign).
playing the intermediary role, is reasonably uncontroversial, and

passed unchallenged into medieval and modern philosophies of
language. That is exactly how it is treated (allowing for differences
of terminology22) by two latter-day Aristotelians, C.K. Ogden and
LA. Richards, more than two thousand years later. In their 'tri-
angle of signification', the path between symbol and referent is
indicated by a dotted line to indicate an indirect relationship.23
REFERENCE
SYMBOL REFERENT
Now it would be perfectly possible to incorporate the distinc-

tion between speech and writing into this structure as follows:
REFERENCE
SPOKEN REFERENT WRITTEN

SYMBOL SYMBOL
That would be both a simple and an elegant solution. But Aris-

totle rejected it in favour of:
22
Ogden and Richards' symbol corresponds to Aristotle's symbol, their
reference to his affection of the soul, and their referent to his object in the
external world.
23
C.K. Ogden and LA. Richards, The Meaning of Meaning, London,
Routledge & Kegan Paul, 1923, p. 11.
REFERENCE
SPOKEN SYMBOL REFERENT
WRITTEN SYMBOL
Thus Aristotle's semiology makes the relation between spoken

and written language quite independent of the relation between
reference and referent. The questions we must ask are: Why does
Aristotle treat writing as no more than the use of grammata? Why
does he think grammata have nothing to do with thought? Why
does he regard grammata as no more than phonetic symbols?
A possible clue lies in the fact that Aristotle was born about
twenty years after an important orthographic reform: the official
introduction of the Ionic alphabet to Athens (403 BC), replacing
the previously used local Attic alphabet. Naturally, documents and
inscriptions in the old Attic alphabet did not disappear overnight.
Every educated Athenian of Aristotle's generation was perfectly
well acquainted with the two systems, and therefore with the fol-
lowing facts. First, it is possible to change alphabets (deliberately)
without changing the spoken language in the slightest. Second,
changing alphabets is a matter of changing a few grammata - noth-
ing more. Third, comparison between alphabets is not just a
matter of comparing different letter-shapes but comparing the
correspondences between letter-shapes and sounds. Finally, the
possibility of changing alphabets shows that there are no intrinsic
links between grammata and sounds: grammata can be invented,
borrowed or adapted to suit any needs. (They are, in the
Saussurean sense, 'arbitrary'.)
It does not take a great deal of perspicacity to realize that these
four propositions between them cover the whole of Aristotle's
semiology of writing. In other words, Aristotle's model is derived

directly from the experience of alphabetic writing as practised in
the Greece of his day, and from this experience he extrapolates to
linguistic communication in general, even though his remarks on
writing are presented in De Interpretatione as a kind of casual after-
thought. One wonders whether the apparent afterthought is not
in reality the source of the whole Aristotelian schema. (Otherwise,
why mention writing at all? It is quite irrelevant to an analysis
of propositions.) If this is so, it becomes easy to explain (i) why
Aristotle recognizes the same (i.e. 'symbolic') relationship in two
apparently quite different cases, and also (ii) why the two sets of
symbols in question (spoken and written) are nevertheless treated
separately, even though both are used in the service of linguistic
communication in one and the same society. In the case of (i)
what matters is that changing alphabets is seen as a paradigmatic
case of the independence of form from meaning when the
semiological relationship is based on complementarity, not on likeness. In the
case of (ii) what matters is that adopting the Ionic alphabet does
not mean that an Athenian has to adopt Ionic words or
pronunciations.
The practical implications of this new pattern of relations
between speech and writing are worth pausing to consider. For the
Athenian writer all that is initially required can be based on rela-
tively simple substitution formulae ('For old spelling x, substitute
new spelling jv'), which will be valid independently of the spoken
forms. So x acquires a new value, in virtue of functioning as inter-
mediary between the unchanged pronunciation and the changed
spelling^. But as familiarity with x recedes (being replaced for new
generations byy) the need arises for putting in place a pattern that
does not rely on x at all. So a new set of written forms has to be
redistributed or re-mapped on to an old set of spoken forms in the
language. Since the new alphabet left many words orthographic-
ally unchanged in any case, that requirement reduced to assigning
a phonetic value to (a few) new letters. Thus the sounds came to be
seen as holding the key to the whole system, not the letters, since
the sounds remained unaltered whichever system was adopted. Here we
have the origin of the phonoptic assumption that became

incorporated into the Western view of writing.
Let us now consider what Saussure says about 'phonetic' writing

against this historical background. Once it is assumed that utter-
ances contain a determinate number of vocal units (or 'sounds'),
each corresponding to a single letter in the corresponding written
form, then any combination of sounds can be represented by the
corresponding combination of letters. Seen from this perspective,
an ideal phonetic alphabet will provide exactly the right number
of letters necessary for writing the language in question (an ideal
already explicitly recognized by Quintilian); and on that basis it
will be possible to construct an orthography in which the
appropriate written form is determined automatically by the
phonetic composition of the oral message. This dispenses with
any need for orthographic 'conventions', spelling being regulated
in advance by the initial correlations between letters and sounds.
Or, more exactly, given a perfect alphabet (see Chapter 4) the only
conventions needed will be those determining the direction of
writing (right-to-left, top-to-bottom, etc.).
With this set of assumptions established early in the Western
tradition, Saussure's pronouncements mark a deviation to note.
His position is one which concedes that spoken sounds - in what
he calls parole - are infinite in number, and hence that it is chimer-
ical to hope to create an ideal form of writing that will succeed in
distinguishing them all.24 Any such idealization, furthermore, is
24
This was already taken for granted by other linguists in the second half
of the nineteenth century. See, for example, Ch. 21 of Hermann Paul,
Principien der Sprachgeschichte, 2nd ed., Halle, Niemeyer, 1886, where it is
recognized that the series of possible speech sounds is 'infinitely numer-
ous'. Where Saussure goes beyond Paul is in trying to explain how this
infinite variety can nevertheless be 'reduced' to a small, fixed inventory
of elements which is not an arbitrary selection from competing phonetic
classifications.
for Saussure an irrelevance, since what counts is not the actual

sound but the image acoustique. Unlike its (variable) realization in
utterance, the image consists of a determinate number of elements.
It is this structure that 'phonetic writing' can — if appropriately
organized - render accurately. There is no harm in calling the
relevant phonetic units 'sounds', provided it is clearly understood
that they are not to be confused with the raw phonic products of
the human vocal apparatus.
Why are these units of the image acoustique determinate in num-
ber and identity? Because they are elements of la langue. That is
why, in principle, the phonetic value of a letter depends essentially
on the language in question, and not - as it might first appear - on
the corresponding activity of the vocal apparatus.
In order to grasp the full force of this answer, it will be
necessary to situate it in the context of Saussure's proposals for
semiology (Chapter 2). But already it is apparent that Saussure's
intellectual strategy for dealing with the problem of writing is
very like Aristotle's in at least one respect: it is a strategy which
interrelates speech and writing by referring both to a more general
linguistic model.
CHAPTER TWO
Structuralism in the Scriptorium
Neither Plato nor Aristotle ever suggested setting up a separate

branch of inquiry devoted to the study of signs and signification.
Their semiological arguments were incidental to other concerns.
Locke, in the seventeenth century, was the first to identify semei-
otike, or 'the doctrine of signs', as one of the three main branches
of science, 'the business whereof is to consider the nature of signs,
the mind makes use of for the understanding of things, or convey-
ing its knowledge to others.'1 But it was not until the very end of
the nineteenth century that any general framework which would
accommodate the study of all signs was proposed.
The proliferation of semiological studies has undoubtedly been
one of the salient features of intellectual life in the humanities
since then. On all sides we have been offered general introduc-
tions to semiology, together with semiological approaches to
almost every subject under the sun. There have been semiological
analyses of works of literature, of paintings, of films, of cere-
monies, of musical compositions, of buildings. Most of these stud-
ies either originated or were eventually published as written texts.
In other words, writing has functioned as the preferred — or even
the indispensable - form for semiological studies: it has become
1
An Essay Concerning Human Understanding, 5th ed., London, 1706, ed.
A.C. Fraser, 1894, repr. New York, Dover, 1969, Bk.IV, Ch.21.
the semiological filter through which the semiologist's exposition

is obliged to pass. It is all the more ironical that during all that
time almost the only applications of semiological theory to writing
itself have been those derived directly or indirectly from Saussure.
It seems, in short, that in the twentieth century most writing
about writing has simply taken the status of the written sign for
granted. In this respect the writers concerned followed faithfully
in the footsteps of their predecessors. This might even be regarded
as the defining feature of utilitarian literacy: one does not ask
questions about those useful writing techniques one is familiar
with. Even those who regard the status of the written sign as
being, on the contrary, highly problematic evidently do not see this
as any hindrance to writing volubly about writing themselves, or
even treating the written sign as a paradigm of the linguistic sign
in general. Thus, for example, Jacques Derrida's ambition to
'destabilize' traditional Western discourse about writing, while
substituting a preferred discourse about an 'arche-writing' which
can never be subject to scientific investigation, would be an
example of the more ingeniously and absurdly self-defeating
extremes to which this kind of intellectual engagement with writ-
ing can be taken.2 So between blind confidence on the one hand
and self-stultifying scepticism on the other, recognition of the
need for taking the semiology of writing any further than Saussure
took it goes by default. That is doubtless because admitting that
there is any such need to rethink writing might be seen as tanta-
mount to challenging an education based on textbooks, or even
calling in question the profession of literate intellectual itself. And
that, in a culture which prides itself on the level of literacy it has
reached, would verge on the unpardonable. (Plato would have
found it all unutterably depressing.)
Implicit in Saussure's semiology of writing is a critique of all

previous thinking on the subject. Before either accepting or
2
J. Derrida, De lagrammatologie, Paris, Minuit, 1967, p.83.
Structuralism in the Scriptorium 41
rejecting Saussure's conclusions, it is important to understand the

basis of this critique, and how, from Saussure's perspective, the
enterprise of a semiology of writing is to be conceived.
The birth of Saussurean semiology was announced, somewhat
prematurely, in 1901 in Adrien Naville's Nouvelle Classification des
sciences? This was a revised and expanded version of a work by the
same author dating from 1888. In the new edition there was a
section devoted to sociology, in which appears the following
statement:
M de Saussure insists on the importance of a very general
science, which he calls semiology, having as its object the laws
governing the creation and transformation of signs and their
meanings. Semiology is an essential part of sociology. As
human conventional language is the most important sign sys-
tem, the most advanced semiological science is linguistics., the
science of the laws governing the life of language/
It is not clear whether this is Naville's own formulation or Saus-

sure's. How semiology would set about its scientific task is not
explained. Nor, apparently, did Saussure ever elaborate it to his
students in his Geneva lectures. However, in the posthumously
published Cours de linguistique generate we are told:
A language is a system of signs expressing ideas, and hence
comparable to writing, the deaf-and-dumb alphabet, symbolic
rites, forms of politeness, military signals, and so on. It is simply
the most important of such systems.
It is therefore possible to conceive of a science which studies the
life of signs as part of social life. It would form part of social
psychology, and hence of general psychology. We shall call it
semiology (from the Greek semeion^ 'sign').*3
!
A. Naville, Nouvelle Classification des sciences. Etude philosophique, Paris,
1901.
4
Naville, op. cit., p. 104.
' Saussure, op. cit., p.33.
What is interesting for our present purposes is to note that, at

the very moment when the curtain rises on modern semiology, we
find a comparison between languages and writing as systems of
signs. This is surprising in view of the exceptional status that
Saussure proceeds to assign to writing: its sole purpose, according
to the Cours, is to represent other systems of signs.6 This can hardly
be regarded - at least from Saussure's point of view - as typical.
It is rather like introducing a science of currency by drawing
attention to the existence of travellers' cheques.
The account of semiology in the Cours does not match that
given fifteen years earlier by Naville. There, as noted above,
semiology was introduced as 'an essential part of sociology'. In
the Cours it is placed under the aegis of a different discipline -
psychology. This is confirmed by the statement on the same page
of the text that 'it is for the psychologist to determine the exact
place of semiology'. It would seem from this that Saussure is not
too concerned about exactly where semiology fits into the aca-
demic scene: what matters much more is to establish the claim
of linguistics to constitute a legitimate branch of the new
science.
Again, whereas in Naville's version this science would be con-
cerned with the 'laws governing the creation and transformation
of signs and their meanings', according to the Cours it would study
'the life of signs as part of social life'. There are two striking
differences here: they relate to (i) the 'creation' of signs, and (ii)
their 'transformations'. Both repay consideration.
The question of the 'creation' of linguistic signs is one which
Saussure explicitly refuses to consider in the Cours: refuses, in fact,
twice over. He rejects it when it arises in the context of the initial
stages of language, both as a problem about how children learn
their native language and as a problem about the early history of
the human race. Why? Because, says Saussure, there is nothing to
be gained by pursuing such questions:
It is quite illusory to believe that where language is concerned
6
the problem of origins is any different from the problem of

permanent conditions.7
He rejects the same question again when it crops up in

connexion with linguistic change:
No individual is able, even if he wished, to modify in any way
a choice already established in the language. Nor can the lin-
guistic community exercise its authority to change even a single
word. The community, as much as the individual, is bound to its
language.8
And on the following page:

In fact, no society has ever known its language to be anything
other than something inherited from previous generations,
which it has no choice but to accept. That is why the question
of the origins of language does not have the importance gener-
ally attributed to it. It is not even a relevant question as far as
linguistics is concerned. The sole object of study in linguistics
is the normal, regular existence of a language already
established.9
The idea that, however far we go back, 'the language has

always been there' is one which has very important implications as
far as writing is concerned. This thesis, which might be called the
doctrine of'linguistic priority', is not to be confused with the more
familiar doctrine of the 'primacy of speech'.10 The two often go
together, but their theoretical status is by no means the same.
7
Saussure, op. cit., p. 24.
8
9
10
This is the doctrine that 'in the natural order of language . . . speech
comes first' (as it was put in the editorial preface to the first volume of the
Oxford English Dictionary). For discussion of Saussure's position with
respect to this doctrine, see R. Harris, Reading Saussure, London, Duck-
worth, 1987,pp.l7ff.
What is clear enough is that whenever the question of the 'cre-

ation' of linguistic signs threatens to come to the fore, Saussure
heads it off. And one wonders whether that would not be Saus-
sure's strategy with respect to any system of signs, i.e. to insist that
what counts, semiologically, is not how the system originated but
its 'normal, regular existence'. Thus if Saussure shows no interest
in questions concerning the origin of writing or the origin of any
particular kind of writing (including alphabetic writing), this
would seem to be not a lacuna in his account but an exclusion on
principle.
That may have some bearing on the second striking difference
between Naville's account of semiology and that given in the
Cours. It would be reasonable to suppose that the laws governing
the 'transformation of signs and their meanings' would fall under
diachronic semiology. For Saussure, however, there presumably
could be no question of embarking on diachronic semiology
before studying the 'normal, regular existence' of signs — in other
words, synchronic semiology. For unless that 'normal, regular
existence' were understood, the investigator would be in no posi-
tion to analyse changes in it. Any other programme would be
tantamount to repeating in semiology the mistake already made in
nineteenth-century historical linguistics. There is no reason, in
brief, to suppose that Saussure would not have carried over from
linguistics to semiology in general his insistence on an absolute
separation between synchronic and diachronic studies. Nor can
we suppose that he would have failed to give priority to the former
in semiology, as in linguistics. The distinction between synchronic
and diachronic, the Cours tells us, is fundamental for any science
that deals with values.''
The next question worth pondering in the account of semiology

presented in the Cours concerns the 'life of signs as part of social
life'. The exact wording here is probably the editors'; but the
11
Saussure, op, cit., p. 115.
double use of the biological metaphor is a deliberate repetition,

not a stylistic blemish. This reference to a life within a life is a way
of insisting on the symbiotic relation between signs and society.
For Saussure, semiology is not a science of signs in abstracto. It
proposes neither a general philosophy of signs nor a logic of their
relations but, quite specifically, a study of their mode of existence
in society. It excludes, in other words, signs which have no publicly
recognized signification. The point comes out even more forcefully
in Albert Riedlinger's notes for the Second Course than it did
subsequently in the published text of the Cours. According to
Riedlinger's version:
We <therefore> recognize as semiological only that part of
the phenomena which characteristically appears as a social
product,<and we refuse to consider as semiological what is
properly individuals12
Thus it is the community which validates and guarantees the

operation of the sign system.
On this important point Saussurean semiology differs markedly
from the semiotics of Peirce.13 According to Umberto Eco, the
12
E. Komatsu and G. Wolf (eds), F. de Saussure, Deuxieme Cours de linguistique
generate (1908-1909), Oxford, Pergamon, 1997, p. 14.
13
Some commentators appear to treat the terms semiology and semiotics as
more or less synonymous. According to S.E. Larsen, the difference merely
marks a 'superficial distinction' (P.V Lamarque (ed.), Concise Encyclopedia
of Philosophy of Language, Oxford, Pergamon, 1997, p. 177.) The super-
ficiality, however, lies not in the distinction but in dismissing it as
unimportant. It is difficult to understand how anyone more than super-
ficially acquainted with the work of Saussure and Peirce could reach the
conclusion that Saussure's semiology and Peirce's semiotics had much in
common. Retaining the Saussurean term, in preference to Peirce's, is a
reminder of the difference between two quite diverse approaches to
signs. Saussure can be seen as continuing the European tradition, where
Peirce marks a notable divergence from it. That is why it is far less
misleading to call the European tradition 'semiological' than to rebaptize
it (and presumably Saussure as well) with Peirce's term.
main difference between Saussure and Peirce is that semiology is

based on the concept of communication, which in turn implies
intentional activity.
[For Saussure] the sign is implicitly regarded as a communi-
cational device taking place between two human beings inten-
tionally aiming to communicate or to express something. It is
not by chance that all the examples of semiological systems
given by Saussure are without any shade of doubt strictly con-
ventionalized systems of artificial signs, such as military signals,
rules of etiquette and visual alphabets. Those who share
Saussure's notion of semiologie distinguish sharply between
intentional, artificial devices (which they call 'signs') and other
natural or unintentional manifestations which do not, strictly
speaking, deserve such a name.14
Eco is wrong about the intentionality, at least as far as Saussure

is concerned. It is true, however, that the idea he draws our atten-
tion to is expressed quite explicitly by certain theorists working
in the Saussurean tradition. Buyssens, for example, proposes the
following definition:
Semiology may be defined as the study of processes of com-
munication; that is to say, of means used to influence another
person and recognized as such by the person one wishes to influence.^
Buyssens's insistence on communication and the intentions of

the participants is very clear. That insistence, in effect, redefines
semiology as the study of speech acts in the sense of modern
analytic philosophy.16 What is highly questionable, however, is
14
U. Eco, A Theory of Semiotics, Bloomington, Indiana University Press,
1976, p.15.
13
E. Buyssens, La Communication et I'articulation linguistique, Paris/Brussels,
Presses Universitaires de France, 1967, p. 11. Italicization in the original.
16
See e.g.J. Searle, Speech Acts, Cambridge, Cambridge University Press,
1969. NB A speech act, in this sense, is not necessarily verbal, much less
oral.
whether that was Saussure's conception of semiology. Nowhere

does Saussure mention the speaker's intentions. Nor do these
intentions play any part in the Saussurean 'speech circuit': there
we find only 'concepts' - that is 'significations' (signifies}. And that is
something quite different. We are dealing with communication
indeed; but not with the intentions of the communicators. Saus-
sure's signifies do not belong to the speakers: they belong to the
language. Intentions, on the contrary, belong not to the language
but to individual speakers. Speakers use the concepts provided by
the language in order to express, as best they can, their own
intentions.
It is important to be clear what is at issue here, because the
question of intentions makes a vital difference to theories of the
written sign. (Is it the writer who is ultimately in control of'what is
written'?) Speech-act theory is indeed a semiological theory in
Saussure's sense; but to conflate it with Saussurean semiology
would be to distort beyond recognition Saussure's thinking about
writing.
The following discussion will proceed on the assumption that
Saussure neither claims nor presupposes that the semiological
function of the written sign has anything to do with the intentions
of the writer. An interesting piece of evidence on this point is
provided by a passage in Constantin's notes. In the lecture he
gave on 4 November 1910, it appears that Saussure spoke of a
psychology of sign systems. Constantin summarizes as follows:
Any psychology of sign systems will be part of social psycho-
logy - that is to say, will be exclusively social; it will involve the
same psychology as is applicable in the case of languages. The
laws governing changes in these systems of signs will often be
significantly similar to laws of linguistic change. This can easily
be seen in the case of writing - although the signs are visual
signs — which undergoes alterations comparable to phonetic
phenomena. 17
17
E. Komatsu and R. Harris (eds), Ferdinand de Saussure., Troisieme Cours de
linguistiquegenerate (1910-1911), Oxford, Pergamon, 1993, p.9.
The first point to note is the proviso 'exclusively social'. In other

words, this is not a matter of individual psychology; so there is no
question of bringing in the intentions of those involved. Thus
what Saussure proposes regarding linguistic signs applies equally
to any system of signs. That is not made very clear in the pub-
lished text of the Cours. (But Bally and Sechehaye did not have
Constantin's notes available when they produced their edition.)
In the second place, one is struck by the explicit comparison
between changes a writing system may undergo and phonetic
changes. That does not appear in the Cours either. The com-
parison is very significant, because Saussure repeatedly makes it
quite clear that phonetic changes have nothing to do with
speakers' intentions. These are changes brought about by mem-
bers of the linguistic community through their speech, but they
have no awareness of what they are doing. (Whether Saussure's
comparison between sound change and changes in writing con-
ventions is valid is another question: the point here is that in
Saussure's eyes it evidently is.)
What does all this amount to? If Constantin is a reliable wit-
ness, it means that Saussurean semiology envisages studying signs
only insofar as they comprise systems with an independent social
life — independent, that is, of their users' intentions. This conclu-
sion has important consequences, to which we must return in due
course. Straight away, however, it is worth noting that this limita-
tion of the domain of semiology fits perfectly within the frame-
work of Saussure's ideas. For if the limitation were not in place, it
is difficult to see how linguistics could serve as a 'model for the
whole of semiology', as we are told it does in the Cours.18 Such a
claim would make little sense if semiology were envisaged as need-
ing also to deal with signs subject to the will of the individual.
Another valuable piece of evidence is the passage in the chapter

on 'linguistic value' in the Cours, where a detailed comparison
18
between languages and writing is presented. This passage deserves

detailed examination, because it attempts to demonstrate that the
'material' element in linguistic value is solely a function of
relations and differences between linguistic signs. Having com-
mented on the latitude which speakers are allowed in the way they
pronounce sounds, the discussion continues as follows:
An identical state of affairs is to be found in that other system
of signs, writing. Writing offers a useful comparison, which
throws light upon the whole question. We find that:
1. The signs used in writing are arbitrary. The letter t, for
instance, has no connexion with the sound it denotes.
2. The values of the letters are purely negative and differen-
tial. So the same individual may write t in such variant forms as:
The one essential thing is that his t should be distinct from his /,
his d, etc.
3. Values in writing are solely based on contrasts within a
fixed system, having a determinate number of letters. This fea-
ture, although not the same as 2 above, is closely connected
with it; for both 2 and 3 follow from 1. Since the written sign is
arbitrary, its form is of little importance; or rather, is of impor-
tance only within certain limits imposed by the system.
4. The actual mode of inscription is irrelevant, because it
does not affect the system. (This also follows from 1.) Whether I
write in black or white, in incised characters or in relief, with a
pen or a chisel - none of that is of any importance for the
meaning.19
With this key passage, structuralism makes its entrance into the
19
Saussure, op. cit., pp. 165-6.
Western scriptorium, at a carefully chosen moment and with a

very specific objective in view.
It must be noted, first of all, that these observations occur under
the heading 'Linguistic value: material aspects'. In the immedi-
ately preceding section we have already had comments on the
conceptual aspects of linguistic value. There, clearly, the thesis
that values are purely negative is more plausible, or at least less
difficult to maintain, since in the conceptual domain we are
automatically dealing with abstractions; whereas in the material
domain we appear to be dealing with concrete, positive facts. And
precisely at that moment - a pivotal one for the argument in
question - Saussure appeals to writing.
Why? Doubtless because in the case of writing the 'concrete'
facts are more evident than in the case of speech. As a medium,
sound is invisible and ephemeral. The identification and re-
identification of particular sounds already requires some kind of
abstract analysis. Writing, on the other hand, presents us with
stable, visible marks that can be examined at leisure and repeat-
edly, without indulging in any analytic jiggery-pokery. (It may be
worth reminding ourselves that in the classroom of Saussure's day
there were no sound spectrograms or video screens to make
speech 'visible'.) So if, in spite of the 'material' nature of writing, it
can be shown that the formal value of the written sign is purely
negative, extending that conclusion to the linguistic sign would be
a natural corollary.
This argument patently relies on the written sign and the lin-
guistic sign being semiologically comparable in the relevant
respects. Hence the first point to establish is the arbitrary charac-
ter of the written sign. Here we encounter an initial difficulty. Is
it beyond doubt that the written sign is in fact arbitrary? Saussure
offers no proof, but merely cites the example of the letter t. Are we
to assume that it is only the alphabet he has in mind as an arbi-
trary system of writing? No such restriction is apparent in the
terms in which the case is presented.
For Saussure the differentiation between what is arbitrary and
what is not is sufficiently important to be worth marking by a
terminological distinction. He proposes reserving the term symbole

for devices that are not entirely arbitrary. (For example, the scales
as a symbol of justice.) Now it is noticeable that Saussure never
applies the term symbole to elements of writing. It is presumably
also significant, given that he must have been familiar with the
widespread belief in his day that writing originated with
pictography. So the question of what Saussure understands by
'arbitrary' assumes a particular importance for his semiology of
writing.
Enough has been said above to indicate how intimately Saussure's

thinking about writing is intertwined with his thinking about signs
in general. That is doubtless what one would expect of a theorist
whose knowledge of languages was so indebted to the study of
ancient texts (as was that of most nineteenth-century European
linguists). Saussure, however, is also regarded as a theorist who
reinstated the notion of the essential orality of language and
emphasized the importance of not equating languages with their
written texts. He accuses the comparative philologists of his day
of working with a mistaken perception of the relationship
between speech and writing. He clearly holds no high opinion of
the way they had treated the written evidence on which they based
their reconstructions of linguistic history. His comments on their
failure to grasp that an oral tradition is not always faithfully
reflected in surviving documents are revealing.
The first linguists were misled in this way, as the humanists
had been before them. Even Bopp does not distinguish clearly
between letters and sounds. Reading Bopp, we might think that
a language is inseparable from its alphabet. His immediate
successors fell into the same trap.20
20
Saussure, op. cit., p.46. Franz Bopp (1791-1867) published a compara-
tive study of the conjugation systems of Sanskrit, Greek, Latin, Persian
and Germanic in 1816. This is often regarded as the beginning of
comparative Indo-European linguistics.
The implications of these remarks should not be under-

estimated. Not distinguishing clearly between letters and sounds is
an accusation that could be levelled at many of Bopp's predeces-
sors, going all the way back to Plato. It is the traditional 'literate'
perspective of Western thinking about language that is being
questioned here. What Saussure is saying is that failure to under-
stand the relation between speech and writing is responsible for
failure to understand what language is. In other words, it is
not just a question of correcting a misguided view of writing
but of setting our whole thinking about language straight. The
semiology of writing is about much more than writing.
For Saussure, the blind spot in traditional Western thinking about

writing (and other forms of communication) is a failure to grasp
the systematicity involved. Writing, for Saussure, is not just an ad
hoc appendage to speech. Writing systems are systems in their own
right, even though they subserve or supplement forms of oral
communication. For Saussure, the beginning of wisdom here is to
grasp the (semiological) fact that writing and speech depend on
different t systems of signs. The writing system is not just a copy or
mirror image of the speech system, and cannot be.
There is nothing in the Aristotelian semiology of writing that
recognizes this. On the contrary, if we take Quintilian or the
grammarians of Port-Royal as representative practitioners of the
Aristotelian theory of writing it seems that one could add to or
simplify one's inventory of written signs almost at will, depending
on the decisions about whether or not it was worth 'recognizing' a
particular sound (i.e. providing it with a 'letter' or mark of its
own). Quintilian's view, as noted in the previous chapter, was that
the Latin alphabet should provide separate letters for all the dis-
tinguishable sounds of speech in Latin, but no more.
For Saussure, the assumption that one might simply 'add' or
'remove' letters as required would be not only a failure to realize
the systematicity of writing, but in addition a manifestation of
closet 'nomenclaturism'. Nomenclaturism, for Saussure - as later
for Wittgenstein - was the original sin of Western thinking about

signs. It is no accident that the foundational mythology of lan-
guage in the Western tradition is the Biblical story of Adam nam-
ing the animals. This is a profoundly nomenclaturist account. The
creatures exist before they are named: Adam merely supplies a
separate vocal label for each. The 'vocabulary' thus created can
extend indefinitely, for as long as Adam can think of new names
for new creatures. The only constraint on the Adamic nomencla-
turist (presumably) is that one should avoid giving different crea-
tures the same name, since that would defeat the purpose of the
exercise. Adam did not say to God 'Why not just call them all the
same?' (That would have been the first semiological heresy.)
Although Saussure does not single out individual linguists by
name, it is not difficult to find examples of naive nomenclaturism
in the work of his contemporaries. A blatant case is one of the
most eminent American linguists of the day, WD. Whitney of
Yale. According to Whitney the basis of all language is 'the name-
making process'. This, says Whitney, is very easy to understand.
When a human being is born into the world, custom, founded
in convenience, requires that he have a name; and those who
are responsible for his existence furnish the required adjunct,
according to their individual tastes, which are virtually a reflec-
tion of those of the community in which they live.21
According to Whitney, this is merely a conspicuous example

('only with variety in the degree of consciousness involved') of the
process of name-making in all its varieties. We give names not
only to people, places and things but to the many and various
conceptions our mind entertains.
First, there is always and everywhere an antecedency of the
conception to the expression. In common phrase, we first have
our idea, and then get a name for it. This is so palpably true
of all the more reflective processes that no one would think of
denying it; to do so would be to maintain that the planet, or
21
Whitney, op. cit. p. 135.
plant, or animal, could not be found and recognized as some-

thing yet unnamed until a title had been selected and made
ready for clapping upon it; that the child could not be born
until the christening bowl was ready. But it is equally true, only
not so palpable, in all the less conscious acts, all the way down
the scale to the most instinctive.22
Quintilian's approach to the alphabet might be described as

naive nomenclaturism of this order. (Another sound? No problem.
Invent another letter.) It would be comforting to think that naive
nomenclaturism was dead; but it has continued to flourish in
various guises in all kinds of scholarly thinking about writing
during the late twentieth century.
In what respects did Saussure think that naive nomenclaturism

failed to recognize the systematicity of writing? A diversion, which
will be less of a diversion than it may seem, is necessary at this
point in order to put Saussure's concept of a writing system in its
proper theoretical perspective.
Systems, for Saussure, create their own semiological values.
That may be as concise a way as any of summing up the approach
to culture that was subsequently called 'structuralism'. For Saus-
sure, as a pioneer of structuralism, society would be impossible
without systems of signs, of which the most important are
linguistic signs. In the Cours Saussure explains his concept of
systematicity - i.e. of'structure' - by constantly invoking compari-
sons between language and chess. There are three passages from
the Cours that are worth examining in detail in this connexion.
(1) The language itself is a system which admits no other order
than its own. This can be brought out by comparison with the
game of chess. In the case of chess, it is relatively easy to dis-
tinguish between what is external and what is internal. The
facts that chess came from Persia to Europe is an external fact,
22
Whitney, op. cit.,p.!37.
whereas everything which concerns the system and its rules is

internal. If pieces made of ivory are substituted for pieces made
of wood, the change makes no difference to the system. But if
the number of pieces is diminished or increased, that is a
change which profoundly affects the 'grammar' of the game.23
(2) But of all the comparisons one might think of, the most
revealing is the likeness between what happens in a language
and what happens in a game of chess. In both cases we are
dealing with a system of values and with modifications of the
system. A game of chess is like an artificial form of what
languages present in a natural form.
Let us examine the case more closely.
In the first place, a state of the board corresponds exactly to
a state of the language. The value of the chess pieces depends
on their position upon the chess board, just as in the language
each term has its value through its contrast with all the other
terms.
Secondly, the system is only ever a temporary one. It varies
from one position to the next. It is true that the values also
depend upon one invariable set of conventions, the rules of the
game, which exist before the beginning of the game and remain
in force after each move. These rules, fixed once and for all, also
exist in the linguistic case: they are the unchanging principles of
semiology.
Finally, in order to pass from one stable position to another
or, in our terminology, from one synchronic state to another,
moving one piece is all that is needed. There is no general
upheaval. That is the counterpart of the diachronic fact and all
its characteristic features. For in the case of chess:
(a) One piece only is moved at a time. Similarly, linguistic
changes affect isolated elements only.
(b) In spite of that, the move has a repercussion upon the
whole system. It is impossible for the player to foresee exactly
23
where its consequences will end. The changes in values which

result may be, in any particular circumstance, negligible, or
very serious, or of moderate importance. One move may be a
turning point in the whole game, and have consequences even
for the pieces which are not for the moment involved. As we
have just seen, it is exactly the same where a language is
concerned.
(c] Moving a piece is something entirely different from the
preceding state of the board and also from the state of the
board which results. The change which has taken place belongs
to neither. The states alone are important.
In a game of chess, any given state of the board is totally
independent of any previous state of the board. It does not
matter at all whether the state in question has been reached by
one sequence of moves or another sequence. Anyone who has
followed the whole game has not the least advantage over a
passer-by who happens to look at the game at that particular
moment. In order to describe the position on the board, it is
quite useless to refer to what happened ten seconds ago. All
this applies equally to a language, and confirms the radical
distinction between diachronic and synchronic. Speech oper-
ates only upon a given linguistic state, and the changes which
supervene between one state and another have no place in
either.
There is only one respect in which the comparison is defec-
tive. In chess, the player intends to make his moves and to have
some effect upon the system. In a language, on the contrary,
there is no premeditation. Its pieces are moved, or rather modi-
fied, spontaneously and fortuitously. [. . . ] If the game of chess
were to be like the operations of a language in every respect, we
would have to imagine a player who was either unaware of
what he was doing or unintelligent.24
(3) Consider a knight in chess. Is the piece by itself an
24
Saussure, op. cit, pp. 125-7.
element of the game? Certainly not. For as a material object,

separated from its square on the board and the other conditions
of play, it is of no significance for the player. It becomes a real,
concrete element only when it takes on or becomes identified
with its value in the game. Suppose that during a game this
piece gets destroyed or lost. Can it be replaced? Of course it
can. Not only by some other knight, but even by an object of
quite a different shape, which can be counted as a knight, pro-
vided it is assigned the same value as the missing piece. Thus it
can be seen that in semiological systems, such as languages,
where the elements keep one another in a state of equilibrium
in accordance with fixed rules, the notions of identity and value
merge.23
If writing depends on writing systems, all this applies pan passu

to writing. Thus the implications of (Saussurean) systematicity
might be spelt out as follows.
1 Writing is not merely making marks, even if the marks happen
to conform to what readers would recognize as a written form.
Playing chess is not just making random moves on the board,
even if they correspond to those recognized by the rules of
chess. Chess is a system. So is writing.
2 Writing does not depend on the shapes of the marks. Any other
shapes could be substituted, provided they were accorded the
same semiological values in the writing system. Just as we could
substitute a button for a chess piece, so we could replace a
letter-form by an asterisk (provided in both cases we did so
'systematically').
3 The identity (and value) of any written character depends on its
differentiation from other characters in the same system.
(Hence Saussure's example of various ways of writing the letter
t as a paradigm case of semiological identity.) It does not matter
whether one knight is exactly the same size and shape as
2j
Saussure, op. cit, pp. 153-4.
another knight, provided it is recognizably distinct from a king,

a queen, a bishop, etc.
4 If we introduce extra characters into our alphabet, or eliminate
others, that automatically restructures the entire writing system.
For there are now new combinations that have to be recog-
nized, new combinatorial possibilities, new spellings.
5 In analysing a writing system, its past history is irrelevant (the
'previous states of the board'). All that matters is the set of
characters currently in use ('in play') and their respective values.
This structuralist semiology of writing challenges the tradi-

tional Western notion at a very basic level, since questions about
formal contrasts within the visual system now take priority over
questions about what 'sounds', if any, the units of the system are
paired with. That is, in any case, left open; for, unlike Aristotle,
Saussure was also concerned to account for writing systems that
cannot plausibly be regarded as operating on the basis of pairing
letters with sounds at all.
The trouble with Saussure's semiology of writing is that, as

applied by Saussure himself, it does not deliver on its promises.
His handwriting example is a case in point. For Saussure, a sign
pairs a form (signifiant] t with a meaning (signifie} in the mind of
whoever is using the system of signs in question. Now Saussure's
handwriting example takes it for granted that the letter t desig-
nates a specific linguistic unit; that the former functions as the
signifiant of the latter. But nowhere does he explain the process of
semiological analysis which led to this conclusion. The question is
rather crucial; for this linguistic unit supposedly corresponding to
the letter t is itself a theoretical abstraction. (In actual practice, the
letter t appears to answer to a variety of possible pronunciations.)
Furthermore, there are different possible ways of identifying this
linguistic unit. How in practice do we identify it? And what of the
numerous cases where the letter t appears in the orthographic
form of a word, but no 'sound [t]' is heard in the corresponding
pronunciation? From a semiological point of view these are

serious questions.
The difficulty does not stop there, however. For exactly parallel
questions arise with regard to 'the letter f itself. Are we quite sure
that this letter in fact is a Saussurean signifiant? Saussure evidently
assumes so, but again gives no semiological justification for the
assumption. This emerges as a problematic lacuna when we con-
sider his examples of the different handwritten variants which the
letter may have. By citing these examples Saussure evidently
wishes to demonstrate that the semiological status of a sign does
not depend on the intentions of its users. In other words, what
links the three variants - indeed makes them semiological variants
- of a single letter is not the fact that the writer intended them to be,
nor the fact that they resemble one another, but the (negative) fact
that in each case the form is distinct from that of an /, a d, etc. ('In
the language itself, there are only differences.'26) So that is 'what
makes it' a t. The identity of the written signifiant and the link
between its variants are — for Saussure — aspects of one and the
same semiological fact.
This much is clear. But it is nevertheless open to question, and
on more grounds than one. First, within a Saussurean theoretical
framework the identification of signs requires a simultaneous co-
determination of both signifiant and signifie. Neither is given in
advance. It is not a question of starting out with a list of signifi-
ants and looking around to find their signifies; nor vice versa. We
are not matching up pairs from separate inventories already
known.
Once this is realized, the example of the three variants of the
letter t becomes particularly awkward for the Saussurean semiolo-
gist. It is an example which appears to suggest that all that needs
to be taken into account is the visual configuration of the marks.
The comparison in question is a visual comparison between these
marks and the only relevant question is whether or not any given
configuration is visibly distinct from the possible variants of /, of d,
26
etc. What remains unexplained is how to make sense of this task (i.e. of
identifying the variants of t) without the assumption that there is
indeed a finite list of letters (including t} determined in advance.
Otherwise we are sifting through the visual haystack for a needle
that may not be there.
How might Saussure have responded to this objection? Pre-
sumably he would not have wished to say that any identification
of the variants of t proceeds on the assumption that we know
there are only a certain number of basic sounds ([t], [d], [1],
etc.) to be represented by any alphabet. That would be a disas-
trous answer from a Saussurean point of view. It would
immediately introduce a concrete and 'positive' factor into the
situation, undermining straight away the claim that the only
thing that counts is avoiding possible confusion between letter-
forms. Thus the last thing Saussure would concede is that we
identify variants of the same letter simply on the basis of
whether or not they all have the same pronunciation. For that
destroys any need to appeal to his somewhat more mysterious
notion of semiological identity defined negatively in terms of
differences.
On the other hand, if the signifie is ignored and all that matters
are similarities and differences between visible marks, it is difficult
to see why anyone should be driven to the conclusion that the
three examples we are offered in the Cours are all versions of a
single letter. Working on the Saussurean principle that a single
sign cannot have more than one signifiant, four other conclusions
are possible: (i) that the first two marks are variants, while the third
represents a separate letter, (ii) that the first and third marks are
variants, while the second represents a separate letter, (iii) that the
second and third marks are variants, while the first represents a
separate letter, and (iv) that all three forms represent different
letters. What we are not told is how to choose between these possi-
bilities. In short, visual similarities and differences alone fail to
yield adequate criteria for the semiological concept 'graphic vari-
ant'. And this is precisely the concept on which the plausibility of
Saussure's example rests.
The concept becomes even more problematic when we

reflect that Saussure's example gives us no information about
the language (English? French? Spanish?) in which the hypo-
thetical text in question is written. Given that not all languages
using alphabetic scripts employ the same inventory of letters,
one might have expected Saussure, in accordance with his own
structuralist principles, to claim that the semiologist needs to
distinguish between a different 'letter f in each such case (on
the ground that the sum total of contrasts within each
system of writing will be different). But no such proviso is men-
tioned in the text of the Cours. The reason why is not difficult
to fathom. That would have ruined the immediate plausibility
of the example, which depends entirely on our lay acquaint-
ance with the fact that handwriting, unlike printed text, does
not reproduce uniform shapes of each letter. We grasp the
point about variants of t without even bothering to inquire
whether it is an English t, a French t, a Spanish t, or any
other t.
From a strictly Saussurean perspective, however, that ought to
matter. It is all very well to claim that Values in writing are solely
based on contrasts within a fixed system, having a determinate
number of letters'; but that does not explain how the number of
letters is determined. To put the point another way, there is no
guarantee that the number of distinct letters in the (semiological)
system corresponds to the number of letter-names currently in
use. (If that could be safely assumed, being a semiologist of writing
would be a sinecure.) The fact that a lay observer might call three
distinct graphic shapes all '/' does not necessarily reflect anything
other than a certain nomenclature derived from the observer's
education: in itself it proves nothing about the semiological
structure of the writing system or about the number of signs it
comprises. Indeed, the lay nomenclature of 'letters' traditionally
permits each letter-name to be applied right across the range of
such distinct shapes as 'the capital letter', 'the small (or lower-
case) letter', "italics', etc. But it is open to question whether all
these function semiologically as 'the same sign'. That might
perhaps depend on the text in question and the purpose of the

analysis.27
The critical point here is that as soon as the possibility of dis-
covering alternative semiological analyses of alphabetic writing is
taken seriously, that automatically undermines any assumption
that the alphabet is 'a fixed system, having a determinate number
of letters'.
It seems difficult to reconcile the notion that 'the alphabet' is
itself a system of signs, determinate in number and having
determinate phonetic values, with the notion that alphabetic units
are to be identified semiologically only on the basis of contrastive
relations. To take an obvious example, in written English the
graphic shape we call 'capital T' occurs initially in names like Tom
which begin with the dental stop [t], but also in names like Thelma
which have no initial [t]. Nevertheless, no literate English person
would claim on the basis of pronunciation that the names Tom and
Thelma are not spelt with the same initial letter. That same letter is
recognized as occurring as the initial letter of the first word in this
sentence. The same word ends with a 'small' t. What is the semio-
logical relation between the 'small' and the 'capital' letter? Clearly
they are not interchangeable, since in English orthography we do
not encounter spellings like thaT. Nor, in general, are 'capital let-
ters' found in word-final position in English unless the whole word
is written 'in capitals': e.g. THAT. These and similar facts about
the dissimilar distributions of T and t in written English, quite
apart from their coincidence and non-coincidence with the occur-
rence of the consonant [t] in corresponding spoken forms, make it
clear that the question of determining how many alphabetic units
we are dealing with, and the question of giving an exact semio-
logical definition of each, is far more complicated than at first
sight might appear. It might even turn out that neither our
27
The Port-Royal grammarians had already pointed out the usefulness
of the distinction between capital letters and 'small' letters, even though
that difference corresponds to nothing in pronunciation (Grammaire
generate et raisonnee, I,v).
traditional 'capital T nor our traditional 'small f unambiguously

represents such a unit.
Evidently, analysing the spelling of English — as even this one
example shows - involves considering similarities and differences
which simply do not arise in the analysis of spoken English. What,
then, are we to make of Saussure's contention that writing exists
solely in order to represent speech? What semiological status can
be assigned to written distinctions that have no counterpart in oral
communication?
Finally, what of Saussure's contention that 'the actual mode of
inscription is irrelevant'? A letter t in black or in white, in ink or in
chalk, is still a letter t. Quite so. But that is a mere tautology, not a
principle of semiological analysis. From the semiologist's point of
view, a black letter or a white letter, an ink mark or a chalk mark,
may well be different signs with quite different significations.
The further we pursue Saussure's claims about writing, the
more inextricably we find him enmeshed in a semiological snare
of his own making. By insisting both that writing systems are semi-
ologically separate systems of signs from those of speech and, at
the same time, that written signs are merely metasigns which serve
to signify the signs of speech, he creates a dilemma for himself. A
theorist can opt for one or the other: what is disastrous is trying to
have it both ways at once.
CHAPTER THREE
Writing off the Page
Anyone satisfied neither with Aristotle's semiology nor with Saus-

sure's is inevitably led to look for another theoretical framework
within which to rethink writing. There are not many options to
choose from.
One possibility would be a framework based on C.S. Peirce's
theory of the sign. This would have to be constructed 'on Peirce's
behalf, so to speak; for in Peirce's voluminous (mure there is
remarkably little on the subject of writing. For Peirce, written signs
are 'symbols', but not symbols in the Aristotelian sense.
A Symbol is a Representamen whose Representative character
consists precisely in its being a rule that will determine its Inter-
pretant. All words, sentences, books, and other conventional
signs are Symbols. We speak of writing or pronouncing the
word "man"; but it is only a replica, or embodiment of the word,
that is pronounced or written. The word itself has no existence
although it has a real being, consisting in the fact that existents
will conform to it.'
There are already enough metaphysical problems in this pas-

sage alone to make it clear that constructing a Peircean 'semiotics
1
J. Buchler (ed.), Philosophical Writings of Peirce, New York, Dover, 1955
p.112.
Writing off the Page 65
of writing' would be a considerable task in itself. It will not be

attempted here. Peirce defines a sign as 'an object which stands for
another one to some mind', or, even more vaguely, as 'something
that stands to somebody for something in some respect or
capacity'. The visible letters in ink on a page (which Peirce calls
'tokens') are in his view signs of, i.e. stand for, invisible graphic
abstractions (which he calls 'types'). But the formula 'x stands forjy'
does not seem to take us any further than Aristotle's lx symbolizes
y. It merely defers the problem. What now has to be explained is
the 'standing for' relation, just as in Aristotle's case what remains
to be explained is the symbolic relation. Peirce's sign is in any case
pre-Saussurean, and reverting to a pre-Saussurean model of the
sign is not an attractive move if we are looking for a framework
which will provide a critical perspective on both nomenclaturist
and structuralist accounts of writing.
The only current framework that has no prior nomenclaturist
or structuralist commitments at all is that of integrational semio-
logy.2 Integrational semiology matches the scope and generality of
both Saussurean semiology and Peircean semiotics, inasmuch as it
covers the whole range of human communication. At the same
time, it leaves open exactly the questions about writing that Peirce
deferred and Saussure foreclosed. On both counts it provides a
viable independent basis for rethinking writing.
For an integrationist, however, the problem of glottic writing
has to be seen in a quite different way from the way Saussure,
Peirce and their predecessors saw it. For integrational semiology
takes a radically different view of what a sign is.
Aristotelian and Saussurean thinking about signs share a serious

flaw in common: both start at a level of abstraction where signs
are already decontextualized theoretical items. The question of
the 'working conditions' under which signs actually function as
2
R. Harris, Signs, Language and Communication, London, Routledge, 1996;
R. Harris, Introduction to Integrational Linguistics, Oxford, Pergamon, 1998.
signs is already bypassed. Essential to this level of abstraction is a

dualist conception of the sign, whereby it is taken for granted that
what identifies any given sign is the pairing of a 'form' with a
'meaning'. The meaning of the sign is what it signifies: the form is
the signal or indicator of that meaning. This binary relationship
itself — and this alone — identifies the sign. A sign, on this view,
may - or may not (in some versions) - have more than one form or
more than one meaning. Nevertheless, differences of form or of
meaning or of both are what distinguish one sign from another.
On the dualist view, signs are distinct from one another only on
condition that they differ in at least one of these respects.
As Michel Foucault observes, dualism (or the 'binary theory of
the sign', as he calls it) is one of the essential foundations of
Western theories of 'representation'. Its ramifications extend far
beyond the domain of language. Saussure was in this respect
a very conservative revolutionary. Thus although Saussure's
definition of the sign
may have appeared to be 'psychologist^' [. . . ] in fact that was
a way of reinstating the classic condition for conceptualizing
the binary nature of the sign.3
Dualism leaves room for many different interpretations of what

counts as form and what counts as meaning; but they all have in
common the notion that in order to recognize the sign we have to
know which forms go with which meanings. The relationship
between form and meaning is usually assumed to be established
by social convention, at least for 'arbitrary' signs, including those
of language. (Aristotle says explicitly that names are established by
convention, no sound being by nature a name.4 Saussure describes
la langue as 'a body of necessary conventions adopted by society'.5)
Thus the flag at half mast, the red traffic light, the 'EXIT' notice,
3
M. Foucault, Les Mots et les chases, Paris, Gallimard, 1966, p.81.
4
De Interpretation, 16A 20-30. Whitaker (op. cit. p. 19) points out that this
is confirmed by what Aristotle says in De Sensu.
3
etc. are all assumed to be forms to which a public meaning has

been assigned by social convention. The traditional assumption is
that anyone who does not grasp the relationship between the form
and the meaning does not understand the sign (and is to that
extent not a communicationally competent member of that soci-
ety). What this dualist way of conceptualizing signs does not
explain is how a sign ever came to signify what it (supposedly)
does. (Aristotle never addresses this question: Saussure, as noted in
the previous chapter, explicitly refuses to address it.)
The integrationist theory of signs begins one stage further back
and invites consideration of precisely that question. In other
words, it does not assume that one can take 'form' and 'meaning'
for granted as established. Why not? Because, for one reason, it
leaves the theorist with no satisfactory way of dealing with cases
where two people disagree about what the sign signifies. The dual-
ist cannot afford to concede that whatever people think a form sig-
nifies is what it signifies; for if everybody is right automatically, that
amounts to abandoning the social-conventionist view altogether.
The alternative is to claim that some people are (sometimes) just
wrong about (some) signs. But that involves spelling out how we
establish the 'right' meaning for any given sign. It is no use saying
at this point that the right meaning is the meaning established by
the social convention, because that is just what is at issue. It is
possible for two people to disagree about what the social conven-
tion is. So do we hold a referendum? Do we defer to some higher
authority? Do we take it that whoever produced the sign in ques-
tion is the person who knows what it means? Or hand over the
decision to whichever of the people in the dispute has a gun? Or
how, exactly? These are unattractive options for the dualist, and
whichever one is chosen leads to further conundrums of its own.
The integrationist proposal is that dualist assumptions about
form and meaning do not provide a very good basis for approach-
ing these questions in the first place. In order to understand signs
and signification, we have to begin at a much more elementary
level of human behaviour. We certainly do not have to start with
the notion of a social convention already in situ. For our own
experience tells us that we attribute significations to things and

events, irrespective of whether there is any social convention
about the matter or not. Signs do not necessarily have a social
dimension at all. Here, straight away, there is a fundamental
difference between integrational semiology and Saussurean semio-
logy; i.e. the integrationist does not accept that one must recog-
nize as semiological 'only that part of the phenomena which
characteristically appears as a social product'. On the contrary,
the integrationist would argue that unless semiology starts below
the social level it will never be able to explain publicly recognized
signs at all.
Take, for example, the case of the familiar landmark. There
are doubtless landmarks that almost everybody in a community
recognizes. But there may also be landmarks recognized by certain
individuals only. I may look out for a particular tree, knowing that
I have to take the first turning on the left after that tree on my
usual way home. (On reaching the tree, I change down into a
lower gear, move into the left-hand lane, etc.) Thus for me the tree
signifies something, has a certain semiological value. Perhaps it
does for others too; but that is strictly irrelevant. It is a landmark as
far as I am concerned, and that is already sufficient. Its value as a sign
arises simply - and solely - from the fact that I rely on it to
integrate certain programmes of activity in my daily comings and
goings. In terms of integrational theory, the tree thereby acquires
an integrational function, i.e. becomes a sign, in virtue of the role
it plays in those activities. Outside that framework, it has no
semiological status (unless it plays a comparable role in some other
programme of activities). But as far as I am concerned it requires
no co-operation from anyone else; that is, the 'tree' sign does not
depend on my interaction or agreement with another person, any
more than tying a knot in my own handkerchief does.
It is important to note that none of this means (a) that somehow
the physical tree has now become a 'form' with its own 'meaning'
(e.g. 'Take the next left'), or (b) that one object, the tree, now
'stands for' another object, the first turning on the left, or (c) that a
mental image of the tree, tagged with the conceptual 'take-the-
next-left' label, has now been added to my brain's stock of equip-

ment. To refer to the tree as a sign — at least in the sense that
integrational semiology construes that term — implies simply that I
recognize and contextualize it in a certain way in relation to cer-
tain activities. (How I manage to do that is another question, but it
is a question for neuropsychologists, not for semiologists.) The tree
is a sign only insofar as I make it a sign.
Signs, for the integrationist, provide an interface between dif-
ferent human activities, sometimes between a variety of activities
simultaneously. They play a constant and essential role in integrat-
ing human behaviour of all kinds, both publicly and privately, and
are products of that integration. Signs are not given in advance,
but are made. The capacity for making signs, as and when required,
is a natural human ability. Some signs (e.g. the landmark) are
recurrently useful and are constantly being remade as often as
required; but others may serve the purpose just for one particular
occasion (as when the reader turns down the corner of the page
before putting the book aside, in order to mark the point at which
to resume reading next time). Signs do not have any superhuman
capacity for outliving their makers. When languages die, it is
because no one is any longer engaged in remaking them. But this
is the case for all signs.
Signs, therefore, in an integrationist perspective are not invari-
ants', their semiological value depends on the circumstances and
activities in which, in any particular instance, they fulfil an inte-
grational function. Thus even for me the semiological value does
not somehow remain permanently attached to my landmark tree.
If I change my usual route home in order to avoid the traffic,
looking out for the tree may cease to play any orientational role at
all in my daily journeys. Its semiological value will then lapse as far
as I am concerned: it will no longer be a sign.
If the basic process by which signs are created and function is as
the integrationist construes it, then the notion of a sign which
integrates the activities of two or more individuals is not difficult
to establish. And from there the notion of a sign with a common
(i.e. public) value is not too difficult either. What it requires
minimally is that A assign a semiological value to x, and B assign a

semiological value to x , and that A and B both carry out mutually
integrated programmes of activity on that basis. The tree, for
example, can become a landmark for you too. Suppose I give you
directions to my house: 'Take the first left after the tree . . . etc.'.
The fact that I told you about the tree does not affect the point at
issue. What matters is how you integrate spotting the tree into
your journey. What makes the tree a landmark/brjow is the use you
make of it in finding your way to where I live. The difference
between us is simply that I worked it out for myself but you did
not. But that does not somehow rob the tree of its landmark
function in your case. That function is put to the test, i.e. estab-
lished, for both of us by exactly comparable procedures. There is
no philosophical puzzle about 'sameness' here: we demonstrate the
integration of our activities by both ending up in the right place
on the basis of planning our route by reference to the tree in
question. (There could be various reasons why this does not work
out - roadworks, a puncture, a heart attack - and also possible
scenarios in which either you or I end up at the right place in
spite of having missed the tree, but these are irrelevant to the
semiological point.)
As this example illustrates, from an integrational point of view
the difference between a private sign and a public sign is not
particularly puzzling. Public signs are public because more people
are involved: i.e. more people assign a semiological value to cer-
tain things or to certain practices and integrate their activities
accordingly. And the more people do this the more they take it for
granted that other people are familiar with these signs too. But
what the signs signify is established in exactly the same way,
irrespective of how many or how few people are involved. Thus
the appeal to social conventions which underlies both nomencla-
turist and structuralist theories of the sign is in the end a red
herring. We do not have to start by presupposing that the relevant
social conventions are already in place.
The integrationist approach to writing is based on the assump-
tion that this applies just as much to the written sign as to any
other kind of sign. The implications of this are far-reaching.

Whereas Saussure assumes, in common with his predecessors in
the Western tradition, that in order to explain the written sign we
have to start from the spoken sign, as already established in some
public code (i.e. the relevant oral language), the integrational
semiologist makes no such assumption. Integrational semiology
offers no warrant for believing that written signs bear some kind
of constant relation to spoken signs anyway. On the contrary, the
integrationist will assume that in any semiology of writing we
need to begin with the written sign itself.
Thus, to illustrate the difference in terms of Saussure's own
example, the identity of a 'letter £ in a particular instance is not
established by reference to the internal contrasts in the writer's
habitual handwriting. For an integrationist, the fact that the writer
has fulfilled all the conditions Saussure mentions (forming his t
differently from his /, his d, etc.) does not prevent someone else,
particularly someone unfamiliar with the writer's hand, from tak-
ing this particular mark for an / and thus, say, reading seat as seal.
Exactly how important this might be will depend on the circum-
stances of the case. But however important or unimportant, the
fact remains that there is only one mark on the paper, but there
are two written signs: one which the writer identifies as 'f and
another which the reader identifies as '/'. The mark itself has no
semiological value other than that attributed to it by writer or
reader; and that value depends on how the written message inte-
grates communication between them. In sum, integrational func-
tion, and that alone, is the criterion for establishing what a sign is.
At first sight this might not seem to yield a very different analy-
sis of the handwriting example from Saussure's. However, it does
differ on three important counts.
1 According to the integrationist analysis, the conflicting inter-

pretations of the mark remain unresolved, unless settled by
negotiation between reader and writer. ('I can't read your writ-
ing. Is it a t or an / at the end of this word?') But (a) it may not
occur to the reader that there is any doubt about the identity of
the letter, so the question may never be asked, and (b) any such
negotiation with the writer is itself a further attempt to inte-
grate the mark in question into some communicational pro-
gramme. The mark is now, in effect, recontextualized (by the
questions the reader asks and the writer's response).
2 In the integrationist account, there is no appeal to the semio-
logical systematicity supposedly manifested in the writer's
formation of the letter, i.e. no recourse to structural criteria for
resolving the problem. (Asking the writer is a pragmatic not a
structural solution. The writer may just insist that it is a t even
if it is patently indistiguishable from an / in the very next
word.)
3 For the integrationist, there is no question of whether it 'really'
is a t or not. In other words, the form of the sign is itself
indeterminate.
It is this third point which brings out the most radical difference
between the Saussurean and integrationist positions. In Saus-
surean structuralism, a sign is always a doubly determinate unit
(determinate in both form and meaning); for that is a condition of
its existence within the system to which it belongs. Integrational
theory, by contrast, treats the intrinsic indeterminacy of the sign as
the foundation of all semiological analysis.6
The integrationist thus takes a quite different view of human

communication from that which underlies traditional Western
thinking on the subject. The basis of traditional thinking is what
integrationists refer to as the 'fixed-code' fallacy. This is the notion
that communication (in particular, linguistic communication)
depends on the establishment of publicly recognized systems of
6
For further discussion of the indeterminacy of the sign, see R. Harris,
'The integrationist critique of orthodox linguistics', Integrational Lin-
guistics: a First Reader, ed. R. Harris and G. Wolf, Oxford, Pergamon,
1998, pp. 15-26.
correlation between forms and meanings. Thus 'a language'

(English, French, Swahili, etc.) is assumed ab initio to be
a fixed code which, by relating entities in a dimension called
'form' to entities in a dimension called 'meaning', provides
language-users with a means of transmitting and receiving
thoughts.7
This in turn relies on the traditional assumption that communi-

cation is telementational, i.e. a process of thought-transference
from A's mind to B's mind.8 The essential role of the (linguistic)
sign is seen as being to facilitate that mental transfer. Thus the
fixed-code fallacy and the doctrine of telementation feed off each
other. For if communication is to be successful, the internal logic
of the process requires that B's mind shall eventually 'receive' the
thought that set out from A's mind. If B's mind receives some
different thought, or no thought at all, then communication has
broken down somewhere along the line. As Locke put it in his
classic formulation of the telementational thesis,
[men] suppose their words to be marks of the ideas in the
minds also of other men, with whom they communicate: for
else they should talk in vain, and could not be understood, if
the sounds they applied to one idea were such as by the hearer
were applied to another, which is to speak two languages.9
Locke's statement provides a brilliantly concise synopsis of the

communicational implications of Aristotle's semiology. The 'affec-
tions of the soul' have to be the same in both speaker and hearer
for verbal communication to be successful. The difference is that
for Locke this is a problem, whereas Aristotle circumvents the
7
N. Love, 'The fixed-code theory', in R. Harris and G. Wolf (eds.),
Integrational Linguistics: a First Reader, Oxford, Pergamon, 1998, p.55.
8
For an integrationist view of telementation, see M. Toolan, 'A few
words on telementation', in R. Harris and G. Wolf (eds), Integrational
Linguistics: a First Reader, Oxford, Pergamon, 1998, pp.68-82.
9
Locke, op. cit., Ill, ii.
problem by simply assuming that the 'affections of the soul' are

the same for all.
As Talbot Taylor notes,
the telementational picture of communication, passed down to
us from Locke, takes the understanding of an utterance to be
an unobservable, private, mental event.10
If we take this privacy condition seriously, Taylor argues, we are

led inevitably to the paradoxical conclusion that 'we can never
know if our hearers understand what we say to them'. Here, in
effect, Aristotelian semiology (at least as far as language is con-
cerned) has its bluff called. Either the communicational question
is being dodged, or else we are missing an explanation which
Aristotle takes to be so obvious as not to require elaboration.
How does all this affect our understanding of written com-
munication? Rather profoundly. To see why we need only follow
through Nigel Love's simple example of the word postman. In
spoken English, unstressed vowels are commonly 'reduced' (as
phoneticians describe it) to such an extent that the singular postman
becomes indistinguishable from the plural postmen. In written Eng-
lish the two are always distinct, unless (for reasons such as those
pertinent to Saussure's example of the letter t) a writer's hand
blurs the visual difference between a and e. So how does that leave
the grammatical distinction between singular and plural? Do we
have one grammar for spoken English but another for written
English? To paraphrase Locke, is this a case of 'two languages'?
Love lists six possible pronunciations of the plural postmen., some of
which overlap with possible pronunciations of the singular post-
man. Some of these might only be articulated when a speaker
wished to avoid a possible ambiguity. In other words, the speaker
10
TJ. Taylor, 'Do you understand? Criteria of understanding in verbal
interaction', in R.Harris and G. Wolf (eds), Integrational Linguistics: a First
Reader, Oxford, Pergamon, 1998, p.207. The whole question of Lockean
'scepticism' about communication is discussed at much greater length in
TJ. Taylor, Mutual Misunderstanding, Durham, Duke University Press, 1992.
can always 'make' the distinction between postman and postmen

phonetically obvious when — but only when — required. Thus a
secretary taking dictation can use forms with 'unreduced' final
vowels to ask 'Was that [pous'man] or [pous'men]?'. But this is not
a question about what was said, because what the speaker said was
neither. It is a question about how the secretary is to integrate the
current scribal operation with the (speaker's) previous utterance
(and, consequentially, with any later optical scanning by readers
of the letter). And that is already a problem for any theorist who
supposes that writing is merely a 'visible' reflection of speech. For
in effect this avoidance of ambiguity reverses the canonical pri-
ority normally assumed, i.e. the assumption about the 'primacy of
speech', as well as undermining the thesis that a language is a
fixed code.
This is why the structuralist language-describer is reluctant to
include within the scope of his description deviations from
canonical forms brought about by awareness of communi-
cational difficulties that might arise from use of the canonical
forms in particular circumstances. For if he were obliged to take
account of such variants, the task of exhaustively enumerating
or in some sense 'defining' a language's stock of form-meaning
pairs would be impossible.11
Anyone who realizes the implications of Love's point will see

that it casts doubt upon a whole set of traditional assumptions
about relations between speech and writing. What it means is that
in a literate society both speakers and writers are free to opt
between alternative ways of trading off speech against writing
and vice versa, not by special dispensation but as a matter of course.
These are, indeed, integrated forms of communication and not just
parallel but separate forms of communication which happen to be
available. Furthermore, the disparity between the grammar of
singularity and plurality reopens the whole question of how these
could fojust parallel forms of communication. From this point on,
" Love, op. cit, p.55.

thinking about writing becomes a different enterprise. Or, at least,

it should do.
But traditional beliefs die hard, and in Saussure's case we have
a remarkable example of the lengths to which theorists will some-
times go to defend them. Saussure was just as well aware as Quin-
tilian that there are cases in which written alphabetic forms do not
'match' the corresponding spoken forms. But instead of conclud-
ing from this that what needs rethinking is the relationship
between the two, Saussure puts the entire blame on inadequate
spelling systems. He complains about their 'inconsistencies' (in-
consequences), 'aberrations' (aberrations] and 'irrational spellings'
(graphics irrationnelles}. For him it is 'irrational' to spell êttel in
German with a double consonant, because it is spelt that way
'simply in order to indicate that the preceding vowel is short and
open'. Similarly, made in English creates the misleading appearance
of having a second syllable, whereas in fact the final -e is simply an
indication of the pronunciation of the preceding vowel. Even
worse are cases like bourru, sottise and souffrir in French, where the
spelling shows 'illegitimate double consonants' (the doubling hav-
ing no phonetic justification at all). In short, Saussure concludes,
'writing obscures our view of the language'. It is 'not a garment,
but a disguise'.12
Saussure's arguments on this score will be considered in more
detail in the following chapter. They are well worth examining not
only because they show how Saussure allowed the traditional view
of alphabetic writing to override his structuralist principles, but
also because these are in any case arguments of the kind that have
often been deployed by linguists since Saussure's day, whether or
not they would count themselves as 'structuralists'. For inte-
grationists, this is a paradigm case of how semiological analysis
can be led astray when written forms are treated as decontextual-
ized items and simply compared, one by one, to their equally
decontextualized 'pronunciations'.
12
'pas un vetement, mais un travestissement'. Saussure, op. cit., pp.51-2.
An important piece of historical evidence in support of the inte-

grationist approach to writing is that writing tends to develop over
time a symbiotic relationship with speech. The evolution of
alphabetic writing in Europe provides some prime examples of
this symbiosis in the form of reciprocal influences. What one
might expect in any society where speech and writing come to be
closely integrated practices is that eventually the interrelationship
would be reflected in changes in both. That is to say, not only
would the way words are pronounced come to affect the way they
are written, but also the way they are written would come to affect
the way they are pronounced. This topic will come up again for
discussion in a later chapter, but it is worth commenting at this
point on Saussure's reluctance to face up to the semiological
implications of the phenomenon.
Although Saussure was well aware of these reciprocal influ-
ences, he regarded the influence of the spoken form on the
written form as natural and desirable, but the influence of the
latter on the former as pernicious. He is quick to castigate such
manifestations as exceptional, or even 'monstrous'. He fulmi-
nates against instances where the written form gives rise to
'erroneous pronunciations'. This phenomenon, according to
Saussure, is 'strictly pathological'.13 He laments the probability
that these deformations will become increasingly frequent, and
that 'more and more dead letters will be resuscitated in pro-
nunciation'.14 He waxes indignant that in Paris 'one already
hears sept femmes ('seven women') with the t pronounced. Dar-
mesteter foresees the day when even the two final letters of vingt
('twenty') are pronounced: a genuine orthographic monstrosity.'13
This remarkable tirade against the baneful influence of writing
concludes:
These phonetic distortions do indeed belong to the language
but they are not the result of its natural evolution. They are due
13
14
1
' Saussure, op. cit., p.54.
to an external factor. Linguistics should keep them under

observation in a special compartment: they are cases of
abnormal development (cos teratologiques).16
But Saussure's conception of 'teratology' touches on one

limited aspect of an area of study which includes a whole net-
work of complex relations. These are relations which inevitably
arise, as an integrationist would point out, from the coexistence
of two forms of communication — speech and writing — which
both compete and complement each other in the life of a soci-
ety. We are dealing in such cases with exactly that 'life of signs
as part of social life' that Saussure elsewhere posits as the
authentic domain of semiology. How ironic that he should rele-
gate these examples to a special laboratory for linguistic
monstrosities!
It would be a mistake, however, to suppose that this is just a
personal idiosyncrasy of Saussure's. We find the same notion of
'monstrosity' invoked in the following passages from Max Miiller's
Rede lecture of 1868:
In the natural history of speech, writing [ . . . ] is something
merely accidental. It represents a foreign influence which, in
natural history, can only be compared to the influence exercised
by domestication on plants and animals.17
This 'artificial domestication of language' by 'literary cultiva-

tion', according to Miiller, produces 'unnatural' results, just as in
the biological world, so that
however perfect, however powerful, however glorious in the
history of the world, - in the eyes of the student of language,
Sanskrit, Greek, and Latin, Hebrew, Arabic, and Syriac, are
what a student of natural history would not hesistate to call
'monstra, unnatural, exceptional formations which can never
16
17
EM. Miiller, Chips from a German Workshop, Vol.iy London, Longmans.
Green, 1875, p.74.
Writing of the Page 79
disclose to us the real character of language left to itself to

follow out its own laws without let or hindrance.18
Here it is not just spelling pronunciations but the whole literary

language which is 'monstrous'. The conception is a direct result of
a certain way of thinking about writing and written texts as
'unnatural' products.
Like Saussure, the integrational theorist proceeds on the assump-

tion that the question of graphic differences lies at the heart of
any semiological analysis of writing; but unlike Saussure, the inte-
grational theorist proceeds on the assumption that the question of
'phonetic values' of letters is pertinent only to those communi-
cational practices involving the integration of writing with speech.
Even there the first requirement is graphic differentiation - i.e. the
establishment of features which distinguish one letter or character
from another. Without an unambiguous identification of its units,
writing cannot fulfil its integrational function. This is a practical as
well as a theoretical requirement, as we are reminded every time
we try to read an illegible hand. To solve practical problems of
this order it is (in most cases) quite irrelevant to concern ourselves
about how the writer might have pronounced the word(s) in ques-
tion. That has nothing to do with the reader's difficulty, which
consists in the first instance in identifying the written forms, how-
ever incompetently or negligently they may have been inscribed.
In other words, even in glottic writing there are structural features
of the written form which are quite independent of any 'phonetic
value'.
A similar conclusion emerges if we consider the case of the so-
called 'braille alphabet' for the blind, which is not an alphabet in
the historical sense of the term, but a transcription system used to
transpose certain visual shapes into patterns of dots in relief,
thus allowing them to be 'scanned' by touch (Fig. 1). These
18
Muller, op. cit., p.74.
equivalences are established at the 'letter' level. Any phonetic or

other values that the letters may have are an irrelevance, and it is
possible to translate quite precisely into braille a text of which one
understands nothing except its alphabetic composition. Which
does not mean that braille lacks any semiological structure of its
own.19 On the contrary, the alphabetic equivalences are 'arbitrary'
inasmuch as the utilization of the six dots available in the braille
system does not correspond in any way to the shape of the letter
translated.
a b e d e f g h i j
k 1 m n o p q r s t
u v x y z w
Figure 1
What is missing from Saussure's account of alphabetic writing

is any theoretical recognition that there are two levels of structure
involved. From an integrational point of view, this is a major
lacuna, the full implications of which will be considered in the
following chapter.
19
The system also allows for a large number of contractions and
abbreviations.
Saussurean semiology, then, is a semiology of fixed codes. The

langue recognized in Saussurean linguistics is always a language
that enjoys the privilege of being 'already there', waiting to be
used, given in advance. The code itself comprises a set of decon-
textualized forms and meanings, together with 'rules' for their
combination into equally decontextualized 'messages'. The first
step towards freeing semiology from this theoretical straitjacket is
to reject the conception of context as something extra, belonging
outside the signs themselves. The basis of integrational semiology
is recognition that all communication is intrinsically context-
dependent. The sign itself is the product of contextualization.
To put it another way, what sets integrational semiology apart
from Saussurean and neo-Saussurean semiology is the premiss
that there are no contextless signs. Given this premiss, it becomes
possible to envisage a semiology of writing which is not tied umbili-
cally to the traditional belief that writing has to be understood -
and analysed - as a mode of 'representing' a spoken language.
The relations between speech and writing, from an integrational
perspective, are construed in a quite different way. First, because
an integrational semiology recognizes that there are forms and
features of writing which have nothing to do with speech at all.
Second, because in those cases where speech is one of the activities
that is integrationally relevant, its connexions with writing can
take a variety of different forms, depending on the circumstances
of the case and on the other activities involved. So there is no
simple, universal relationship between the written sign and the
spoken sign of the kind that Saussurean semiology postulates.
Signs, seen from this perspective, are not pairings of form and
meaning, already set up in some pre-established code. They arise
from actual events and circumstances in which the participants
are involved. Thus, in a particular situation, any object, action, etc.
can acquire a semiological value. Furthermore, it is not essential
that all the participants should agree on what that value is, or even
agree on which particular features of a given communication
situation have such a value. On the contrary, everyday life is full
of situations which are of such semiological complexity as to defy
any definitive interpretation, whether by the participants or by an

observer-analyst. Writing is not magically exempt from these
problems.
So the sign of integrational semiology will be not the signifiant-
signifie of Saussure (a postulated psychological pairing), nor Peirce's
token (an existing physical entity), nor Peirce's type (a purely abstract
invariant) but the working sign; in other words, the sign with all
the practical limitations, doubts and problems that its function in
an actual communication situation may bring with it. The con-
stant creation of such signs in human interaction is what Saussure
called their 'social life' - the social life that his semiology proposed
to study. But whereas the Saussurean sign is put away back in its
system after each occasion of use in parole, rather like a chess piece
being returned to its box after the game, ready to come out again
when next needed, the integrational sign is not a discrete autono-
mous entity. It has no continuous identity outside the contextual-
ization that brought it into existence. Anyone who supposes that
writing is not subject to the same ephemerality as speech is mistak-
ing the document or the inscription for the written sign. And that
mistake has been the other major obstacle (i.e. apart from the
traditional misconstrual of the relationship with speech) standing
in the way of setting up a serious semiology of writing.
An integrational perspective thus reverses the priority taken for

granted in traditional accounts of writing. Attention is focussed
not on some system of supposed correlations between written and
spoken signs, but upon the particulars of written communication
in specific circumstances. But this reversal of priorities entails
others.
Rethinking writing from an integrational point of view involves
recognizing how powerfully modern thinking on the subject has
been moulded by one of the outstanding post-Renaissance pro-
ducts of utilitarian literacy: the printed book. Written com-
munication, from an integrational point of view, is a form of
communication in which contextualized integration relies in the
Writing of the Page 83
great majority of cases on a visual framework and visual analo-

gies. The full implications of this are in part obscured because we
find it nowadays difficult to escape from the intellectual legacy of
a literacy in which the mechanically produced printed book is
presented as the paradigm case of a written text. Typically, we are
dealing with a work comprising scores or hundreds of pages,
reproduced in hundreds or thousands of more-or-less identical
copies. Typically, it is portable. It is a work designed so that its
reading shall be entirely controlled by the individual reader, who
can pick it up or put it down at will, turn the pages at leisure, go
back and forth within its confines in whatever idiosyncratic pat-
terns personal interest dictates. It will be the reader who decides
whether to read aloud or silently, when, where and - in the case of
reading aloud - to whom. In short, this paradigm case offers a
product which, of all forms of writing, is the least context-bound
and has the highest degree of autonomy (at least, if these proper-
ties are measured in terms of the distance - physical, temporal,
cultural - separating writer from potential reader). But the book is
also a somewhat unusual example of writing in that it presupposes
no particular visual framework, except - in certain cases - one
that is provided by accompanying 'illustrations'. So the visual ana-
logies that predominate are the internal analogies emerging from
the written text itself: these are the analogies that link one ink
mark to another, one configuration of such marks to another, that
enable the reader to recognize 'the same letter' or 'the same word'
twice on one page, etc. These are the analogies that underlie
Peirce's distinction between 'types' and 'tokens'. With such a text,
there is no need for the reader to look for any other visual frame-
work: the book is self-sufficient. It can be read anywhere (provided
there is light to read by). One could sum up all these salient fea-
tures of the printed book by saying that it represents the limiting
case in which, insofar as it is humanly possible, the written text is
made to supply its own context.
However, that is not the case for the written sign in general; and
that is why it would be a mistake for any semiologist of writing to
treat the printed book and its familiar features as a tacitly accepted
model, to which all other forms are imperfect approximations. If

we wish to rethink writing, perhaps the first thing we need to think
about is the full extent of this atypicality.
The image of the printed book hovers over the term context itself,
and has done since the sixteenth century. It is no accident that
context is commonly used, and not only by linguists, to designate
the immediate verbal environment, the words preceding and/or
following some other word or words: thus, for example, in a big
man and a big mistake the word big is said to appear in two different
'contexts'. A quotation taken 'out of context' is one removed from
the rest of the text in which it originally appeared. These and
similar usages tacitly appeal to - and bolster - the notion that the
written text is self-sufficient: it is the whole which contextualizes its
own parts.
From an integrational point of view, this is a quite inadequate
notion of what context is. A written text does not appear, nor is it
read, in a communicational vacuum but in a specific communica-
tion situation. It is this situation, with all its constituent circum-
stances, verbal and non-verbal, which provides the basis for
contextualization. A written text must be written by someone: its
production is necessarily integrated into some wider pattern of
events in the life of the writer. Similarly, reading a text is neces-
sarily integrated into some wider pattern of events in the life of
the reader. These are not just banal truisms but statements of the
conditions of existence for every written text; and it is these condi-
tions which, in any given instance, supply the basis for contextual-
izing what is written. When a readable form is produced under
these conditions - and only then - a written sign appears. That is
why the notion of a contextless sign is, for integrational semiology,
a contradiction in terms.
The term context calls for another preliminary observation. It is

often used in the singular. But in practice we are never dealing
with a context or the context, but always with multiple contexts.

There is not only a spatio-temporal or physical context but a
psychological context. The psychological context of the writer is
not that of the reader. This plurality of contexts must always be
borne in mind. From an integrational point of view, the sign is
invariably at the centre of a whole series of actual and possible
contexts, i.e. surrounded by other potentialties for con-
textualization.
The most important factor in the contextual integration of the
written sign, if we leave aside for the moment the case of writing
by and for the blind, is visual location. A written message has to
be written somewhere. But since writing is not always a question of
marking a surface already available, it is preferable to adopt a
less specific term and refer to the installation of the written sign.
Anyone who wishes to communicate by means of writing is
obliged to find or prepare an installation for the written message.
From this there will follow automatically a whole range of inte-
grational constraints, since not anything can be written anywhere or
anyhow.
But the constraints which apply to writing are not those which
apply to speech, because different biomechanical activities are
involved. All signs have a biomechanical basis, provided by the
human body and its sensory equipment. (Unless I could see the
tree, there would be no question of its becoming a landmark in my
journey home.) No semiologist can afford to ignore these biome-
chanical factors, since in the end they determine limits beyond
which communication is not possible. The point being made here
is simply that, at the biomechanical level, writing requires a form
of contextualization which is quite different from that of speech,
and that difference has far-reaching semiological consequences.
In the case of speech, we have what might be regarded as a default
installation already supplied by Nature. Silence is the background
against which the sounds of the human voice are best heard. Now
in the case of writing it might perhaps be urged that the surface of
the blank page offers a default installation equivalent to silence.
But the comparison will not do, for the page has been artificially
prepared as a setting for the text. (The psychological and socio-

political implications of this are by no means negligible.) In
short, unlike the spoken sign, the written sign requires installation.
The presentation of writing most commonly depends on an
artifact deliberately prepared for that purpose. The exceptions
one can think of confirm rather than disconfirm the foregoing
generalization. Shakespeare could doubtless have written a sonnet
on the sea shore by tracing words with his finger in the damp sand;
but the text would have vanished with the next tide. The whole
development of writing presupposes right from the beginning a
society that has reached a level of technological advancement
which allows the preparation of durable surfaces, ready to receive
more or less permanent marks, made by means of instruments
designed specifically for such a purpose. In the development of
oral language, by contrast, technology played no comparable role.
But that difference is the source of much confusion about the
relationship between the two. It is what underlies the thesis that
writing is 'unnatural' - a thesis which already conflates the sign
with its installation.
Writing and speech, then, have quite different biomechanical

bases. (Leroi-Gourhan made this the foundation of his theory of
primitive 'mythological' writing, insisting on the difference
between the role of the hand in tool-making and the role of the
face in the elaboration of oral language.20) What is of interest
from an integrational point of view is how the biomechanical
difference underpins various semiological differences. In particu-
lar, the notion of 'context' is not identical in the two cases, since
the contextualization of a written message typically assumes a set
of preparatory conditions involving the availability of writing
materials, writing instruments, etc.
These 'material' requirements play a role which has long been
20
A. Leroi-Gourhan, Le Geste et la Parole, I. Technique et Langage, Paris,
Albin Michel, 1964, Ch.6.
recognized, insofar as historians of writing have acknowledged

that the visual form of the written sign often depends on what is
available for the writer to work with: paper, stone, reed, wax tab-
let, baked clay, pattra, bronze, bamboo, silk, tortoise shells, etc.
According to Beatrice Andre-Salvini:
There are two necessary conditions for the invention of writing
in any civilization: the existence of an established society -
which presupposes awareness of collective unity - with symbols
recognized by all which can be materialized, 'transcribed', and
the discovery of a durable basis which can be easily obtained
and used.21
Where the integrationist takes issue with claims of this kind is

over the suggestion that the starting point in the development of
writing is a social convention ready and waiting to be imple-
mented, i.e. an embryonic inventory of signs already agreed in
advance, but waiting for the appearance of a suitable technology
which will allow them to be 'materialized'. The precise form of
this 'materialization' will then depend on the technology that
becomes available. But is this a viable hypothesis?
It seems rather obvious that people do not write on paper as
they write on soft clay or on wax. But it would be a mistake to
conclude that the material is no more than an adventitious or
'external' factor (to adopt Saussurean terminology) in the birth of
a writing system. In other words, it is implausible to suppose that
the written sign exists from the beginning at a level of abstraction
which is independent of its biomechanical realization, i.e. as a
pure geometric configuration.
For the integrationist, the importance of the material elements
involved in writing is of quite a different order. Their semiological
function is complex. The entire syntagmatics of the written text
may be determined by limitations imposed by these elements. But,
more important still, it is the interplay of these material elements
21
B. Andre-Salvini, L'Ecriture cuneiforme, Paris, Editions de la Reunion des
musees nationaux, 1991, pp.2-3.
and the biomechanical factors involved that acts, semiologically, to

transform marks into written messages.
It is at this level that we encounter important semiological dif-
ferences between the tree that functions as a private 'signpost' for
me on my journey home and the public signpost that has been put
up by the municipal authority. A commonplace but instructive
example is provided by the old-fashioned finger-post, nowadays
confined to rural roads (Fig. 2). Here the material support, the post
Figure 2
itself, acts as a semiological vehicle indicating how the text it bears

is to be understood. The form of the post is usually distinctive
(traditionally in Great Britain a white post of wood or metal, with
alternating black and white bands on the upright, and black letter-
ing on a white ground along the horizontal arm). The occurrence
of the text in this position and on this kind of object already
determines to a large extent the range of possible intepretations.
In other words, the written marks are only part of a larger semio-
logical complex.
The function of this complex whole is, manifestly, of an inte-
grational character: it serves to integrate certain programmes of
anticipated activities. Travellers coming from a certain direction
are assumed to be in need of certain information in order to get
their bearings and continue their journey. In a world where no one
travelled by road, this form of integration would be unnecessary

and the finger-post would not exist.
The crucial semiological point to grasp is that this integrational
function depends on the traveller's understanding of the written
forms. In the case of the 'tree' sign, there is nothing to understand
other than the topographical relationship between the location of
the tree and my destination. But in the case of the finger-post, that
understanding (of one's present location in relation to one's
destination) is mediated by writing. It would be simplistic, how-
ever, to treat the writing as a semiological device which transforms
what would otherwise be just a metal upright and an arm into a
source of topographical information. That would be to ignore the
other side of the relationship. For it is also true that the traveller's
understanding of what a signpost is contributes to the semio-
logical value of the written forms. The metal upright and the
horizontal arm provide the installation for the text, thus turning
what would otherwise be just sequences of letters into a message.
This is not a simple conjunction of elements but a functional
complementarity.
The written sign must not be confused either with the signpost
or with the visible configuration of letters (e.g. 'GODSTOW') that
it bears. The sign, for an integrationist, is not a physical item of
either kind, nor the mental correlate of either. Discarded sign-
posts, complete with their lettering, still survive as material objects
long after having ceased to play a role in any system of communi-
cation (Fig. 3). But they do not carry their semiological status
around with them. For the sign qua sign is neither the support nor
the marks that appear thereon.
There is no theoretical contradiction between recognizing the
radical impermanence of the written sign and the (relative) permanence
of the visible marks. Traditional wisdom of the kind that utili-
tarian literacy promotes tends to confuse the two, thus giving rise
to the mistaken belief that what distinguishes written sign from
spoken sign is that the former is not subject to the intrinsic
ephemerality of the latter. And this in turn, as Plato saw, gives rise
to the illusion that writing 'fixes' words, whereas speech cannot.
Figure 3
But once we see semiological value as depending on integrational

function the role of writing appears in a quite different light. In
the case of glottic writing there are typically at least three bio-
mechanically different operations to be integrated: (i) graphic
(manual), (ii) vocal, and (iii) optical. No dualist theory of the
written sign can possibly accommodate all three. Freeing writing
from this traditional Procrustean bed is the first requirement if we
wish to understand it as a mode of communication.
CHAPTER FOUR
Notes on Notation
The first step in any rethinking of writing within an integrational

framework must be to recognize the validity of a distinction that
Saussure failed to draw. His failure to draw it is entirely attribut-
able to following the traditional wisdom about the phonoptic
function of letters. Although the distinction in question is com-
monly recognized by virtually every European child who learns at
school to read and write a 'foreign language', there are no tradi-
tional pedagogical terms for referring to it (another lacuna which
points to the blind spot in traditional thinking). In integrational
terms it is the distinction between a notation and a script.
This distinction corresponds to typically different patterns of
integration between the activities associated with reading and
writing, and also to different questions that may be asked about
them. 'Is it a t or an /?' is a question about a notation. 'How do you
spell [kat]?' is a question about a script. The relationship between
script and notation can be stated generally and informally as fol-
lows: the same notation may serve as a basis for more than one
script (in fact, theoretically, for any number of scripts). That is why
it is important not to confuse the elements of a notation with the
(homographic) elements of a script.
It is through its incorporation into a script that an element of
notation acquires its value as a written sign in the texts of that system.
This is the conclusion Saussure ought to have reached if he had
been faithful to his own structuralist principles.1 That conclusion

would have allowed him to recognize and account for such facts as
the existence of interlingual homography: e.g. that the letter-
sequence c-h-a-i-r spells one word in English but a quite different
word in French. The members of this homographic pair are dif-
ferently pronounced and have different meanings, but are never-
theless orthographically indistinguishable.
The integrationist's rationale for drawing this distinction again
rests on the premiss that semiological values are determined by
the structure of the activities integrated. In order to copy accu-
rately a text containing the letter-sequence c-h-a-i-r a medieval
scribe familiar with the alphabet does not need to know whether
this is a French text or an English text; nor what the texts 'says'.
Modern printers can and do set type in languages with which they
are quite unacquainted. The viability of such activities as these
already shows that there is something missing from Saussure's
semiological analysis of the written sign. What is missing turns out
to be a basic level that we can identify if we consider more ele-
mentary forms of semiological structure than writing. Examples of
such structures are provided by children's games that are based on
simple matching of shapes and/or colours from a given inventory
of pieces. (Children are sometimes, as Quintilian recommended,
introduced in this way to the letters of the alphabet; but the point
being made here is of far wider application.) It would be quite
mistaken to dismiss such games as 'meaningless'; on the contrary,
they exemplify particularly clearly how meaning emerges from the
integration of activities. Nowadays games of this kind are quite
carefully devised to develop the child's skills of recognition and
1
A remark from the Second Course, recorded by Patois, but omitted
from the published text of the Cours, bears on this question. On the
subject of letter-forms, Saussure notes that 'P' does not denote the same
consonant in the Latin alphabet as in the Greek. But this is because he
takes it as signifying [p] in one case and [r] in the other. He draws no
general distinction between notation and writing system on this basis.
(Komatsu and Wolf, op. cit., p. 113.)
Notes on Notation 93
manipulation. But to determine the value of each piece in the set

of pieces we do not need to look further than the way it functions
in the game. Thus, for instance, a circle will mean something
different from a square because it does not fit into the same slot on
the child's board. However, if the relevant matching criterion
were colour, not shape, the circle and the square might be equiva-
lent. Some games, such as the card game commonly called 'Snap',
depend entirely on matching two samples of the same configura-
tion. Others, such as jigsaw puzzles, depend entirely on fitting
together parts of a pre-determined configuration. Saussure's
myopia about the alphabet, seen from an integrational perspec-
tive, is like that of someone who insists that a jigsaw does not make
sense unless it fits together to form a picture. (Landscapes by Con-
stable and interiors by Dutch masters seem to be traditional jigsaw
favourites.) But the fact is that Constable's landscape provides an
extra layer of meaning, superimposed on the more basic geometry
of the jigsaw. In the Constable jigsaw there are two semiological
structures involved, not just one.
In retrospect it may seem astonishing that Saussure failed to see
the implications of interlingual homography and draw the
appropriate semiological lessons. But it is worth pointing out that
he also failed to see a very similar distinction in the domain of
spoken language; namely, the distinction between phonetic units
and phonological units. This latter distinction turned out to be
indispensable for the development of modern structural phono-
logy. Consequently retrospective attempts were made to trace it
back to Saussure. In this connexion a great deal was made of his
early Memoire sur le systeme primitif des voyelles dans les langues indo-
europeennes, which Saussure had published in 1878 at the age of 21.
Some Saussurean scholars detected in this essay a concept of the
phoneme which anticipated that of the Prague school half a cen-
tury later. The suggestion is certainly an attractive one and, if it
were right, would fit admirably into the general framework of
Saussure's ideas about language. What throws doubt upon it,
unfortunately, is that nowhere in Saussure's Geneva lectures, or in
the text of the Cours, do we find anything approaching an
unambiguous statement of the distinction in question. If Saussure

had already grasped the phoneme principle as early as 1878, it is
difficult to explain why he left it undeveloped for the next thirty
years of his academic career. One would have expected its appli-
cations, on the contrary, to have become one of the major themes
of his teaching. It would also have supplied what is conspicuously
missing in the Cours. As has been pointed out in Chapter 2, it is in
the section on the 'material aspects' of linguistic value that the
exposition of Saussure's ideas runs into all kinds of difficulties
through its question-begging appeal to the analogy of writing.
This could have been avoided if a clear formulation of the
phoneme principle had been available. So there seems to be a
significant connexion between Saussure's failure to draw the
semiological distinction between a notation and a writing system
on the one hand and his failure to distinguish phonetic from
phonological units on the other.
The general idea that a notation must not be confused with any
superimposed system of values it is used to express is perhaps
more important in mathematics than it is in linguistics. In lay
mathematical terms, the basic point is easily made: figures must not
be confused with numbers. There is no such simple and perspicu-
ous statement available in the case of language. ('Letters must not
be confused with sounds' makes a quite different point. Perhaps
where Saussure went astray was in assuming that this was the
appropriate linguistic parallel.) In the first mathematical treatise to
be printed in Europe, the Practica of Treviso (1478), the author
begins with this distinction as the foundation for all mathematical
reasoning: 'numeration', as he puts it, is the 'first operation':
Numeration is the representation of numbers by figures. This is
done by means of the following ten letters or figures: i [sic], 2, 3,
4, 5, 6, 7, 8, 9, 0?
2
D.E. Smith (ed.), A Source Book in Mathematics, New York, McGraw-Hill,
1929,pp.2-3.
This 'first operation' is, in effect, the choice of a notation, and

the notation here presented has become one of the most familiar
in modern Western culture: the so-called 'Arabic numerals'.
From this 'first operation' there will follow a whole range of
decisions that need to be taken concerning the system(s) of
mathematical writing to be based upon it. For example, if it is
desired to express more than ten numbers, some convention will
have to be determined for combining the individual notational
marks (i, 2, 3, etc.) into complex mathematical signs. Any such
convention, however, will belong to that particular system of
mathematical writing, not to the notation as such. Many such
conventions might be devised, all based on the same original
notation, and irrespective of the numerical values assigned to the
basic notational marks.
For non-mathematicians, however, it is very easy to fall into
confusion about what belongs to the notation and what does not.
For example, there may be a temptation to suppose that 5 always
designates the number five, this value being somehow built in to
the definition of that notational unit. That temptation arises, if it
does, simply because of the arithmetic value most commonly
assigned to the figure 5 in the calculations of daily life. Worse still,
it is not uncommon to speak of 'the number five' to refer either to
the number or to the figure, as if the two were indissolubly asso-
ciated. Thus National Lottery results are announced as winning
'numbers'. This and similar usages may encourage people to sup-
pose that the figure 5 always has a numerical value, even when
it occurs in post-codes or telephone 'numbers'. (This is exactly
parallel to the assumption that the letter t always designates the
sound [t] or the phoneme /t/.)
Such confusions have to be set aside if we are ever to grasp the
(semiological) fact that 5 retains its identity as a notational unit
regardless of what numerical value it may be assigned in any given
instance, and regardless of whether it has any numerical value at
all. This fact is reflected (although somewhat indirectly) in our
acquaintance with the possibility of two ways of reading certain
arithmetic expressions aloud. To read 555 as 'five five five' is to
read it as a figure, whereas to read it as 'five hundred and fifty five'

is to read it as a number.
Most of the above observations about 'Arabic numerals' have
their counterparts with respect to those other familiar notational
forms in Western culture which we call 'letters of the alphabet'.
(Thus, for example, like the figure 5, the letter E retains its no-
tational identity irrespective of how it is pronounced or whether it
is pronounced at all; we read letter-sequences aloud differently
depending on whether we treat them as words or not, cf. reading
CAT as [si ei ti] or as [kat].)
A notation is not just a haphazard collection of autonomous
marks. It has an internal systernatization. 5 has its own place in the
numeral series ( . . . 3, 4, 5, 6. . .) Similarly E has its own place in
the alphabetic series ( . . . C, D, E, F. . . ) There is no notational
series which combines the two. The 'places' that individual no-
tational marks occupy in their respective series are features of
notational structure, and are not to be confused with the variable
'places' that may be occupied in written signs; e.g. the place of 5 in
the numeral 5000, or the place of E in the word Edinburgh. These
latter places are features of the structure of particular scripts, not
of notational structure. In the case of letters, notational place-
sequence is commonly referred to in English as 'alphabetical
order'. It has nothing whatsoever to do with place-sequence in
the signs of written English. Similarly, we recognize for figures
what might be called a 'counting order', although that term is
not in common usage. Thus the instruction 'Count up to five'
requires one to recite the names of the relevant figures in a
certain order: the place of each in that traditional order is
independent of the place(s) it may occupy in the expressions of
any particular system of mathematical writing. (European chil-
dren commonly start to learn 'counting order' before they can
actually count or write.) Considerations of this kind point again
to the necessity for a semiological distinction between notations
and scripts. Just as a supply of metal discs of various shapes
and sizes does not in itself constitute a currency system, even
though it may provide the necessary materials, so a notation
does not in itself constitute a script, but may provide the basis
for one.
The distinction between notation and script is also relevant to
understanding an ambiguity in the everyday use of the term letter.
When we talk, for example, about a word beginning with a certain
letter or about how many letters there are in a certain word (as, for
instance, in connexion with crossword puzzles or other word
games) we are referring to letters as notational units and we refer
to them either by their letter-names (in speech) or by instantiating
them (in writing). We can also, although perhaps less frequently,
refer to them by their place in alphabetical order, as e.g. 'the first,
second, third, etc. letter of the alphabet'. This notational sense of
the term letter, however, does not capture the sense in which there
are important differences between capitals and 'small' letters,
between roman and italic letters, etc. These contrasts between
corresponding shapes also have to be taught and learnt in an
apprenticeship to alphabetic literacy. So there is a sense in which a
capital A and a small a are indeed different letters (as typographers
and printers insist), even though in the notational sense both are
'the same letter'. Differences of the kind that distinguish capitals
from small letters, roman from italic letters, etc. are features of
scripts, not notational features. Capital A and small a do not
occupy different places in alphabetical order: both have the same
place and share the same letter-name.3
By insisting on a semiological parallel between the way writing

works and the way speech works, Saussure overlooked a profound
3
It is perhaps worth pointing out that the distinction between letters as
notational units and letters as units of scripts is not to be confused with
the Peircean distinction between 'types' and 'tokens'. In Peirce's sense,
every letter-shape, whether capital, lower-case, roman, italic, etc. will
have its own type and innumerable tokens. The number of letter-tokens
on this page remains the same, regardless of how many letter-types are
distinguished.
structural difference between the two, a difference which tends to

develop in all civilizations where writing becomes a widespread
form of communication. This difference hinges on the fact that
writing may be based on units which have a certain degree of
independent organization; or, in other words, on structures
which to some extent live a social life of their own, not geared
to the immediate demands of communication. This is because a
tradition of writing requires mechanisms of apprenticeship, con-
servation and transmission to be put in place, mechanisms which
are far more complicated and specialized than those of speech,
and hence become a focus of social attention in themselves.
In this way, notational units acquire a social existence that is
denied to individual speech sounds. The consonant [t] on its own
has no existence at all except one conjured up by linguists and
phoneticians. It has no history either, any more than the sound of
coughing has a history. But the history of the letter / is quite a
different matter: we can trace it back over the centuries to the
original emergence of alphabetic writing. To ask for the history of
the sound [t], on the other hand, is to ask a nonsense question.
One can describe the mechanism of articulation of the sound and
its acoustic properties; but that is all.
The history of alphabetic letters is often illustrated in textbooks
by tables showing the variations of form at different chronological
stages. Thus the modern A, for example, turns out to be an
inverted form of an original pictograph representing the head of
an ox. There is nothing wrong with tabular histories of this kind,
provided the information they contain is accurate and provided
we do not treat them as anything more authoritative than sche-
matic guides to the evolution of certain visual configurations.
Similar tables, after all, can be drawn up to illustrate the evolution
of patterns which have nothing to do with writing; for instance, in
textiles or ceramics.4 What would be a mistake, however, would be
to ascribe a similar validity to the tables of sound changes that
4
See R. Harris, The Origin of Writing, London, Duckworth, 1986, p. 125
for a table illustrating the evolution of corner decorations on Delft tiles.
authors of textbooks on historical linguistics are fond of compil-

ing. These have a quite different status. The 'sounds' represented
in such tables are abstractions reconstructed by linguists on the
basis of certain hypotheses and the application of certain analytic
methods. The supposed 'evolution' of these 'sounds' is equally
hypothetical, and metaphorical to boot. Yet the misleading
impression is given — and accepted by many students — that the
table of letter-form changes and the table of sound changes are
somehow on an equal footing, and merely exhibit different facets
of the same linguistic story. This only serves to obscure a fact
which the semiologist of writing must bring out into the open and
place in the forefront of any investigation: letters and figures are
cultural artifacts. Sounds are not (except insofar as their identifica-
tion is a second-order activity which itself depends on familiarity
with writing).
The history of metalanguage is very revealing in this con-
nexion. In most European languages, letters have names but
speech sounds do not; with the result that a letter-name is fre-
quently pressed into service in order to identify a particular vowel
or consonant. ('He slurred his esses.' 'French people can't say tee
aitch properly.') Throughout the European tradition there is a
tendency to conceptualize sounds in terms of letters. Greek and
Roman authors so frequently refer to sounds as 'letters' that one
might be forgiven for supposing that they failed to distinguish
between the phonetic phenomena and the marks on the page. It
was this pervasive tendency to think of the sound by reference to a
written form that Saussure had in mind when he warned his stu-
dents against looking at language through the grid imposed upon
it by conventional orthography. He speaks of the 'tyranny' of the
letter.3 More explicitly:
the written word is so intimately connected with the spoken
word it represents that it manages to usurp the principal role.
As much or even more importance is given to this representa-
tion of the vocal sign as to the sign itself. It is rather as if people
3
believed that in order to find out what a person looks like it is

better to study his photograph than his face.6
In spite of these warnings, Saussure himself does not seem to

see that, as cultural artifacts, letters lead a life of their own 'within
the life of society', whereas speech sounds do not. Nor does he see
that for this very reason the semiologist of writing cannot take it as
a basic premiss that the sole function of letters is - or should be -
to designate sounds. This would be like starting from the assump-
tion that the alphabet was the creation of the International
Phonetic Association.
Even odder, however, is Saussure's failure to recognize in writ-
ing what he recognizes very promptly in speech: the existence of
dual articulation. Although consonants and vowels are units in the
chain of speech, they are not ipso facto signs. But when it comes to
writing, Saussure treats individual letters not only as units in the
graphic chain but as signs, each signifying a given sound. The
bizarre result of this theoretical move is that he is then obliged to
devote a long discussion to explaining away the 'discrepancies'
between orthography and pronunciation.
The causes of these discrepancies, according to Saussure, are
many and various. He mentions only the most important. The
first and most general is that 'a language is in a constant process of
evolution, whereas writing tends to remain fixed'.7 Thus, he points
out, the French word for 'king' is still spelt roi, which corresponds
to its thirteenth-century pronunciation.
This example is worth pausing over, since it epitomizes some of
the basic problems with the utilitarian view of writing. On the one
hand, writing is praised for its capacity to 'fix' the spoken word: on
the other hand, it is blamed for 'fixing' it too permanently. Thus
the chimerical ideal of a spelling system which would somehow
automatically adapt to changes in pronunciation is conjured up.
Saussure seems to have regarded current proposals for French
6
7
spelling reform with caution, if not with contempt (presumably

because he thought they were advocated by people linguistically
incompetent to judge the 'real' relationship between the writing
system and la langue). But the explanation he himself offers for the
'discrepancy' between the spelling of the word roi and its modern
pronunciation does not inspire confidence in the ability of lin-
guists to deal with the problem either, since it raises as many
questions as it answers. He cites it as an example of the 'immobil-
ity' of writing. But this alleged immobility is clearly extrapolated
from a comparison between modern texts and medieval texts,
from which it emerges that the modern spelling roi and the
medieval spelling roi coincide. Now to conclude from such a
comparison that here we have a typical example of inertia or
conservatism in the writing system involves assuming that the
same orthographic principles are operative in both periods. In
other words, we are tacitly invited to judge the twentieth-century
spelling by the practice of the medieval scribe. Does this make
sense? Where do these panchronic criteria come from?
What is no less puzzling is how to reconcile Saussure's 'immobil-
ity' claim with the obvious counterevidence that this example
itself provides, /^letters represent sounds, as Saussure (bearing in
mind his interpretation of sounds as elements of the image acous-
tique) maintains, then the expected conclusion would be that what
has changed over the centuries, to judge by the case of roi, are the
rules of representation. On the other hand if, as Saussure seems to
be insisting here, the modern spelling fails to reflect the modern
pronunciation, what this calls in question prima facie is the thesis
that letters represent sounds. The case of roi could be argued
either way; but what Saussure again seems to be trying to do is to
have it both ways at once. In other words, he wants it both to
illustrate and to be an exception to his main semiological thesis
about the alphabet. Saussure evidently thinks that a better (i.e.
more accurate, more rational) spelling for the twentieth-century
word would be rwa, but he offers no explanation of why French
people do not adopt that spelling, as presumably they could if
they too thought that. What is one to conclude? One possibility is
that, contrary to Saussure's claim, the modern spelling roi is not

recognized by most people as an orthographic 'discrepancy' at
all.
To sum up, Saussure seems to be guilty here of an inconsistency
which he criticizes severely in other linguists; namely, mixing syn-
chronic and diachronic perspectives. A 'discrepancy' between
spelling and pronunciation - if there is any such linguistic phe-
nomenon - must be a synchronic phenomenon and therefore
requires a strictly synchronic analysis. How the word roi was
written — or pronounced — in the thirteenth century is simply an
irrelevance as far as the semiological status of the twentieth-
century spelling is concerned.
The second cause of 'discrepancies' that Saussure lists is
borrowing.
Another cause of discrepancy between spelling and pronun-
ciation is the borrowing of an alphabet by one people from
another. It often happens that the resources of the graphic sys-
tem are poorly adapted to its new function, and it is necessary
to have recourse to various expedients. Two letters, for
example, will be used to designate a single sound.8
The example he cites is the voiceless dental fricative of the

Germanic languages being spelt th because the Roman alphabet
had no corresponding letter.
This is another puzzling case of an illustration undermining the
main thesis it is invoked to support. In the first place, the very
possibility of adapting and augmenting an alphabet shows that we
are dealing with nothing more than a notation. Otherwise, it
would be impossible to borrow an alphabet without borrowing its
pronunciation at the same time. In the second place, the spelling th
would be inexplicable if it were true that letters represent sounds,
since the Germanic dental fricative was not a [t] followed by [h].
Saussure adds, for good measure, that King Chilperic tried to
introduce a special letter for this fricative sound, but without
8
success. So what we end up with is borrowing as a variation on the

theme of the 'immobility' of writing.
The third cause of 'discrepancies' that Saussure cites is 'etymo-
logical preoccupation':
It is not infrequently the case that a spelling is introduced
through mistaken etymologising; d was thus introduced in the
French word poids ('weight') as if it came from Latin pondus,
when in fact it comes from pensum. But it makes little difference
whether the etymology is correct or not. It is the principle of
etymological spelling itself which is mistaken.9
It is difficult to know what to make of this authoritarian pro-

nouncement. For an integrationist it would be self-evident that
there is no point in trying to establish a semiology of writing (or of
anything else) on the basis of prescriptive claims of this kind.
Etymological spellings are obviously to be condemned if one
believes that letters 'ought' to stand for sounds and nothing else.
But for anyone not committed to this belief, the very existence of
etymological spellings is an important piece of evidence against the
thesis that letters are just phonetic signs.
There is little to be rescued from this disastrous chapter of the
Cours. Even the idea of grapho-phonic 'discrepancy' does not sur-
vive. In order to make any sense at all, it would have to be con-
strued as a synchronic concept. But now suppose we are dealing
with a writing system where certain spellings or certain pronun-
ciations cannot be explained by general orthoepic rules. The ana-
lyst then faces a simple choice. Either (i) the principle that letters
designate sounds is to be upheld - in which case the orthoepic
rules have to be altered in order to accommodate the previously
recalcitrant examples - or else (ii) the principle is abandoned and
one concedes that there must be other factors at work. In the first
case there is, by definition, no 'discrepancy', since the rules
are reformulated in such a way as to take care of the former
exceptions. In the second case there is no 'discrepancy' either,
9
since the original principle is now abandoned. Either way, the

notion of'discrepancy' itself vanishes - a mirage projected by an
inadequate theory of writing.
Saussure's failure to distinguish between script and notation in his

analysis of alphabetic writing is not just an 'accidental' blind spot,
but a consequence of his own structuralist approach to semio-
logical questions. This becomes apparent if we go back to his own
favourite example of chess. He points out - quite rightly - that the
relations between successive states of the board are such that any
passer-by (who understands the rules) can take in the current situ-
ation at a glance, without knowing anything about antecedent
states that led to it. Such a person can even take over from one of
the players and carry on the game from there. What Saussure
omits to mention, however, is that in order to step in in this way
the newcomer must not only know how to move the pieces but,
before that, be able to recognize them as they stand on the board.
Now chess pieces are recognized by their form. That is why
Saussure's remarks about replacing a knight cannot be accepted
without serious reservations.10 According to Saussure, even 'an
object of quite a different shape' will do 'provided it is assigned
the same value'. That is indeed possible. But it would be a mistake
to conclude on that account that the 'normal' shape of a knight
counts for nothing. Replacing the knight by a spare bishop, for
example, would certainly run the risk of confusing the players in
the long run. And if the knight were replaced by an object having
nothing to do with the game — a coin, for example — a passer-by
would no longer be able to judge the state of the board simply on
inspection: it would be necessary to explain the substitution.
Here again the importance of the integrational concept of
notation makes itself felt. In order to play chess at all, we have first
to learn certain 'notational' distinctions. That is to say, before
there is any question of how the knight moves, we have to grasp
10
Saussure, op. cit., pp. 153-4.
Motes on Notation 105
which pieces are knights, which are bishops, and so on. Saussure
makes no mention of the fact that in chess there is a long tradition
of forms and names. These cannot be dismissed as merely
'external' elements of the game, since they articulate an essential
part of the rules. The distinction between the two colours ('white'
and 'black') is basic to the way chess is played, i.e. as a competition
between two opposing players, each in command of a certain
team. Furthermore, in the traditional chess set equivalent pieces
have the same shape. The beginner grasps straight away, for
instance, that all pawns move in the same way. There is no need to
ask what the moves are for each pawn individually, and no need to
explain that white pawns behave according to exactly the same
pattern as black pawns. (Nevertheless, as chess signs, a white pawn
and a black pawn mean something different.) In short, the icono-
graphy of chess is not merely 'decorative' even if it is 'arbitrary',
but reflects certain crucial features of the game.
There is an interesting parallel between the way Saussure plays
down the significance of these allegedly 'material' aspects of the
game of chess (the colours, the characteristic shapes of the pieces)
and his failure to recognize notation as a basis for alphabetic
writing. In both cases what has been missed — or dismissed — is the
existence of a dual structure within the semiological organization
of the system.
What is the explanation? Any search for one brings us back to
Saussure's notion of semiological 'differences'. The importance
Saussure attached to this is a reflection of his dislike of the spirit
of positivism which, he believed, had dominated nineteenth-
century linguistics. The fundamental error engendered by this
positivism, in Saussure's view, was the assumption that systems of
signs arose from the conjunction of series of objects and thoughts
given in advance. Against this Saussure argues that a word such as
arbre ('tree') is not a mere vocal label attached to an object supplied
ready-made for us by Nature. On the contrary, both the signifiant
and the signifie of this word are products of a vast network of
phonic and conceptual differences which constitutes the French
language.
This emphasis on the importance of differences and differen-

tiation, at the expense of positive relations, is one of the originali-
ties and strengths of Saussure's thinking about signs. But it is at
the same time the Achilles' heel of Saussurean semiology. For
there are systems of signs based on something other than a network
of differences; and scripts are among them. This is not to say that
in such systems differences do not count: they do indeed. But
nevertheless such systems cannot simply be reduced semiologi-
cally to series of internal oppositions, in the way Saussure believed
was both possible and necessary in the case of languages.
The integrationist distinction between script and notation is only

the application to glottic writing of a more general semiological
distinction. Notations exemplify a type of structure which, far
from being confined to writing, is one of the most basic structures
in the domain of signs. It is the structure characteristic of any set
of items fulfilling the following conditions.
1 Each member of the set has a specific form which sets it apart
from all others in the set.
2 Between any two members there is either a relation of equiva-
lence or a relation of priority. Thus every member has a
determinate position with respect to all other members in the
set.
3 Membership of the set is closed.
Such a structure constitutes what in integrationist terms is an

emblematic frame. The very simplest emblematic frames comprise
just two members: examples are an on-off switch with only two
possible positions and a red-green traffic light where the two
colours alternate but never show simultaneously. Slightly more
complicated examples are the following.
A The Japanese game of shenken is based on a three-member

emblematic frame. The three members of the set are: Knife,
Paper, Stone. The priorities are: Knife beats Paper, Paper
beats Stone, and Stone beats Knife. The two players simul-
taneously choose an emblem. The winner is the player choos-
ing the emblem with the higher priority. It would be possible,
obviously, to play with a different set of priorities; e.g. Knife
beats the other two, Stone and Paper are equal. But then the
game would presumably lose its interest, since the players
would always choose Knife.
B In the traditional pack of playing cards, each suit has the
structure of an emblematic frame. There are thirteen mem-
bers: ace, king, queen, jack, etc. In certain games, the deuce
takes priority over the ace and the joker takes priority over all
other cards. There is no priority between suits except in cer-
tain games, including games where 'trumps' are declared. The
notion 'trump' is an interesting example of a semiological
concept: a priority is assigned where 'normally' there is no
priority.
C The traditional Chinese calendar is based on an emblematic
frame. The emblems are: Rat, Ox, Tiger, Rabbit, Dragon,
Snake, Horse, Sheep, Monkey, Cock, Dog and Boar. Here the
priorities are chronological. The cycle repeats every twelve
years, always in the same order. Thus Horse is always pre-
ceded by Snake and followed by Sheep.
D A more complex calendrical example is the pelelintangan of Bali
(Fig. 4). This is an astrological calendar comprising thirty-five
emblems, of which there are a number of variants. *' One of
these goes: kola sungsang, demon upside-down; gajah, elephant;
patrem, dagger; uluku, plough; laweyan, headless body; kelapa,
coconut tree; kukus, smoke; kiriman, gift; lembu, bull; pedati, cart;
kuda, horse; jyuyu, crab; asu, dog-Jong sarat, full boat; sidamalung,
sow; tangis, tears; gajah mina, mythological beast with the body
of a fish and the head of an elephant; lumbung, rice store;
kartika, the Pleiades; tiwa tiwa, death rites; sangkatikel, broken
hoe; salah ukur, faulty measurement; bade, cremation pyre;
" F.B. Eiseman,Jr., Bali: Sekala and Nisakala. Vol.1. Essays on Religion, Ritual
and Art, Berkeley/Singapore, Periplus, 1989, Ch. 18.
Figure
kumba, urn; naga, snake; banyak angrem, brooding goose; bubu

bolong, fish trap; prahu pegat, shipwreck; magelut, embrace; udang,
prawn; begoong, headless ghost; ru, arrow; sungenge, sunflower;
puwuh atarung, fighting quails; pagelangan, stare.
The order in this case is not a simple chronological succession.
The place of each member of the pelelintangan depends on the fact
that the Balinese calendar consists of several weeks that run con-
currently. The two most important weeks are thepancawara of five
days and the saptawara of seven days. Coincidences between days
in these two weeks, of which the above emblems are allegorical
representations, are considered particularly significant. The pelelin-
tangan, in other words, summarizes the total set of possibilities of
coincidence between days of the pancawara and days of the sapta-
wara. Each of these is a bintang. So there are thirty-five possible
bintang. The character of a child is believed to be determined by
the bintang of the day on which it was born.
It should be noted that although the pelelintangan is traditionally
displayed as a five-by-seven grid, the arrangement of the thirty-
five emblems is not determined by the chronological succession of

bintang, but by the sequence of days in the two basic weeks. At the
same time, however, the succession of bintang results in
rearranging the days of the weeks in a new order: 1, 6, 4, 2, 7, 5, 3
for the sapatawara, and 1, 3, 5, 2, 4 for the pancawara. (The semio-
logical importance of this point - which will not be pursued here —
is that, unlike some calendars, the pelelintangan is not a kind of
spatial diagram of time. There is in this sense no answer to the
question of 'what' the pelelintangan grid 'shows': any more than
there is an answer to the question of what the game of shenken
'means'.)
In the four examples cited above, the emblematic frame is a struc-

ture determined by tradition. Its images may originally have had
some rationale of a practical, religious or magical nature which
has since been obscured. But, once established, the emblematic
frame becomes a cultural artifact in its own right, irrespective of
what purpose it may originally have served. It can be described
quite precisely without reference to its beginnings or to the social
practices that maintain it. A description of a Balinese pelelintangan
could just as easily be the description of a gaming board, or a
description of a pack of playing cards, or that of a cosmic symbol-
ism. It is no coincidence that emblematic frames are often the
subject of mystical or superstitious interpretations which have
nothing to do with their social history.
Emblematic frames of some kind turn up in every culture that
has been studied by anthropologists. They may serve very diverse
functions: they feature in calendars, military uniforms, archi-
tecture, rituals and games of all kinds. The point here is that it
should come as no surprise — anthropologically speaking — to find
them at the basis of writing systems. For they offer just what a
script needs: a notation. A notation simply is an emblematic frame
adapted to or devised for the purpose of writing. But it is not in
itself a script, any more than a set of emblems constitutes in itself
a calendar.
It is sometimes possible to utilize or devise emblematic frames

in such a way that the emblems correspond on a one-to-one basis
with units of some other structure. In such cases the result is an
isomorphism between the two structures. That possibility is cer-
tainly available, in principle, in the case of notations. But, as far as
the semiologist is concerned, it cannot be taken for granted that
such isomorphisms regularly occur, or are even common.
It is ironic that Saussure, of all theorists, should have over-

looked the distinction between a notation and a script, for this is
bound to make a great difference to any semiology organized on
structuralist principles. Once that difference is given its due, there
will be two levels of structure to be taken into account in every
case, and the most basic mistake the analyst can make would be to
conflate them. The notational structure and the structure of the
script need to be kept separate because some features of a written
text are to be explained by reference to the former and some by
reference to the latter. In order to distinguish them for purposes of
discussion let us call them notation features and script features
respectively.
Usually we find that the script superimposes a more complex
layer of organization on that of the notation it deploys. But that
does not mean that notation contributes nothing to the final struc-
ture of the written message, or so little that it can be safely
ignored. On the contrary, notation features are typically among
the most prominent features distinguishing one form of writing
from another.
A clear illustration of this is provided by forms of arithmetical
writing that uses Arabic numerals. Consider what can be done
with the classic notation 1, 2, 3, 4, 5, 6, 7, 8, 9, 0. Given this
notation, we next have to choose a set of arithmetic values. In
order to express the number 'thirteen' we have a variety of scripts
available. Depending on which one we choose, the appropriate
numerical expression will be, for example, 13, 1101, 111, 31, 23,
21 or //. The binary system uses only two units of the notation
whereas the denary system uses all ten. But in all cases the struc-
ture of the expression is grafted on to a more basic structure
provided by the notation.
Take, for example, the sign for 'thirteen' as expressed in the
binary system: i.e. 1101. The syntagmatic organization of this
expression and the numerical values are supplied by the binary
system; but the two figures, the contrast between them and their
relative priority come from the notation itself and have nothing to
do with the binary principle. Similarly, when we compare the two
signs 13 and 31, both meaning 'thirteen', we see that the figures
are the same but their position in the syntagmatic chain is differ-
ent. This has nothing to do with the notation, which is the same in
both cases, but is entirely due to the different value systems.
The number 'thirteen' can also be expressed in Roman
numerals. Roman arithmetic notation uses the seven letters 7, F,
X, L, C, D and M. As regards their shape, these are recognizable as
letters of the alphabet. But their order of priority is not that of
alphabetical order. In this case the numerical system imposes an
order of its own, on the basis of the values 'one', 'five', 'ten', 'fifty',
'hundred', 'five hundred', 'thousand'. 'Thirteen' is written XIII.
Again, the syntagmatic structure is not determined by the
notation.
It is clear from these cases, which could be multiplied ad infini-
tum, that semiological analysis requires recognition of a distinction
between notation structure and script structure. How would this apply
in the case of alphabetic writing? What difference would it make
to the way written forms are analysed and classified? It is instruct-
ive to consider one of Saussure's examples: the French word for
'bird'.
The modern orthographic form is oiseau. Saussure takes this as
a flagrant example of how it is possible for a writing system to
obscure entirely the structure of the linguistic sign. 'Not one of
the sounds of the spoken word (wazo},' he complains, 'is repre-
sented by its appropriate sign'.12 But that complaint presupposes
12
Saussure, op. cit, p.52.
that oiseau should indeed be spelt wazo; which is an entirely gra-

tuitous assumption from a semiological point of view. Indeed,
doubly gratuitous; for it is difficult to see what authority Saussure
has for invoking a hypothetically 'correct' spelling which has no
antecedents in the history of French orthography, and no less
difficult to see what justification he has in any case for assuming
that the written form should reveal the phonetic composition of
the corresponding spoken signifiant. The idea that a written sign
might signify simply in virtue of expressing a particular concept
(in this case 'bird') is one Saussure is evidently not prepared to
consider, even though that is exactly what he assumes in the case
of the corresponding oral sign.13 The result is that his semiological
analysis has nothing to say about oiseau as a written sign at all
except that it is bizarrely - even perversely - 'incorrect': i.e. it has
no synchronic explanation and totally violates what Saussure takes
to be the current structure of the French orthographic system.
A quite different picture emerges if instead we consider the
form in the light of the distinction between notation features and
script features. It is difficult to see in the case of oiseau that the
notation contributes anything apart from the letter-shapes, and
therefore it must be in the structure of the script that any explana-
tion of its graphic form is to be sought. A synchronic rationale at
this level - where Saussure never looked for it - is not difficult to
find, and can be found without employing any analytic methods
other than those Saussure himself recommends. It begins to
emerge as soon as we consider the following two series of forms:
1. oiseau, oisif, oiseux, Oise, toise, ardoise, etc.14
2. oiseau, beau, eau,peau, sceau, etc.15
The spelling oiseau clearly conforms to certain orthoepic pat-
terns quite abundantly exemplified in matching sets of modern
13
Discussion of a neo-Saussurean theory (glossematics) which allows for
the possibility Saussure rejected will be reserved for a later chapter.
14
Phonetically: [wazo], [wazif], [waz0], [waz], [twaz], [ardwaz].
15
Phonetically: [wazo], [bo], [o], [po], [so].
French words. These patterns, to be sure, are not without

competitors; but such patterns rarely are (except in the case of
artificially constructed languages). The existence of competing
paradigms is one of the conspicuous features of linguistic
structure at all levels of analysis.
When we look for patterns that reflect the meaning of the word,
we find:
3. oiseau, oiselle, oisellerie, oisillon, oiselet, oiseleur, oiselier,16
4. oiseau, corbeau, moineau, becasseau, etourneau, vanneau.17
Thus, quite independently of its pronunciation, oiseau also fits
certain lexical and semantic patterns which find their expression
in the written language. It is ironic that what prevents Saussure
from reaching these eminently Saussurean conclusions is his mis-
guided insistence on treating individual units of notation as signs
of sounds.
That the concept of notation is not a mere artifact of integra-

tional theory is shown by a great variety of literate practices. An
interesting example is that of the lipogram: the literary form
which is based on the systematic avoidance of certain letters. It is
quite easy for the unforewarned reader to read a page or a whole
chapter of Georges Perec's novel La Disparition without realizing
that the text contains no examples of the letter <?.18 What is of
interest here is not the ingenuity of the enterprise, nor the ques-
tion of its literary value, but the simple fact that the rules would be
incomprehensible if a written text could not be analysed at the
level of notation. It has nothing to do with pronunciation. It has
nothing to do with spelling either, except insofar as the exclusion
of a particular letter automatically entails the exclusion of certain
"' 'bird', 'hen-bird', 'aviary', 'fledgling', 'small bird', 'fowler', 'bird-

fancier'.
17
'bird', 'crow', 'sparrow', 'sandpiper', 'starling', 'lapwing'.
18
G. Perec, La Disparition, Paris, Denoel, 1969.
words. In short, the lipogram would be inconceivable unless there

were writing systems based on fixed inventories of graphic units,
and unless it were possible to classify written texts on the basis of
the presence or absence of one of those units irrespective of any
phonetic value it might have or any function in the script.
The same notion is the cornerstone of the modern dictionary,
which arranges words in 'alphabetical order'. But lipograms are
much older than dictionaries. Trephiodorus in the fifth century
wrote an Odyssey in 24 books, of which each one omitted a letter:
alpha from the first book, beta from the second, and so on. Before
him Nestor of Laranda had composed a lipogrammatic Iliad. In
Latin, Fulgentius is the author of an Absque litteris de aetatibus mundi
et hominis, of which only fourteen books have survived. The oldest
known lipograms are those of the poet Lasos from the sixth cen-
tury BC. So there can hardly be any doubt that from a very early
period in the European tradition people understood, even if they
did not have a technical term for it, what a notation was: a set of
graphic units with its own structure. And the proof is that they
amused themselves by inventing a literate game which consists,
essentially, in superimposing the structure of a notation on the
structure of texts.
The distinction between notation and script also underlies the

familiar use of letter-forms and letter-names for metalinguistic
purposes. Thus 'Church begins with c' is false or nonsensical
unless the first six letters of the written form are taken as identify-
ing metalinguistically the subject of the proposition in question
and the last letter as identifying the same notational unit as the
first letter. All the rest of the sentence is to be interpreted non-
metalinguistically. The general phenomenon whereby, depending
on the circumstances of the case, letters stand to be interpreted in
one way or the other may be called 'notational ambiguity'. In
order to avoid notational ambiguity, writing commonly employ
devices such as inverted commas, underlining and italicization.
(E.g. '"Church" begins with "c"', or 'Church begins with <;'.)
Philosophers distinguish here between the 'formal mode' and the

'material mode' of expression, or between 'mention' and 'use'.19
Of particular interest for our present purposes is the asym-
metry between speech and writing that is thrown into relief by
cases of notational ambiguity. In order to avoid corresponding
ambiguities in speech, it is frequently necessary to make the point
more explicitly. (Thus, read aloud, 'There are twelve letters in this
sentence1 can still be misinterpreted, whereas 'There are twelve let-
ters in the words this sentence' deals with that potential problem.) In
short, a purely graphic device (italicization, inverted commas, etc.)
can function as the equivalent of a metalinguistic term. The fact
that this strategy is available - and commonly resorted to in liter-
ate cultures - means that we can if we wish, in written communi-
cation, choose to ignore or bypass the structure of the script and
have direct recourse to the notational structure that underlies it.
Nothing like this is possible in speech, for the simple reason that
spoken discourse has no notational structure. (An impression to
the contrary may sometimes be fostered by pedagogical tech-
niques such as 'sounding out' a word, but such techniques are in
fact examples of orthography projected into the domain of phon-
etic education, '[k] [a] [t] says [kat]' is a pseudo-sentence that has
no existence outside the classroom, and it is manifestly calqued on
'C-A-T spells cat'. The proof of this is that when asked how to
pronounce the word written cat, one answers by pronouncing it,
not by attempting the curious vocal feat of articulating three
phonetic isolates - unless, of course, one has been brainwashed at
school by a teacher of so-called 'phonies'. In a literate culture
based on alphabetic writing, 'How many letters are there in your
name?' is a genuine question. 'How many sounds are there in your
name?' is not.)
19
For a critical discussion of the way in which philosophers have used
such distinctions and the corresponding graphic devices, see R. Harris,
The Language Connection, Bristol, Thoemrnes, 1996. The case could be
made that were it not for writing conventions Western philosophy would
never have attributed such importance to these matters.
The semiological explanation of this asymmetry, once again, is

that writing incorporates two structural levels, which it is possible
to distinguish whenever that becomes necessary for purposes of
communication. There is, strictly speaking, no oral counterpart to
the written sentence 'C-A-T spells caf. If I say '[si ei ti] is pro-
nounced [kat]' that is self-evidently false, since I have just pro-
posed a three-syllable pronunciation of a monosyllabic form; and
even if I try to utter the elements of this form as three separate
articulations separated by pauses I fare no better, since that is
clearly not how I pronounce the word in question. I can, to be sure,
utter the phonetic tautology '[kat] is pronounced [kat]'. But this
still fails to match 'C-A-T spells cat', since 'C-A-T spells caf is no
tautology at all, but states a non-trivial fact of English orthog-
raphy. (There is no orthographic principle that requires C-A-T to
spell anything at all, any more than Z-X-T does.)
Centuries ago Quintilian came closer than Saussure ever did to

acknowledging the distinction between script and notation. He
offers the following observations on the methods used in his day to
acquaint Roman children with the alphabet:
I do not approve the practice which is followed in many cases:
teaching infants the names and the order of the letters before
their shapes. This makes it difficult for them to recognize the
letters, since the children do not pay attention to the form but
rely on their memory. Thus, when the teacher thinks the child
has mastered the letters in their customary order, they are
written out in reverse sequence, and then in all kinds of per-
mutations, in order to get the child to recognize them by their
form ( = facie) rather than their place (= ordine). The best way
is, as with persons, to introduce them by their appearance
( = habitus] and by name (= nomina) at the same time. But what
is an awkward method where letters are concerned is not to be
rejected when it comes to syllables. I do not disapprove the
practice of giving children ivory letters to play with, or any
such method that will encourage children's interest at that age,
when they love to handle things, look at them and name

them.20
What Quintilian's recommendation comes down to is this: first

teach the notation, and only when that has been mastered pro-
ceed to teach the child how to use it. What is interesting from a
semiological point of view is that Quintilian has no difficulty in
recognizing that notational structure as such is quite separate from
the structure of the writing system and can be taught
independently.
A general question about notation might appropriately be
raised at this point. Why does not physical configuration alone
suffice to identify the units of a notation? The answer might be
that a shape, as such, lends itself to various interpretations. Given
a prototypical example to copy, this can always be analysed visu-
ally in more ways than one, copied in more ways than one, and
hence give rise to variants. The possibility of confusion is reduced
if some alternative means of identification is added. In practice,
the expedients commonly employed are to give the shape its own
name or its own place in a series. What Quintilian is complaining
about in the passage cited above is that Roman schoolmasters in
practice commonly gave these ancillary forms of identification
priority over the actual letter-shapes when it came to teaching
their pupils. What he recommends instead is identifying each let-
ter ab initio by a combination of physical appearance and name.
For it is thus, he says, that we learn to recognize human beings.
The validity of the comparison, however, is open to question.
In the case of human beings, there is no particular limit on the
number we may need to be able to recognize, and there is no
guarantee that two will not share the same name. Accustoming
children to recite the letter-names by rote in alphabetical order
serves a purpose that Quintilian does not mention. What this
practice captures for the learner is the important semiological fact
that the letters constitute a small closed inventory, the members of
"° Quintilian, Institutio Oratoria, I, i, 24—26.

which stand in strict one-one correlation with their names. Quin-

tilian's recommendation would make more sense if the written
language to be learnt were Chinese.
Is the distinction between script and notation in any way parallel

to that which linguists draw for spoken languages between
morphology and phonology? Do not letters as notational units
correspond, in effect, to phonemes as phonological units?
Yes and no. The whole history of modern phonology shows
that the phoneme, as a theoretical concept, is based on the letter.
However sophisticated the phonological system, its basic seg-
mental units are invariably represented in transcription by alpha-
betic letters of some kind. Furthermore, the development of
phonemic transcription is inextricably bound up with efforts by
missionaries and others to devise writing systems for hitherto
unwritten languages. In this sense, the parallel between letter
and phoneme is just another manifestation of the scriptism that
pervades traditional Western thinking about language.
Semiologically, on the other hand, the parallel may easily give
rise to an illusion. The illusion consists of thinking that the phon-
emes are already there, like the pile of bricks waiting for the house
to be built. Whereas it is the other way round. The phonological
bricks depend on the construction of the building, not vice versa.
The building in question is oral discourse. And when that building
is up, it takes a great deal of sophisticated analysis to distinguish
the bricks. How many went into the construction and exactly what
shapes they had even the builders do not know, and these are
questions that may be answered differently by different analysts.
In the case of writing, on the other hand, we find nothing parallel
to this. Writing starts with bricks - alphabetic bricks in the case of
the languages of Europe - and no constructional progress is made
without first mastering that inventory. Learning to speak is not like
that: there is no list of phonemes to learn off by heart.
Perhaps someone will object that phonological analysis, even if
it tells us little directly about the processes of language learning,
must nevertheless reflect a speaker's cognitive organization 'at

some level or other'. For, in the end, do we not have to be able to
produce and recognize the acoustic differences between words like
pin, bin, tin, din, sin, etc.? Are not these phonemic differences? And
are not these minimal differential units, which 'mean' nothing in
isolation, exactly parallel in that case to the corresponding letters
in the written forms? So do not the phonemes of a language in
effect constitute its oral alphabet?
The equation is deceptive on all counts, precisely because
spoken discourse has no structural basis that corresponds to a
notation. It has no list of units, determinate in number, drawn up
in advance and handed down by tradition, constituting in itself an
independent cultural artifact. The phonological systems recog-
nized by modern phonologists are abstract systems of synchronic
oppositions derived from - and hence valid only in respect of -
the contrasts manifested in the morphology, syntax and vocabu-
lary of a particular language. Thus the notion that the same
phoneme might occur in two different languages is theoretically
incoherent. Whereas the notion that the same alphabet with the
same letters and letter-forms might be used in two different scripts
is not incoherent at all. And this is because a notation is indepen-
dent of a script and has its own existence.
But there is also a more basic reason for rejecting the equation
between letters and phonemes. Phonological analysis is based on
series of comparisons such as that cited above (pin vs. bin vs tin vs
din vs sin . . . ) in which isolated words are compared in order to
determine minimal differences. The reasoning the phonologist
then deploys, on the basis of this decontextualized 'evidence',
involves assuming that such minimal pairs of words must differ by
just one phoneme. But this logic is clearly debatable. That is to say,
from the fact that pin is consistently distinguishable from bin, tin,
din, sin, etc. it does not follow that the English language must
contain a phoneme /p/ which 'accounts for' that phenomenon.
No one would accept similar reasoning in chemistry, geography or
botany. Minimal differences do not prove the existence of corre-
sponding minimal units. If they did, it would not be difficult to
show that a ten-pence coin consists of two five-pence coins, or

large potatoes of smaller potatoes. That would be an atomistic
sophistry worthy of the scientists of the Grand Academy of
Lagado.
We are certainly not obliged to accept this kind of sophistry in
linguistics, unless we are anxious to make phonological analysis
conform to an a priori atomic model. And that seems, in fact, to be
just what modern phonologists have been anxious to do. Now the
lineage of their atomic model can hardly be in doubt. It becomes
obvious as soon as we consider that nowhere in the world except
in certain literate communities do we find an analysis of speech
that ends up by postulating an inventory of some twenty or thirty
separate atoms of sound. And those are communities whose
traditional form of writing is alphabetic.
Perhaps, nevertheless, there is a temptation to justify all this by
saying that the alphabet itself is an intellectual triumph of analy-
sis, capturing a truth about speech that pre-literate communities
have not yet recognized; namely that spoken discourse consists in
constant repetition and recombination of the same very limited
set of sounds. Here one recognizes another variant of the ancient
topos about savages and their primitive mentality. But today it can
hardly be taken seriously. No pre-literate community has so far
been discovered in which speakers had no grasp of oral repetition
and could not mimic or make mock of certain forms of speech
and idiosyncrasies of pronunciation. The notion of pre-literates
as phonetic simpletons does not stand up to a moment's investiga-
tion: in communities without writing the ear for niceties of speech
and accent is as well developed as in any literate elocution class.
What is beyond pre-literate comprehension, on the other hand, is
why anyone should suppose that the human speaking voice is
restricted to producing sounds from a mysteriously fixed inventory
of phonemes. For that reduction of their natural oral capacity is
comprehensible only when people have been conditioned, as
Europeans are, to thinking about speech in terms of writing.
CHAPTER FIVE
Alphabetical Disorder
Perhaps the most heretical implication of an integrational semio-

logy of writing is that it requires us to rethink the alphabet. Since
establishing the history of the alphabet is generally counted as one
of the crowning achievements of language studies in modern
times, it might seem outrageous to suggest that this great monu-
ment to learning is built on semiological foundations of sand. But
this is the conclusion to which one is led. It is not merely that
certain details of historical evolution and transmission are obscure
in the canonical history, but something much more serious. The
concepts 'alphabet' and 'alphabetical' are themselves not well-
defined. Worse still, different authorities interpret them differently
to suit their own arguments.
Saussure, as already noted in Chapter 2, held no very high
opinion of the way in which his predecessors had treated the
written evidence on which they based their reconstructions of
linguistic history. In order to avoid what he called the 'trap' into
which Bopp and his successors had fallen, Saussure made a point
of insisting from the outset that a language and its written form
constitute two quite different systems of signs.
But this decision brought immediately in its wake a problem of
academic strategy. Bopp's error was not fortuitous but in certain
respects necessary; without it the early development of Indo-
European linguistics would never have taken place. So Saussure,
intent on setting up an intellectual charter for modern linguistics,

faced an awkward choice. Either reject the study of 'dead' lan-
guages as falling outside the scope of the new science (which
would have been the more honest course); or else find a theoretical
justification that would allow the linguist to use ancient texts as
linguistic evidence, but at the same time steer clear of the 'trap'.
The solution was to postulate that the two systems of signs,
spoken and written, are bound together by relations which lend
themselves to semiological analysis.
It has often been noted that Saussure needed semiology (even if
the discipline did not yet exist) in order to provide a guarantee of
the autonomy and specificity of linguistics among the sciences.
Which is true enough. But that is only, so to speak, an external
reason. There was also an internal reason. Saussure also needed
semiology in order to explain how the linguist can be in a position
to examine a system of vocal signs which no longer exists and
therefore is not amenable to observation. The answer was to pro-
ceed via the intermediary of another system of signs, of which
evidence that is subject to direct scrutiny still survives. The need
for that strategy was pressing if the new structural linguistics was
not to deprive itself of being able to utilize whatever it needed
from the labours of previous generations of linguists.
Saussure thus found himself from the start in the position of
having to solve his own problem of the relationship between two
systems of signs, one spoken and the other written, which could
not be treated as completely unrelated. Given his theory of the
sign as a bi-partite entity, established by structural oppositions
internal to the system, the obvious solution was to make one of
the two systems a system of metasigns. This seemed to be the only
semiological connexion that could be invoked.
The range of possibilities for a Saussurean typology of meta-
signs is quite restricted. Given that each metasign must have its
own signifiant and signifie, this yields just three hypothetical types:
1 The signifie of the metasign is the signifiant of the first-order sign.

2 The signifie of the metasign is the signifie of the first-order sign.
Alphabetical Disorder 12 3
3 The signifie of the metasign is both signifiant and signifie of the

first-order sign, i.e. the first-order sign as a whole.
For all practical purposes, the third possibility inevitably merges
with the first two. Thus it should be no surprise to find that when
Saussure proposes a typology of writing systems it has just two
branches. Most commentators seem to regard this typology as
reflecting either a deliberate oversimplification on Saussure's part
or else his ignorance of the variety of existing writing systems,
whereas in fact it is the logical outcome of his theory of signs. His
typology of writing is part of his semiological theorizing, not a
synoptic description of the world's writing practices.
There are only two systems of writing:
1. The ideographic system, in which a word is represented by
some uniquely distinctive sign which has nothing to do with the
signs involved. This sign represents the entire word as a whole,
and hence represents indirectly the idea expressed. The classic
example of this system is Chinese.
2. The system often called 'phonetic', intended to represent
the sequence of sounds as they occur in the word. Some phon-
etic writing systems are syllabic. Others are alphabetic, that is to
say based upon the irreducible elements of speech.
Ideographic writing systems easily develop into mixed sys-
tems. Certain ideograms lose their original significance, and
eventually come to represent isolated sounds.1
The problem of 'ideographic' writing will be dealt with in due

course. For the moment, let us focus on matters more directly
relevant to the alphabet; namely, the semiological features of what
Saussure classifies as writing 'commonly called "phonetic"'.
Saussure's typology makes no mention of 'phonetic' writing
based on the analysis of distinctive features. Apart from modern
systems of stenography, the example usually cited is Han'gul, the
Korean writing system dating from the fifteenth century. It is a
1
remarkable system on several counts, originally introduced by

King Sejong and called Hun Minjong Urn ('Correct Sounds for the
Instruction of the People'). The royal decree promulgating
Han'gul claims that the writing system previously used, based on
Chinese characters, was not suitable for most people's purposes.
The new system was not a simplification of the old one, but a
radically different type of writing. It purported to have a meta-
physical foundation, the five basic vowels corresponding to five
universal elements: water, fire, wood, metal and clay. Thus, given
that the material world consists of various combinations of these
elements, it would follow that any vocalic sound could be repre-
sented by some combination of the corresponding symbols. The
logic of this explanation may not be altogether convincing, but
it is of interest from a semiological point of view as an attempt
to elaborate principles of phonetic representation within a
cosmological framework.
The feature of Han'gul that has attracted the interest of lin-
guists, however, is the formation of the letters, where certain
marks appear to be used systematically to indicate phonetic rela-
tions. For example, aspirated consonants are marked by a stroke
added to the character for the corresponding non-aspirate. Palat-
alization is likewise marked by another added stroke. It is evident
that the inventors of Han'gul had made a careful study of
articulatory phonetics, since they adopted letter-shapes represent-
ing schematic diagrams of the positions of the vocal apparatus -
another semiological complication, since this makes the charac-
ters simultaneously phonetic and pictographic. More interestingly
still, the royal decree claims that the system could faithfully repre-
sent even the howling of the wind, the cries of birds and the
barking of dogs. In short, this was an ambitious attempt to create
not just a system for recording speech but a map of the whole
universe of sound.2
2
For further details, see S. Lee, A History of Korean Alphabet and Movable
Types, Seoul, Ministry of Culture and Information, Republic of Korea,
1970 and G. Sampson, Writing Systems, London, Hutchinson, 1985, Ch.7.
Alphabetical Disorder 125
A number of writing systems utilize diacritic marks to indicate

tonal differences in a systematic way. Thai writing employs four
such marks to distinguish the five tones of the spoken language.
The Pahawh Hmong of Laos has no less than seven diacritics,
each corresponding to a tone when superposed to a vocalic letter,
but in other cases having a purely differential function with no
constant phonetic correlate. For instance, the letter H without a
diacritic stands for the diphthong [ai] with high level tone, where-
as the same letter with a superscript dot indicates the same vowel
with low tone. This tonal distinction is marked in the same way for
all vowels. The same diacritic dot is also found with consonants,
but in these cases has no fixed value. Thus R with no dot indicates
the consonant [m], but with a dot the aspirated affricate [tsh].3
These examples suffice to indicate why it is difficult to accept
Saussure's characterization of alphabetic writing as based on 'the
irreducible elements of speech'. Even for languages like French
and English, vowels and consonants can hardly be regarded as
'irreducible elements': they are phonetic segments of the 'speech
chain' or, more exactly, phonetic contours which characterize cer-
tain segments. When we look further afield we find writing
systems that recognize 'smaller' component elements into which
consonant and vowel sounds can be analysed. It should be noted,
however, that neither in the East nor the West do we ever find a set
of characters entirely based on phonetic distinctive features: at
most there are traces of the recognition of distinctive features in
certain systems.
There are problems too with Saussure's statement that phonetic
writing 'aims to reproduce the sequence of consecutive sounds in
the word'. One is that syllabic systems of writing usually give no
indication of the sequence of sounds within the syllable. (Saussure
can hardly be defended here on the ground that in syllabic writing
the syllable itself is treated as a single indivisible sound. For if that
3
W.A. Smalley, C.K. Vang and G.Y Yang, Mother of Writing: the Origin and
Development of a Hmong Messianic Script, Chicago, University of Chicago
Press, 1990, Ch.4.
argument were accepted, a system in which each character stood

for a whole word - or even a sentence - could likewise be classed
as 'phonetic', provided the order of characters in the written text
followed the sequence of corresponding units in the spoken utter-
ance.) Another problem is that if phonetic writing did always aim
to reproduce the sequence of consecutive sounds in the spoken
word there should be no systems in which the order of graphic
units fails to correspond to the order of phonetic units. However,
this non-correspondence is by no means a rare phenomenon. In
Thai writing, for example, the position of vowel marks in relation
to non-vocalic characters varies a great deal. The vowel mark may
be placed above, below, behind, in front of, half in front of and
half behind - or even half above and half behind - the nearest
consonant character. The 'actual' sequence of sounds can always
be worked out if one knows certain orthographic conventions:
nevertheless, it can hardly be claimed that the order of graphic
units 'reproduces' this sequence.
Saussure's typology of writing also leaves in doubt the status of
so-called 'consonantal' systems, such as early Phoenician. This has
attracted the attention of scholars because of its relevance to the
question of the history of the alphabet. On this, opinions are
much divided. It is the semiological rather than the historical
aspect of the controversy which is of interest here. However, it
should be said straight away that it is by no means easy to give a
summary of the accepted historical 'facts' without entering into
semiological interpretations. Even the term consonantal is suspect,
inasmuch as it might be taken to imply an alphabet without
vowels. It seems preferable, therefore, to approach the whole
problem by considering how it has been formulated by the
experts.
IJ. Gelb claims that Semitic writing of the so-called 'con-
sonantal' type is in fact syllabic writing.4 However, unlike more
typical syllabic systems, it omits any indication of the syllabic
vowel, because to mark it would have been superfluous for the
Gelb, op. cit. pp.76ff., pp.!47ff.

languages in question. Thus interpreted, each character in the

system has a syllabic value, albeit a more abstract value than is
usually found in syllabic writing, since it includes a whole range of
possible pronunciations. Thus k, for instance, may represent
[k] + [a], [kj + [o], [k] + [e], etc., as well as just [k].
According to James Barr, on the other hand, this is a para-
doxical way of looking at it:
From a certain point of view, indeed, one might conceivably
grant that in a Semitic consonant writing, with no vowels
marked, each sign represents 'a consonant plus any vowel or no
vowel' and is in that sense theoretically 'syllabic'. It is, however,
difficult to see any sense in insisting on the term 'syllabic' when
the script does absolutely nothing to tell you what vowel. A
Semitic writing like dbr or mlk determines graphically only the
consonants and in no way registers which vowels are in the
spoken form. I would say therefore that it is better to follow the
traditional terminology and call a script 'syllabic' only when it
docs something to specify the vowel of the syllable. It is more
economical and sensible to regard a Semitic consonant writing
as an alphabetical writing in which the vowels are not marked.5
At first sight it might appear that Barr is merely expressing a

terminological preference. The reasons he advances are not par-
ticularly compelling, and one wonders why the issue is being taken
that seriously. It is difficult to see exactly what the difference is
between the two propositions (a) 'Semitic writing is syllabic writ-
ing in which the vowels are not marked', and (b) 'Semitic writing is
consonantal writing'. But the question lurking in the background
emerges when Barr goes on to comment:
Professor Abercrombie writes: 'The invention of a system of
writing based on segments of the syllable has taken place once,
but only once: it was the brilliant discovery of the Greeks, and it
:>
J. Barr, 'Reading a script without vowels'. In Writing Without Letters,
ed. W. Haas, Manchester, Manchester University Press, 1976, Ch.4,
pp. 7 4-5.
gave us alphabetic, as distinct from syllabic, writing.' On page

168, note 6, he adds that 'Although the word "alphabet" is
Semitic in origin, Semitic writing systems are not alphabetic;
they are syllabic systems of a somewhat unusual kind.' I would
prefer to say that the alphabetic principle, in the sense of a
principle of marking in writing segments and not entire syl-
lables, was basically Semitic in origin; but the earliest Semitic
writing of this kind marked only certain segments, i.e. basically
the consonants, and left others unmarked. The Greek innova-
tion in this respect was to extend the principle and to mark all
segments alike. Reasons which may have influenced them in
this direction include the facts that their language structure had
(a) many words beginning with vowels and (b) sequences of
vowels with no consonant coming between.6
The mist lifts and one sees that this is not a trivial question of
terminology after all, because what is at issue is who shall get the
credit for having discovered the 'alphabetic principle'. This is
implicitly regarded as a great triumph in the history of the human
sciences, on a par with the discovery of the law of gravitation in
the history of the physical sciences. In short, from a Western
academic perspective, which is that adopted by most modern his-
torians of writing, the alphabet represents what Joseph Vendryes
once called 'the final perfection of writing'.7
In Barr's view, there is no doubt that the credit should go to the
Semites. He proposes to redefine the term consonantal as follows.
It can be very roughly said that the scripts of languages like
Arabic and Hebrew are 'consonantal' scripts and that the indi-
cation of vowels (a) historically was added at a relatively late
stage; (b} graphically is clearly additional to the consonantal
6
Barr, op. cit, pp.75-6. The reference to Abercrombie is to D.
Abercrombie, Elements of General Phonetics, Edinburgh, Edinburgh Uni-
versity Press, 1967, p.38.
7
'le dernier perfectionnement de 1'ecriture'. J. Vendryes, Le Langage,
Paris, Renaissance du Livre, 1923, p.356.
writing of a text, consisting in a series of points or marks above

and below; and (c] is optional, in that these points or marks may
be inserted or left out. Actually, however, even the so-called
'consonantal text' of such a Semitic writing commonly includes
the marking of some vowels. This was done, from quite ancient
times, in the following way: that certain of the consonant mar-
kers were also used to mark certain vowels. The consonant signs
mainly used for this purpose were four: h, w,y and aleph (or alif,
a glottal stop). Consonant characters used in this way are com-
monly called 'vowel letters' or (in Hebrew grammar) matres lec-
tionis. In the major Semitic texts of historic times, as opposed to
the early origins of Semitic alphabetic writing, this is an
important phenomenon. In central languages like Hebrew or
Arabic it is in fact omnipresent, and this is so alike in the text of
the Bible or the Quran and in the modern newspaper. In fact,
in the average Hebrew or Arabic text, quite apart from the
marking of vowels by points and additional marks, one in four
or one in five of the 'consonants' is in fact a vowel indicator. To
be strict, therefore, the usual statement that a Semitic script is
consonantal in nature should be amended somewhat as follows:
such a script is composed of signs of which all may stand for
consonants; of which the majority can stand only for conson-
ants but a minority may stand for either a vowel or a consonant;
and of which none can stand only for a vowel and never for a
consonant.8
The Greeks, it seems, are up against an adversary who does not

hesitate to exploit all the terminological tricks of the trade. Hav-
ing started off by rejecting the term syllabic, and insisted that con-
sonantal is the only apt designation for forms of writing that do not
mark vowels, he then claims that Semitic writing marks them after
all, at least sometimes. Clearly, writing in which vowels appear or
disappear according to which side of the argument you are on
must be alphabetic.
8
Barr, op. cit., pp.76--7.
The Greek cause is not championed either by David Diringer,

author of the best-known scholarly book on the history of the
alphabet.
Some scholars believe that, as the North Semitic script did not
possess vowels, it cannot be considered a true alphabet; accord-
ing to them, only the Greeks created an alphabetic writing. This
opinion is erroneous. The North Semitic was from the first
moment of its existence a true alphabet; at least, from the
Semitic point of view. It was not perfect. But perfection has not
yet been reached by any alphabet [. . . ].9
This claim raises the question: what would a perfect alphabetic

system be? Diringer's answer runs as follows:
Perfection in an alphabet implies the accurate rendering of
speech-sounds; each sound must be represented by a single
constant symbol, and not more than one sound by the same
symbol.10
Here we recognize virtually the same conception of alphabetic

writing as Saussure's, except that Diringer expresses it even more
bluntly. So this is an appropriate point at which to say why it is
semiologically unacceptable. The problem lies in the presupposi-
tion that the pronunciation of a word or phrase combines a finite
number of discrete elements which are its constituent 'sounds'.
This is a misconception, whether considered from an articulatory
or an acoustic point of view. To the extent that the notion has any
psychological plausibility at all, that is probably a postfacto product
of acquaintance with alphabetic writing itself. The phonetic
events that occur in the course of any given utterance are highly
complex: they can be described and analysed quite exactly in
various ways, depending on the criteria adopted. Experimental
phonetics leaves no room for doubt on this score. But the impres-
sion we may have of repeating exactly the same sounds when we
9
Diringer, op. cit., p.217.
10
Diringer, op. cit., pp.217-218.
repeat a word is an illusion. To ask 'How many sounds are there in

this word?' is to ask a nonsense question (for the same kind of
reason as it is nonsense to ask how many movements it takes to
stand up): a continuum can be described and analysed, but it does
not consist of a finitely denumerable concatenation of single
elements. It follows that the idea of an optimally 'correct' written
record as one that indicates the exact number of sounds occurring
in spoken discourse is nonsense too.
Diringer's naive belief in a complete set of one-to-one corre-
lations between letters and sounds should not be confused with a
more subtle misconception that is popular among linguists.
Although rejecting as absurd the idea that a perfect alphabet
would allow for the separate representation of every single sound,
or even of all the phonetic similarities and differences the human
ear can recognize, many linguists suppose that systems of alpha-
betic writing can be judged according to whether or not they give
an accurate representation of all the phonologically pertinent
oppositions in a language. This too is an absurdity if coupled - as it
usually is - with the theoretical assumption that there is a finite
number of such oppositions. The fact, for instance, that the words
pin and bin are phonologically distinct in English certainly provides
a good reason for distinguishing them orthographically in the
written language (i.e. in order to avoid confusion for the reader).
But from this it does not follow — and it would be an illusion to
suppose — that it must be possible to count exactly how many such
distinctions there are in English and provide a distinctive way of
marking each one alphabetically. The method of counting
depends on the principles of phonological analysis adopted. Here
too a perfect alphabet (one - and only one - letter for each
member of a phonological opposition) is a nonsensical ideal.
Diringer's notion of alphabetic perfection runs into another
problem. Any study of the human voice will detect certain resem-
blances between the many and various sounds it produces.
Although the initial sounds of pin and bin are different, they are
nevertheless similar. These similarities are problematic in the
sense that one is tempted to say that a common phonetic element
is shared. Yet Diringer's principle lays down that the same sound
must always be indicated by the same symbol and that two differ-
ent symbols must never be used to indicate one sound. Now within
the resources of an inventory of letters of the kind that is tradi-
tionally called 'alphabetic' it is difficult to see how to deal with
cases of the pin / bin type, where difference and similarity are
equally recognizable. Diringer's perfect alphabet, it seems, would
work only with a language where the sounds were as different
from one another as, say, [p] from [r] or [b] from [s]. In many
languages, however, this is far from being the case: contrasting
sounds tend to fall into pairs or sets sharing common features. So
the paradoxical side of the perfect alphabet turns out to be that if
Diringer's requirements were strictly applied the expected conclu-
sion would be the rejection of the alphabet altogether and the
adoption of a graphic matrix of phonetic features instead.
Thus, far from throwing any light on the status of writing sys-
tems like early Phoenician, discussions of the so-called 'alphabetic
principle' reveal a total failure by the historians of writing to think
through the logic of their own criteria.
This is already bad enough; but there is worse to come. Not
everyone agrees that 'consonantal' writing systems are phonetic
anyway. J.G. Fevrier in his book Histoire de I'ecriture expresses
serious doubts. Having pointed out that archaic Phoenician, with
its alphabet of 22 letters, does not use matres lectionis and retains no
vestiges of ideograms, determinatives or syllabic writing - and is
thus a perfect example of a 'pure' consonantal system — Fevrier adds:
This is writing that has rejected the ideogram but remains ideo-
graphic to some extent, since it notes only the root, irrespective
of the vowels.
In order to understand the birth of a conception of writing
which seems so strange to us Europeans, it is necessary to com-
pare the structure of an Indo-European word with the structure
of a Semitic word. In both we have a root, which gives the
sense, and modifications of this root which indicate the function
of the word in the proposition. But in Indo-European this root,
developed into a radical, forms a compact and relatively stable
block, to which prefixes and suffixes are added as functional

indicators. In French, starting from the radical pad ('speak'), we
have parl-er ('to speak'), parl-ant ('speaking'), parl-e ('spoken'), parl-
ous ('let us speak'), etc. In Hebrew, by contrast, starting from the
root QTL ('kill'), we have QeTol ('to kill'), QoTeL ('killing'),
QaTuL ('killed'), QaTaLnu ('we have killed'). Thus what remains
stable in the word and corresponds to the idea, not the function,
is not a kind of solid block of consonants and vowels but an
abstract group of consonants. Every Semite who hears a word
decomposes it, by instantaneous mental gymnastics, into con-
sonantal root and vocalic flexion. In writing, he is careful to
avoid anything which might occasion confusion between root
and flexion.
Considered from this point of view, Phoenician writing does
not seem to be so defective. Completely consonantal, it high-
lights with admirable clarity the consonant skeleton of the
word. Punctuation marks also help to isolate each root [. . . ]
Forms of Semitic writing derived from Phoenician have
inherited this desire to indicate clearly, above all, the consonant
skeleton, the root of the word. This is what explains their
reluctance, so strange in our eyes, to create for vowels signs like
those the consonants have. The example set by the Greek and
Latin alphabets was tempting; but they obstinately refused to
follow it because it would have drowned the consonants amid
the vowels. They preferred to have recourse to the system of
vocalic dots (Hebrew, Syriac, Arabic), or to the rather special
kind of syllabic writing found in Ethiopia, because in that way
the root retains its graphic autonomy.
One wonders whether such a conception of writing, even
though it was to lead to the comprehensively phonetic writing
of the Greeks, is not basically much nearer to primitive
ideography than the syllabic writing into which the various
cuneiform systems tended to develop.''
"J.G. Fevrier, Histoire de I'ecriture, 2nd ed., Paris, Payot, 1984, pp.210—
212.
Thus according to Fevrier the idea of consonantal writing as a

half-way stage between syllabic writing and alphabetic writing is a
twmpe-l'osil effect created by adopting a Eurocentric perspective. In
other words, we are dealing with a classification of written signs
and writing systems based on the assumption that the history of
writing consisted in a long, laborious progress towards a final peak
of achievement. The culminating point was the invention of the
alphabet in roughly the form we become acquainted with from
our childhood onwards; that is to say, as applied to the languages
of Europe. This prejudice is widespread among historians of lin-
guistics and of writing. (A typical example is the typology of writ-
ing proposed by Pedersen.12 It is also the typology adopted more
recently by Higounet.Li)
The same prejudice is to be found, although it may be less
obvious, among scholars who dispute the importance of the
Greek contribution and give the accolade for the invention of the
alphabet to the Semites. Their view is that consonantal writing
already is alphabetic, whereas according to their opponents that
was merely a tentative groping towards the alphabet. But on both
sides there is, in effect, agreement that alphabetic writing sur-
passes all previous systems. Where they differ is over the point at
which the alphabet emerges.
What is a semiologist to make of this sterile debate? From an
integrational point of view, the so-called 'alphabetic principle',
in spite of all that has been written about it, lacks a valid
semiological definition. And this lacuna is not fortuitous. It is no
coincidence that the alphabetic principle floats in this curious
theoretical vacuum: an alphabet is only a notation. On the basis of a
single notation it is possible in principle to construct an infinite
number of sign systems which have nothing in common as regards
12
H. Pedersen, Linguistic Science in the Nineteenth Century. Afothods and Results,
trans. J.W. Spargo, Cambridge, Mass., Harvard University Press, 1931,
p. 142.
13
Ch. Higounet, L'Ecriture, 7th ed., Paris, Presses Universitaires de
France, 1986.
their semiological structure other than using that particular no-

tation. In itself a notation does not determine the structure of
a script.
That is why in the end the only sense that can be made of the
term alphabetic writing is: 'a form of writing that uses notational
units derived historically from the series traditionally known as
"alphabetic" and employed in various cultures from the second
millennium BC onwards'. This is not as circular as it may sound.
For the only way to check whether a script is or is not alphabetic is
to trace its historical affiliations: the variations of form are now so
great that it is no longer possible to recognize 'the alphabet' on
inspection. To say 'Alphabetic writing is writing that uses the
alphabetic principle' is vacuous in the absence of any coherent
definition of the principle in question.
To test this out, let us go back to Saussure's famous example of
the word oiseau. Is this written form an example of alphabetic
writing? How do we determine whether it is or not? It would be
folly to try to resolve the question on the basis of pronunciation
(i.e. to determine whether the spelling conforms to any 'alphabetic
principle'). Everyone agrees that however oiseau may nowadays be
pronounced that pronunciation is the result of a whole series of
historical accidents in the history of French that could not possibly
have been predicted. So whatever the original justification for the
spelling might have been, it no longer holds. The form survives
today as a graphological fossil. And yet, it will be said, oiseau
remains an 'alphabetical' spelling. But what does that amount to
other than saying that it perpetuates the historical employment of
letters belonging to a certain written tradition?
One could propose on the basis of exactly the same consider-
ations the opposite conclusion: oiseau is no longer (if it ever was)
an alphabetical form, because it fails to meet the conditions of
correspondence between letters and sounds that the 'alphabetical
principle' requires. This line of argument would have to be
backed up by providing a list of the correspondences claimed as
being in accord with the principle. That could doubtless be done
(although Saussure fails to do it in defence of his condemnation of
the spelling oiseau): the trouble is that it could be done only too
easily, and differently by different arbiters. What is less clear is how
the listing could be anything other than an attempt to impose
normative standards on the orthography of French. But a more
serious objection is the risk of finding as many different - and
conflicting - 'alphabetic principles' as there are demonstrable
correlations between letters of the alphabet and the pronunciation
of French words. (Thus, for instance, the letter s correlates with a
voiceless sibilant in sire and a voiced sibilant in bise. But we cannot
have both. How does the arbiter decide which of the two corre-
lations is in accordance with the alphabetic principle and which
is not? And likewise for countless other cases?)
None of this means that it is impossible to take an alphabetic

notation and construct a script based on one-to-one correlations
between letters and phonetic units. On the contrary, that is often
done and the results utilized in modern dictionaries to give a
rough indication of how words are pronounced. It can indeed be
done for any language and at almost any pedagogic level. What
does not follow, however, is that this rather specialized use of
alphabetic notation is more 'correct' than its use in everyday,
traditional spelling. Even less does it prove retrospectively that this
was what the original inventors of the alphabet (whoever they
were) had in mind.
From a strictly phonetic point of view, every syllable is a con-
tinuum with a certain duration, which can be divided for purposes
of analysis into as many consecutive segments as may be wished.
By means of modern electronic equipment it is possible to obtain
detailed acoustic information about any one of these segments.
But the question 'How many such segments need to be repre-
sented in a faithful transcription of the syllable?' is self-defeating.
For the only kind of representation which is 'faithful' to the phon-
etic facts would be one which did not divide the continuum into
segments at all - as we see in a sound spectrogram. Alphabetic
transcription inevitably misrepresents speech to the extent that it
Alphabetical Disorder 13 7
is obliged by its own conventions to mark a series of subdivisions

that do not exist. For the semiologist, any belief that an optimally
accurate alphabetic transcription mirrors the structure of the
utterance is rather like supposing that the best kind of drawing of
a jet of water must be one in which each droplet is separately
shown.
The problem cannot be resolved by making a few minor
emendations to currently held theories of the alphabet, or correct-
ing overambitious claims that historians have made for it. Once
the integrational distinction between script and notation is recog-
nized, the whole question of the alphabet appears in a different
light. We are led inevitably to the conclusion that, from a
semiological point of view, there is no such thing as alphabetic writing.
In other words, in spite of the importance that is —justifiably —
ascribed in cultural history to what are described as systems of
'alphabetic' writing, the heterogeneous scripts thus classified do
not in fact have any special status at all in the semiology of writing.
It is not even clear that they belong unequivocally in the domain
to which Saussure (along with others) unhesitatingly assigned
them — that of 'phonetic' writing. The alphabet is the keystone to
Western thinking about writing. Without that keystone securely in
place, everything else about writing needs rethinking too.
CHAPTER SIX
Ideographic Hallucinations
Reasons for rethinking the notion of 'phonetic writing' inevitably

go hand in hand with reasons for rethinking other forms of writ-
ing. Saussure's other main category is ideographic writing. In
ideographic writing, according to the Cours:
a word is represented by some uniquely distinctive sign which
has nothing to do with the sounds involved. This sign represents
the entire word as a whole, and hence represents indirectly the
idea expressed.l
This definition calls for examination, being in certain respects

remarkably un-Saussurean. Here the term sign (signe) seems to do
duty for what Saussure elsewhere calls the signifiant. So what are we
to make of this signe unique?
Are ideographic systems those in which each word has only one
written form? That is, are there no graphic variants in ideo-
graphic writing? Or are ideographic systems those in which each
word has a written form which belongs to it alone? More exactly,
are there no homographs in ideographic writing? These ques-
tions are not made any easier to answer in the light of the fact
that Saussure chooses Chinese as his 'classic example' of ideo-
graphic writing. For it would be hard to maintain that Chinese
1
Ideographic Hallucinations 139
characters have no variants or that Chinese writing has no

homographs.
Another possibility would be that Saussure subscribed to the
common (Western) view that Chinese is a 'monosyllabic' lan-
guage. And since each syllable has its own written character, it
would then be easy to assume that in Chinese the number of
words equals the number of syllables. Is this what he means?
Whatever the answer, it is difficult to see why ideographic writing
as such should be forced to comply with any principle of one-to-
one correspondence between word and ideogram.
A certain amount of light is thrown on the enigma by compar-
ing the text of the Cours with the corresponding passage in Con-
stantin's notebooks. There we read, in the notes on Saussure's
lecture of 6 December 1910:
Two main systems of writing are known.
(1) the ideographic system which attempts to represent the word
without bothering about the constituent sounds <(but the aim is
indeed to represent the word, not the idea)>, using therefore a
single sign, which can only relate to the idea contained.2
This at least allows us to follow Saussure's train of thought a

little more clearly. He is trying, apparently, to distinguish between
(a) systems of writing in which each word is represented by a
signifiant which comprises a single monolithic block, and (b) sys-
tems of writing in which each word is represented by a sequence
of signifiants. In other words, signe unique is to be interpreted as
implying a simple, autonomous graphic unit. This is the exact
opposite of alphabetic writing, where each word is normally com-
posed of a sequence of individual letters which are themselves (in
Saussure's view) signs. Hence the otherwise puzzling 'therefore'
(done] in the latter part of Constantin's sentence. In short, the idea
is that in ideographic writing the whole word is represented by a
single sign, because this kind of writing pays no attention to the
vocal signifiantas such.
2
Komatsu and Harris, op. cit., pp.41-2.
What also emerges clearly from Constantin's note on this point

is Saussure's reluctance to endorse the common view of an ideo-
gram. The term itself was invented by Champollion in connexion
with the decipherment of Egyptian hieroglyphs. Given their
etymological associations, ideogram and ideographic) seem appro-
priate words for discussing a 'writing of ideas'.
This is evidently what Champollion thought. In his now cele-
brated Lettre a M. Dacier (1822), Champollion commented on the
hieratic and demotic forms of Egyptian script:
these two kinds of writing are, both of them, not alphabetic, as
had generally been supposed, but ideographic, like the hiero-
glyphs themselves, that is to say depicting the ideas and not the
sounds of a language.3
Brilliant Egyptologist as he was, Champollion had little interest

in any general theory of writing: he merely took on trust a tradi-
tional distinction that goes back to the Greeks, who regarded
hieroglyphs as a 'pictorial' form of script (unlike their own). That
is precisely why Champollion thought he had made such an
important discovery in demonstrating the existence of 'phonetic
hieroglyphs' (as he called them). These were for him 'exceptions'
to 'the general nature of the signs of this writing', because they
could 'express the sounds of words' (exprimer les sons des mots}.4
According to Saussure, however, the whole notion of being able
3
Italicization reflects the original text as published in J-F. Champollion,
Lettre a M. Dacier, Fontfroide, Bibliotheque Artistique et Litteraire, 1889,
p. 1. (Dacier was the secretary of the Academic Royale des Inscriptions et
Belles-Lettres.)
4
Champollion's lack of (modern) theoretical sophistication is evident
throughout the Lettre a M. Dacier, which introduces the category of 'semi-
alphabetic' alongside the notion of 'strictly alphabetic' (proprement dit),
without giving any definition of the latter. Saussure, we must not forget,
had an uphill struggle against this kind of muddle, which was still current
in his day. (Champollion died in 1832, and his decipherment of Egyp-
tian hieroglyphs was one of the monumental 'modern' achievements for
linguists of Saussure's generation.)
to write down an idea is a misapprehension, due simply to the fact

that ideographic writing is not phonetic. There is no 'writing of
ideas', strictly speaking: there are only forms of writing which pay
no attention to the signifiant, and this gives rise to the inference that
the signs must somehow represent ideas instead. Ideographic writ-
ing, in Saussure's view, is also designed to record speech, but does
so less effectively than phonetic writing because it has no way of
indicating the structure of the image acoustique.
Here again we see how Saussure's typology of writing is deter-
mined in advance by his semiological assumptions. The bipartite
division into 'phonetic' and 'ideographic' is a direct projection of
the postulated relationship between written sign and vocal sign. In
one case it is the vocal sign which is the signifieof the written sign,
whereas in the other case it is not. We also see why syllabic writing
has to fall together with alphabetic writing, in spite of failing to
indicate the sequence of sounds in the word. And finally the
unhappy terminology chosen for the two main divisions ('phon-
etic' versus 'ideographic') falls into place: 'phonetic' not because
sounds as such are signified (Saussure takes some trouble to point
out that this is not so) but because that kind of writing reflects the
structure of the image acoustique] and 'ideographic' not because
ideas as such are signified but because that kind of writing ignores
the image acoustique altogether. In short, the essential difference for
Saussure between the two categories of writing is that between (i)
writing in which the minimal units signified are themselves signs
and (ii) writing in which the minimal units signified are not signs.
He would have done better to call them (i) ecriture semique and (ii)
ecriture asemique respectively.
Although Saussure's typology of writing systems is patently dic-

tated by theoretical elegance rather than empirical observation,
Saussure does not hesitate to draw psycholinguistic conclusions
from it.
The written word [ . . . ] tends to become a substitute in our
mind for the spoken word. That applies to both systems of
writing, but the tendency is stronger in the case of ideographic

writing. For a Chinese, the ideogram and the spoken word are
of equal validity as signs for an idea. He treats writing as a
second language, and when in conversation two words are iden-
tically pronounced, he sometimes refers to the written form in
order to explain which he means. But this substitution, because
it is a total substitution, does not give rise to the same objection-
able consequences as in our Western systems of writing. Chi-
nese words from different dialects which correspond to the
same idea are represented by the same written sign.5
On the basis of this evidence, Saussure clearly qualifies for

inclusion in H.L. Chang's list of Western scholars who betray
their ignorance of the Chinese language by making rash general-
izations about Chinese writing.6 The list of guilty parties begins
in the seventeenth century with John Wilkins and goes down to
Derrida in the twentieth: it includes Leibniz, Descartes and Hegel.
Derrida describes Western views of Chinese writing as split
between 'ethnocentric scorn' and 'hyperbolic admiration', and
opines that we are dealing with a kind of 'European hallucina-
tion'. Chang agrees about the hallucination, but regards Derrida
himself as a victim of it. Those under the spell of this hallucina-
tion see Chinese writing as having a quite different relationship
with the spoken language than that which obtains between writing
and speech in Western countries. They are fascinated on the one
hand by the gap which they perceive between written and spoken
forms in China, and on the other hand by the 'visible' connexion
(in certain cases) between the Chinese character and its meaning.
Leibniz thought this a great advantage for a writing system and
predicted that it would become a universal means of communica-
tion. For Hegel, quite the opposite: this was a great disadvantage,
3
Saussure, op.cit., p.48.
6
Han-Liang Chang, 'Hallucinating the other: Derridean fantasies of
Chinese script', Center for Twentieth Century Studies, Working Paper No.4,
1988.
because it made it impossible to indicate in writing all the nuances

of speech. Derrida's illusion, according to Chang, follows in the
same tradition; for Derrida cites Chinese as proof that Chinese
culture has succeeded in avoiding the phonocentrism which
dominates and distorts European thinking.
This is not the place to pursue Chang's interesting argument,
but it raises an interesting question about the nature and extent of
Saussure's 'hallucination' about Chinese writing. Although the
distinction between phonetic and ideographic systems is funda-
mental to Saussure's semiology of writing, he admits that the
difference is not always as clear in practice as it may appear in
theory, because
Ideographic writing systems easily develop into mixed systems.
Certain ideograms lose their original significance, and eventu-
ally come to represent isolated sounds.7
Thus although he admits the existence of 'mixed' systems of

writing (i.e. systems which combine phonetic and ideographic
signs), Saussure does not place Chinese writing in this class. For
him, it remains the 'classic example' of the ideographic type. The
admission that there are 'mixed' systems is nevertheless theoretic-
ally significant. It implies that the criteria which distinguish the
various types apply in the first place at the level of the individual
sign, and only consequentially at the level of the system.
There is a problem here on which Saussure does not comment;
how, in the case of a mixed system, can the analyst determine —
sign by sign - which are ideograms and which are not? This may
seem a minor methodological point. Within a Saussurean frame-
work, however, it is far from trivial. For what is under threat here is
the entire Saussurean notion of a semiological 'system', i.e. a
whole held together by an internal structure of oppositions. It is,
in principle, easy to see how one phonetic sign contrasts with
another or one ideogram with another; but in what sense a phon-
etic sign can contrast with an ideogram within the same system is by
7
no means evident. If it can do so, however, that raises far-reaching

and (for Saussure) awkward questions. (For instance, why cannot
written signs contrast with spoken signs within a single system?
But then the whole edifice of Saussurean linguistics would
collapse.)
Before pursuing Saussure's conception of Chinese writing any

further, it is apposite to say something about the way in which
Chinese characters are classified in the indigenous Chinese tradi-
tion. Six categories are usually recognized and the typology dates
back to at least the second century. What is summarized below is
the account given by Y.R. Chao: his translation equivalents are
also adopted.8
The first category of characters are shianqshyng ('pictographs'),
the second jyyshyh ('simple ideographs'), the third hueyyih ('com-
pound ideographs'), the fourth jeajieh ('borrowed characters'), the
fifth shyngsheng or shyesheng ('phonetic compounds') and the sixth
joanjuh ('derivative characters'). It is interesting to note that Chi-
nese authorities themselves do not always agree in which category
a given character should be placed. However, Chao provides the
following as typical examples.
1 shianqshyng: the character meaning 'sun' in early Chinese writ-
ing is a circle with a dot in the centre.
2 jyyshyh'. the character meaning 'three' comprises three hori-
zontal strokes.
3 hueyyih. the character meaning 'bright' is a compound of the
characters for 'sun' and 'moon'.
4 jeajieh: the character meaning 'come' is a pictogram represent-
ing a kind of wheat. This cereal has a name that happens to be
pronounced in the same way as the word for 'come'.
5 shyngsheng or shyesheng. the character meaning 'burn' is a com-
8
Y.R. Chao, Mandarin Primer, Cambridge, Mass., Harvard University
Press, 1948, Gh.4.
Ideographic Hallucinations145
pound of the two characters 'burn' + 'fire', the former having

been taken over as the character for the word 'thus', which
happens to have the same pronunciation as 'burn'.
6 joanjuh: the character meaning 'enjoy' is formed by adding an
extra stroke to the character meaning 'propitious', to which it is
felt to be related in meaning.
It is not difficult to see how some of these categories might
cut across one another. To make matters more difficult, some of
the categories have alternative definitions. (For example, the
second category is sometimes said to be the category of charac-
ters 'indicating an action'.) But without going any further it
would seem that Chinese writing is rather more complicated
than Saussure's cursory remarks allow for. If Chinese is indeed
the 'classic example' of ideographic writing that Saussure
claims, and if Chao's descriptions of the processes of character
formation are correct, it begins to look as though ideographic
writing is something of a disaster area for Saussurean semio-
logy. In particular, it appears that not all ideographic signs are
arbitrary.
Chao, however, like Saussure, insists that Chinese characters
represent spoken words, not ideas:
[ . . . ] from very ancient times, the written characters have
become so intimately associated with the words of the language
that they have lost their functions as pictographs or ideographs
in their own right and become conventionalized visual repre-
sentations of spoken words, or "logographs". They are no
longer direct symbols of ideas, but only symbols of ideas in so
far as the spoken words they represent are symbols of ideas.9
Chao goes on to say that this has been recognized by Sinologists

since the first half of the nineteenth century, and he refers to the
publication in 1838 of A Dissertation on the Nature and Character of the
Chinese System of Writing by P.S. du Ponceau. (In which case, it is
9
Chao, op. cit., pp.60-1.
possible to see Saussure's remarks about Chinese as merely sum-

marizing a communis opinio held by the scholars of his day.)
The matter can hardly be left there, however, even if all the
world's Sinologists spoke with one voice on this question, since it is
not at all clear what is meant by the claim that Chinese characters
represent spoken words 'directly' and ideas only 'indirectly'. Is this
a psychological claim or a semiological claim?
If it is a psychological claim, is it open to verification? If so,
how? Is it a matter of determining for how many Chinese the charac-
ters represent in the first instance spoken words and ideas only
secondarily, and for how many the priority is the other way round?
And would this be an investigation of Chinese people's opinions,
or of their neurolinguistic processes? Are all Chinese included, or
only those who can read and write? And Chinese fluent in several
dialects, or only monoglots? How, exactly, is the linguist or the psy-
chologist to set about collecting evidence bearing on any of this?
As soon as these questions are posed it begins to emerge just
how obscure any psychological 'priority' claim is. The obscurity is
particularly problematic for Saussure, who champions the thesis
that it is la langue which furnishes speakers with their stock of ideas.
One wonders how to make psychological sense of the notion of
'indirect representation' of ideas, given that the linguistic sign
itself, according to Saussure, unites concept and image acoustique in
the mind, and that the two are inseparable.
Perhaps, on the other hand, it is a mistake to treat the claim as a
psychological thesis. Does it make any better showing as a semio-
logical thesis? Hardly. For Saussure, an ideogram is a metasign: it
is the written sign of a vocal sign. But in that case, it is hard to see
any justification for 'indirect representation' of ideas, given that
the Saussurean vocal sign combines a concept with an image acous-
tique. Where in this analysis is there any room for a second level of
representation? The concept is already present in the vocal sign.
However, if we are to regard that state of affairs as constituting
eo ipso 'indirect representation', then by the same token it would
seem inevitable that the metasign also includes an indirect repre-
sentation of the image acoustique. For that was also originally present
in the vocal sign. This conclusion would be entirely compatible

with Saussure's admission that sometimes ideograms end up as
representations of sounds. (It is difficult to see how that could
happen unless the metasign began by signifying a combination
which included an image acoustique.}
To that extent, the logic of the metasign seems consistent with
the thesis of'indirect representation'. That is to say, the metasign,
in virtue of signifying a vocal sign, is held to 'represent indirectly'
both the concept and the image acoustique associated with that vocal
sign. So far, so good. The trouble is that this doctrine, although
internally consistent, at one stroke destroys the basis of Saussure's
taxonomy of writing systems. In other words, every writing system
now emerges with the capacity to represent indirectly both the
meaning and the pronunciation of vocal signs. The semiological
raison d'etre for distinguishing between 'phonetic writing' and
'ideographic writing' has vanished.
Is there any evidence that Saussure was aware of this theoretical
impasse? Constantin's notes report an observation which did
not survive into the published text of the Cours. In his lecture of
20 December 1910, Saussure remarked in connexion with
orthography:
One must not forget that the written word eventually becomes,
through force of habit, an ideographic sign. The word has a
global value < independently of the letters of which it is
formed>. We read in two ways: spelling out unfamiliar words
and reading familiar words at a glance.10
This seems to confirm the conclusion that, for Saussure, the

ideogram is defined negatively with respect to alphabetic writing,
i.e. by its failure to give an analysis of the image acoustique. But it
is nevertheless a surprising admission for someone wedded to
Saussure's account of writing.
The notion of a gradual degeneration of alphabetic writing
through habit is to be found in Hegel, as Derrida points out:
10
Komatsu and Harris, op. cit., p.64.
Acquired habit also eventually suppresses the specificity of

alphabetic writing, namely of taking the eye a roundabout way
through the sense of hearing in order to reach what is repre-
sented, and turns it into a kind of hieroglyphics, so that we no
longer need to be aware of the intermediary role of sounds
when using it.11
But Derrida does not seem to notice that whereas this kind of
admission poses no theoretical problems at all for Hegel, in Saus-
sure's case it certainly does. Saussure is doubtless right to say that
a reader does not need to 'spell out' words that are familiar. But if
that means that all the familiar words in an alphabetic text have
become ideograms, and only the unfamilar ones retain the status
of phonetic writing, it follows that whether or not a written sign is
an ideogram no longer has anything to do with its actual graphic
form. And that, at one stroke, puts paid to Saussure's typology of
writing. Theoretically even more damaging - if that were possible
- is that the semiological status of the metasign becomes depen-
dent on the competence of the reader, i.e. the individual; and that
puts paid to any theory of writing systems. (It belongs to the
graphic counterpart of a study of parole, not of langue.} In short,
the remark reported by Constantin threatens to demolish the
entire edifice of a Saussurean semiology of writing. (It confirms,
however, albeit unwittingly, exactly what an integrational theorist
would claim; namely, that the semiological status of any given
graphic configuration depends on how it is contextualized in par-
ticular cases. The reader who has to 'spell out' unfamilar forms is
engaged in a different integrational programme from the reader
who has no need to do so.)
On first inspection it is not obvious how Saussure has come by
this self-inflicted injury. Did he have to concede that alphabetic
writing can 'become' ideographic? The explanation lies in his
conception of the relation between alphabetic writing and speech.
He needs to account somehow for the existence of cases like oiseau
11
Cited by Derrida, op. cit., p.40.
and, more generally, for the fact that alphabetic forms are not
altogether reliable evidence about pronunciation. As noted earlier,
Saussure inveighs against the alleged 'immobility' of writing and
holds it responsible for most of the 'discrepancies' between spell-
ing and pronunciation. This in turn raises the question of what
causes 'immobility' in writing.
Saussure's cryptic answer is contained in the observation Con-
stantin reports: the written sign tends naturally to become an
ideogram. This is an inevitable consequence of the social use of
writing and hence familiarity with written forms. It supplies the
reason why the French word for 'king' has been able to keep its
medieval spelling in spite of all the sound changes that have
occurred in the interim. The form roi has become in practice an
ideogram and can be accommodated to any pronunciation, just as
Chinese characters can. The history of phonetic writing is a
constant struggle between establishing the orthographic analysis
of the vocal signifiant at a given period and the countervailing
tendency towards ideography. This latter tendency is inherent in
writing and has nothing to do with diachrony: hence the term
conservatism is inappropriate and immobility is preferable.
If this interpretation of Saussure's thinking is on the right lines,
it suggests either that Saussure thought it more important in the
end to explain the mismatch of alphabetic spelling with pronunci-
ation, or else that he simply failed to distinguish at crucial points in
his theorizing between the semiology of writing and the psycho-
logy of reading. In this connexion it is remarkable that Saussure
had nothing to say about abbreviations. For it is in this area of
writing that he would have found the most plausible examples of
alphabetic signs being reduced to ideograms. Abbreviation, he
could have argued, is the clearest indication that the word is so
familiar that we can dispense with the usual representation of the
image acoustique.
This, however, would have taken Saussure much closer to the
integrationist position than he would (or should) have been willing
to move. The whole phenomonenon of graphic reduction (in the
sense of substituting a less complex for a more complex graphic
form) is one which demonstrates the validity of the integrationist

principle that signs articulate the co-ordination of activities.
Graphic reduction starts from one written form and 'reduces' it to
another: it essentially involves a comparison between the original
and the substitute, seen as alternative ways of achieving the
writer's purpose. The common methods of alphabetic abbrevia-
tion are well known; but theorists of writing have failed to realize
their theoretical implications. Saussure must have been perfectly
familiar with the practice of writing Madame as Mme\ but seems
never to have reflected on the circumstances in which this
abbreviation is found, or on the significance of the fact that there
is no corresponding oral form *[mm]. Had he done so, it might
have led him to revise or qualify his view that the basic function of
alphabetic letters is to 'represent' the image acoustique. For if that
required Madame in the first place, it is hard to see how leaving out
the three letters ada makes no difference. Here there is no question
of appealing to the alleged 'immobility' of writing (as in the case
of oiseau): for this is a case of deliberate contraction of the original
orthographic form. In order to contruct a 'phonetic' explanation
of the abbreviation, one would presumably need to make out a
case for saying that some letters give more information about the
image acoustique than others. But even if such a case could be made,
only someone brainwashed by the phonocentricity of the Western
tradition is likely to believe it.
For the integrationist, 'phonetic' explanations of alphabetic
abbreviations simply obscure what is taking place in such cases.
Once it is realized that the sole function of written signs (in glottic
writing) is to integrate visual, manual and oral aspects of com-
munication, one would in general expect simplification of the
written form to occur when there is no risk of compromising the
integrational function of the sign in question, irrespective of the
phonetic structure of the corresponding utterance. The validity of this
assumption is independently corroborated by numerous pieces of
evidence. Among them the following might be mentioned: (i) the
whole history of Western shorthand systems from Graeco-Roman
antiquity down to the present day, (ii) the simplification of Chinese
characters in writing reforms (aimed primarily at reducing the

number of strokes without obscuring the visual identity of the
character), (iii) the common European practice of using figures to
replace corresponding number words (e.g. 'Louis XIV), (iv) the
'borrowing' of alphabetic abbreviations from other languages
(usually Latin in the case of English), such as 'cf for 'compare'
and 'lb.' for 'pound', (v) the use of personal initials to substitute for
a full name, but only when the identity of the person in question is presumed
known to the reader, (vi) cases where abbreviations are treated as
having a 'grammar' of their own, independently of the grammar
of the words they 'stand for' 12 , and (vii) the current proliferation
of acronyms in newspapers and other printed publications. In this
last category we find examples which would constitute yet another
kind of 'monstrosity' for Saussure. These are instances like GATT
( = General Agreement on Tariffs and Trade) which have actually given
rise in the spoken language to a re-phoneticized form of the acro-
nym (pronounced [gat]). Some of these new formations enter into
circulation so rapidly (e.g. AIDS) that current speakers may have
forgotten or not even know what they originally 'stood for'. All
these cases illustrate various aspects of the integrational function
of the written sign in a literate society.
One further point calls for special comment here. The develop-
ment of shorthand systems in the West, from Greek tachygraphy
and Latin notae Tironianae1^ onwards, while presupposing an
already established tradition of 'full' writing, is motivated by the
need, as Peter Bayles put it in his Arte of Brachigraphie in the late
sixteenth century, 'to write as fast as a man speaketh'. Modern
shorthand systems actually enable experts to write considerably
12
Quite common in Latin epigraphy. Sandys (Latin Epigraphy, 2nd rev
ed., London, 1927, p.292) cites the remarkable example of DDDD
NNNN FFFF LLLL for dominis nostris Flaviis quattuor, i.e. the siglae are not
merely pluralized (by reduplication, which is a frequent practice) but
'quadruplized', because here there are four emperors in question.
13
Traditionally named after Cicero's secretary Tiro, who supposedly
invented the system or adapted an earlier one.
faster than that (up to about 250 words a minute). But the relevant
point here is that writing 'as fast as a man speaketh' is a special
integrational requirement based on the biomechanical factors
involved in the two different activities. In other words, if we ignore
the integrational function of the written sign, the invention of the
scripts we call 'shorthand' becomes theoretically inexplicable.
Writing is commonly discussed in the Western tradition as if both
writing and speaking were 'timeless' activities. Shorthand is a
reminder that they are not. Nor can we suppose that there are no
temporal contraints at all on scripts other than shorthand systems.
Had it taken the speediest scribe in antiquity an hour to write each
letter of the alphabet, we can be fairly sure that, at the rate of
approximately two-words-per-working-day, neither Cicero nor
any other orator of antiquity would have bothered to dictate his
speeches.
Would it not be possible for a theorist who outdid even Saussure
in phonocentricity to dismiss these problems by roundly declaring
all abbreviations to be ideograms? Yes, it would. Whether it is
plausible to regard Mme (for Madame) as falling into the same semi-
ological category as a Chinese character is another question. Both
do indeed fail to offer any detailed 'phonetic analysis' of the signifi-
ant. But that observation simply brings us back via another route
to the original fault line in Saussure's typology, where ideographic
writing was defined negatively with respect to phonetic writing.
Is there, then, a semiologically viable definition of the term ideo-

gram - a definition of the kind that Saussure failed to provide?
Theorists are far from unanimous.
There are some, like J.G. Fevrier, who use this term for the
symbols of a Wortschrifi which designate objects but give no indica-
tion of how the words in question are pronounced.14 This usage,
although not coinciding with Saussure's, is not far removed from
it. There are other authorities who reject the term altogether.
14
Fevrier, op.cit., p. 103.
Ideographic Hallucinations 15 3
There are others again who admit the existence of ideograms but
maintain that ideograms are devices that cannot serve as the basis
for a system of writing. In other words, such theorists reject not
the notion 'ideogram' but the notion of 'ideographic writing'.
One of these is John DeFrancis, for whom an ideogram (or ideo-
graph) is 'a symbol representing a meaning without indicating a
pronunciation'.13 In his view individual ideograms can serve cer-
tain communicational purposes (for example, an ideogram may
mean 'No smoking') but cannot be used as the foundation of
writing: the term 'ideographic writing' is simply a misnomer.
Although this sounds like a challenging thesis, it turns out to be of
no semiological interest whatever, being based on DeFrancis's
refusal to count 'strictly' as writing any graphic system which lacks
the capacity to record the totality of utterances in a spoken lan-
guage. Having made this arbitrary decision, DeFrancis finds no
difficulty in convincing himself that all traditional forms of writ-
ing are more or less phonetic, but some systems are phonetically
better than others. Hence the title of his book: writing, as he
understands it, is nothing other than 'visible speech'.
According to Geoffrey Sampson, the term ideographic is to be
avoided because it is not clear what it means and, furthermore, it
blurs a fundamental distinction between 'semasiographic' and
'logographic' writing.16 What Sampson calls a 'logographic' sym-
bol, however, seems to correspond to what other theorists call an
'ideogram'. Sampson sees no reason to deny the possibility of
developing a logographic system for any spoken language. As an
illustrative example, he proposes as a logographic rendering of the
English sentence The cat walked over the mat the following sequence
of seven logograms: (i) a finger pointing right, (ii) a cat's face, (iii) a
pair of legs walking, (iv) a clock face with an arrow pointing anti-
clockwise, (v) a rectangle with an arrow pointing right, (vi) another
finger pointing right, and (vii) a mat. Sampson explains that 'the
pointing hands in first and sixth place are being used to represent
'' DeFrancis, op. cit., p.279.

16
Sampson, op. cit., p.34.
the word the, the walking legs in third place represent the root
walk and the clock with anticlockwise arrow in the fourth place
represents the past-tense morpheme -ed.^1
A related issue over which authorities disagree is the distinction
between ideograms and 'pictograms' (or 'pictographs'). This is
another term that Sampson rejects.18 Others, however, recognize
pictography as one of the important stages in the history of writ-
ing, or at least as an important preliminary to the emergence of
'true' writing. Florian Coulmas claims that both Chinese writing
and Mesopotamian cuneiform, quite independently of one
another, went through a pictographic phase. 'Both systems started
with pictographs which were interpreted as logograms.'19 What
Coulmas calls the 'pictographic principle' is the adoption of
written characters based on the visual image of an object, as with
the Chinese category of shianqshyng.
A related and no less controversial classification is 'picture-
writing'. Leonard Bloomfield dismisses picture-writing as a mislead-
ing term on the ground that it confuses writing with something
else.20 Examples Bloomfield cites include graphic signs used by
American Indians either for purposes of trade or as mnemonic
aids in the recitation of sacred chants. But these Bloomfield does
not accept as 'real writing'. Real writing, according to Bloomfield,
requires a determinate relationship with the sounds of a spoken
language and also a limited inventory of characters.
Marcel Cohen takes the view that pictography, although not
actually writing, is nevertheless 'proto-writing' (protoecriture}.2] He
refuses to accept it as writing on the ground that pictography is
independent of any spoken language. He draws a distinction
17
18
19
F. Coulmas, The Writing Systems of the World, Oxford, Blackwell, 1989,
pp.99-100.
20
Bloomfield, op. cit., p.284.
21
M. Cohen, La grande invention de I'ecriture et son evolution, Paris, Klinck-
sieck, 1958, pp.27ff.
between 'pictographic signs' and 'pictographic signals'. The for-

mer provide a visual representation of what they signify, while the
latter are merely aids to memory and presuppose a prior
acquaintance with the message in question.
David Diringer, on the other hand, counts pictography among
the categories of 'true writing'. He distinguishes it from 'embry-
onic writing', which lacks the capacity to represent continuous
discourse. Pictography, in Diringer's view, makes it possible to
represent a 'simple story': that the story could equally well be
translated into several spoken languages makes no difference.22
A variant version of ideographic hallucination makes the status of

the written form depend on the variety of possible vocalizations.
Thus the hallmark of ideographic writing, according to some, is
the possibility of different oral readings which nevertheless
express the same 'content'. The case of Chinese has already been
mentioned above. Another example often cited is that of mathe-
matical notation: 2 + 2=4 will be read aloud quite differently by
an English child and a French child, but both readings express the
same mathematical proposition.
For Coulmas, however, diverse oral renderings prove nothing
about the ideographic character of the system.
It is noted with wonderment by many Westerners that what is
written in Chinese characters can be read throughout China, in
spite of the pronounced differences between the dialects which
in speech are mutually unintelligible. For two reasons this
argument is not very pertinent. First, until recently, mastery of
the Chinese script was the prerogative of a very small elite, and
this mastery was invariably acquired in conjunction with learn-
ing Mandarin. Second, to find mutually unintelligible dialects
sharing a common written norm one does not have to restrict
one's attention to non-alphabetically written languages; English
is a perfect example. A speaker of Indian English from Bombay
22
D. Diringer, Writing, New York, Praeger, 1962, p.21.
will be hard put to understand the broad drawl of a southern

Texan, and the latter will find the dialect of Glasgow quite
difficult to comprehend. Yet neither of them has any problems
reading standard British or American English and relating it in
some way to their own dialect. From this observation, no one
would want to draw the conclusion that English orthography is
ideographic.23
In spite of what Coulmas maintains, the conclusion that Eng-

lish writing has become (at least for some readers) ideographic is
by no means absurd and would be fully in accord with the Saus-
surean thesis that 'the written word eventually becomes, through
force of habit, an ideographic sign'. It is interesting to note in
passing that the conclusion that English writing had become at
least in part ideographic was reached many years ago by an emi-
nent English philologist and lexicographer, Henry Bradley, who
had never read Saussure.24
Any further examination of the diversity of opinion that reigns
on the subject of ideographic writing would be superfluous here,
since what has been noted above already makes it obvious that
what are lacking in this controversial area are any well-founded
semiological criteria. Experts select whatever suits their immediate
purpose as a reason for accepting one view or rejecting another,
but make little attempt to justify their choice within a broader
semiological framework. The student thus gains the unfortunate
impression that in the end all positions on the question are equally
arbitrary, or else determined by considerations which have
nothing to do with semiology at all.
Worse still, one rarely encounters any analysis of such basic
notions as 'representation of ideas', 'representation of objects',
'representation of words', or 'representation of sounds'. Discus-
23
Coulmas, op. cit, pp. 106-7.
24
H. Bradley, On the Relations between Spoken and Written Language, with Spe-
cial Reference to English, Oxford, Clarendon, 1919. (The original paper
dates from before the publication of the Cours de linguistique generate.)
sion is conducted as if all these notions were entirely perspicuous

and mere inspection sufficed to determine whether or not they
applied to a particular form of writing.
Such notions, on the contrary, are far from clear in semiological
terms. In particular, the notion of'representation' cannot be satis-
factorily reduced to binary relations. In other words, we cannot
explicate 'representation' within the framework of a dualist theory
of signs. The Procrustean bed of dualism inevitably results in
taxonomies of writing which attempt to classify a written sign by
reference to a prior classification of possible candidates for 'what
is represented'. The categories of written sign are thus deter-
mined by what the signs are deemed to signify. Signs signifying
ideas will be placed in one class, signs signifying sounds in another,
and so on. The problem with this programme is that it remains
quite unclear how to determine what a sign signifies. Does, for
instance, the written form & signify the letters and, or the sound
[and], or the English word so pronounced, or the concept 'and',
or the relation of conjunction, or some combination of these, or
none of them? In spite of the definitions commonly given of such
terms as ideographic and phonographic, in practice the determining
factor is often the form of the written sign. That, indeed, is the
basis of Saussure's paradoxical complaint about written forms like
oiseau: they look as if they should be phonographic when in fact
they ostensibly flout expected phonographic principles. (The
complaint is paradoxical in that it is precisely the assumption that
the alphabet is a phonetic system which gives rise to the problem.)
Similarly, for non-alphabetic writing, if the written form looks
anything like a schematic drawing of an object, and the object in
question corresponds to what the sign is taken to mean, theorists
are happy to claim without more ado that the written form 'repre-
sents' the object signified. If there is no recognizable object-image,
they are equally happy to claim that the written form 'represents'
something more abstract: the word or the idea signified.
This simplistic approach breaks down when dealing with cases
like the ampersand (&), which historians tell us is the vestige of
a former Latin abbreviation. Presumably only those with an
education that includes the paleography of Western scripts can

actually 'read' it in this way. The rest are left to make sense of it as
best they can. Those with sufficient imagination might perhaps
see the ampersand configuration as a looped knot and thus con-
strue it as a metaphorical pictogram. ('Knot' = 'joined together'
= 'and'.) Those with less imagination will perhaps construe it as an
interlingual ideogram. Others, focussing on its occurrence in texts
belonging to a particular language, may treat it as a logogram.
Few will be tempted to classify it as a phonogram, since they will
fail to see the shape as an alphabetic abbreviation. Now there is an
element of rationale in all these possible interpretations, but noth-
ing to tell us which is the 'right' one. And that is already an
indication that we are trying to squeeze the ampersand into a
taxonomy which is itself semiologically inadequate.
It may perhaps be objected that a sign like the ampersand is a
marginal case, since it does not function as an intrinsic unit in any
established system of writing. But similar problems arise with
graphic signs which cannot be dismissed in this way. The Chinese
character for the number 'three' consists of three horizontal
strokes, arranged one above the other without ligature. It is impos-
sible to say whether these are - or were originally - supposed to be
three sticks, three planks, three fingers, three swords, etc. Three
unidentified long thin objects? No matter, it will be said, since the
important thing is that there are just three of them, and that is
sufficient to make the numerical meaning clear. But that reply
misses the point. The difficulty does not reside in counting the
strokes, but in interpreting the fact that there are three of them.
For it does not follow that a character with three strokes must
mean 'three'. The Chinese character meaning 'forest' appears to
show three schematic trees, each with its trunk, branches and
roots. In both the character meaning 'three' and the character
meaning 'forest' we see three units grouped together. But in the
case of the trio of trees it looks as though three means 'many' not
'three' (nor even 'a few'). Analogously, it might have been sup-
posed that the trio of horizontal strokes meant 'a pile' or 'a lot' (or
perhaps 'a few'). In neither case does what the sign 'shows'
automatically make clear what it signifies: it makes no difference

whether we can recognize what is shown (as in the tree case) or
whether we cannot (as in the numeral case).
In the traditional Chinese classification, the character meaning
'forest' belongs to the 'pictographic' category of shianqshyng, while
the character meaning 'three' belongs to the 'ideographic' cate-
gory of jyyshyh. Does this difference correspond to two different
ways of 'reading' the marks? Can one say that in the former case
we have a picture, but in a latter case merely an abstract represen-
tation of a number? The suggestion is not very convincing, given
the way Chinese characters are distributed into these two cate-
gories. But the very possibility of envisaging different solutions
depending on the visual interpretation of the marks illustrates the
kind of problem that may arise even in apparently simple cases.
It is interesting to compare this case with that of the Arabic
numeral 3. Arabic numerals are often cited as typical examples of
ideograms, and much is made of their 'international' character.
They are said to represent not words but mathematical abstrac-
tions. As regards the individual shapes of the figures, opinions are
divided. For Sampson, 0 represents an 'empty hole' and 1 a 'single
stroke', whereas 6 and 7 are 'arbitrary' shapes.25 But what about
3? In angular writing it is not difficult to detect three horizontal
strokes, linked by a vertical to the right. But the rounder the hand
becomes, the more this image tends to disappear and be replaced
by a curly figure not unlike 8. So can the same notational
character be 'ideographic' in one person's handwriting, but not
in another's?
Those acquainted with the Saussurean notion of semiological
'motivation' may feel inclined, like Sampson, to say that written
shapes can be more or less motivated; so that it is not always
possible to give a clear 'yes' or 'no' to the question of whether or
not we are dealing with a motivated shape. This concession mani-
festly fails to produce any solution to the more general problem of
validating the (Western) categories of non-alphabetic writing
23
(even supposing there were any general agreement about what

those categories were). More awkwardly still, it apparently leads
straight to the conclusion that what for one individual may be a
pictogram may for another be just an arbitrary shape. In short,
the categories themselves are hallucinatory artifacts produced by
the imaginative eye of the beholder.
CHAPTER SEVEN
On the Dotted Line
A test case for any semiology of writing is the signature. Signing

one's name is a topic about which historians of writing have very
little to say, possibly because they suppose that all there is to be
said about it is already obvious to members of a literate com-
munity. But there is a great deal more to the signature than meets
the eye. The signature has taken many different forms in different
ages and cultures, including some (such as impressing one's
thumbprint 1 ) which are not traditionally regarded as forms of
writing at all. But in many literate societies the signature involves
signing one's name with one's own hand, and this is the insti-
tutionalized practice that will be the focal point of discussion here.
The signature in this sense deserves a chapter of its own in any
theoretical account of writing, since it utilizes the semiological
resources of writing in a particularly revealing way. It also, as it
happens, illustrates clearly the extent to which an integrational
analysis of the written sign differs from traditional accounts.
1
Although the thumbprint is ancient, fingerprinting as a general tech-
nique of identification, and the scientific evidence to back it, is relatively
modern. It was introduced by the British in India in the late 19th century
in order to provide a form of personal identification that was viable for a
largely illiterate subject population. In that sense the fingerprint is not an
ancestor of but a substitute for the signature.
Although written signs are often referred to as signs used by the

writer, from an integrational point of view it would be less mis-
leading to say that in writing the writer creates written signs. The
signature provides a perspicuous example of this: each and every
signature constitutes a new sign in its own right, uniquely created.
A signature can be copied by someone else; but its status as a
signature does not transfer to the copy. In this respect signing one's
name exemplifies in an immediately obvious way a characteristic
of all writing. Writing - at however humble a level - is a creative
act in the same sense that a painter creates in the very act of
applying each brush-stroke to the canvas. Not every text is signed,
and many paintings are not signed either. But just as every paint-
ing is the potential bearer of a signature, so too is every text. The
signature is the visible confirmation of its status as a creative act
by a certain individual.
Within the domain of writing, the signature is the reflexive sign par
excellence', it signifies by reference to its own making and the identity
of its maker. That is the difference between A's signature and B's
copy of A's signature, however indistinguishable to the eye the two
sets of marks may be. The activities integrated in the production
of the signature are not those integrated in the production of the
copy, and the crucial difference is that in one case they are A's
activities and in the other case not. Although this is a necessary
condition for a mark to be A's signature, it is clearly not sufficient.
What, then, are the linguistic requirements?
It might perhaps be argued that the signature as such is not
strictly a linguistic phenomenon at all, even though in certain
cultures it may require the writing of a certain linguistic form, i.e.
a name. From an integrational point of view the exact linguistic
status of the signature matters little. For integrational analysis is
not exclusively concerned with linguistic phenomena; nor is it
concerned with every aspect of communication. Its focus is on
communication as a means of articulating human relations and
human experience. The reason why it pays particular attention to
On the Dotted Line 163
language is that language is the area in which all matters pertinent

to human affairs - questions of advantage, disadvantage, kinship,
love, hate, work, play, duty, war, peace, etc. - come into the reck-
oning and affect communication. Language, in short, reflects
human relations in all their complexity and, in turn, serves to
establish and develop them. This applies at every level, from the
interpersonal to the international, where human beings interact
with one another. Language is something that no individual, alone
and unaided, could ever have developed. From our birth, we are
initiated into language by others, and it is through contact with
others that, as individuals, we develop our own linguistic capacities
and our own linguistic identity. Having a name and (in literate
communities) the practice of signing one's name are interrelated as
manifestations of one's linguistic identity. That, from an integra-
tional point of view, is warrant enough for including the signature
among the topics that fall within the scope of a theory of writing.
The point of departure for an integrational analysis of the signa-

ture (as of any other written form) is that in order to make sense of
any episode of human communication we have to recognize an
integration of activities being carried out by particular individuals
in a particular set of circumstances. Signs are created in the
course of this integrational process. They subserve understanding
and negotiation within the limits imposed by that situation, both
understanding and negotiation being efforts that human beings
make to achieve a more satisfactory organization of their mental
and social world.
Whatever activities are integrated in the course of human
communication are always integrated into a temporal framework
of some kind. Communication does not somehow lie outside the
time-track of other events but is cotemporal with them. This
cotemporality is not the theoretical fiction of Saussurean 'syn-
chrony' but the cotemporality of human experience, in which
there is always a past and a (possible) future. A past and a future
are implicit in every semiological phenomenon. When we abstract
from time, as Saussurean semiology does, we abstract from the

process of communication altogether.
How does all this relate to the signature? Our semiological
experience involves recognition of our own temporal existence.
When we sign a document, we are (if in full command of what we
are doing) aware of doing something that, formally and in
accordance with established convention (as recognized by Austin
and other speech-act theorists), integrates our past with our future.
We may subsequently complain that we were acting under duress,
the influence of drugs, that the balance of our mind was disturbed
etc., in order to opt out of the consequences. But the very
requirement that such justifications or excuses have to be made -
and can be made - bears witness to the fact that in a literate
society signing often involves signing away certain freedoms. It
may also create certain entitlements (that we hope others will
honour). In any case it integrates our lives into the life of society in
a very special way that requires very special procedures to undo or
retract. But if that were all, it would be no different in principle
from taking a solemn spoken oath or going through certain kinds of
initiation ritual. A satisfactory theory of the signature (or, at least,
an integrational theory of the signature) requires something more.
A first clue to locating this 'something more' emerges if we
reflect on why it is that whereas the rest of a signed document can
be read aloud, when we come to the signature there is, strictly, no
oral equivalent that can be produced. Reading out the name at
the foot of the document is just that and nothing more: the read-
ing of a name. But a name is not a signature. Names of all kinds
can be appended to or included in documents without being eo ipso
signatures. Reading aloud fails to distinguish phonetically between
name and signature. This is a case of graphic heteronymy.2 Or, to
2
Heteronymy is a technical term usually applied to cases in which two
words are pronounced the same but spelt differently, or pronounced
differently but spelt the same. I here extend it to cover the case in which
we have identical spelling and identical pronunciation but different writ-
ten forms. For Pessoa (see below), heteronymic works are those written
under a pseudonym which reflects a personality invented by the writer.
put the point another way, signatures have no pronunciation,

other than the pronunciation of the name they exemplify. In this
sense, the signature is essentially a phenomenon of writing: it
cannot be 'transferred' into another medium.3 Unless we grasp
this, we shall fail to see what is sui generis about the semiology of
the signature.
In order to clarify the relationship between the signature and the

name of the signatory, it is first necessary to say something more
about names. Inasmuch as one's name is a badge of social iden-
tity, we are here dealing with macrosocial factors. These may be
very complex, since an individual may be known by various
names, and not all of these are necessarily acceptable for all pur-
poses as the name one signs. It would be odd, to say the least, to sign a
love letter in the same form as one signs a cheque. Banks and
businesses tend to be somewhat stricter than friends or casual
acquaintances in the matter of what they will accept as one's
signature. For certain purposes it may suffice to 'sign' with one's
initials; but in those cases the initials are taken as 'standing for'
one's name. However tenuous or specialized the macrosocial link
between name and signature becomes, it is never severed entirely.
Writers who decide to write under a nom de plume are regarded as
deliberately concealing their identity. It is interesting to note that a
writer like Fernando Pessoa, who writes under several pseud-
onyms, feels it necessary to apologize for this: he denies that it
amounts to 'insincerity'.
I call 'insincere' anything done to impress, and also - please
note, it is important - whatever does not harbour any funda-
mental metaphysical thought; that is, where there is no inkling,
even fleetingly, of the mystery and seriousness of life. This is
why I consider to be serious everything I have written under the
names of Caeiro, Reis and Alvaro de Campos. Into each of
3
When I read my own signature aloud, I am not eo ipso doing something
orally equivalent to signing.
them I have put a profound concept of life, different in the

three cases, but always mindful of bearing witness to the
important mystery of existence.4
Such an apology would be inconceivable outside of tradition in

which the equation 'signature = name of signatory' was a stand-
ing assumption, and without a logic in which a signature is taken
to guarantee the bona fides of the signatory. (Compare giving a false
name and address, which is regarded by many as a prima facie case
of dishonesty.) The fact that a writer may be allowed by special
dispensation, as it were, to assume another name and 'sign' work
under it is not counterevidence to but confirmation of the unique
cultural role of the signature. A signature based on an alias is
intrinsically suspect. In short, we have no plausible semiology of
the signature at all unless we recognize and account for its moral
dimension. From an integrational point of view the morality of
the signature is not something adventitious or external to it, but
something that has to be explained in terms of what this particu-
lar written sign means. Sincerity is not just a personal quality that
individual writers choose to put into — or withhold from — what
they write (as one might choose to enter a competition 'seriously'
or 'for a lark'). It is something that is endemic to certain forms and
devices deployed in writing; and the linchpin to all of these is the
signature.
Although anthropologists tell us that in all societies children are

given names (often in accordance with very precise and elaborate
customs), and that in many societies one's personal name may
change during the course of one's lifetime (again, in conformity
with quite specific macrosocial criteria), they more rarely point
out that as literacy develops in a culture the very concept of a
personal name undergoes a significant change. Put briefly, the
4
Letter to Armando Cortes-Rodrigues, 19.1.15. F. Pessoa, Sur les hetero-
nymes, trans. R. Hourcade, Le Muy, Editions Unes, 1985, pp. 13-14.
On the Dotted Line 16 7
change may be described as follows. The written form of the name

comes to be regarded as its authentic form, at the expense of the
oral form.
This is particularly noticeable in societies where the writing
system makes available more than one possible way of rendering
the spoken name. Thus even in announcing one's name orally it is
sometimes advisable to refer to its spelling. (White 'with an f is not
the same family name as Whyte 'with aj>'; nor is Green the same as
Greene 'with an e on the end'.) This is not a merely occidental
peculiarity. In Chinese, a family name is often given orally (to
strangers) with an explanation that specifies the corresponding
characters: 'My name is Li (as in 'wood' and 'son')'.3
Conceivably in some preliterate cultures (although the author
knows no examples) it might be customary to announce one's
name by giving a quasi-etymological explanation to avoid homo-
nymic confusion. (Hypothetically: 'My name is Trog, after the sun
god, not the crocodile spirit.' Or: 'My name is Sharp, meaning
'clever', not 'pointed'.) But this would be a quite different kind of
onomastic elucidation from 'My name is White, not Whyte'.
The point of telling someone that your name is White, not
Whyte, is to enable that person to integrate successfully a whole
range of future activities that might involve correct identification
of the name; including forwarding mail, making hotel bookings,
charging expenses to your account, and so on. The whole problem
of 'mistaken identity' in a literate culture is anchored to the notion
that a name is a written form. It is only when personal identifica-
tion needs to be pursued further that the name qua identification is
superseded by the passport, the National Insurance number, the
DNA test, etc. But even then the immediate evidence required in
any particular case is usually documentary, backed by the signa-
tures of experts; because a literate society cannot conceive of any
better form of evidence.
' V Alleton, L'Ecnture chinoise, Paris, Presses Universitaires de France, 4th

ed., 1990, pp.62--3.
If it is the written name which counts as the authentic designation

of the individual in a literate society, then it is understandable that
the signature - which involves that name being inscribed by the
individual in person - should count for much more. It becomes
not only the ultimate guarantee of authentication by the indi-
vidual, but the most basic form of individual self-expression as
well. We see manifestations of the latter at all levels of society and
over many centuries. How many European potentates and
noblemen have had their personal monograms decoratively
incorporated into the architecture of their palaces, the wallpaper
of their public and private rooms, the crockery on their dining
tables, or the upholstery of their furniture? These are not, to be
sure, instances of personal inscription, but only because the high
and mighty do not stoop to manual labour. It is quite the opposite
at the other end of the social scale, where countless otherwise
undistinguished citizens have left the record of their initials per-
sonally carved or scratched on tree trunks, garden gates, school
desks and lavatory doors; or, as is more fashionable nowadays,
sprayed their 'tags' on walls and other public surfaces (Fig.5). One
is reminded of the way in which certain animals spray-mark their
'territories'. Nor can it seriously be doubted that the princely
monogram itself belongs to that vast class of signs which are
sometimes called by historians of writing 'property marks'. What
is interesting here is that, according to the same historians, pro-
perty marks are among the important 'forerunners' of writing.
Diringer draws our attention to the widespread use of property
marks among pre-literate peoples:
Property marks have been found amongst the Lapps in Sweden,
the Votiaks, a Finnish people of north-eastern Russia, the
Cherkessians, the Kadiueo of South America, the Ainus of
Yezo Island, on the Moresby archipelago, in Australia, amongst
the Masai of eastern Africa [. . . ]6
As to how property marks originated (granted the relative
6
Diringer, The Alphabet, p.29.
Figure 5
sophistication of the notion of 'property' itself), Gelb offers the

following speculative explanation:
Suppose that a primitive man drew on his shield a picture of
a panther. This drawing originally may have had the magic
purpose of transmitting the strength or the swiftness of the
panther to the man who owned the shield. But in the course of
time the panther on the shield became also a symbol which
communicated to everyone the fact that the shield was owned by
a certain person. The symbol of the panther therefore became a
property mark, whose aim was that of utilitarian writing. The
drawing of a panther as a property mark is, of course, not yet
real writing, even though it stands for a personal name and may
be habitually associated with one certain person, because it does
not yet form a part of a well-established and conventional
system. But it is an important step in the direction of writing.7
' Gelb, op. cit., p.36.

There are points here on which an integrational theorist would

disagree fundamentally with Gelb: it takes more than 'habitual
association with one person' to transform a sign into a property
mark. Associations count for little unless there are specific pro-
grammes of activity to be integrated. Nor does mere association
explain how a panther-mark could 'stand for a personal name'.
Nevertheless, what is of interest here is that if Gelb and Diringer
are right in regarding the primitive property mark as an important
forerunner of writing, what we see in the princely monogram of
later centuries is writing returning to its origins. More exactly, we
see the reinvention, within the framework of a literate society, of
an extremely ancient semiological device. All that has altered is
that literate property-owners not only have names but can spell
them.
The case of the modern spray-gun signature is somewhat dif-
ferent, but intimately related - at least, from an integrational point
of view — to the foregoing. Here we have a society in which many
people own no property apart from a few personal effects. Yet they
live in cities which are physically constituted by houses, cities,
streets and factories owned by someone else. An integrationist will
point out that part of what a sign means depends not just on
where you put it but on whether you were authorized or entitled to
put it there at all. The spray-gun signature is not only a form of
self-expression but a form of social protest that both mimics and
mocks the property mark from which it derives.
Consideration of the spray-gun signature leads on to another

basic aspect of what a signature means. We nowadays find it
natural to suppose that signatures are important because of what
may follow - legally,financially,etc. - from our signing a docu-
ment (or, as speech-act theorists would put it, from the 'perlocu-
tionary' act8). Such considerations are not only important but
8
J.L. Austin, How to do things with Words, Oxford, Clarendon, 1962,
pp.lOlff.
essential in any integrationist account of the signature. There is

no doubt that by signing one document I may sign away my
inheritance, by signing another I may pay my gas bill, and so on.
The doctrine of perlocution is applicable to the signature, and fine
as far as it goes. Nevertheless, like all speech-act theory, it misses
out something more basic. To see what this is, we need to consider
cases in which we have a bona fide signature, but no legal contract
entered into, no debt paid, no letter signed, etc.
To begin with a trivial but nevertheless instructive example,
there is the (relatively modern) practice of'collecting autographs'.
When the captain of the Australian cricket team signs the small
boy's autograph book at Lord's he enters into no legally binding
agreement. But he enters into a moral contract, nevertheless. If he
signs 'Marilyn Monroe' (having had a bad day at the crease) he
abuses a procedure, disappoints a hero-worshipper and renders
himself liable to censure; for what he produces is neither Marilyn
Monroe's signature nor his own. Every autograph-hunter knows
enough to realize that. And every episode of autograph-hunting
illustrates the integrationist lesson that in order to get the desired
autograph in your book, you have to get your book to the signa-
tory. Two potential progammes to be integrated (yours and the
signatory's). That you have to know in order to be an autograph
hunter in the first place.
So what bond exists between the captain of Australia and this
small boy who, if lucky, comes away the proud - and potentially
permanent - possessor of the autograph of a famous cricketer?
What communicational 'rules' were tacitly invoked? The
encounter itself lasts no more than a few seconds. One party
knows who the other is, but the other has doubtless never seen the
autograph-hunter in his life before and may never see him again.
No word needs to be spoken. All is implicit in the book and the
pencil thrust before the great man. The cultural bond is not just
cricket but cricket plus minimal membership of a literate
community.
What do we mean by 'minimal' here? If the signature occupies
a very special position in the semiology of writing, this is in part
because of one specific macrosocial implication attaching to it. In

most literate societies, being able to sign one's name is recognized
as the very least qualification for literacy. The signature, in itself,
says: 'I belong to the literate community; and here is my member-
ship card.' In the course of history there must have been millions
of individuals for whom being able to sign their name was the
only act of writing they could manage.
Among the chronicles of the 'wild child' family that have
aroused so much interest on the part of educationists and psycho-
logists doing research into language acquisition a place of honour
is reserved for the well-known case of Kaspar Hauser.9 In the sad
story of this young man from Nuremberg, the detail worth noting
here is that this hapless individual spent hour after hour filling
pages with his signature. It is by no means absurd to see this, in the
first instance, as a desperate attempt to establish his own identity,
but also as an affirmation that he had the right to participate in
the life of that society to which he had been denied access, and
into which he so much wished to be integrated. The message of
Kaspar Hauser's endlessly repeated signature can be read as:
'Here I am. This is my name. I am like you: look, I can write!
What more do you want?'
But there is no need to appeal to pathological cases in order to
make the point that signing one's name counts — and has always
counted - among the most important rites of passage ever
practised by the tribes of literacy.
The signature is subject, as far as the integrational theorist is con-

cerned, to the usual circumstantial provisos that govern all acts of
writing. One cannot write anything anywhere. Where do you sign
(given the graphic space available)? That depends on what you are
signing and the circumstances. But there are rather conspicuous
9
First reported by von Feuerbach in 1832. For a bibliography covering
similar cases, see L. Malson, Les Enfants sauvages, Paris, Union Generale
d'Editions, 1964.
On the Dotted Line 17 3
regularities. For instance, the usual practice in Western letters,

contracts, etc. is to sign at the foot of the text and either in the
middle of the line or on the right-hand side. It is possible to find
exceptions to this10, but more significant than the exceptions is this
tendency for the signature to gravitate towards a particular end-
position. For instance, my cheque (a rectangular slip of paper
measuring approximately 3" by 6") has a line reserved for my
signature at the lower right-hand side. But why there? Why not in
the top left hand corner? Or right in the centre of the rectangle? A
plausible answer seems to be that in Western alphabetic scripts
one reads from left to right along the lines and from top to bottom
down the page. The position of the signature marks the conclu-
sion of the document (the assumption being that the signatory has
written it, or at least read it, before signing). This is important
inasmuch as anything added below the signature line may be dis-
counted as not belonging to what the signatory signed. So, again,
there are potential programmes of activity implicit in the pre-
ferred place for putting one's signature, and these tie in with
assumptions about when a document will be signed, i.e. after the
text it contains has been settled to the satisfaction of the signatory
or signatories. One can, to be sure, sign a blank cheque; but the
risks one takes in so doing in themselves demonstrate that the
signature is a sign which is subject, like all other signs, to integra-
tion into a temporal sequence. What distinguishes signing a blank
cheque from the corresponding oral promise ('I will pay whatever
sum you may subsequently decide') is precisely that the temporal
logic is different in the two cases. The oral promise is only a
promise: it remains to be seen whether you will pay up. But by
signing the cheque you have already fulfilled the only biomechani-
cal condition required (provided there is enough money in your
10
In the U.S.A. there seems to be a preference for the left-hand side. I
have been unable to find an explanation for this. It may have something
to do with the fact that on the old manual typewriter it was difficult to
centre type precisely, and rather inconvenient to go for right-hand justifi-
cation along an otherwise blank line.
account: but that applies to all cheques, not just to blank

cheques).
When what is signed is not a document but a painting, some-
what different considerations apply. In principle, the painter could
sign anywhere on the canvas, and there are well-known examples
of signatures being ingeniously inserted into the pictorial subject
itself. Nevertheless, here too the tendency in Western art for the
signature to gravitate towards the bottom right-hand corner is
remarkable. Although art historians rarely comment on this, the
primafacie explanation is that the practice is borrowed from written
documents. It took many centuries in the West for painting to be
recognized as an art form on a par with the verbal arts. The
practice of signing a painting as one signs a document may be
seen as an implicit affirmation of the painter's parity of status
with the writer.
The usual explanation that is offered for the gradual extension in
Europe of the custom of signing art works at all is one that appeals
to social and commercial factors. According to Michel Butor:
Previously (in the Middle Ages and the Renaissance) painters
signed their work only when they were sufficiently proud of it to
regard it as an advertisement for themselves or their studio. But
the change in the situation of the painter within the economy
of the West, whereby works were executed not at the behest of
some church or prince but as objects for sale through com-
mercial channels, objects needing a maker's mark or guarantee
of authenticity, resulted in the signature becoming increasingly
usual and increasingly important.''
This provides a classic example of semiological value derived

from the way a certain type of written sign comes to be macroso-
cially required as the formal sign on which hinges the successful
integration of various activities: in this case, those of painter,
dealer and purchaser. It is not simply that the signature identifies the
painter (that could be done in other ways) but that the unsigned
11
M. Butor, Les Mots dans lapeinture, Geneva, Skira, 1969, p.80.
picture is increasingly regarded as being rather like an unsigned

document - of dubious value because no one is prepared to take
responsibility for it. Here again we are dealing with the morality
of the signature. The key question is no longer 'Why sign it?' but
'Why isn't it signed?' That shift is as important a landmark in the
semiology of the signature as in the history of painting.
No such considerations, it is clear, weighed with the artist
responsible for the celebrated medieval tympanum of the cathed-
ral of Autun (Fig. 6), who nevertheless 'signed' his work (insofar
as one can produce a signature with a chisel). The Latinized form
of his name (Gislebertus} may even have been one unfamiliar to the
artist in daily life, and the evidence provided by the words
GISLEBERTUS HOC FECIT doubtless does not satisfy the
usual modern criteria for signing. There may be sceptics who
doubt whether Gislebertus (or Gilbert) held the chisel himself.
They are welcome to their scepticism. Much more important than
these irrecoverable historical details is the conclusion, which it is
difficult to escape, that the whole conception of the Autun tym-
panum is that of a signed work. And the most persuasive detail is
the position of the name Gislebertus. It is placed in the centre of the
composition, right at the feet of the majestic figure of Christ. A
coincidence?
The inescapable reflexivity of the signature is what makes it a

magnet for modern graphologists. The graphologist sees the sig-
nature as an important source of information about the character
of the signatory. What is of interest here to the semiologist is not
the validity of graphological interpretation as such but, in the first
place, the rise of graphology itself. The first published work in this
field dates from the 17th century and graphology can reasonably
claim to be one of the first forms of semiological inquiry bearing
on some aspect of daily behaviour in Western civilization. Also of
interest are the particular features of the signature that grapho-
logists regard as significant.
The expansion of graphology in Europe during the 18th and
Figure 6
19th centuries may be regarded as in part a reaction against the

'impersonal' character of the printed word. Typography to a large
extent suppresses the individual features of written forms in
favour of uniform reproduction of copies. The notion of a 'printed
signature' verges on self-contradiction. This emerges particularly
clearly in cases where signatures are mechanically reproduced in
order to 'authenticate' documents. A paradigm case is the 'signa-
ture' of the Chief Cashier of the Bank of England that appears
(much reduced) on a five-pound note, alongside the picture of the
head of the reigning monarch. The position of the signature sug-
gests we are dealing with a signed document, for above it appear
the words: 'I promise to pay the bearer on demand the sum of five
pounds'. It is interesting to speculate what would happen if one
confronted the Chief Cashier with this and demanded to be paid
five pounds.
According to Michel Moracchini, no serious graphological
study can be undertaken at all without taking the writer's signa-
ture into account. He warns us, however, that
the question of determining what the basic meaning of the
signature is must be approached with great caution. For some,
the signature reveals the depths of the personality, its most
essential nature, including unconscious tendencies, insofar as it
is the product of a spontaneous mark, free from any calli-
graphic constraint. For others, on the contrary, the signature is
an advertisement, a social label carefully produced, which the
writer inscribes as a 'representation' in Klages' sense; that is, in
conformity with an anticipated image and quite the opposite of
a spontaneous expression.12
There are obvious problems with the notion of 'spontaneity' in

any form of activity that relies on habits inculcated by long train-
ing. But let us set those on one side. What is of interest here from a
semiological point of view is that, even with a sign as reflexive as
the signature, and in the course of a work which does not hesitate
12
M. Moracchini, ABC de graphologie, Paris, Grancher, 1984, pp. 148-9.
to treat writing as revealing the personality of the writer, there is

recognition of a certain tension between the demands of the indi-
vidual and those of the community. However idiosyncratic the
signature, it never loses its social character. Like other written
signs, it is meant to be recognized by the reader.
It is consequently no surprise to find that graphologists attach
considerable importance to comparison between the signature
and the signatory's 'normal' writing. According to Moracchini, if
there are no divergences this indicates
a certain 'unity of the Self, or well-integrated personality.
Hence indications of authenticity, genuineness, simplicity, self-
possession, balance, constancy and equanimity. The writer
accepts what he is, and entertains a good opinion of himself.
The situation is different when discrepancies between the
signature and other writing appear. These indicate conflict or
conflicting tendencies within the writer [. . . ] 13
Among the discrepancies Moracchini lists are: (i) the signature

written in larger letters than normal, indicating a tendency to
pride, vanity, self-satisfaction, etc., (ii) the signature written in
smaller letters than normal, indicating modesty, self-effacement,
diffidence, etc., (iii) a sloping signature contrasting with upright
normal writing, indicating concealed sensitivity and emotion, and
(iv) an upright signature contrasting with normally sloping
writing, indicating a tendency towards repression of feelings or
adoption of a false cordiality.
Whether these supposed correlations have any empirical foun-
dation is another matter. The same might be said of the grapho-
logical interpretation of flourishes and superfluous marks
accompanying the signature: for example, a signature incorporat-
ing an underline is taken as indicating a tendency to self-
promotion.
Such analyses are often supported by reference to the signatures
of famous people. According to Moracchini, Napoleon's signa-
13
Moracchini, op. cit, pp. 15 8-9.
ture, with its emphatic crossed flourish, is indicative of a 'brusque,

curt, aggressive' person, while Hitler's, with its short flourish cross-
ing the final letter of the first name, indicates a 'high degree of
pugnacity'. In fact, to the untrained eye, the signatures of Napo-
leon and Hitler do not look remotely alike (Fig. 7); or at least by no
means as similar as the accompanying psychological descriptive
terms ('brusque', 'curt', 'aggressive', 'pugnacious') might lead one
to expect. Sometimes graphologists see more direct pictorial ana-
logies in the form the signature takes. For instance, in the signature
of Marat with its prominent looped flourish descending from the
final letter we are invited to see 'a rope and dagger', appropriate
for 'the bood-stained hangman of the French Revolution'.14
Figure 7
However, this is not the place to attempt a detailed criticism of

the assignment of psychological descriptions on the basis of
graphological evidence. What is more relevant to the present dis-
cussion is the fact that graphological analysis, from its very begin-
nings, adopted a structuralist methodology, and in so doing
anticipated by some considerable time the 'official' emergence of
structuralism in linguistics. That is to say, graphologists always
sought to reduce handwriting to a relatively small number of
presumably significant oppositions (letter size - large or small;
letter angle - upright or sloping; pressure of writing instrument -
heavy or light, etc.). Furthermore, these contrasts operated within
and up to the limits imposed by the writing system itself. In this
connexion it is worth recalling Saussure's observations on vari-
ations of handwriting discussed in an earlier chapter. According
14
E. Singer, Graphology for Everyman, London, Duckworth, 1949, p.70.
to Saussure, it is the structure of the system which accounts both

for the range of possible variants and for the limits beyond which
variation cannot go. There are dozens of ways of writing a letter t
they are all valid provided that, under the pen of the same writer, t
is always distinct from /, d, etc. That one writer's t may look like
another's /, or like the d of a third, does not matter. All, provided
they maintain the internal oppositions between the various letters,
are using 'the same system'. Here we have structuralist principles
pushed to their logical conclusion and applied at the level of the
individual.
What Saussure seems to have overlooked is that the writer's
own signature frequently violates these principles. These are the
cases of what are commonly described in lay terms as 'illegible'
signatures; or what graphologists would classify as extreme
inconsistency between the writer's 'normal' hand and the signa-
ture. The point to note here is that whereas Saussure would pre-
sumably have had to dismiss such cases as random 'exceptions' to
the observation of the writer's own structural rules, for an integra-
tional theorist they are not exceptions at all but exactly what inte-
grationist theory would predict. That is to say, the occurrence of
such cases demonstrates that the written sign is identified in the
first instance by 'global' contextualization, not by analysing its
constituent units in terms of a postulated series of internal opposi-
tions. In the case of the signature, its distinctive position in the
graphic space available is usually sufficient to identify it as a signa-
ture. The issue of whose signature is a quite separate matter, from a
semiological point of view; i.e. a question of the relationship
between signature and name. That this is so is confirmed by the
existence of graphic practices where the two functions are visibly
divorced. A current example is the modern business letter, where
the name of the signatory is typed out immediately below the
signature. To suppose that this practice evolved because so many
businessmen had bad handwriting would be putting the cart
before the horse. On the contrary, what it demonstrates is recogni-
tion of two quite distinct semiological functions, each of which
can be assigned to a separate written sign. The 'illegible' signature
is not an 'exception' to the rules of writing, or to anything else: it is

simply one more development in the functional differentiation of
writing.
As to why this development should have followed this particular
path, it is relevant to point out that there was always a tension, if
not a contradiction, between the two macrosocial purposes that
the signature was traditionally called upon to serve as an authenti-
cating device. Needing to be recognized as belonging to the signa-
tory alone, it needed also to be distinguishable from that of all
other possible signatories: this precaution against forgery put a
premium on developing unusual, idiosyncratic forms of signature.
On the other hand, as a sign identifying a certain member of the
community by name, it needed to be readable by readers who
were not necessarily familiar with the signatory or the signatory's
writing. These two requirements are in conflict inasmuch as the
more idiosyncratic a signature becomes, the more it is likely to
confound to the average reader's ability to decipher the name.
(This relationship between idiosyncratic writing and readers'
expectations applies not just to signatures but to handwriting in
general.) The solution of adopting different signs for the two func-
tions and juxtaposing them in graphic space is both a rational
solution and one that fits a division of labour already in place:
the secretary types the name (as well as the document), and the
signatory supplies the signature in the appropriate place.
One further piece of evidence from the history of the signature
corroborates the theory of semiological differentiation presented
above. It is the development of procedures to enable even those
who are totally illiterate to 'sign' a document. The classic device in
Western culture is 'signing with a cross'. Thus even those who
cannot write their own name (provided they can hold a pen)
are admitted by graphic proxy to the benefits and disadvantages
of literacy. The history of colonial expansion provides many
examples of preliterate people who were duped into 'signing
away' their lands and rights by a piece of graphic magic they
never understood. But since one 'X' looks very much like
another, such documents had to be validated (in Western eyes) by
appending beside each cross a note identifying the signatory. Here

we see the underlying logic of the signature laid bare. Marking a
cross, even though it is totally undistinctive as a mark, counts as
'signing' simply in virtue of being made by a particular individual
(who, supposedly, has the authority to do whatever the document
in question requires to be done). The problem is that the cross
itself defeats the purpose of the exercise by making the signatory
anonymous. So this anonymity has to be counteracted by supply-
ing the individual's name separately. Thus there is de facto recogni-
tion of two semiological functions and their formal separation.
The semiology of the signature has no counterpart in the domain

of speech. Its uniqueness highlights in a particularly interesting
way the semiological independence of writing. We sometimes
have occasion to ask someone to repeat their name (because we
did not catch it the first time). But this is not because they uttered
it in some deliberately idiosyncratic way. There is no social or legal
occasion in which protocol demands or allows an inarticulate
grunt to substitute for the 'normal' pronunciation of one's name
(which would be the equivalent of the 'illegible' - but proper -
signature). Bank officials do not expect you to announce a spe-
cially distorted version of your name in order to identify yourself
over the phone, but they may well expect something visually
analogous to that on paper if you enter into correspondence with
them. (The day may soon come when they will check your identity
over the phone by means of a voiceprint, but it has not come yet.)
But, as Beatrice Fraenkel points out, there is another paradox
about the signature, which consists in requiring that it be not only
distinctive but also recognizably uniform; and this as a legal
requirement dates back at least to the Middle Ages.15 If you keep
changing your signature, your bank may well start returning your
cheques. So although the signature is of all written signs the most
ineradicably reflexive, nevertheless reflexivity cannot be carried to
15
B. Fraenkel, La Signature. Genese d'un signe, Paris, Gallimard, 1992, p.205.
the point where the individual is allowed totally idiosyncratic free-

dom of decision and action. Odd though it may sound, you are
not in charge of your signature, even though no one else can sign
in your place: there are powerful macrosocial constraints opera-
tive. The signature is the only written sign which has to meet both
the requirement that it be written by one specific individual and at
the same time the requirement that the individual, in so doing,
conform to a previously established graphic habit. Only in respect
of the signature does society expect its literate members to be
consistent in their personal forms of writing. It makes no compar-
able demand in the case of speech. But in the case of the signa-
ture, in effect, the individual is expected to behave like a graphic
replicating device. As Fraenkel notes:
The use of seals allowed the production of impressions similar
in every detail to their common matrix. In order to forge a seal,
a false matrix must be made. The signatory is deemed to pro-
duce a signature as if he himself were a matrix capable of
replicating a form. 16
What the semiology of the signature tells us is something about

the society responsible for its evolution as a graphic practice. It is
evidently a society with great respect for the individual, and the
gradual extension of the signature as a formal procedure goes
hand in hand with the development of the rights of the indi-
vidual, in both political and economic matters. At the same time,
however, the individual is expected to respond by conforming to
consistent behavioural patterns that will, by sacrificing a small
amount of freedom on particular occasions, facilitate and safe-
guard greater freedoms in the long run. And, in order to lock both
signatorial functions into place, the signature is made morally
sacrosanct. The validity of your own signature is thus integrated
with your not faking the signature of anyone else.
16
Fraenkel, op. cit., p.206.
CHAPTER EIGHT
Beyond the Linguistic Pale
Is writing language? In modern linguistics, an intrinsically con-

troversial discipline, this question rates as one of the most contro-
versial. What is at issue, as often, is unfortunately obscured by
uncertainties of terminology. According to some theorists, it is legiti-
mate to apply the term writing in both a broad and a narrow sense.
Writing, in the broad sense, is any visual and spatial semiotic
system; in the narrow sense, it is a graphic system of linguistic
notation.'
The trouble is that this 'narrow' sense itself depends on a no

less narrow interpretation of the term language. For in a 'broader'
sense, language too can be taken to include many forms of com-
munication. It would beg the question to dismiss this by saying
that the broader sense must involve metaphor (as in 'the language
of architecture', 'the language of flowers', 'the language of bees',
and so on); for 'metaphor' itself is a semiologically contentious
concept.2 In short, it would be naive to imagine that the relations
1
O. Ducrot and T. Todorov, Dictionnaire encyclopedique des sciences du langage,
Paris, Seuil, 1972, p.249.
2
For a discussion of metaphor from an integrationist perspective, see
M. Toolan, Total Speech. An Integrational Linguistic Approach to Language,
Durham, Duke University Press, 1996, Chapter 2.
Beyond the Linguistic Pale 185
between writing and language can be elucidated by appeal

to dictionary definitions or etymologies or supposedly 'literal'
meanings of the terms in question. For this whole metalinguistic
terrain is a quicksand.
Those writers who invoke a 'narrow' sense of writing and a
correspondingly narrow sense of language seem to take for granted
the legitimacy of equating language with speech. Histories
of writing provide numerous examples of this. To take one at
random, according to James Fevrier writing is
a process used to immobilize, to fix articulated language, which
is, in its very essence, ephemeral.3
Here, it has to be assumed, 'articulated language' must be

speech, unless Fevrier takes a more cosmic view of ephemerality
than most of us. What Fevrier is doing seems to be summed up by
Julia Kristeva when she writes:
Writing is considered a representation of speech, its double; not a
separate material form whose combinations give rise to lin-
guistic functions different from those of sound. Thus the
science of writing is trapped in a conception of language which
equates language with spoken language [ . . . ] .4
What Kristeva describes covers the whole Western tradition

of assumptions about writing. These assumptions have their
roots in pedagogic programmes developed for initiating children
into the rudiments of alphabetic literacy, where a basic one-to-
one correspondence between 'letters' and 'sounds' has been
taught for many centuries, with cases of non-correspondence
being dismissed as 'exceptions'. However effective such pro-
grammes may have been pedagogically, that does not make
3
Fevrier, op. cit, p. 9.
4
J. Kristeva, Le Langage, cet inconnu, Paris, Seuil, 1981, p.35. Italicization
as in the original.
the assumptions Kristeva describes any more acceptable

semiologically.5
As soon as one begins to probe the traditional 'representation'
story at all insistently, both terminological and conceptual embar-
rassments are revealed. For if language is what writing represents,
then writing can hardly be at the same time language. Unless we
are being asked to accept that what writing represents is all lan-
guage, including itself. Yet we hear not only of 'written language'
but of'written languages'. How could there be any such thing(s) if
writing were no more than representation? If language is by
nature audible and ephemeral, how could anything which is nei-
ther phonetic nor transient be a form or variety of language? And
how could it, as Fevrier claims, 'fix' the ephemeral flux of the
spoken word? It is rather like maintaining in all seriousness that
the meteorological chart fixes the weather; that, moreover, it
makes the weather visible; andfinally,for good measure - as some
television weather-forecasters seem to believe - that the isobars
actually are the weather (or at least a cause or manifestation of it).
In the case of writing we find ourselves straight away plunged into
similar semantic gobbledegook: the intrinsic ephemerality of lan-
guage is contrasted with the non-ephemerality of writing which,
nevertheless, allegedly 'fixes' the unfixable, while at the same
time what started out only as a visible representation of language
eventually emerges as a form of language itself.
Is this making too much of mere fafons de parler? Hardly. For
none of the accepted authorities in the Western tradition, from
Plato onwards, has ever proposed a way of discussing writing that
does not pitch one straight away into these contradictions.
Perhaps someone will object that it is perfectly reasonable to say
that a painter 'fixes' on the canvas one fleeting visual impression
of a landscape. If that example were at all pertinent, it would be
3
Their pedagogic effectiveness has been, and continues to be, a debat-
able issue. But that is irrelevant to the point being made here. All kinds
of effective training programmes may be based on simplifying assump-
tions which turn out to be or are known to be false.
an argument from analogy. Does the analogy hold for language

and writing? Before we bother about that, there is a prior question
to be addressed. Does thisfafon de parler even hold for a painting?
Or for a photograph? Getting people to 'sit still' was indeed a
major practical problem for early portrait photographers, who
devised neck rests and other contraptions to ensure the (unnatural)
immobility of the body. But does this paragraph on the printed
page bear a relation of that kind to the sounds you might hear if
someone read it aloud? It is very doubtful. At least a photograph
of the landscape is (supposedly) a visual record of a visible subject.
But does it make sense to talk in the same breath of a visual record
of the invisible? For this is what is involved in the case of writing.
When our comparisons cross the border between sense modalities,
we risk talking nonsense. What would a photograph of a smell
look like?
It is clear that if the traditional 'representational' account is to

pass muster at all, it must be given a much more careful formula-
tion than the muddled one which historians of writing are still
evidently happy to perpetuate. Can this be done in such a way as
to rescue the story from its own incoherences?
One strategy might be to maintain an absolute distinction
between writing and language, and avoid as far as possible talking
about 'written language(s)' altogether. For instance, the author of
an influential linguistics textbook of the late 50s and 60s, C.F.
Hockett, declares that the term written language 'is not desirable'.6
But it has always been one of the pious frauds of modern lin-
guistics to refuse writing admission at the front entrance while
letting it slip in quietly by the back door. Thus we find the same
linguist in the same book announcing that because what speakers
actually say is often marked by hesitations, gaps and inconsisten-
cies of various kinds, the scientific study of language should be
6
C.F. Hockett, A Course in Modern Linguistics, New York, Macmillan, 1958,
p.549.
based 'exclusively on edited speech'.7 Where the notion of'editing'

speech comes from, unless from the forbidden 'written language',
he does not explain. The casuistry is blatant: the linguist will study
the spoken language 'scientifically', but only insofar as it approxi-
mates to standards of clarity and coherence expected in writing.
One of the best-known American linguists of the interwar
period, Leonard Bloomfield, was particularly adamant on the
non-linguistic status of writing. 'Writing is not language, but
merely a way of recording language by means of visible marks.'8
This still leaves unresolved the problem of how the ephemeral
phonetic phenomena of language can be 'recorded' by means of
written forms, i.e. static visible marks. The notion of 'recording' is
explicated by Bloomfield in terms of 'symbols'. But Bloomfield's
symbol is neither Aristotle's nor Peirce's, much less Saussure's.
Symbols for Bloomfield are defined in terms of 'representation',
which in turn is interpreted as follows.
A symbol "represents" a linguistic form in the sense that people
write the symbol in situations where they utter the linguistic
form, and respond to the symbol as they respond to the hearing
of the linguistic form.9
This curiously unconvincing a priori account was presumably

dictated by Bloomfield's intellectual commitment to behaviourism
rather than by observation of the actual behaviour of writers. But
at least it can be said in Bloomfield's favour that what he means by
'representation' is reasonably clear; which is an advance on those
accounts where 'representation' is left in limbo as a theoretical
mystery term. (In Bloomfield's account, the traditional notion that
writing 'fixes' language has dropped out of the picture
altogether.10)
7
Hockett, op. cit., p. 144.
8
9
10
It might even be argued that Bloomfield's account implicitly dismisses
this as misconceived.
A different move from Bloomfield's would be to concede that

writing does acquire linguistic status, but only insofar as it succeeds
in 'representing' language. This would allow for writing including
both linguistic and non-linguistic features. But does that help? It is
hard to see that it does; for what now remains to be explained is
the puzzle of how something which itself is 'not language' can
nevertheless acquire linguistic status in virtue of merely 'represent-
ing' language. A map does not become a town or acquire the
status of a town simply because it shows - however accurately -
where the streets are. We drive along the streets, but not along the
map. Furthermore, the notion that it is the representational
function which confers on writing whatever linguistic status it has
cannot be squared with the notion that writing somehow 'fixes'
language. On the contrary, it would be the other way round: a
case of language 'fixing' writing, i.e. determining the linguistic
values of the letters or characters. (Likewise, it is the town
which determines the accuracy of the map, not the map which
determines the accuracy of the town.)
The belief that writing - or, at least, an ideal writing system, if
one existed - could actually be an exact visual 'representation' of
language, i.e. exhibit visually all and only the features present in
the oral mode in language (whatever they might be) is a fantasy
which has haunted the imagination of many modern theorists. It
does not bear serious scrutiny. There is not - as is often supposed -
a merely practical difficulty; that is, a difficulty that could be sorted
out if linguists sat down and worked out a sufficiently minute
'transcription system', albeit one too complex to stand any chance
of acceptance by the general public. What stands in the way of
devising an ideal writing system in this sense is not any practical
obstacle of detail but the requirements of the enterprise itself.
There could be no complete isomorphism between any system of
visible marks and any system of sounds for a quite fundamental
reason already mentioned above; namely, the incommensurability
of the sensory modalities involved. It is not merely that the choice
of particular configurations to 'represent' particular sounds (as,
for example, the letter p to represent the sound of a voiceless
bilabial stop) will be arbitrary in the Saussurean sense, but that the
basis on which visual configurations are identified and dis-
tinguished has no phonetic counterpart at all. In other words, saying
that a certain shape 'represents' a certain sound is not like saying
that on this diagram, drawn to scale, an inch 'represents' a foot.
The sense in which an inch can represent a foot is explicable by
reference to a common system of measurement and appeal to
physical proportions. The sense in which the letter p can represent
the sound of a voiceless bilabial stop (if indeed it can) is not. It is
not that choosing that shape to correspond to that sound is odd,
unmotivated, capricious, etc. but, much more fundamentally, that
to ask what phonetic relationship is represented by the way the
downstroke of the letter p joins the loop, or whether that is a
correct representation, are questions that do not make sense.
C.E. Bazell pointed out more than forty years ago that linguists
deceive themselves about the alleged correspondences between
the structure of writing systems and the structure of phonological
systems.'' In the case of alphabetic writing, it is supposed that to
the phoneme there must correspond a unit called the 'grapheme'
(which always turns out to be a born-again version of what was
traditionally called the 'letter'). But when one examines examples,
blatant discrepancies appear. For example:
By definition the phoneme cannot contain smaller distinctive
features unless these are simultaneous. The corresponding
graphic unit should equally have no smaller features except
such as are spatially superimposed. But letters are normally
distinguished from each other by features (dots, curves, etc.)
located in different positions, these positions themselves being
relevant (e.g. b/d). Hence it is, for instance, the bar and the loop
of b and d, not these whole letters, that answer to phonemes.12
Thus the prospect that one might validate the traditional notion
of 'representation' by actually demonstrating isomorphisms
11
C.E. Bazell, 'The grapheme', Litera, vol.3, 1956, pp.43-46.
12
Bazell, op. cit.
between writing and what it is supposed to represent collapses

almost as soon as investigation begins.
Bazell did not conclude from his analysis that writing systems are
'autonomous' (at least, as regards the areas of language structure
falling under phonology). However, such a conclusion might be
drawn. More generally, the thesis that writing is 'autonomous'
with respect to language has been identified by Jacques Anis as
one of three possible theoretical positions on the question of
how writing and language are related.13 This third position is
important for the present discussion, because it dispenses with the
notion of 'representation' altogether.
However, it is far from clear exactly what the 'autonomist'
position is. The term seems to imply something radically different
from the traditional view that writing merely 'reflects' speech. But
does it? That depends on what other assumptions are being made
at the same time.
To illustrate the problem, consider the following pro-
nouncement:
Writing, as we have defined it, does not imply any unique,
determinate relationship with language (langage); it can be
placed at any level: a mere reflection of a linguistic system
(systeme linguistique), or entirely autonomous with respect to the
latter, even though the content of the signs (sign.es), as we have
said, is necessarily the same in the two systems; that is to say,
human experience (êxperience humaine].^
J. Anis (with J.L. Chiss and Ch. Puech), L'Ecriture: theories et descriptions,
Bruxelles, de Boeck, 1988, p.77. The other two possible positions are
'phonocentric' and 'phonographic'. The former treats writing as an
imperfect representation of speech (but a representation nevertheless),
while the latter treats writing as representing the structure of (spoken)
language, but incorporating into this representation features of its own.
14
E.A. Llorach, 'Communication orale et graphique'. In Le langage, ed.
A. Martinet, Paris, Bibliotheque de la Pleiade, Gallimard, 1968, p.518.
This is Emilio Llorach, whose account of writing will be exam-

ined in more detail below. 'Autonomy' is patently here conceived
of as one possible relationship between writing and language. For
Llorach writing and language are separate semiological systems,
but linked by having a common 'content'.
The two systems of communication, language and writing,
share an identical content: general human experience. They
are distinct in that their signifiants are different: writing uses
graphic elements and language vocal elements.15
This appears neo-Saussurean, inasmuch as it adopts a Saus-

surean view of the sign as comprising a signifiant linked to a signifies
but it is otherwise quite unSaussurean, for it seems to imply the
possibility that a spoken and a written sign could have the same
signifie. Here Llorach begins to hedge his bets, admitting that in
principle 'it is not essentially necessary that writing should be a
graphic copy of language (caique graphique du langage}\ Exactly what
a caique graphique would be he never explains, but his discussion
seems to assume that the possibility of this copying is something
that writing makes available.
It is difficult to know what to make of Llorach's concept of
'autonomy'. He insists on the notion that language and writing
have a common 'content' (contenu). But in the sense in which a
book or speech may be said to have a content, it is difficult to see
that writing as such or language as such have any. The 'content'
problem is complicated rather than simplified by appealing to
'general human experience'. Presumably human experience
varies enormously, depending on the circumstances of individual
human lives. If there is indeed a common residuum that might
supply the 'content' of both language and writing, it would be
interesting to know who has discovered this and how. And even if
there is, Llorach offers no reason for supposing that language and
writing necessarily relate to the same aspects or parts of that
experience.
10
Llorach, op.cit., p.518.
When we come to consider what Llorach proposes by way of

the analysis of writing systems, it seems he assumes that for most
practical purposes the writing system of a community will be
employed for expressing the same range of messages as speech.
Thus the 'autonomy' of writing which was admitted initially as a
theoretical possibility remains just that: a theoretical possibility.
Llorach even goes so far as to admit that any graphic system that
organized the 'content' of human experience in a way that bore
no relation to any form of language could not be counted as
writing. In practice, then, we are left with analysing as 'writing'
only those systems of marks which in some way or other reflect
linguistic organization. Which is a roundabout - even tortuous -
way of reasserting the dependency of writing on speech.
The autonomy of writing, as a semiotic system, would exist if,
as we have seen, it were an immediate and direct expression of
the content. In that case, we should have a system of signs each
of which had a graphic expression and a content which could
coincide, or not coincide, with each of the contents dis-
tinguished by the expressions of the linguistic system, since
there would be an analysis independent of that made by
language (langage).^b
So now you see it, now you don't. The autonomy of writing
exists as a theoretical possibility; but if that possibility were actu-
ally realized, then it would not actually count as glottic writing (i.e.
would fall outside any enterprise of linguistic analysis).
The 'autonomy' option reappears in a different guise in the work

of theorists of'medium-transferability'. Their basic assumption is
that semiological relations remain constant across material
instantiations. This translates into such apparently 'common-
sense' pronouncements as:
16
Llorach, op. cit., pp.521-2.
Language, as we know it today in most parts of the world, exists

in two main forms: speech and writing.17
Who - we are implicitly invited to ask - could possibly disagree?

(Other than cantankerous academics like Bloomfield?) So straight
away writing is accorded the linguistic status that others have
denied. It is a 'form' of language. Furthermore, written language
is immediately granted 'some degree of independence' (i.e. from
other forms of language and communication).18 This is not
exactly full autonomy, but goes some way towards it. The basis on
which this status is granted depends on (i) a theoretical decision
and (ii) a piece of dubious metaphysics (not, it should be noted, on
the mere fact that everyday parlance sanctions talk about 'spoken'
and 'written' language). The theoretical decision is the decision
not to treat sound as 'an essential feature of language'.19 The
metaphysics in question, introduced to prop up that decision, boils
down to this:
Sound (and more particularly that range of sound which can be
produced by the human 'speech organs') is the 'natural' medium
in which language is realised: written utterances or texts result
from the transference of language from the phonic to the graphic
medium.20
There are already various problems here, including the inter-

pretation of 'natural' and the notion of 'transfer'. But even if we
bypass those, there remains the question of what it is that is,
allegedly, being 'transferred'. What could it possibly be? Mani-
festly, not sounds. A mark on paper is not a sound transferred. Or
is it? ('See, my dear Watson, I have transferred that sound on to
paper!' 'Really? You astound me, Holmes.')
17
J. Lyons, 'Human language'. In R.A. Hinde (ed.), Non-Verbal Communi-
cation, Cambridge, Cambridge University Press, p.62.
18
Lyons, op. cit., p.65.
19
20
No, it turns out to be not the sounds that are being transferred
from one medium to another, but just 'the words', or at least most
of them. Exactly what 'a word' is that makes it transferable is
never explained, but we are reassured that
people can learn, fairly easily and successfully for the most part,
to transfer from one medium to another, holding invariant
much of the verbal part of language.21
Here, at last, the metaphysical conjuring trick is revealed, and

very disappointing it is too. All depends on the notion of 'invari-
ance'. If I read this written sentence aloud, I am deemed to be
holding 'the words' invariant. But it is difficult to see that I am
doing anything of the kind: I am simply reading it aloud, i.e.
engaging in a vocal performance of a kind I was laboriously
trained in as a child. One might as well say that when I play the
piano from a score I am 'holding the notes invariant' too. The
alleged 'invariants' turn out to be theoretical abstractions derived
from the correspondences they are invoked to explain, not actual
items transferred at all. In short, a certain kind of linguistic
analysis is being passed off as if language were a form of energy
and the analyst were explaining the processes of energy
conversion.
A more interesting version of the 'autonomy' thesis, which dis-

penses with the specious fiction of 'medium transferability', is
proposed by glossematicians (notably Louis Hjelmslev and HJ.
Uldall). Glossematic analysis is based on the assumption that
linguistic units are independent of their expression in speech, writ-
ing, or any other material form. This is held to follow logically
from accepting Saussure's dictum that a language is a form, not a
substance.
The system is independent of the specific substance in which it
is expressed; a given system may be equally well expressed in
" Lyons, op. cit., p. 65.

any one of several substances, e.g. in writing as well as in

sounds. [ . . . ] The fact that articulated sound is the most
common means of expression is not a consequence of any
particularity inherent in the system, but is due to the anatomic-
physiological constitution of man.22
According to Uldall,
it is only through the concept of a difference between form and
substance that we can explain the possibility of speech and
writing existing at the same time as expressions of one and the
same language. If either of these substances, the stream of air
or the stream of ink, were an integral part of the language itself,
it would not be possible to go from one to the other without
changing the language. [ . . . ] ink may be substituted for air
without any change in the language [ . . . ] When we write a
phonetic or a phonemic transcription we substitute ink for air,
but the form remains the same, because the functions of each
component form have not been changed.23
More explicitly:
The system of speech and the system of writing are [. . . ] only
two realizations of an infinite number of possible systems, of
which no one can be said to be more fundamental than any
other.24
One objection sometimes raised against the glossematic posi-

tion is that in fact the written language does not always corre-
spond exactly to the spoken language: so it becomes implausible to
treat both as equivalent manifestations of exactly the same under-
lying abstract system. There are often, for instance, words that are
spelled alike but pronounced differently. Hjelmslev acknowledges
22
B. Siertsema, A Study of Glossematics, 2nd ed., The Hague, NijhofF,
1965, pp.111-112.
23
Siertsema, op. cit., p. 113.
24
that 'not all orthographies are "phonetic"!.25 But he claims that

this is irrelevant: 'it does not alter the general fact that a linguistic
form is manifested in the given substance.'26 He goes on to add
that
the task of the linguistic theoretician is not merely that
of describing the actually present expression system, but of
calculating what expression systems in general are possible as
expression for a given content system, and vice versa.21
It is rather difficult to see how Hjelmslev proposes to deal with

cases of heteronymy. For example, the fact that the English noun
entrance and the English verb entrance are identically spelt does not
automatically obliterate any difference of meaning. All that hap-
pens in such cases is that those familiar with written English learn
— to put it in traditional terms — that there are quite different
usages, grammatical constructions and pronunciations associated
with a single written form. However, according to Hjelmslev, the
fact that some orthographies are not phonetic shows that 'differ-
ent systems of expression can correspond to one and the same
system of content.'28 The problem is that if cases like entrance are
admitted as actual examples of this state of affairs, it becomes
theoretically possible to imagine writing systems which economize
on their inventory of orthographically distinct words by allowing
homographs to proliferate. The reductio ad absurdum would be a
writing system which had only one word.
The difficulty is compounded by Uldall, who claims:
If we keep the units of content constant, we shall have the same
language whatever system is used to make up the corresponding
units of expression. [ . . . ] a system of any internal structure
23
L. Hjelmslev, Prolegomena to a Theory of Language, trans. FJ. Whitfield,
rev. ed., Madison, University of Wisconsin Press., 1961, p. 104.
26
Hjelmslev, op. cit., p. 105.
21
28
will do, provided that a sufficient number of units can be made

up from it to express the units of content.29
This seems to imply that at least as many distinct units of

expression are needed as there are units of content in the lan-
guage. Which in turn means that any vocal or graphic systems
which allow cases of homophony or homography automatically
misrepresent the content system of the language. But that in turn
leads to the paradoxical conclusion that neither English speech
nor English writing properly expresses English. It is paradoxical
because without English speech or English writing it is difficult to
see what kind of existence the English language would have.
Furthermore, if it is possible in principle that speech and writing
may misrepresent the structure of a language, there seems to be
no a priori reason to assume that we can with any assurance detect
which elements of phonic or graphic manifestation correctly
represent the structure of the language and which, on the other
hand, do not. In brief, within the framework of glossematics, the
'autonomy' of writing turns out to be a direct consequence of
the assumption that language systems can exist independently of
any specific materialization whatsoever. Whether one can make
any theoretical sense at all of that assumption is another matter.
The glossematic approach has been criticized by the Prague

school theorist Josef Vachek for not bringing out sufficiently 'the
autonomous character of written language, as opposed to spoken
language'.30 The basis of his criticism is that speech and writing
fulfil quite different social functions. This, in his view, is 'more
29
30
J. Vachek, 'Some remarks on writing and phonetic transcription', Ada
Linguistica, vol.5, 1945-9, pp.86-93. Reprinted in E.P. Hamp, F.W.
Householder and R. Austerlitz (eds), Readings in Linguistics II, Chicago,
University of Chicago Press, pp. 152-7. The remark cited occurs on
p. 152, fn.2, of the latter.
profound and more essential' than the 'difference of material'

between them. Here we have yet another interpretation of'auton-
omy'; i.e. writing is regarded as an autonomous form of com-
munication in that written messages are not merely duplications
of or substitutes for spoken messages, but are employed by the
linguistic community for quite different purposes.
Other Prague school linguists, concerned with practical prob-
lems of language teaching, have pointed out that the notion of
written language representing spoken language is in any case illu-
sory, inasmuch as the norms characteristic of each are recogniz-
ably different. Thus while it is possible to read aloud a leader
article from The Times no native English speaker hearing this
would mistake it for an oral discussion of the topic. Written com-
position has its own hallmarks, selecting a grammar and a vocabu-
lary distinct from those of everyday speech. Written French, for
example, uses tenses and modal forms of the verb that are never
encountered orally except when a written passage is being read
aloud. In some literate cultures, reading aloud a particular text
may even be treated as calling for a particular form of expertise
(as is the case in the recitation of the Koran in the Islamic
tradition).31
For an integrationist, 'disparities' of this order are unsurprising.
They reflect the fact that glottic writing is not based on any uni-
form set of correspondences with an independent mode of orality
called 'speech'.
Possibly for this reason, the integrationist position is also some-

times regarded as 'autonomist'.32 This characterization, however,
is misleading on various counts.
M
K. Nelson, The Art of Reciting Qur'an, Austin, University of Texas Press,
1985.
'2J. Anis and C. Puech, 'Autonomie de 1'ecriture'. In J.G. Lapacherie
(ed.), Proprietes de 1'ecriture, Pau, Publications de 1'Universite de Pau, 1998,
pp.79-87.
For the integrationist, written communication of the kind that

takes place if you read and understand this chapter is a form of
communication in its own right. It does not become so simply in
virtue of and insofar as written forms can be correlated with
spoken forms, even though you are perfectly at liberty to read the
chapter aloud if you so wish. In order to communicate in written
English it is not essential that one should also know how to speak
it, any more than reading or writing Latin requires one to know
how Caesar, Cicero and their contemporaries pronounced it. As
Vachek points out, those who suppose otherwise are conflating
writing with phonetic transcription.
That is to say, in deciphering a text put down in writing no
detour by way of spoken language is necessary to make out its
content, as is the case in deciphering a phonetically transcribed
text. A clear proof of this assertion is the well-known fact that
there are many people who can, for instance, read English
without having any idea of how the written text should be
pronounced.33
However, that this can be done does not alter the fact that there
are also biomechanical and socially institutionalized practices
known as 'reading aloud' and these are systematically integrated
with optical scanning of the text. That form of integration, pre-
cisely, is one of the criteria available for distinguishing between
'glottic' and 'non-glottic' writing. Anyone who can read and
understand an English text without, as Vachek says, 'having any
idea' of how to pronounce it (and presumably having no co-
relative idea of how to write down in English orthography an
English utterance) has simply not mastered English as a form of
glottic writing. It is fair to say, in short, that there is something
important missing from such a person's grasp of communication-
in-English. So to adduce the fact that written English can be treated
as non-glottic as a reason for claiming it to be an autonomous
form of communication would be rather beside the point.
33
Vachek, op. cit., p. 155.
Furthermore, from the fact that written English can be treated

as non-glottic it does not follow that it could exist independently
of any other form of communication, but simply that it could exist
independently of spoken English. If that is all that being an
'autonomist' amounts to, then integrationists are autonomists. But
then so are many other theorists - at least, all those who accept
that there are forms of writing (e.g. musical notation) that have
nothing to do with speech at all. The whole point of the term
'glottic' writing is to recognize that there are forms of writing
intimately connected with spoken language. So if anyone wants to
insist that glottic writing has to do with speech, they are merely —
as far as the integrationist is concerned - harping on a tautology.
The question of what exactly glottic writing has to do with speech
is another matter, and here the integrationist proposes an account
which is quite different from any 'representational' account.
If anyone should feel inclined to object that none of the above
considerations provides a theoretical justification for including
written English as a form of linguistic communication, the inte-
grationist reply is that this is how its practitioners regard it (i.e. as a
form of their language). This assimilation holds across all literate
societies for which reliable evidence is available. The evidence in
question is metalinguistic: i.e. the name of the spoken language is
automatically transferred to the associated form of writing. This
seems to be universal in societies that have reached the stage of
utilitarian literacy. What stands in need of justification, on the
contrary, is the position of those theorists who insist on equating
language with speech and treating writing as 'a kind of highly
unsatisfactory pseudotranscription'.34
It is a mistake to regard the integrationist position as 'autono-
mist' in any other than the respect already indicated above, since
within an integrational framework there are no autonomous signs
and no autonomous sign systems. Nor could there be. For signs
exist only as contextualized products of particular communication
situations. This applies as much to the written sign as to any other
34
Vachek, op. cit.,p.!53.
kind of sign. In the case of glottic writing the debate about 'auton-
omy' is particularly futile, since the whole function of glottic writ-
ing is to act as an interface between biomechanically different
modes of communication.
Some of the points made above may be illustrated by reference to

one of the most curious episodes in the history of writing: the
attempt by Jean Itard, resident physician at the National Institute
for Deaf Mutes in Paris, to introduce the alphabet into the educa-
tion of Victor, the languageless 'wild boy of Aveyron', whose case
aroused such interest in the early nineteenth century. Itard's own
account of this runs as follows.
I had the twenty-four letters of the alphabet printed in large
type, each on a piece of cardboard two inches square. I had an
equal number of slots cut in a plank one-and-a-half feet square,
into which I had the pieces of cardboard inserted, but without
glueing them, so that they could be rearranged as required. An
equal number of letters of the same size were made in metal.
These were to be compared by the pupil with the printed let-
ters, and placed in the corresponding slots. The first trial of this
method was made in my absence by Mme Guerin. I was greatly
surprised to learn from her on my return that Victor could
distinguish all the letters and arrange them properly. This was
straight away tested and he did it without the slightest mistake.
Delighted with such rapid success, I was still far from being able
to explain it; and it was only a few days later in my room that
the way our pupil managed to sort the letters was revealed. To
make it easier, he had devised a little trick which alleviated the
work of memory, comparison and judgment. As soon as the
board was in his hands, he did not wait for the metal letters to
be removed from their slots; he took them out and piled them
up on his hand in order; so that the last letter of the alphabet
was on top of the pile when the board had been cleared. Then
he started with this one and finished with the last in the pile,
thus beginning at the foot of the board and proceeding always
from right to left. That is not all: he managed to perfect this

procedure; for often the pile collapsed and the letters fell down;
and then he had to sort it all out by paying careful attention to
each one. The twenty-four letters were arranged in four rows of
six; so it was simpler just to take out each row and put it back
again, leaving the second row until the first had been dealt with.
I do not know if his reasoning was as I suggest; but he cer-
tainly carried out the operation as I have described. So it was
indeed a routine, and one of his own invention, which perhaps
did as much credit to his intelligence as a methodical classifica-
tion was soon to do to his discrimination. It was not difficult to
put him on the right track, by giving him the letters all jumbled
up each time he was offered the board. Eventually, in spite of
my frequently changing round the printed characters in their
slots, and presenting tricky juxtapositions, such as puttting G
next to C, E next to F, etc., his identification was impeccable. By
this training in all the letters, my aim was to prepare Victor to
make some, albeit elementary, use of them to express the needs
he could not express by means of speech. Far from believing
that I was already so close to that important stage in his educa-
tion, more out of curioisity than hope of success, I tried the
following experiment:
One morning when he was waiting impatiently for his daily
milk at breakfast, I took the four letters L A I T from his set and
arranged them on a board I had specially prepared. Mme
Guerin, whom I had alerted, comes forward, looks at the letters
and then gives me a cup full of milk, which I pretend to take for
myself. A moment later I go to Victor and give him the four
letters I have just taken from the board: I point to the board
with one hand while offering him the jug of milk with the other.
The letters were immediately replaced, but in exactly the
reverse order, so that they read TIAL instead of LAIT. I showed
him the corrections needed, pointing to each letter and where it
should go. When these changes had produced the sign of the
thing, I did not deprive him of it any longer.
It may be difficult to believe that five or six such trials sufficed
not just to get him to arrange the four letters of the word lait
systematically, but also, I would say, to give him the idea of the
relation between the word and the thing. That is at least
strongly suggested by what happened eight days after this first
experiment. We noticed that while getting ready to leave one
afternoon to go to the Observatory he took these four letters on
his own initiative, put them in his pocket and, as soon as he
arrived at citizen Lemeri's, where, as I said earlier, he goes for a
drink of milk every day, arranged them on a table to form the
word lait.35
It cannot be supposed that Itard was an integrationist seeking

experimental confirmation for any hypothesis about the semi-
ology of the alphabet; but if he had been, he could hardly have
done better. Victor first of all treats the letters as units of a nota-
tion (in the sense of Chapter 4), because this is how they feature in
his 'lessons'. At the beginning, the only activities to be integrated,
as far as he is concerned, are those involved in matching the
twenty-four letters visually with their exemplars and putting them
in the right slots. What the letters signify, as far as he is concerned,
is exhausted by their role in this matching game, which he soon
learns to play. But this is no longer the case at the next stage,
which concerns just four letters and their arrangement in a fixed
order, this invariable combination being matched with access to a
drink of milk. To assume, as Itard did, that this new performance
is explained by Victor's having understood the relation between
the word and the thing is already to impose a dualist psychological
interpretation on the case, and quite unnecessarily so. The sys-
tematic integration of the activities involved is all that is required
to account for the semiological value of the combination L-A-I-T
in this new game. That is all Victor needed to grasp, just as he had
previously done in the letter-matching game.
According to Harlan Lane, Victor's failure to make significant
35
This passage from Itard's original text (1801) is reprinted in L. Malson,
Les Enfants sauvages, Paris, Union Generale d'Editions, 1964, pp. 181-4.
linguistic progress at this stage in his education was due to 'mis-

placed stimulus control', and specifically to his inability to grasp
the difference between 'tacting' and 'manding'.36 (This, in Skin-
nerian jargon, is the difference in verbal behaviour between nam-
ing and requesting.37) But the behaviourist explanation hardly fits
the evidence cited above. To suggest that 'the response may have
been evoked by the many cues that accompany milk as presented
in this setting' is hardly a convincing diagnosis of Victor's exploit
at citizen Lemeri's. Carrying letters in your pocket in anticipation
of the need to 'name' an object is not prima facie explicable as a
response to a stimulus: it looks too suspiciously like conceiving a
plan of action. Victor's initiative makes more sense if seen as an
experiment in creative contextualization. Put in integrationist
terms (although these are obviously not those Victor would or
could have used), the question to be answered was whether the
previously established value of the sign 'L-A-I-T' could be carried
over to a new situation with new participants. Its outcome does
not have to be interpreted in terms of any theoretical distinction
between 'tacts' and 'mands', and in itself it neither proves nor
disproves Victor's understanding of the concept of 'requesting'.
We are reckoned to understand concepts like 'requesting' when
we show we have grasped by participation the biomechanical,
macrosocial and circumstantial limits within which integrational
patterns of give-and-take operate over a whole range of situations.
Itard eventually taught Victor the rudiments of writing and
reading, but never had any parallel success in teaching him to
36
H. Lane, The Wild Boy of Aveyron, London, Allen & Unwin, 1977,
p. 124.
3/
The following definitions are given in A.R. Reber, Penguin Dictionary of
Psychology, Harmondsworth, Penguin, 1985. 'Tacting is verbal behavior
that is most clearly under the control of its antecedents; it results from or
is linked with that which has gone before. Naming is the classic tact.'
(p.757.) '[ . . . ] manding represents verbal behavior primarily under
control of its consequences. Classic examples of mands are "Please pass
the salt" and "Get me my book".' (p.416.)
speak. The reasons for this failure have been much debated. But
clearly the case speaks to the question of writing as 'representa-
tion'. For Victor, writing never was - and never could have been -
a 'representation' of speech. But here too the traditional view of
the 'primacy of speech' prevailed - to Victor's irrecoverable
disadvantage. As Lane observes:
In teaching Victor to understand and produce written lan-
guage, Itard inexplicably left off once Victor had mastered
strings of verb plus noun. Small steps indeed lay between this
performance and the control of elementary French sentences
[ . . . ]. If Itard had been less committed to oral language,
Victor might have realized Bonnaterre's ambition for him and
gone on to master the written language and to become a
Massieu after all.38
One of the most original thinkers of the nineteenth century,

Wilhelm von Humboldt, recognized that in a literate society
speech and writing cannot remain indefinitely in a state of
developmental apartheid.
The needs, limitations, merits and peculiarities of both influ-
ence each other. Changes in writing may lead to changes in
language; and even though we write as we do because that is the
way we speak, it is also the case that we speak as we do because
1 * 1 • 3 9
that is the way we write.
It is interesting that it was not until more than a hundred and

fifty years later that the latter part of Humboldt's observation was
38
Lane, op. cit, p. 170.
39
W. von Humboldt, 'Uber die Buchstabenschrift und ihren Zusam-
menhang mit dem Sprachbau', Abhandlungen der koniglichen Akademie der
Wissenschqften zu Berlin, 1826, pp.161-8. Quoted and translated by T.C.
Christy, 'Humboldt on the semiotics of writing': in I. Rauch and G.F.
Carr (eds), The Semiotic Bridge, Berlin / New York, Mouton de
Gruyter, 1989, p.340.
taken up seriously by any academic linguist: this was done by F.W.

Householder in his book Linguistic Speculations.*0 He argued that in
(American) English cases in which pronunciation adapts to spell-
ing are more numerous than the other way round and, further-
more, the 'rules' which relate written forms to their pronunciation
are much simpler than the 'rules' which relate the pronunciation
of words to their spellings. One way of explaining this would be to
suppose that although American children learn to speak before
they learn to write, the written forms, once learnt, serve to explain
and regularize the spoken forms the child is already familiar with
orally. In short, written English would, on this view, provide a first
elementary metalanguage - the letters of the alphabet - which
made it possible for the learners to describe, compare and analyse
the oral forms they heard, and thus the continuous process of
learning new spoken words would come to be guided by fitting
them into the framework supplied by spelling. Householder called
his thesis (provocatively, given the views then current in orthodox
linguistics) 'the primacy of writing'.
This thesis was later developed and generalized by David
Olson.41 According to Olson, we are dealing with a universal pro-
cess underlying the history of all literate societies. It is illusory to
suppose that writing systems owe their origins to the need to
'represent' oral structures already recognized. On the contrary,
it is the development of a writing system which facilitates concep-
tualization of a corresponding oral structure:
[ . . . ] writing systems provide the concepts and categories
for thinking about the structure of spoken language, rather
than the reverse. Awareness of linguistic structure is a product
of a writing system, not a precondition for its development.42
40
EW. Householder, Linguistic Speculations, Cambridge, Cambridge Uni-
versity Press, 1971.
41
D.R. Olson, 'How writing represents speech'. Language & Communica-
tion, Vol.13 No. 1, 1993, and The World on Paper, Cambridge, Cambridge
University Press, 1994.
42
Olson 1994,p.68.
On this view of literacy, what it involves is remaking one's con-

ception of speech in the image of the writing system one has been
taught. Thus the alphabet would not have been invented in order
to 'represent' the word as a sequence of sound units already recog-
nized. The boot is on the other foot: acquaintance with alphabetic
writing would have induced those familiar with it to regard the
stream of speech (even when not 'written down') as comprising
such a sequence of units. This relationship does not apply merely
at the level of alphabetic literacy. In Olson's view, the word itself
as a discrete linguistic unit is likewise a graphic conception.
If this is so, the relationship between speech and writing - as far
as the lay members of the literate community are concerned - is
exactly the reverse of what is assumed by orthodox theorists. The
invention of writing gave birth to a new conception of the utter-
ance, a projection of the graphic image on to the screen of oral
awareness.
Evidence that can be mustered in support of this thesis is not
wanting. Most obvious, perhaps, is the fact that the basic linguistic
terminology of Western education, including the term grammar,
presupposes acquaintance with writing. There is no vestigial
vocabulary of speech analysis surviving from an earlier 'oral' trad-
ition. On the contrary, as was noted in Chapter 1, from Plato
onwards one is struck by the absence of any rigorous termino-
logical distinction between letters and sounds. This is a conflation
which lasts right down to the nineteenth century, as Saussure's
critical remarks about the comparative grammarians remind us.
The 'failure', however, must not be overinterpreted. It does not
mean, as was already noted in Chapter 1, that we have to attribute
to Plato and his contemporaries the superstitious notion of
'armies of little black signs issuing forth from human mouths, like
the diamonds or toads of fairy stories.'43 Nor did Saussure
imagine that Bopp thought this. It remains true, nevertheless, that
1
F. Desbordes, 'La pretendue confusion de 1'ecrit et de 1'oral dans les
theories de 1'antiquite'. In N. Catach (ed.), Pour une theorie de la langue ecrite,
Paris, CNRS Editions, 1989, p.2 7.
for the Greeks as for Bopp the alphabetic letter was not only an
important model for their conceptualization of the unit-structure
of speech, but the only model available.
The whole problem of phonetic analysis, from antiquity down
to the present day, turns on a supposed correlation between the
letters of the alphabet on the one hand and 'sound segments' on
the other. It is obvious that the former are far more clearly defined
than the latter. In fact, the puzzle is how one would ever set about
segmenting a continuous sound sequence at all without appeal to
an independent series of discrete graphic marks. The task itself
calls for projecting imaginary discontinuities on to a continuum.
The dangers (for linguistic analysis) of this projection were
recognized by Saussure when he admitted that 'the written word is
so intimately connected with the spoken word whose image it is
that it manages to usurp the principal role.'44 From this usurpation
Saussure did not draw the same conclusion as Olson; namely, that
to think of the relationship in these terms is to see it the wrong
way round. Olson's conclusion is that for literate communities, it is
not the written word which is the image of the spoken word, but
the spoken word which is the image of its written counterpart.
Interesting support from this comes from an area where it
would be least expected if the orthodox view were right. Debates
have raged endlessly in twentieth-century phonology over the
'phoneme' and how to determine how many such units any given
spoken language has. For instance, it has been pointed out that for
'standard French' the number of vowel phonemes has been reck-
oned by different linguists as being as low as eight and as high as
twenty.45 Such disparate totals evidently imply different methods
of counting, but that is the point: no comparable debate has arisen
among linguists over analysing the letters of the French alphabet
(in spite of the complications caused by acute, grave and circum-
flex accents). What emerges as a theoretical problem in phonology
44
45
T.B.W. Reid, Historical Philology and Linguistic Science, Oxford, Claren-
don, 1960, p. 12.
is generated by the assumption that just as any schoolboy can tell

you how many letters there are in a given written word (provided
his arithmetic is up to counting them), so the linguist should be
able to tell you how many phonemes there are in its oral form.
The answer to the second question turns out to be much more
elusive than the answer to the first. But the point here is that the
form of the problem is dictated by looking for a determinate
number of segmental phonemes in the first place. That itself sup-
ports Olson's thesis: the phonological word is being conceptual-
ized after the image of the written word, not vice versa.
Olson also draws attention to various research findings which
suggest that, psychologically, writing systems serve as a model for
speech or even determine the way in which speech is perceived.
For example, studies of 'phonological awareness' indicate that
ability to divide words into sub-syllabic segments depends on
acquaintance with alphabetic writing. Chinese familiar with the
(alphabetic) pinyin system show an ability to do this which is not
matched by those Chinese who can read only traditional Chinese
characters.46 Similarly, Vai literates, familiar with the Vai syllabic
script, were found to be much better than non-literate Vai people
at putting together separate syllables into phrases and splitting up
phrases into their constituent syllables. Olson comments:
This suggests that the learning of a syllabary is a matter of
coming to hear one's continuous speech as if it were composed
of segmentable constituents.47
Olson's claims raise many intriguing questions which cannot be

pursued here. And the issue on which Olson calls in question the
traditional Western wisdom about writing as a 'representation' of
speech must not be oversimplified. No one argues that, prior to
the advent of writing, speakers were quite oblivious to features of
46
C.A. Read, Y. Khang, H. Nie and B. Ding, 'The ability to manipulate
speech sounds depends on knowing alphabetic reading', Cognition, vol.24,
1986, pp.31-44.
47
Olson, op. cit, p.82.
oral language. This would be like supposing that musicians were

tone deaf before the invention of musical scores. It is clear from
the evidence of poetry in pre-literate cultures that awareness of
quite subtle patterns of rhythm, rhyme and assonance did not
escape either the pre-literate poet or the pre-literate audience.
What is at issue, however, is the role of writing vis-a-vis speech in
that symbiotic relationship which characterizes literate societies.
Two hypotheses stand in confrontation. According to one, writing
serves to record linguistic units and structures already recognized
by speakers of the language in question. According to the other,
writing itself serves as the model for the recognition of such units
and structures.
What neither hypothesis offers is a direct answer to the question
of how the written sign comes to play the role hypothetically
assigned to it. Integrational theory is in a position to offer a
semiologically motivated answer:
Any graphic configuration acquires a certain linguistic value insofar as it
serves to articulate the integration of one form of verbal activity with
another, or verbal activities with non-verbal activities.
In short, the very integration with speech that glottic writing

affords in such practices as 'dictation' and 'reading aloud' is what,
inevitably, leads the participants to treat these as alternative or
correlative forms of linguistic expression. When this happens the
linguistic units comprising such messages - whether previously
recognized as such or not — can no longer be defined by purely
oral criteria. The integration of speech with writing irreversibly
transforms social conceptions of language. Language becomes
subject to other conditions than those of orality. It becomes pos-
sible to ask for the identification of linguistic forms in visual as
distinct from auditory terms, or even to demand, for particular
purposes, that they shall be so expressed.
The cultural history of the signature (Chapter 7) provides a
particularly clear illustration. In a pre-literate society, a personal
name serves to identify an individual, enabling others to call or
refer to that person: if A's name coincides phonetically with B's
name, then A and B both have 'the same name'. In a literate

society, on the other hand, the name can serve other purposes; for
example, in signing a document. Here the name remains, as
before, a means of identification. But, as Beatrice Fraenkel aptly
remarks, through the signature recording one's name also
becomes a form of action.48 This extension in the functions of the
name is bought at a certain price: its phonetic characteristics
become secondary. What is important is no longer how it sounds
but how it is written. So Mr White and Mr Whyte are no longer
counted as having the same name. This change is a change in the
concept of the name. And for so long as that change is supported
by the social and legal practices of a literate society, there is no
prospect of reverting to the old oral concept of what a name is.
Nor can it be supposed that the old oral concept survives along-
side the new concept. When someone asks your name, you do not
say: 'Which one? My written name or my spoken name?' In a
literate society you have only one name, even though it has both
an oral and a written form.
That, in a literate society, goes not only for names but for
words, phrases, sentences and for language in general. What dis-
tinguishes a literate culture from a pre-literate culture is not so
much the addition of a quite separate mode of verbal communica-
tion as the incorporation of oral communication into a higher-order
semiological synthesis involving the written sign. In that syn-
thesis, however, it is increasingly the graphic element which
dominates.
Something similar can be observed in the history of Western
music. The crucial development is not the invention of new
instruments with new 'sounds', or new ways of playing old
instruments, important though these may be, but the introduction
of systematic and detailed forms of music notation. Originally an
aid to performance, written notation made possible compositions
of an extent and complexity that could not be undertaken before,
lust as in the case of language, the price paid for this is a change in
48
Fraenkel, op. cit, p. 12.
the conception of the musical work. The score becomes its

'authentic' form.
In both the linguistic and the musical case, the underlying
reasons are not difficult to see. The integration of writing with
both speech and music produces a new material object: the
document. From the moment of its creation, the document leads
an independent material existence. It thus provides an effective
means of controlling the future of whatever text or composition it
contains. It supplies a basis for replication, analysis and verifica-
tion which did not exist in the pre-documentary era.
The macrosocial advantages of writing appear no less evident.
Documentary records facilitate forms of control over individuals
and whole populations that are impossible in a pre-literate society.
These do not appear overnight. (It took, for example, several cen-
turies in Anglo-Saxon England to establish the validity of written
wills.49) But once the effectiveness of these forms of control is
recognized, it is only a matter of time before they are extended.
We are destined to live in an ever more minutely documented
society, even though the documentation will be electronic docu-
mentation, not paper documentation.
For societies that have reached this historical turning point, the
choice is between an 'old' semiology of writing in which writing is
treated simply as one possible form for the expression of a mes-
sage and a 'new' semiology in which writing is treated as the
creation of textualized objects. For reasons indicated in the pre-
ceding chapters, only an integrational approach can do justice to
the latter. The essential difference between the two is that in one
case semiological values will depend on the 'adequacy' of the
written form to express the given message, whereas the other will
see semiological values as derived from the role of the textualized
49
B. Danet and B. Bogoch, 'From oral ceremony to written document:
the transitional language of Anglo-Saxon wills', Language & Communica-
tion, Vol. 12 No.2, 1992.
object in integrating the activities of those who participate in its

making and interpretation. These are two quite different concep-
tions. The first leads to a reification of 'the message' as something
which exists in a prior form, either physical or mental, needing to
be reconciled with the exigencies of a particular writing system.
The second leads to recognizing 'the message' not as something
given in advance - or given at all - but as something created
by interaction between writers and readers as participants in a
particular communication situation.
But if we are to envisage how this may affect the future of a
literate society, it is important not to go on thinking of 'writer' and
'reader' as occupying sempiternally the fixed roles that were
allotted to them under the old regime of utilitarian literacy.
CHAPTER NINE
Mightier than the Word
Even before the publication of Saussure's Corns de linguistique

generale in 1916, a far more revolutionary rethinking of writing was
already under way. The revolutionary forum was not linguistics
but literature, and its most spectacular early manifestation was
Mallarme's poem Un coup de des. At first sight it might appear that
its novelty consists merely in being a work in which 'the type is an
integral part of the enterprise'.1 In his brief preface, Mallarme
admits that perhaps the most upsetting aspect of the work for the
average reader is likely to be the absence of the regular lines that
define the visual form of poetry (Fig. 8). Mallarme's meticulous
concern with the exact positions of the words on the page is
revealed in the proofs he corrected shortly before his death for the
(subsequently unpublished) Lahure edition, which was to have
been illustrated by Odilon Redon.2
A recent critic describes Mallarme's engagement with
typography as follows:
So as to indicate the structure of this fairly lengthy and
extremely complex piece of prose made up of some 650 words
1
R.G. Cohen, Mallarme's Masterwork. New Findings, The Hague/Paris,
Mouton, 1966, p. 78.
2
Cohen, op. cit, reproduces Mallarme's corrected proofs and also
Redon's lithographs.
Figure 8
covering 21 pages Mallarme uses different kinds of lettering.

The main clause, printed in bold capitals, is interrupted after
'jamais' by a subordinate clause in smaller capitals which, in
turn, is interrupted by a long and intricate passage in ordinary
roman type. Only after these two parentheses does the verb
'n'abolira' appear and it too is followed by a long parenthesis in
italics before the object of the verb, 'le hasard', makes its
appearance. A final qualifying clause is then introduced, at first
in italics and then in roman type, to bring the work to a close.3
If that were all, it would already suffice to show that at least one
poet in the nineteenth century had grasped the fact that writing
has a potential for going beyond the resources of oral poetry, and
beyond the resources of conventional speech altogether. For on
Mallarme's page the reader sees the visual articulation of a syntax
3
C. Chadwick, The Meaning of Mallarme, Aberdeen, Scottish Cultural
Press, 1996, p. 12.
Mightier than the Word 217
that is orally 'impossible'. (To call it 'ungrammaticaP would be on

a par with saying that Picasso's Les Demoiselles d'Avignon shows a
poor grasp of perspective.) But there is more to the typography
than that.
The words and sentences are sparingly distributed across the
comparatively large area of the double page, which Mallarme
uses as his 'frame' instead of the single page, so that the lines of
print, sometimes trailing across the paper like a drawing of the
wake of a ship, sometimes grouped together like black dots on
white ice, and sometimes more widely scattered like black stars
in a white sky, reinforce the three kinds of imagery which dom-
inate Un coup de des [. . .]
It is interesting and no doubt significant that the concluding
lines of Un coup de des are set out not only in the shape of two
dice which have finally been thrown and have fallen to reveal
three and four dots respectively, but also in the shape of the
constellation of the Great Bear with its slanting line of three
stars and its rectangle of four stars.4
But when in his preface Mallarme explains what he is trying to

do, he has to fall back on comparing blank spaces with silences - a
comparison which immediately reinstates the assumptions under-
lying the phonoptic tradition that Un coup de des deliberately sub-
verts. He speaks of 'this copied distance which mentally separates
groups of words or one word from another' and how this some-
times accelerates and sometimes slows down the movement of the
text.3 What else can the 'copied distance' refer to, if not the inter-
val which may separate two words or phrases in the spoken
4
Chadwick, op. cit, pp. 12-13, p. 165.
5
'L'avantage, si j'ai droit a le dire, litteraire, de cette distance copiee qui
mentalement separe des groupes de mots ou les mots entre eux, semble
d'accelerer tantot et de ralentir le mouvement, le scandant, Fintimant
meme selon une vision simultanee de la Page.' S. Mallarme, (Euvres
completes, ed. H. Mondor and G. Jean-Aubry, Paris, Gallimard (Biblio-
theque de la Pleiade), 1945, p.455.
sentence? Unexpected spaces in the text correspond to unexpected

pauses in speech. Here is a striking example of how difficult it is
even for the iconoclast, within a phonoptic tradition, to conceive
of writing as anything other than a substitute for speech.
It is even more striking that when Mallarme addresses the
semiological status of his poem as writing, he appeals to a musical
analogy and says that his text is, for anyone willing to read it
aloud, a score.6 To suppose that Mallarme is seriously thinking
here of an oral delivery of his poem would miss the point of the
metaphor. The score, for Mallarme, is the paragon of a written
text which presents directly, without ambiguity, what it signifies -
the musical work, the composer's thoughts; whereas poets, on the
contrary, are perpetually obliged to struggle against the con-
straints of language through which they must express themselves.
Dualist theories of communication feed all forms of artistic
schizophrenia. Mallarme's schizophrenia takes the form of want-
ing to make poetry an artistic medium that can render ideas in all
their pristine purity, uncontaminated by the words employed to
express them. Saussure, who died just before the ne varietur edition
of Mallarme's revolutionary graphic work was published, would
have regarded that ambition as a kind of linguistic dementia
resulting from a total failure to understand the mechanism of la
langue.
It is even more difficult to imagine what Saussure would have

made of Guillaume Apollinaire's poem Lettre-Ocean, published in
1914 (Fig. 9). Mallarme's Un coup de des at least has a beginning a
middle and an end, a sequential order of some kind, but with
Apollinaire's poem the problem is 'where to start reading'; for the
work
imposes no point of entry and offers no consistent order of
Aj outer que de cet emploi a nu de la pensee avec re traits, prolonge-

ments, fuites, ou son dessin meme, resulte, pour qui veut le lire a haute
voix, une partition.' Mallarme, op. cit., p.455.
LETTRE-OCEAN
Figure 9
reading. It even undermines conventions of figure and

ground, surface and depth, frame and canvas that traditional
easel painting in the West has made second nature for most of
us.7
In other words, even if we treat it as a picture rather than a poem,

we find it no less difficult to 'read' pictorially than verbally.
The impact of this first modern calligram underlines just how
restricted writers of the eighteenth and nineteenth centuries had
been by the traditional conception of writing. Where even Mal-
larme had paid a modicum of respect to the 'linear' conventions
dear to publishers and printers, Lettre-Ocean flouts them openly and
enthusiastically.
A less obviously visual challenge to the tradition came in 1939

with the publication of James Joyce's novel Finnegans Wake. Here
the attack is taken into the opposition's home territory: since the
external appearance is that of the conventional Western book —
until one starts reading.
riverrun, past Eve and Adam's, from swerve of shore to bend
of bay, brings us by a commodius vicus of recirculation back to
Howth Castle and Environs.
Sir Tristram, violer d'amores, fr'over the short sea, has pas-
sencore rearrived from North Armorica on this side the scraggy
isthmus of Europe Minor to wielderfight his penisolate war: nor
had topsawyer's rocks by the stream Oconee exaggerated them-
selse to Laurens County's gorgios while they went doublin their
mumper all the time: nor avoice from afire bellowsed mishe
mishe to tauftauf thuartpeatrick: not yet, though venissoon
after, had a kidscad buttended a bland old isaac: not yet, though
all's fair in vanessy, were sosie sesthers wroth with twone nath-
andjoe. Rot a peck of pa's malt hadjhem or Shen brewed by
7
R. Shattuck, The Innocent Eye, New York, Washington Square Press,
1986, pp.296-7.
arclight and rory end to the regginbow was to be seen ringsome

on the aquaface.
The fall (bababadalgharaghtakamminarronnkonnbronnton
nerronntuonnthunntrovarrhounawnskawntoohoohoordenenth
urnuk!) of a once wallstrait oldparr is retaled early in bed and
later on life down through all Christian minstrelsy. The great fall
of the ofrwall entailed at such short notice the pftjschute of
Finnegan, erse solid man, that the humptyhillhead of humself
prumptly sends an unquiring one well to the west in quest of his
tumptytumtoes: and their upturnpikepointandplace is at the
knock out in the park where oranges have been laid to rust
upon the green since devlinsfirst loved liwy.
And so on, for six hundred pages and more.

It has been said that the writing of Finnegans Wake marked the
end of English.8 Manifestly untrue. But doubtless Finnegans Wake
was one of the events that marked the end - or the beginning of
the end - for a theory of writing that had gone virtually unchal-
lenged in Western literature since the days of Aristotle. Before
Joyce produced his contentious masterpiece, it still seemed super-
ficially plausible to maintain that all literary prose works were, in
the final analysis, surrogates for speech; not because they were
transcriptions of what the author had actually at one time said,
but in the sense that they had been written in such a way as to
suggest or invite a reconstruction of what the author or the
author's characters would have said, had they spoken. But this
traditional view was hard put to it to explain a work like Finnegans
Wake, peppered with orthographic forms corresponding to no
known words of anyone's spoken language. Joyce's text was - and
is — a form of writing which defies oral rendition (which, as might
have been expected, did not prevent the author from recording his
own 'readings' from it). But, leaving such isolated virtuoso vocal
performances on one side, for most readers the most salient fact
about the book is, as one recent editor notes, that it is 'unreadable'.
8
Philippe Sollers, Theorie des exceptions, Paris, Gallimard, 1986, p.80.
In order to pay it the attention it so impertinently and endlessly

demands, the reader must forego most of the conventions about
reading and about language that constitute him/her as a reader.
The advantage to be gained from doing so is considerable; the
conventions survive but they are less likely thereafter to dwindle
into assumptions about what reading or writing is.9
Whatever doubts about the 'advantage' sceptics may entertain,

there can be no doubt about the perspicacity of the remark in the
latter part of the sentence. That the book is 'unreadable', as the
same commentator remarks, is 'the first thing to say about' it. And
perhaps not only the first thing - since it is obvious from the
opening pages - but the last thing too. For those who survive the
gruelling exercise of grappling with Finnegans Wake, there can be
no doubt that the 'normal' process of reading will never seem the
same again. This is not just a shock tactic that the author has
decided to employ for attention-grabbing purposes. The 'unread-
ability' of the book is a focal point of the author's message about
language and about literature. Line by line, we have to guess at — or
ignore — the pronunciation of words we have never seen before.
And yet, behind this mish-mash of neologisms, derived etymo-
logically from everywhere and nowhere, we seem to hear - some-
times very clearly - the lilt of an Irish accent. This alone should
suffice to convince us that 'unreadability' is, after all, only a rela-
tive term.
It was not English that failed to survive the earthquake of
Finnegans Wake but the conception of written English as the
accredited graphic transcription, however imperfect, however
conventional, of a pre-existing oral English, a form of speech with
which one was already acquainted. What cannot be said, in the
case of Joyce, is that 'the language had always been there'. It had
not. Joyce is the first writer in the Western tradition to reject the
doctrine of 'linguistic priority' in its totality. Joyce's language is
9
J.Joyce, Finnegans Wake, ed. Seamus Deane, London, Penguin, 1992,
p.vii.
created by Joyce, and created on the basis of resources offered by writing.

What would Saussure have made of that? Conceivably, he might
have dismissed it as a playful attempt to create an imaginary lan-
guage. But that would have been to miss the point. The language
of Finnegans Wake is not an extended exercise on the model of
More's Utopian.10 Since, for Saussure, languages as systems are
entirely autonomous systems of arbitrary signs, the resemblances
between Joyce's language and English would have had to be
entirely fortuitous. Furthermore, since writing systems, in Saus-
sure's view, exist only as systems of metasigns, a language (langue)
created on the basis of writing would be tantamount to the mir-
acle of the tail wagging the dog.
This has not prevented French critics from foisting Saussurean
interpretations on the language of Finnegans Wake. In what Joyce
writes, claims Sellers, deliberately echoing a well-known Saus-
surean dictum, 'there are only differences'. The comment would
have been more apposite if 'differences' had been replaced by
'analogies'. Sellers himself provides the following etymology for
the Joycian word traumscrapt.
TRAUMSCRAPT. Traum ('dream'), trauma ('trauma'), script
('script'), rapt ('abduction'). Development: what's an ecrit ('writ-
ten work')? the rapt of a trauma; what's a trauma ? the rapt of an
ecrit, what's a rapt ? the script of a trauma. [. . . ]
This is what Joyce calls his 'trifid tongue' - an expression in
which you can hear terrific (terrible) but also trifle (joking).11
The chains of lexical analogy stretch out in all directions, ignor-

ing linguistic boundaries. What emerges is the profound semio-
logical originality of writing. Only writing is in a position to
10
Appended to Thomas More's Utopia (1516) are four lines of verse 'in
the Utopian tongue', together with a 'translation'. The printer apolo-
gizes for printing the passage transliterated in the conventional Euro-
pean manner, explaining that he does not have available 'the Utopian
alphabet'.
11
Sellers, op.cit., pp.87—8.
orchestrate, simultaneously and with equal resonance, this com-

bination of phonetic, syntactic and etymological images, across a
whole range of languages and dialects. And that is the basis of
Joycian language-making. Words are still, apparently, its means of
articulation; but they are words freed from the constraints of any
fixed code.
It would be beside the point to dismiss Finnegans Wake as show-
ing a bizarre, perverse, experimental approach to writing, just as it
would be beside the point to dismiss analytic cubism as showing a
bizarre, perverse, experimental approach to painting. If the ana-
lytic cubists accomplished nothing else, they demonstrated once
and for all that the semiological limits of painting do not coincide
with the representation of the visible world. Joyce did the same for
the semiology of writing and the representation of the spoken
word.
The lessons of Un coup de des, Lettre- Ocean and Finnegans Wake were
not easy to assimilate for a public educated to respect the norms
of the Western literary tradition. What exactly was going on here?
What was going on - and many commentators on the modernist
movement still have not realized it — was that the long-lived West-
ern concept of glottic writing was being tested to the limit. It was
being deliberately subjected to intellectual and aesthetic pressures
under which it was bound to fracture, as did the 'representational'
conventions of academic painting when challenged by the avant-
garde of the ecole de Paris. The historical irony is that this was
happening at the very time when Saussure was mounting his
structuralist defence of the old phonoptic view of writing. Struc-
turalism here emerges as the reactionary wing of modernism,
committed to a rearguard action that was already doomed to
failure.
Once it becomes obvious - by practical demonstration - that
written communication does not depend either on the existence
of an oral language which it transcribes, or on the existence of
orthographic conventions which govern it, only the persistence
of earlier and more rigid ways of thinking can prevent recognition

of the conclusion that writing can create its own forms of expression. It
took the genius of Mallarme, Apollinaire and Joyce, as it had
taken the genius of Picasso and Braque, to turn theoretical possi-
bilities into semiological realities.
These bold innovations must also be seen as symptoms of a
deeper malaise that makes itself felt during the nineteenth century
in a society increasingly unhappy with - but nevertheless edu-
cationally committed to — the assumptions of utilitarian literacy.
One of the striking differences between eastern and western atti-
tudes to writing and writers is the far higher awareness in the east
of the artistic dimension of the visual text. The same utilitarian
logic which sees writing as no more than a convenient way of
recording speech relegates the written text itself to a fairly low
position in the hierarchy of artistic endeavour. It is not without
significance that the art of calligraphy never attained in Europe
the importance accorded to it in oriental and Islamic cultures.
Western aesthetics clung on for centuries to the notion that what is
written is merely a substitute for the author's voice. This idea
survives long after the introduction of printing. We find a remark-
able example in Hegel, who says of poetry:
Now if we ask where we are to look, so to say, for the material
basis of this mode of expression, the answer is that, since speak-
ing does not exist, like a work of visual art, on its own account
apart from the artist, it is the living man himself, the individual
speaker, who alone is the support for the perceptible presence
and actuality of a poetic production. Poetic works must be
spoken, sung, declaimed, presented by living persons them-
selves, just as musical works have to be performed. We are of
course accustomed to read epic and lyric poetry, and it is only
dramatic poetry that we are accustomed to hear spoken and to
see accompanied by gestures; but poetry is by nature essentially
musical, and if it is to emerge as fully art it must not lack this
resonance, all the more because this is the one aspect in virtue
of which it really comes into connection with external exist-
ence. For printed or written letters, it is true, are also existent
externally but they are only arbitrary signs for sounds and
words. Earlier we did regard words as likewise means for indi-
cating ideas, but poetry imposes a form, at least on the timing
and sound of these signs; in this way it gives them the higher
status of a material penetrated by the spiritual life of what they
signify. Print, on the other hand, transforms this animation into
a mere visibility which, taken by itself, is a matter of indiffer-
ence and has no longer any connection with the spiritual
meaning; moreover, instead of actually giving us the sound and
timing of the word, it leaves to our usual practice the trans-
formation of what is seen into sound and temporal duration.12
Here we have repeated emphasis on the oral nature of poetry

combined with manifest disparagement of the printed word -
evidently not a coincidence. What lies behind it? No one would
wish to argue against Hegel that European poetic forms do not
spring from a tradition of song and recital, based on rhythms,
rhymes, assonances and other auditory — even musical — patterns.
But that hardly explains Hegel's appeal to a mysticism of orality
which credits the poet with the ability to confer on mere sounds
'the higher status of a material penetrated by the spiritual life of
what they signify'. This is not just another manifestation of Aristo-
telian dualism, nor a resuscitation of Platonic scepticism about
writing, but the backward-looking polemic one might expect to
find in an age when the poets have in fact lost touch with their
bardic heritage and are increasingly concerned with words on the
page. The writer-poet is for Hegel, it would seem, a fallen crea-
ture, if not a traitor to the cause of poetry.
It is a far cry from this to the oriental perspective, in which
poetry has always been seen as closer to painting than to music.
Chinese characters are not the inert marks that Hegel evidently
regards alphabetic letters are being. Nor is it an accident that
Chinese civilization never developed its own system of phonetic
12
G.W.E Hegel, Aesthetics, trans. T.M. Knox, Oxford, Clarendon, 1975,
p.1036.
writing (in the sense in which Western historians understand that

term). The semiology of writing in Chinese has a pictorial dimen-
sion that is altogether lacking in the west.
Hegel's refusal to recognize a visual aesthetics of poetry has
something obscurantist about it; for in the European tradition the
poem on the page always adopted a visual form that echoes its
oral structure. (That is why Mallarme feared that his readers
would find Un coup de des difficult.) The line of verse, as it appears
in western manuscripts, has never been anything other than a
graphic unit mimicking an auditory unit. It is the work of writers
who can hear what they write and see what they hear: a monu-
ment to glottic writing.
But by Hegel's time all literature except for drama was increas-
ingly being confined to expression in written form. 13 This sits
uncomfortably within a literary tradition which has deep roots in
orality. The result is a crisis in which all relations between sound
and written form are called in question, as is the authorial 'voice'
itself. The nineteenth century sees the beginnings of 'free verse', a
term which to some extent sums up the problems at issue. For the
'freedom' was a question of breaking the bondage which kept the
line of poetry tied to the traditional metrical conventions, rather
than being at liberty to follow the (alleged) cadences of speech.
These cadences, in fact, were from the outset a graphic mirage,
which never corresponded to those of everyday talk. The rhythms
of free verse were always those of an imaginary language, con-
structed by poets. We are dealing with a kind of literary illusion, in
which the resources of writing are deployed to install and authen-
ticate an orality that never existed (except for the poets concerned
and their poetry-readings). It is an interesting example of a cul-
tural narcissism that has - and could have - no analogue in a
pre-literate society: a phenomenon entirely generated by writing.
A related development, associated with Romanticism, is that
those who adopt the literary role of 'writer' suddenly begin to
13
Hegel even complains that 'no play should really be printed'. (Hegel,
op. cit., p. 1184.)
realize that they are 'free' to write in a so-called 'oral style': hence
the linguistic aesthetics of Wordsworth and his school. This new
interest in orality is in part prompted by the fear that, in a society
still largely illiterate, the writer - unlike the bard of old - risks
losing touch with the majority of the population. The category
'oral style' still survives today in odd corners of contemporary
stylistics. A curious example is Francois Richardeau's analysis of
the style of Marguerite Duras.14 Richardeau's so-called 'rhythmo-
typographic' analyses of texts adapt a technique borrowed from
Marcel Jousse in his work on Arabic and Hebrew material.15
Jousse, an anthropologist of note, regarded writing as a 'bastard'
and 'parasitic' variant of natural language (i.e. speech), causing
cerebral fatigue and exhaustion of thought. A precursor of
Marshall McLuhan, Jousse was already in the 1920s accusing the
ancient Greeks of 'ocular hypertrophy' and predicting the day
when the page would be replaced by the disc. All these are mani-
festations of the same literacy crisis. Its history can be traced from
the beginning of the nineteenth century down to the spread of
radio and television, which simultaneously mask and accentuate it.
In retrospect, Mallarme, Apollinaire and Joyce appear as visionar-

ies who had glimpsed the future of writing, but lacked the
appropriate writing instrument to put what they knew into prac-
tice. The instrument in question arrived in the second half of the
twentieth century: the computer.
There is an old chestnut about how long it would take a mon-
key with a typewriter to produce, serendipitously, the complete
text of a Shakespeare sonnet. Perhaps one reason why it has
dropped out of circulation is that computers have now taken over
from monkeys. We no longer have to imagine the bizarre occur-
rence of texts produced independently of the operation of human
intentions.
14
F. Richardeau, Ce que revelent kursphrases, Paris, Retz, 1988, p.40.
13
M. Jousse, L'Anthropologic dugeste, Paris, Gallimard, 1974.
The point of the monkey example was to raise the question of

literary value. Would the sonnet have been any the worse for
having been produced fortuitously by a chimpanzee? Is it not the
product that matters, rather than the process of production? Is not
value conferred by the reader, rather than being intrinsic to the
activities of the author?
According to current gipsies' warnings, the computer is a tech-
nological innovation that threatens to undermine the educational
virtues that literacy formerly stood for. This undermining is not
peripheral, but apparently threatens the expertise of professional
writers themselves. We hear that, because of the computer, news
reporters are no longer learning how to rewrite their copy, or,
more alarmist still, that 'the word processor is erasing literature'.16
We are invited to contemplate a future in which our descendants
will be counted 'illiterate' if they cannot operate a word processor;
but, at the same time, becoming slaves to the word processor is just
what will rob them of the benefits of the ('old-fashioned') literate
classes. So on either count, the future of literacy is doomed.
Before we succumb to pessimism of this kind, we should stop to
consider more carefully what the role of technology in the spread
of literacy has been hitherto. Is it clear that the 'literate' values
that are now held to be threatened depend on writing at all?
Where exactly does literacy begin and end?
In his book entitled On Literacy Robert Pattison argues elo-
quently against what he regards as the misleading equation of
literacy with the practical skills of reading and writing, and
against what he sees as a consequential series of educational falla-
cies based on that equation. For Pattison, it is obvious that Homer
was a paragon of literacy, and yet Homer could neither read nor
write. Therefore those who define literacy in terms of the ability
to read and write, and thus by implication relegate Homer to the
ranks of the world's illiterate masses, merely demonstrate their
own cultural myopia and a complete failure to understand what it
is that makes literacy worth having.
10
Gore Vidal.
It might perhaps be objected to Pattison that Homer would

never have become a paragon of literacy had his poems not been
reproduced in written form by a later generation of Greeks; and
consequently Pattison's prime historical example proves the very
opposite of what he takes it to demonstrate. Nevertheless Patti-
son's argument, even if we do not agree with it, is sufficiently
provocative to compel attention. In effect, what he is saying is that
it makes no more sense to dismiss Homer as illiterate than to judge
musicianship by whether a player can play by sight from a music-
al score. If that were the test, then there could be no blind musi-
cians. The mistake in both cases is to erect what is only a second-
ary, ancillary technique into the primary criterion of a complex
ability which that technique happens to serve. Just as it makes no
sense to suppose that playing from a score is the only kind of
musical performance, likewise it makes no sense to suppose that
reading and writing are the only kinds of activity that manifest
literacy.
It is interesting to consider how Pattison's argument fares when
applied to the specific form of literacy which is sometimes called
'numeracy'. The modern concept of numeracy treats it as essen-
tially involving the manipulation of quantification symbols
according to principles and patterns which have to be studied and
learnt. When we come across examples of early civilizations cap-
able of undertaking the building of large temples and palaces and
engaging in extensive enterprises of irrigation or town planning
we are reluctant to accept that they could have done all this on the
basis of a purely oral culture, even if they have left behind no
record of their writing system. We prefer to think that they must
have developed a written mathematics and a geometry now lost,
because we cannot otherwise imagine how they could have man-
aged the complex calculations apparently required by their feats
of planning and construction. But let us hypothesize for a moment
that somehow they did manage precisely that. Would we then
dismiss them as having failed to achieve numeracy, on the same
ground that some people would describe Homer as illiterate? Or
would we not rather be forced to conclude that they must have
been so highly numerate that they did not need to commit their
calculations to papyrus or to clay tablets?
In The Domestication of the Savage Mind, Jack Goody recalls that he
found he could not count cowrie shells as quickly or as accurately
as boys of the LoDagaa in Northern Ghana who had had no
schooling at all. And this was because, in the manner of Western
literates, he counted the shells one by one. The native boys, on the
other hand, counted them according to a traditional method in
successive groups of three and two, which was both faster and
easier to check. Furthermore, they had different methods of
counting different objects. Counting cowrie shells had its own
special technique. When it came to multiplication, however,
Goody found that he could easily manage calculations that were
beyond the expertise of the local cowrie counters. And this
superiority Goody attributes to the fact that multiplication, as dis-
tinct from addition, is essentially a literate operation. The native
boys, he says, 'had no ready-made table in their minds' which they
could use for purposes of calculation.
What might be questioned about Goody's explanation, how-
ever, is precisely what the connexion between multiplication and
literacy is. In some cultures an illiterate can use an abacus to
calculate at a speed which will match any literate mathematician's
pencil and paper. Perhaps Goody would reply that using an aba-
cus involves an operational technology which is in all respects
equivalent to that of manipulating figures on paper. But that sim-
ply brings us back to Pattison's argument about what literacy is.
For however we may describe using an abacus, we certainly do not
call it 'writing'.
A conclusion similar to Pattison's might perhaps be reached by
a different argumentative strategy. Suppose the first explorers
from Earth to arrive on Mars reported the existence of a curious
reversal of the familiar terrestrial relationship between speech and
writing. In other words, let us suppose it was discovered that Mar-
tians communicated primarily for everyday purposes by means of
making visible marks on surfaces, and were biologically equipped
to do this because their fingers constantly exuded a coloured
liquid which they used in much the same manner as we use ink.
With this coloured liquid Martians from their earliest years were
used to tracing graphic symbols on any convenient surface that
came to hand. But only few Martians ever learnt to make sounds
corresponding to these graphic symbols, because this involved
learning to use a special piece of equipment invented for the
purpose, which looked like a small box and was worn strapped
round the throat. This box was known in Martian as the 'vocal
apparatus', but very few Martians could afford to buy one, and in
any case this apparatus could be used effectively only after years
of special training in the correspondences between Martian
graphic symbols and the sounds that the box could produce. Per-
haps the first question that might occur to an anthropologist on
the mission from Earth is: 'Why do these Martians bother with
this clumsy vocal apparatus at all?' Taking this science-fiction
story one stage further, let us suppose that the Martian answer to
this question turned out to be that whereas any fool on Mars could
write, using the vocal box required a special form of intelligence,
and furthermore conferred certain advantages on speakers over
writers. For instance, speakers could communicate to one another
by means of sound even in the absence of a writing surface.
Furthermore, speakers could communicate even when doing
something else with their hands and eyes, whereas writers could
not. Third, speakers could communicate with one another audibly
in ways that those who could only write were quite unable to
understand. Fourth, vocalization had the inestimable practicality
of leaving no trace, so that it was impossible for anyone sub-
sequently to prove what had been said. It thus required mental
alertness to engage in vocal communication, and quick reactions
of an order quite beyond the average slow-witted writer. In short,
speaking was a privileged form of communication shared by an
elite, but beyond the grasp of the masses. Any sensible Martian,
therefore, could see that it was well worth buying a vocal appa-
ratus and learning to use it, because being able to use a vocal
apparatus brought with it all kinds of communicational
superiority.
This allegory, it will be self-evident, is constructed so as to

reverse, on every crucial point, the usual terrestrial assumptions
about the relationship between speech and writing. The lessons to
be drawn from it do not need to be spelled out in laborious detail.
Whether writing is judged to be a better form of communication
than speech depends on one's point of view, and that point of
view will be shaped by certain biological and cultural presupposi-
tions. But in one sense it does not matter at what point and in what
precise form technology enters the picture. As the allegory illus-
trates, it is possible to imagine a culture in which speech depends
on the availability of certain tools, just as in our more familiar case
writing depends on the availability of certain tools. What matters
in both cases is the use made of those tools. And the utility of the
tools is always measured against what could be done without
them. This, at least as one reader understands it, is precisely
Pattison's point about Homer.
In brief, the identification of intellectual progress with mastery
of the technology of writing is an error typical of utilitarian lit-
eracy; in fact, a double error. The first mistake is to have confused
a merely contingent use of artifacts with the cognitive con-
sequences of their use. The second mistake is a failure to see that
the technology would be pointless unless subserving other goals,
which are not set by the technology itself. Both of these are essen-
tial points to grasp if we hope to discuss the future of human
communication without falling into the pitfalls of nonsense.
Unfortunately, this is an area of speculation where prestigious
nonsense abounds. Much of it derives from two sources. One of
these sources is the endemic scriptism of the Western tradition;
that is, the tendency to analyse spoken language as if it were
written language avant la lettre. How deeply scriptism is entrenched
in Western thinking about language emerges in various para-
doxes. Perhaps the most striking is that linguistic theorists who
subscribe to the doctrine of the primacy of speech as the charac-
teristic form of human communication (Saussure is the classic
example) nevertheless feel constrained to analyse speech in such a
way that its units correspond in a quasi-miraculous fashion to the
units of writing, even while proclaiming that writing practices are

quite extrinsic to language. This may even carry over to meta-
linguistic terminology. A quite remarkable case is the paper pub-
lished by Leonard Bloomfield in 1927, in which he applied the
distinction between 'literate' and 'illiterate' speech to a pre-literate
speech community, the Menomini of Wisconsin.17 The so-called
'International Phonetic Alphabet' is another example of scriptism
disguised as linguistic analysis, based as it is on one culturally
localized system of writing. Scriptism in all its forms encourages
the view that literacy involves some kind of transference of lin-
guistic skills from a natural medium (namely, that of speech) into
an artificial medium (namely, writing).
Scriptism is defined by Florian Coulmas as:
the tendency of linguists to base their analyses on writing-
induced concepts such as phoneme, word, literal meaning and
'sentence', while at the same time subscribing to the principle
of the primacy of speech for linguistic inquiry.18
Linguists do indeed provide egregious examples. They have

even proposed that human beings are equipped by their vocal
apparatus with what Max Miiller called a 'physiological alphabet'
and AJ. Ellis 'the alphabet of nature'. One could hardly imagine
more blatant cases of projecting the concepts of (one system of)
writing on to the analysis of speech.19 The underlying rationale is
clearly: alphabets distinguish one sound from another - ergo such
distinctions must exist already in Nature. ('How else could humans
have recognized them?') But it is not simply linguists who are
17
L. Bloomfield, 'Literate and illiterate speech', American Speech, 1927,
Vol.2 No. 10,pp.432-39.
18
F. Coulmas, The Blackwell Encyclopedia of Writing Systems, Oxford, Black-
well, 1996, p. 455.
19
Max Miiller's chapter on 'The physiological alphabet' is to be found in
his Second Series of Lectures on the Science of Language (London, Longman,
Green, Longman, Roberts and Green, 1864). AJ. Ellis's 'The alphabet
of nature' was published in the Phonotypic Journal (1844—5).
guilty of such non sequiturs: in this respect they simply follow -

although indeed they have contributed to - the phonoptic bias
that has prevailed in Western culture for many centuries.
The other principal source of nonsense about literacy can be
traced to a body of pronouncements from what may be termed
the anti-scriptist school. The members of this school include Walter
J. Ong, Eric Havelock and Marshall McLuhan. Their major thesis
is that writing, far from being merely speech made visible in the
guise of inscriptions, constitutes a radically different cognitive
enterprise. In their account, writing is not just a convenient way
of recording speech, but involves a restructuring of thought.
Anti-scriptism in its various forms encourages the view that lit-
eracy is a profoundly different mental condition from that of pre-
literate humanity. Adherents to the anti-scriptist school often
emphasize this radical change by speaking of the 'literacy
revolution'.
As the terms scriptist and anti-scriptist suggest, there is certainly
a deep divergence of views here. While accepting this, one
may nevertheless wish to argue that the conflict between these
views is often presented as a conflict over the wrong issue.
The important difference between before and after the advent
of utilitarian literacy is not essentially a difference between
typical ways of thinking about the world, of classifying and
ordering, of overcoming memory limitations, or of strategies
for acquiring knowledge, although all these differences doubtless
correlate with the spread of writing. But they are all mani-
festations of something more fundamental; and this some-
thing more fundamental is a shift in conceptions of language
itself.
The change comes about for two reasons. One is that the intro-
duction of writing destroys, once and for all, the former equation
between language and speech. McLuhan was overstating the case
when he claimed that 'until writing was invented, man lived in
acoustic space'. What he should have said was that until writing
was invented, language lived in acoustic space. In a preliterate
culture the world of language is the world of sound. Writing
changes all that. With writing, language invades the world of vis-
ual communication. It enters into competition - and partnership
- with pictorial images of all kinds. The integration of writing
with speech is what ushers in the misguided concept of language
as something that is medium-transferable: words, it is supposed,
can be spoken, 'transferred' into a different form where they are
visible but no longer audible, and then 'transferred' back again
into speech. This is what sometimes appears to preliterate com-
munities, on their initial acquaintance with writing, as a form of
'magic'; but it does so only because their preliterate conception of
language cannot immediately cope with the forms of integration
involved.
The second reason is no less important. By making it possible to
divorce the message both from its sender and from the original
circumstances of its formulation, writing cognitively relocates lan-
guage in an 'autoglottic' space. That is to say, the text takes on a
life of its own, which is ultimately independent of the life or
intentions — of its author. It becomes an 'unsponsored' linguistic
object, to which there is no parallel in a preliterate culture. And
with this etiolation of personal sponsorship comes a fundamental
change in the notion of meaning. Instead of tracing back mean-
ing to the speaker or writer, as the authenticating source of the
message, people come to regard meaning as residing in the words
themselves. Plato's worries about writing are based on his recogni-
tion of this fallacy. But exactly what he feared came about. A
culture in which writing has become 'internalized' has already
prised open a conceptual gap between the sentence, on the one
hand, and its utterance or inscription on the other. The sentence,
being what lies behind and 'guarantees' both utterance and inscrip-
tion, is itself neither. And this requires a conception of language
which is necessarily more abstract than any that is required in a
preliterate culture. Once this view of language is adopted, it is
hard not to slide into adopting a parallel view of literature.
Derrida's championship of the autonomy of the text is not just
a philosophical aberration but the logical terminus of a
questionable view of literacy that has become progressively
established in Western culture over the centuries since Plato first

objected to it.
The currently popular scenario for the future of writing gives

pride of place to the development of techniques of word process-
ing. It has already been claimed that our traditional concept of
reading has been outdated by the advent of 'dynamic text' and
'hypertext'. The basis of this claim is a transformation in the role
of the reader from passive recipient to active participant in the
process of information transmission. Instead of merely receiving a
message pre-determined by the sender in respect of both form
and content, the reader is now able to control and access whatever
information and information-sequences are deemed relevant for
particular communicational purposes. Text is presented not in a
traditional monolinear format, but as a simultaneous configura-
tion of choices, from which the reader must make a selection.
Depending on that selection, further selections become available,
as the reader explores possible ways through the maze of informa-
tion available. In this process of exploration, individual readers
construct their own text instead of accepting a text dictated to
them.
But this is not all. If it were, one might object that what it
amounts to is simply a formalization and mechanization of read-
ing strategies that have been available to the traditional reader for
centuries. Western culture developed a special form of book and a
special form of verbal deixis based precisely on such reading
strategies. (The book is known as an encyclopedia, and the deictic
device is known as cross-referencing.} So a stronger claim must be
made if we are to be convinced that there is anything new here
other than the technological format.
This stronger claim is based on the fact that programs may be
set up in such a way that it may not be possible for the reader to
retrieve a previously selected portion of text in exactly the same
form as the first time: thus it may not be possible for the reader to
turn back the page and find everything that was written on it
before. Likewise, it may be impossible for the writer to foresee

exactly the sequences and contextualizations that may arise as the
result of selections made by different readers. Jay David Bolter
sums it up as follows:
Unlike printing, which lends fixity and monumentality to the
text, electronic writing is a radically unstable and imperma-
nent form, in which the text exists only from moment to
moment. . . 90
Under these conditions, we may ask, what exactly is the status of

the text? Is it any longer a text at all?
Whatever it may be, it is clearly no longer a static object; and in
that sense it may be claimed that we are dealing with a genuinely
new form which actually requires us to rethink writing, whether
we like it or not. For the primary characteristic of the writing
process, as traditionally understood, was precisely that it produced
a fixed form of words, available for inspection, re-inspection,
interpretation and discussion as required. It was this fixity and
quasi-permanence which was seen as contrasting with the ephem-
erality of speech. These non-ephemeral qualities were what made
writing suitable for the recording of information and its transmis-
sion over space and time. It removed verbal communication from
intrinsic dependence on the particular circumstances of a face-
to-face situation or the vagaries of memory and uncheckable
repetition. Hence, from Biblical times onwards, not only the Ten
Commandments but edicts and laws of all kinds were set down
in writing. Writing became the guarantee of authenticity because
and insofar as it guaranteed in turn the invariance of the text.
While it was true that a written text was only as durable as the
material surface on which it had been inscribed, and on some
surfaces writing is easily erased or altered, nevertheless it was also
true that until altered by material decay or human interference
the text remained static.
20
J.D. Bolter, 'Beyond word processing: the computer as a new writing
space', Language & Communication, 1989, Vol.8 Nos.2/3, p. 129.
Again, however, if this were all, it might appear that what elec-
tronic writing has done is simply reintroduce via technology a
rapprochement between writing and speech; or rather has endowed
writing with the ephemerality which was formerly treated as
characteristic of speech. And a sceptic might well ask what the
point is of employing the latest technological innovations in
order to revert to a more primitive type of communication.
A text which is not a static object not only defeats the storage
function generally regarded as one of the primary utilitarian pur-
poses of writing, but introduces something radically novel into our
whole model of verbal communication, whether spoken or writ-
ten; namely a discontinuity between the initial act of verbalization
and its end product. And this discontinuity is of a different order
from the material transformations which a message may undergo
in its journey from, say, oral dictation to printed page. Simul-
taneously, it casts the text of Shakespeare's sonnets (as established
by pre-computational editors) as merely an arbitrary reification
within a range of possibilities. Compare the tedious dispute about
the A-sharp in the first movement of Beethoven's Hammerklavier
Sonata, opus 106. Did Beethoven 'intend' this? And does it
matter?
Word processors, as everyone agrees, and as the term itself indi-

cates, are essentially machines for manipulating verbal signs. Yet
most of our current theories of verbal signs were formulated long
before the advent of the computer. A serious question to be
addressed is: are those theories now out of date? Has the semi-
ology of the pre-computer era now been superseded?
A preview of how writing by machine may alter our concept of
literacy is already offered by the pocket calculator. What is
expected of students in a three-hour mathematics examination
has changed dramatically since the days when all calculations had
to be done by human brainpower. One can foresee analogous
changes in the assessment of language skills when word proces-
sors automatically correct errors of spelling, grammar and
punctuation. What is interesting about the word processor is that

it provides a machine which enables the user to exploit systematic-
ally the potentialities afforded by the indeterminacy of the linguistic
sign and the open-endedness of sign systems. Traditional writing
does not do this because traditional writing is constrained by the
writer's personal network of word associations, plus the orthologi-
cal legacy of generations of educational convention. The word
processor, on the other hand, is a piece of equipment for linguistic
engineering. It can systematically invent new words, new para-
digms, new constructions, new meanings and new languages if we
wish. We can easily imagine a future in which 'Write a poem' is a
standard examination question in the 'English' examination. (The
vocabulary will be specified in the examination paper, and the test
will be one of imagination in the deployment of the student's
word processor's resources.) But this cannot go without repercus-
sions on such concepts as 'style', 'text' and 'literature'.
What has passed almost unnoticed is that a tool with the power
and ubiquity of the computer has the potential of reversing the
twentieth century's received wisdom on the basic relationship
between language, speech and writing. There are various reasons
for this.
One is quite simply the sheer increase in the amount of written
material generated. It is now confidently predicted that, with the
internet explosion, written communication will quantitatively out-
strip oral communication in the foreseeable future. If it does, that
will certainly be a landmark in human history: speech will for the
first time be the 'minor' form of communication.
But there are more important reasons which have to do with
our grasp of the basic processes of verbal communication. As a
writing machine, the word processor is already redefining our
concept of what a 'word' is. The word is no longer a static lexical
unit belonging to an inventory pre-registered in a dictionary.
Implicitly, for a word processor, the word can be any symbol or
symbolic unit which plays a role in the processing and can be
controlled by well-defined keyboard operations. It is important to

note that that role is not confined by the syntagmatic and para-
digmatic relations which govern what we now recognize as the
conventional words of ordinary language. Nor is it restricted by
the conventional boundary lines which treat iconic symbols as
non-words. Furthermore, such units and combinations can be
invented by the writer as needed. The constraints on their inven-
tion are not conventions in the outside world, but constraints
internal to the machine. The basic operational units are no longer,
as in traditional scripts, either the word or the letter, but the sepa-
rate keys provided by the keyboard; and the operational syntax is
the combinatorial logic of pressing them in sequence or simul-
taneously. Thus far that logic is heavily indebted historically to the
Roman alphabet (as one might expect in a transitional stage
between forms of literacy); but there is nothing at all that requires
it to remain so in the future. The alphabetic letter, like the
Egyptian hieroglyph, is not indispensable.
Today, technology puts us in the position of projecting 'writing'
as something altogether different from the ancillary system for
recording, which was its traditional basic role. It opens up the
possibility of treating writing as the essential creative process and
speech as a marginal commentary on what has been written. That
radical reversal of roles, we may reasonably speculate, will hold
the key to the psychology of education in the next century. At the
same time, it holds the key to what literature will be 'seen as'. Not
a spontaneous linguistic expression of the writer's thought or
emotion (which, arguably, it never was anyway, pace certain emi-
nent theorists) but an exploration of verbal possibilities made uni-
versally available by the electronic writing machine. Electronic
music is the relevant analogy here. In the computerized world of
music, composers are less and less required to be able to play any
instrument at all (including the human larynx).
It is within this perspective that it becomes relevant to return to

Pattison's argument about Homer. Pattison may be right in
insisting that traditional literacy had to be defined in terms of pre-

existing goals or models, which were not set by the technology of
writing itself but were already in place before the technology
became available. Even if this was true in the past, however, will it
be true in the future? Arguably not: and this is where the word
processor not only makes a crucial difference, but turns Pattison's
proposition upside down. The limits of literacy will be set by the
technology, not in the sense of restricting the title 'literate' to those
who read and write electronically, but by the exploitation of new
linguistic possibilities which would not have been available with-
out the technology to hand. And there is no way those limits can
be set in advance; for we have as yet only an inkling of what
computers of the future may make possible or even render
commonplace.
The computer is the most powerful contextualization device
ever known. Its capacity for creating and developing new contexts,
visual and verbal, far outstrips that of the human mind. That is a
far more important fact about the computer than its superhuman
capacity for information storage. We are dealing with a machine
which offers not only the possibility of integrating a simultaneous
presentation of written, auditory and pictorial information, but of
linking that information across languages and cultures as well.
When future generations are quite accustomed to sitting at a key-
board and 'typing' an audio-visual product that incorporates
sounds, letter-forms and pictures systematically interrelated, they
will have acquired a new concept of writing, a new concept of
literature, a new concept of language.
Bibliography
Abercrombie, D., Elements of General Phonetics, Edinburgh, Edinburgh

Alleton, V, L'Ecriture chinoise, Paris, Presses Universitaires de France, 4th
ed., 1990.
Andre-Salvini, B., L'Ecriture cuneiforme, Paris, Editions de la Reunion des
musees nationaux, 1991.
Anis, J. (with Chiss, J.L. and Puech, C.), L'Ecriture: theories et descriptions,
Bruxelles, de Boeck, 1988.
Anis, J. and Puech, C., 'Autonomie de Pecriture'. In Lapacherie, J-G.
(ed.), Proprietes de I'ecriture, Pau, Publications de 1'Universite de Pau,
1998, pp.79-87.
Apollinaire, G., Alcools et Calligrammes, ed. C. Debon, Paris, Imprimerie
Nationale, 1991.
Aristotle, De Interpretatione, ed. and trans. H.P. Cook, London,
Heinemann (Loeb Classical Library), 1938.
Arnauld, A. and Lancelot, C., Grammairegenerate et raisonnee, Paris, 1660.
Austin, J.L., How to do things with Words, Oxford, Clarendon, 1962.
Barr, J., 'Reading a script without vowels'. In Haas, W. (ed.), Writing With-
out Letters, Manchester, Manchester University Press, 1976, pp.71-100.
Barthes, R., Le degre zero de I'ecriture, Paris, Seuil, 1953.
Basso, K.H., 'The ethnography of writing'. In Bauman, R. and Sherzer,
J., Explorations in the Ethnography of Speaking, Cambridge, Cambridge
University Press, 1974, pp.425-32.
Bazell, C.E., 'The grapheme', Litera, vol.3, 1956, pp.43-46.
244 Bibliography
Bloomfield, L., 'Literate and illiterate speech', American Speech, 1927,

Vol.2 No. 10,pp.432-9.
Language, London, Allen & Unwin, 1935.
Bolter, J.D., 'Beyond word processing: the computer as a new writing
space', Language & Communication, 1989, Vol.8 Nos.2/3, pp. 129—42.
Bradley, H., On the Relations between Spoken and Written Language, with Special
Reference to English, Oxford, Clarendon, 1919.
Buchler, J. (ed.), Philosophical Writings ofPeirce, New York, Dover, 1955.
Butler, S., Essays on Life and Science, ed. R.A. Streatfield, London, Fifield,
1908.
Butor, M., Les Mots dans lapeinture, Geneva, Skira, 1969.
Buyssens, E., La Communication et I'articulation linguistique, Paris/Brussels,
Presses Universitaires de France, 1967.
Chadwick, G., The Meaning of Mallarme, Aberdeen, Scottish Cultural

Press, 1996.
Champollion, J-F., Lettre a M. Dacier, Fontfroide, Bibliotheque Artistique
et Litteraire, 1889.
Chang, H-L., 'Hallucinating the other: Derridean fantasies of Chinese
script', Centerfor Twentieth Century Studies, Working Paper No.4, 1988.
Chao, Y.R., Mandarin Primer, Cambridge, Mass., Harvard University
Press, 1948.
Christy, T.C., 'Humboldt on the semiotics of writing'. In Rauch, I. and
Carr, G.F. (eds), The Semiotic Bridge, Berlin/New York, Mouton de
Gruyter, 1989, pp.339-45.
Cohen, M., Lagrande invention de I'ecriture et son evolution, Paris, Klincksieck,
1958.
Cohen, R.G., Mallarme's Masterwork. New Findings, The Hague/Paris,
Mouton, 1966.
Coulmas, F, The Writing Systems of the World, Oxford, Blackwell, 1989.
The Blackwell Encyclopedia of Writing Systems, Oxford, Blackwell,
1996.
Danet, B. and Bogoch, B., 'From oral ceremony to written document:

the transitional language of Anglo-Saxon wills', Language & Communi-
cation, Vol.12 No.2, 1992, pp.95-122.
DeFrancis, J., Visible Speech. The Diverse Oneness of Writing Systems,
Honolulu, University of Hawaii Press, 1989.
Derrida,J., De la grammatologie, Paris, Minuit, 1967.
Bibliography 245
Desbordes, E, 'La pretendue confusion de 1'ecrit et de 1'oral dans les

theories de 1'antiquite'. In Catach, N. (ed.), Pour une theorie de la langue
ecnte, Paris, CNRS, 1989, pp.27-33.
Idees romaines sur I'ecriture, Lille, Presses Universitaires de Lille, 1990.
Diringer, D., The Alphabet, 2nd ed., London, Hutchinson, 1949.
Writing, New York, Praeger, 1962.
Ducrot, O. and Todorov, T, Dictionnaire encyclopedique des sciences du langage,
Paris, Seuil, 1972.
Eco, U., A Theory of Semiotics, Bloomington, Indiana University Press,

1976.
Eiseman, F.B. Jr., Bali: Sekala and Nisakala. Vol.1. Essays on Religion, Ritual
and Art, Berkeley / Singapore, Periplus, 1989.
Fevrier, J.G., Histoire de I'ecriture, 2nd ed., Paris, Payot, 1984.

Foucault, M., Les mots et les chases, Paris, Gallimard, 1966.
Fraenkel, B., La Signature. Genese d'un signe, Paris, Gallimard, 1992.
Gelb, I.J., A Study of Writing, 2nd ed., Chicago, University of Chicago

Press, 1963.
Goody, J., The Domestication of the Savage Mind, Cambridge, Cambridge
Gray, B., 'Language as knowledge: the concept of style', Forum Linguisti-
cum, Vol.3 No.l, 1978, pp.29-45.
Harris, R., The Origin of Writing, London, Duckworth, 1986.

Reading Saussure, London, Duckworth, 1987.
Signs of Writing, London, Routledge, 1995.
Signs, Language and Communication, London, Routledge, 1996.
The Language Connection, Bristol, Thoemmes, 1996.
'The integrationist critique of orthodox linguistics'. In Harris, R.
and Wolf, G. (eds), Integrational Linguistics: a First Reader, Oxford,
Pergamon, 1998, pp. 15-26.
'Making sense of communicative competence'. In Harris, R. and
Wolf, G. (eds), Integrational Linguistics: a First Reader, Oxford, Pergamon,
1998, pp.27-45.
'Three models of signification'. In Harris, R. and Wolf, G. (eds),
Integrational Linguistics: a First Reader, Oxford, Pergamon, 1998,
pp. 113-25.
246 Bibliography
Harris, R., Introduction to Integrational Linguistics, Oxford, Pergamon, 1998.

Havelock, E.A., The Literate Revolution in Greece and its Cultural Consequences,
Princeton N.J., Princeton University Press, 1982.
Hegel, G.W.E, Aesthetics, trans. T.M. Knox, Oxford, Clarendon, 1975.
Higounet, Ch., L'Ecriture, 7th ed., Paris, Presses Universitaires de France,
1986.
Hjelmsley, L., Prolegomena to a Theory of Language, trans. FJ. Whitfield, rev.
ed., Madison, University of Wisconsin Press., 1961.
Hockett, C.F., A Course in Modern Linguistics, New York, Macmillan, 1958.
Householder, F.W., Linguistic Speculations, Cambridge, Cambridge
Humboldt, W. von, 'Uber die Buchstabenschrift und ihren Zusam-
menhang mit dem Sprachbau', Abhandlungen der kb'niglichen Akademie
der Wissenschaften zu Berlin, 1826, pp. 161-8.
Jousse, M., L'Anthropologie du geste, Paris, Gallimard, 1974.

Joyce, J., Finnegans Wake, ed. Seamus Deane, London, Penguin, 1992.
Komatsu, E. and Harris, R. (eds), Ferdinand de Saussure, Troisieme Cours de

linguistique generale (1910-1911), Oxford, Pergamon, 1993.
Komatsu, E and Wolf, G. (eds), Ferdinand de Saussure, Deuxwme Cours de
linguistique generale (J 908-1909), Oxford, Pergamon, 1997.
Kristeva,J., Le Langage, cetinconnu, Paris, Seuil, 1981.
Lallot, J., La Grammaire de Denys le Thrace, Paris, CNRS, 1989.

Lane, H., The Wild Boy ofAveyron, London, Allen & Unwin, 1977.
Larsen, S.E., 'Semiotics', Concise Encyclopedia of Philosophy of Language, ed.
P.V Lamarque, Oxford, Pergamon, 1997, pp. 177-90.
Lee, S., A History of Korean Alphabet and Movable Types, Seoul, Ministry of
Culture and Information, Republic of Korea, 1970.
Leroi-Gourhan, A., Le Geste et la Parole, I. Technique et Langage, Paris, Albin
Michel, 1964.
Liddell, H.G. and Scott, R., Greek-English Lexicon, rev. ed., Oxford,
Clarendon, 1996.
Llorach, E.A., 'Communication orale et graphique'. In Martinet,
A. (ed.), Le langage, Paris, Bibliotheque de la Pleiade, Gallimard, 1968.
Locke, J., An Essay Concerning Human Understanding, 5th ed., London, 1706,
ed. A.C. Fraser, 1894, repr. New York, Dover, 1969.
Logan, R.K., The Alphabet Effect,1 New York, Morrow, 1986.
Bibliography 247
Love, N., 'The fixed-code theory'. In Harris, R. and Wolf, G. (eds),

pp.49-67.
Lussu, G., La lettera uccide, Viterbo, Nuovi Equilibri, 1999.
Lyons,J., 'Human language'. In Hinde, R.A, (ed.), Non-Verbal Communica-
tion, Cambridge, Cambridge University Press, 1972.
Mallarme, S., (Euvres completes, ed. H. Mondor and G. Jean-Aubry, Paris,

Gallimard (Bibliotheque de la Pleiade), 1945.
Malson, L., Les Enfants sauvages, Paris, Union Generale d'Editions, 1964.
McLuhan, M., The Gutenberg Galaxy, Toronto, University of Toronto
Press, 1962.
Moracchini, M., ABC de graphologie, Paris, Grancher, 1984.
Miiller, EM., Lectures on the Science of Language. Second Series, London,
Longman, Green, Longman, Roberts & Green, 1864.
Chips from a German Workshop, Vol. IV, London, Longmans, Green,
1875.
Murdoch, I., The Fire and the Sun, Oxford, Oxford University Press, 1977.
Naville, A., Nouvelle Classification des sciences. Etudephilosophique, Paris, 1901.

Nelson, K., The Art of Reciting Qur'an, Austin, University of Texas Press,
1985.
Ogden, C.K. and Richards, I.A., The Meaning of Meaning, London,

Routledge & Kegan Paul, 1923.
Olson, D.R., 'How writing represents speech', Language & Communication,
Vol.13 No.l, 1993, pp.1-17.
The World on Paper, Cambridge, Cambridge University Press, 1994.
Ong, WJ., Orality and Literacy, London, Methuen, 1982.
Pattison, R., On Literacy, Oxford, Oxford University Press, 1982.

Paul, H., Principien der Sprachgeschichte, 2nd ed., Halle, Niemeyer, 1886.
Pedersen, H., Linguistic Science in the Nineteenth Century. Methods and Results,
trans. J.W. Spargo, Cambridge, Mass., Harvard University Press,
1931.
Pellat, J-C., 'La conception de 1'ecriture a Port-Royal'. In Lapacherie,
J-G. (ed.), Proprietes de 1'ecriture, Pau, Publications de PUniversite de
Pau, 1998, pp. 153-60.
Perec, G., La Disparition, Paris, Denoel, 1969.
248 Bibliography
Pessoa, E, Sur les heteronymes, trans. R. Hourcade, Le Muy, Editions Unes,

1985.
Plato, Phaedrus and Letters VII and VIII, trans. W. Hamilton, London,
Penguin, 1973.
Potter, R.K., Kopp, G.A. and Green, H.C., Visible Speech, New York, Van
Nostrand, 1947.
Pound, E., ABC of Reading, New York, New Directions, 1960 [1934].
Priestley, J., A Course of Lectures on the Theory of Language and Universal
Grammar, Warrington, 1762.
Quintilian, Institutio Oratoria, ed. and trans. H.E. Butler, London,

Heinemann (Loeb Classical Library), 1920.
Read, C.A., Khang, Y, Nie, H., and Ding, B., 'The ability to manipulate
speech sounds depends on knowing alphabetic reading', Cognition,
Vol.24, 1986, pp.31-44.
Reber, A.R., Penguin Dictionary of Psychology, Harmondsworth, Penguin,
1985.
Reid, T.B.W., Historical Philology and Linguistic Science, Oxford, Clarendon,
1960.
Richardeau, E, Ce que revelent leursphrases, Paris, Retz, 1988.
Richards, I.A., How to Read a Page, London, Kegan Paul, Trench,
Trubner, 1943.
Rousseau, J-J., Essay on the Origin of Languages, trans. J.H. Moran,
Chicago, University of Chicago Press, 1966.
Sampson, G., Writing Systems, London, Hutchinson, 1985.

Sandys, J.E, Latin Epigraphy, 2nd rev ed., London, 1927.
Saussure, E de, Cours de linguistique generate, 2nd ed., Paris, Payot, 1922.
All page references are to this edition. Translated passages are from
E de Saussure, Course in General Linguistics, trans. R. Harris, London,
Duckworth, 1983.
Searle,J., Speech Acts, Cambridge, Cambridge University Press, 1969.
Shattuck, R., The Innocent Eye, New York, Washington Square Press,
1986.
Siertsema, B., A Study of Glossematics, 2nd ed., The Hague, Nijhoff,
1965.
Singer, E., Graphology for Everyman, London, Duckworth, 1949.
Smalley, W.A., Vang, C.K. and Yang, G.Y., Mother of Writing: the Origin and
Bibliography 249
Development of a Hmong Messianic Script, Chicago, University of Chicago

Press, 1990.
Smith, D.E., (ed.), A Source Book in Mathematics, New York, McGraw-Hill,
1929.
Sellers, P., Theorie des exceptions, Paris, Gallimard, 1986.
Taylor, TJ., Mutual Misunderstanding, Durham, Duke University Press,

1992.
'Do you understand? Criteria of understanding in verbal inter-
action'. In Harris, R., and Wolf, G., (eds), Integrational Linguistics: a First
Reader, Oxford, Pergamon, 1998, pp. 198-208.
Toolan, M., Total Speech. An Integrational Linguistic Approach to Language,
Durham, Duke University Press, 1996.
'A few words on telementation'. In Harris, R. and Wolf, G. (eds),
pp. 68-82.
Trench, R.C., English Past and Present, London, Parker, 1855. Repr.
Everyman's Library, London, Dent, 1927.
Tylor, E.B., Anthropology, London, Macmillan, 1881.
Vachek, J., 'Some remarks on writing and phonetic transcription', Acta

Linguistica, vol.5, 1945-9, pp.86-93. Reprinted in Hamp, E.P., House-
holder, F.W and Austerlitz, R. (eds), Readings in Linguistics II, Chicago,
University of Chicago Press, 1966, pp. 152-7.
Vendryes,J., Le Langage, Paris, Renaissance du Livre, 1923.
Wells, H.G., The Time Machine, London, 1895.

Whitaker, C.W.A., Aristotle's De Interpretatione, Oxford, Clarendon, 1996.
Whitney, W.D., The Life and Growth of Language, New York, Appleton,
1875.
Index
abbreviations, 149ff., 157 Bloomfield, L., 154, 188f., 194, 234

Abercrombie, D., 127f. Boethius, 26
acronyms, 151 BolterJ.D., 238
alphabet, vii et passim; alphabetical Bopp, E, 5If., 121,208f.
order 96f., I l l , 114, 117; alpha- Boswell,J., 1
betic principle 128, 132, 134ff. Braque, G., 225
Andre-Salvini, B., 87 Bradley, H., 156
AnisJ., 191, 199 braille, 79f.
Apollinaire, G., 218f., 225 Butler, S., 3f.
arbitrariness, 49ff., 145, 190, 223 Butor, M, 174
Aristotle, xiv 17-39, 52, 58, 64-7, Buyssens, E., 46
73f., 188,221,226
Austin, J.L., 164 calligraphy, 225
autograph, 171 Cicero, 10, 15 If.
autonomy of writing, 191-202 Chadwick, C., 215ff.
ChampollionJ-E, 140
Bally, Ch., 48 Chang, H-L., 142f.
Balzac, H. de, 9f. Chao,YR., 144f.
BarrJ., 127 ff. Chilperic, 102
Barthes, R., 9ff. codes 7Iff., 75, 81,224
Basso, K.H., 16 Cohen, M., 154
Bayles, P., 151 comparative philology, 51, 208
Bazell, C.E., 190f. consonantal writing, 126—9, 132f.
Beethoven, L. van, 239 Constantin, E., 47f., 139f, 147ff.
behaviourism, 188, 205 context, 69, 76, 81-6, 148, 180, 205,
Bell, A.M., xii 238, 242
Bible, 12,53, 129,238 cotemporality, 163
252 Index
Coulmas, E, 154ff., 234 Green, H.C., xii

counting order, 96
crypto-literacy, xf. Hauser, K., 172
cuneiform, 133 Havelock, E.A., 1,235
Hegel, G.W.E, 142, 147f, 225ff.
Deane, S., 222 heteronymy, 164, 197
DeFrancis,J., xii, 153 hieroglyphs, 140, 148, 241
Derrick, J., 40, 142f., 147f., 236 Higounet, Ch., 7, 134
Desbordes, E, 20, 208 Hitler, A., 179
Descartes, R., 142 Hjelmslev, L.T., 195ff.
Dionysius Thrax, 19f. Hobbes, T, viif, xi
Diringer, D., 5, 130ff., 155, 168, 170 Hockett, C.E, 187f.
Disraeli, B., llf. Homer, 229f, 233, 241
dual articulation, 100 homography, 92f, 138f, 197f.
dualism, 66f., 157, 204, 218, 226 Horace, 17
du Ponceau, P.S., 145 Householder, F.W., 207
Duras, M., 228 Humboldt, K.W. von, 206
dynamic text, 237 hypertext, 237
Eco, U, 45f. ideographic writing, 123, 133, 138-60

Ellis, A.J., 234 installation, 85-8
emblematic frame, 106-110 integrationism, 65-72,75-82,84-92,
etymological spelling, 103 103f., 106, 113, 121, 134, 137,
148fT., 152, 161-4, 166f, 170f,
FevrierJ.G, 132ff., 152, 185f. 180, 199ff., 204f, 211,213f, 236
figures, 94ff., 99, 151, 159 ItardJ-M-G, 202-6
fingerprinting, 161
Foucault, M., 66 Johnson, S., If., 14
Fraenkel,B., 182f, 212 Jousse, M., 228
free verse, 227 Joyce,J.A.A., 220-5
Fulgentius, 114
full literacy, xi, xiv, 14 Kopp, G.A., xii
Kristeva,J., 185f.
Gelb,IJ., 7, 126, 169f.
Gislebertus, 175f. Lane, H, 204ff.
glossematics, 112, 195-8 Larsen, S.E., 45
glottic writing, viii et passim Lasos, 114
Goody, J., 231 Leibniz, G.W. von, 142
grapheme, 190 Leroi-Gourhan, A., 86
graphic reduction, 149f. letters, xii et passim
graphology, 175, 177f., 180 linguistic priority, 43, 81, 222
Gray, B., 6 lipogram, 113f.
Index 253
literacy, ix et passim Perec, G., 113

literature, 8ff. perlocution, 170f.
Llorach, E., 19Iff. Pessoa, E, 164ff.
LockeJ., 39, 73f. phoneme principle, 94
logogram, 145, 153f., 158 phonics, 115
Logan, R.K., 14 phonogram, 158
Love, N., 73ff. Picasso, P., 217, 225
Lussu, G., xi pictography, 51, 98, 124, 144, 154f,
Lyons, J., 194 158ff.
picture writing, 154
McLuhan, M., 14, 235 Plato, xiv, 17ff, 21, 27, 39f., 52, 89,
Mallarme, S, 215-8, 220, 225, 227 186, 208, 226, 236f.
Marat, J.P., 179 Port-Royal, 27ff, 52, 62
medium-transferability, 193ff., 234, Potter, R.K., xii
236 Pound, E., 10
monograms, 168, 170 Priestley, J., xii, 2f.
Moracchini, M., 177f. primacy of speech, 43, 75, 193, 233
More, T, 223 primacy of writing, 207
Muller, EM., 78f., 234 printing, xi, 61, 82ff, 92, 177, 225ff.
property marks, 168ff.
Napoleon I, 178f. punctuation, 133
Naville, A., 4Iff.
Nestor of Laranda, 114 Quintilian, viii, 30ff, 37, 52, 54, 76,
nomenclaturism, 30, 33, 52ff, 65, 70 92, 116ff.
notation, 91-120, 134-7, 204; math- Quran, 129, 199
ematical, 94ff, 110f, 155; musical,
ix, 201,212, 218, 230 reading, vii et passim
notational ambiguity, 114f. Redon, O., 215
numeracy, 230f. Richardeau, E, 228
numeration, 94 Richards, I.A., 10,34
Riedlinger, A., 45
Ogden, C.K., 34 Romanticism, 227
Olson, D.R., 207-10 Rousseau,J-J., If, 14
Ong,WJ., 13,235
oral literature, 11 Sampson, G., 153f, 159
orthoepic rules, 103 Saussure, F. de, xivf, 15f, 19, 27-30,
32f, 35, 37f, 40-68, 71, 74,
Patois, Ch., 92 76-82,91-4,97,99-106,110-113,
Pattison, R, 229ff, 233, 24If. 116,121ff.,125f,130,135,137-52,
Paul, H., 37 156f, 159, 163f, 179f, 188, 190,
Pedersen, H., 134 192, 195, 208f, 215, 218, 223f,
Peirce,C.S., 15,45f, 64f, 82,97, 188 233
254 Index
script, 91, 96f., 104, 106, 11 Off., 114, Taylor, TJ., 74

128, 135ff, 155, 158 Tiro, 151
Sechehaye, A., 48 Trench, R.C., xii
Sejong, 124 Trephiodorus, 114
semiology, xiv et passim Tylor, E.B., 4f.
semiotics, xiv, 45, 64
Shakespeare, W, 86, 228, 239 Uldall,HJ., 195-8
Shattuck, R, 220 utilitarian literacy, xf., xiiif., 10, 14f.,
shorthand, 123, 150ff. 89, 100,214,233,235
signatures, 161-83,211
signs, xiv et passim; indeterminacy of, VachekJ., 198, 200f.
72, 240 values, 48ff., 54-8,61, 70,81,89-95,
Skinner, B. E, 205 111, 114, 127, 147,204,211,213
Sellers,?., 221,223 VendryesJ., 128
speech, vii et passim; speech acts, 46f, Victor of Aveyron, 202-6
164, 170f; speech circuit, 47; vis- Vidal, G., 229
ible speech, xii, 18, 27, 50, 75, 235 visual analogies, 83
spelling reform, xiii, 35f. Voltaire, F-M. A., xii
structuralism, xiv, 16, 39-63, 65, 70,
72,75f.,92,104,110,122,179,224 Wells, H.G., 7f, 11, 14
SwiftJ., 25 WilkinsJ., 142
symbols, xiv, 21,2 3ff.,34ff, 51,64,87, Wittgenstein, L., 53
130,132,145,152f, 188,232,240 Whitaker, C.W.A, 23, 66
Whitney, W.D., 21, 53f.
telementation, 73f. Wordsworth, W, 228

Roy Harris-Rethinking Writing-Athlone (2000)

Uploaded by

Copyright:

Available Formats

Roy Harris-Rethinking Writing-Athlone (2000)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Roy Harris-Rethinking Writing-Athlone (2000)

Uploaded by

Copyright:

Available Formats

RETHINKING WRITING

This page intentionally left blank

Reprinted in paperback 2001 by Continuum

© Roy Harris 2000

All rights reserved. No part of this publication may be reproduced

British Library Cataloguing-in-Publication Data

ISBN 0-485-11547-6 (hardback)

Typeset by RefmeCatch Limited, Bungay, Suffolk

Foreword: Writing and Civilization 1

A profitable Invention for continuing the memory of time

Hobbes's matter-of-fact description summarizes centuries of tradi-

glottic system. Furthermore, glottic writing is not the only kind of

difficulty involved in trying to reconsider the question of writing

awe, but as a practical tool or technique for doing what would

utility of the substitution is seen as being that writing can transmit

counterpart. But that has not as yet undermined the standing or

such by physiologists and psychologists. Reading aloud and read-

to question. If he did not, it seems important to understand his

Most of the chapters began life as lectures given at the Sorbonne

was published in Paris by the Centre National de la Recherche

Writing and Civilization

When Boswell objected to Johnson's calling the Chinese 'barbar-

characters to stand for words and sentences, while the third is

Rousseau's example of the writing of a barbaric people is Chinese.

Priestley himself came to the conclusion that 'the imperfection

This kind of a priori reasoning about writing systems is by no

Underlying the notion that links the invention of writing to the

instantly without material trace, and if it lives at all does so only

This in turn leads to the conclusion than when the bonds of

By this criterion, the Chinese are at last upgraded from barbar-

ism to civilization. (Tylor magnanimously attributes their failure to

This claim appeals to a different rationale from Tylor's. But it

The key factor in the formation of a specialist literate class in

Put more simply, the point is that it would be naive to construe

The views we encounter in histories of writing to this day still

than a generalization from the same scholar's view of the position

What is questionable here is not simply how 'advanced' the

One could go on adding to the list of dubious reasons that

This is one of the great missed opportunities in English litera-

Comment is called for at this point on one of the 'less-than-bare'

This is one of those truisms of such striking banality that they

Similarly when Ezra Pound, many years before Barthes, pro-

ity one senses an attempt at definition. The suggestion seems to be

There is a certain historical irony in the fact that Wells's prophecy

taught to read and write. In other words, at the very moment

concerning the superiority of alphabetic writing. According to

'independent' in that it will be free from the more or less ethno-

writing began to suffer at the hands of linguists.' It still does.21 Any

Here we glimpse the beginnings of what was to become a cross-

otherwise is to make the mistake of equating language with its

Part of the background to Plato's complaint concerns the per-

day.6 There we find a section devoted to the stoikheion which starts

relationship in semiological terms, as Aristotle, for one, certainly

which Aristotle's account is situated. For it is probably only the

As a communicational metaphor, Aristotle's use of the term is

representational. Unlike the relationship between the affection of the

country village or a foreign land or another planet. All that mat-

In the same passage in De Interpretatione Aristotle distinguishes

Aristotle's solution became incorporated into the body of received

This is unadulterated (albeit simplified) Aristotle, with the topos