Speculative Grammatology

Deconstruction and the New Materialism

Deborah Goldgaber

FINAL

GOLDGABER 9781474438339 PRINT.indd 1 01/10/2020 09:30

 Edinburgh University Press is one of the leading university presses in the UK.
We publish academic books and journals in our selected subject areas across
the humanities and social sciences, combining cutting-edge scholarship with
high editorial and production values to produce academic works of lasting
importance. For more information visit our website: edinburghuniversitypress.
com

FINAL
© Deborah Goldgaber, 2021

Edinburgh University Press Ltd

The Tun – Holyrood Road
12(2f) Jackson’s Entry
Edinburgh EH8 8PJ

Typeset in 11/13 Adobe Sabon by

Servis Filmsetting Ltd, Stockport, Cheshire,
and printed and bound in Great Britain.

A CIP record for this book is available from the British Library

ISBN 978 1 4744 3833 9 (hardback)

ISBN 978 1 4744 3835 3 (webready PDF)
ISBN 978 1 4744 3834 6 (paperback)
ISBN 978 1 4744 3836 0 (epub)

The right of Deborah Goldgaber to be identified as the author of this work has
been asserted in accordance with the Copyright, Designs and Patents Act 1988,
and the Copyright and Related Rights Regulations 2003 (SI No. 2498).

GOLDGABER 9781474438339 PRINT.indd 2 01/10/2020 09:30

 Contents

Series Editor’s Preface iv

Acknowledgements viii
Preface: The (Un-)Timeliness of Grammatology x

Introduction: To Speculate – with Derrida 1

1 Materialism and Realism in Contemporary Continental

FINAL
Philosophy 16
2 From Ancestral Events to Posthumous Texts: Two
Critiques of Correlationism 39
3 Texts without Meanings: Deconstructing the
Transcendental Signified 66
4 Rewriting the Course in General Linguistics: From Sign
to Spacing 94
5 On the Generality of Writing and the Plasticity of the
Trace 135

Bibliography 170
Index 179

GOLDGABER 9781474438339 PRINT.indd 3 01/10/2020 09:30

 2

From Ancestral Events

to Posthumous Texts:
Two Critiques of Correlationism

For my own part, I think that if one were looking for a single phrase
to capture the stage to which philosophy has progressed, ‘the study of
evidence’ would be a better choice than ‘the study of language’.
A. J. Ayer, Philosophy in the Twentieth Century

FINAL
Derrida and Meillassoux: Two Critiques of
Correlationism
Today the problem of correlationism and its pertinence to contem-
porary continental philosophy is well known. The term owes much
of its current prominence, of course, to Quentin Meillassoux’s
After Finitude (2008). However, as Husserlian Dan Zahavi help-
fully underlines, its philosophical use can be traced back to an
earlier, less pejorative context.
As Zahavi recalls, Maximillian Beck in 1928 characterised phe-
nomenology in terms of a solution to a series of philosophical
antinomies:

‘Correlationism’ can here serve as the term for a position devel-

oped by Husserl and Dilthey, according to which the old alternative
between idealism or realism, subjectivism or objectivism, philosophy
of immanence and phenomenalism or philosophy of the real, must
be overcome in favor of the following [negative] claim: Neither does
a world in itself exist independently of consciousness, nor does only
consciousness or a conscious subject exist and the world merely as a
mode (experience, function, content) of consciousness or the subject.
Neither do we know the world as it is in itself, i.e. independently of
our consciousness, nor do we merely know an illusory world, behind
which the real, true, world exists in itself. The positive counter-thesis
of correlationism is the following: Consciousness and world, subject

GOLDGABER 9781474438339 PRINT.indd 39 01/10/2020 09:30

 40 Speculative Grammatology

and object, I and world stand in a correlative, i.e. mutually depend-

ent context of being, such that the disjunctions mentioned above are
meaningless.1

If, according to Beck, ‘correlationism’ names a positive solution to

a philosophical problem – a solution to the ontological indetermi-
nateness of what is meant by ‘perceived object’ – for Meillassoux
it names the overarching anti-realist framework for a modern
philosophy that stymies its development.
Meillassoux – no doubt unwittingly – rewrites Beck’s ‘positive
[correlationist] counter- thesis’ as ‘there are no objects, no events,
no laws, no beings which are not always-already correlated with
a point of view, with a subjective access’.2 This correlationist
thesis is a problem because it consumes and exhausts philosophy
with the insoluble problem of access. ‘Experience’ and ‘conscious-
ness’ are indices for access problems and testify to a form of
philosophical forgetting. What is urgent to think is the world in its

FINAL
radical difference from and indifference to the human subject. For
Meillassoux, correlationism’s inability to conceive of the radical
absence of the human from most of the Earth’s past (and its prob-
able absence from its future) is sufficient grounds for rejecting the
correlationist thesis. This rejection implies the rejection of most
post-Kantian philosophy:

The central notion of modern philosophy after Kant seems to be

that of correlation. By ‘correlation’ we mean the idea according to
which we only ever have access to the correlation between thinking
and being, and never to either term considered apart from the other.
We will henceforth call correlationism any current of thought which
maintains the unsurpassable character of the correlation so defined.
Consequently, it becomes possible to say that every philosophy which
disavows naïve realism has become a variant of correlationism.3

Importantly, Meillassoux does not expect the sort of critique he

proposes to lead back to naive realism. That is, if the naive realist
takes the world disclosed through our perceptions to be the real
world, the world as it is independent of our perceptions, then
correlationism will properly point out that this effaces the role
and activity of the perceiver, which must be factored. The corre-
lationist critique will not fall short of the transcendental critique.
As a critique of the critique it seeks a path beyond correlationism

GOLDGABER 9781474438339 PRINT.indd 40 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 41

that will allow us to again conceive of thought as adequate to

the world. Such a critique, however, must first demonstrate its
necessity. If, on Beck’s telling, correlationism was necessary to
overcome the problems of naive realism and to give an account of
the relation between mind and world, what philosophical impasse
motivates overcoming correlationism?
The title of this chapter mentions two critiques of correlation-
ism. There are no doubt other candidates, but my aim in this
chapter is to argue that, while it has gone strangely unremarked,
Derridean deconstruction is a radical critique of correlationism
and proceeds by a similar argumentative strategy to the one pro-
posed by Meillassoux.
Both Meillassoux and Derrida motivate their critique of cor-
relationism by identifying what I will call an ‘extra-correlational
instance’. As Meillassoux argues, a living body is a necessary
condition for the instantiation of a transcendental subject, which
accounts for its finite point of view on the world. An extra-

FINAL
correlational instance, then, is one that ‘takes place’ under condi-
tions that logically preclude the taking place of the transcendental
subject. Any instance is extra-correlational which is in principle
unwitnessable or unrecognisable. Of course, the correlationist will
deny that there could be any such instances in principle, or will
at least claim that any such instances are transcendental illusions.
The argumentative strategy I attribute to Meillassoux and Derrida
lies in demonstrating the necessity of such instances.
Meillassoux argues that correlationist epistemology founders on
‘the ancestral’. The ancestral is an unwitnessable event; by defini-
tion, it takes place prior to the appearance of life. For example, the
accretion of the planet Earth five billion years ago would be such
an ancestral event. The correlationist, Meillassoux argues, cannot
literally believe in unwitnessable events and processes that occur
prior to the appearance of life. Belief in ancestral events requires
an epistemology that does not entail that objects or events outside
of a possible experience are unthinkable.
Derrida, on the other hand, argues that the structure of texts is
extra-correlational. Texts, he argues, remain ‘structurally readable’
in the absolute absence of any possible reader. Correlationism,
wrongly, identifies the conditions of ‘readability’ with the tran-
scendental subject. Only an extra-correlational account of these
conditions will reveal this structure, which Derrida refers to as
sur-vival (sur-vivance).

GOLDGABER 9781474438339 PRINT.indd 41 01/10/2020 09:30

 42 Speculative Grammatology

Meillassoux’s critique of correlationism has been especially

influential in contemporary continental philosophy. In particular,
it has energised debates about the limits of phenomenology for
thinking life, embodiment and the material world. As Catherine
Malabou writes of Meillassoux: ‘in search of noncorrelationist
modes of thinking, [he] elaborates a notion of the absolute which
would not be “ours,” which would remain indifferent to us’.4
She suggests that the idea of material processes radically disjunct
from the transcendental subject – self-informing, ungoverned and
ungovernable – is at the centre of contemporary new materialisms.
In particular, these modes of thinking reject correlationism’s atten-
tion to ‘givenness’, the centrality of the category of first-person
experience, and the problem of the relation of thought to world.
Derrida’s critique of correlationism, as already mentioned, is
less well known. Indeed, it is usually not read as a critique of cor-
relationism at all, but rather as endorsing an anti-realist variant
of correlationism. As a result, its impetus to materialist thought

FINAL
goes unrecognised. As I argued in the previous chapter, Derrida’s
account of textuality is often misread as a post-structuralist variant
of the correlationism that Meillassoux diagnoses. However, as a
reading of Derridean textuality this interpretation is hard to credit
as soon as we factor in that Derrida defines textuality in terms of
its extra-correlational status. Indeed, he could not have been more
explicit that written texts – properly understood – can be cor-
related neither with a subject, nor with lived experience, nor with
transcendental activity.
Just as ‘everyone’ knows, for example, that for Foucault the
‘author is dead’, so too does ‘everyone’ know that for Derrida the
text remains ‘readable’ after its author’s death. However, in both
cases there is a rather large gap between recognising the dictum and
understanding its meaning. The text, Derrida famously writes in
‘Signature, Event, Context’ (1972), survives in the radical absence
of any possible recipient, ‘of any possible reader determined in
general’.5 The text (and its survival) transcends its recognition; it
is not relative to a reader or any enabling acts.
The parallels between Derrida’s claims about a written message
being still ‘structurally readable’ in the ‘absolute absence’ of
any recipients and Meillassoux’s claims about ancestral events
being still evidenced in the ‘absolute’ absence of any witnesses
are unmistakable. Yet many of Derrida’s readers deflate his claim
about texts because it seems prima facie implausible. Yes, it is true

GOLDGABER 9781474438339 PRINT.indd 42 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 43

that the text survives the death of this or that author or reader.
After all, we regularly read the words, thoughts and intentions of
those long dead. Their words survive thanks to writing. However,
the same intuitions which will readily grant the latter will lead us
to say that texts cannot be readable in the radical, let us even say
terminal, absence of readers and writers in general.
The aim of this chapter is twofold. First, it demonstrates that
Derrida’s account of textuality is part of a radical critique of
correlationism, or what he calls the ‘metaphysics of presence’.
Second, it explores the significance of the structural parallels and
disjunctures between the critiques of Derrida and Meillassoux.
The parallels are easily evidenced. Meillassoux’s ancestral event
and Derrida’s posthumous texts both press the correlationist to
admit a radical outside to transcendental activity. Both argue that
the absolute exteriority of ancestrality and textuality, respectively,
cannot be understood in terms of a possible relation to a knowing
subject. Each of them wagers that demonstrating the necessary

FINAL
transcendence of these instances will present the correlationist –
committed to the unsurpassability of the epistemic relation – with
a crisis of their most basic philosophical commitments.
At this level of generality, the structural similarities between
these two thinkers may tempt us to say that the main difference
between their approaches is the placement of the temporal rup-
tures with the human that they introduce. Meillassoux’s ancestral
events point to a time radically anterior to the human (indeed to
any life on Earth), whereas Derrida’s posthumous texts point to a
time radically posterior to human life. However, closer attention
to the differences between the extra-correlational instances breaks
the homology I have just set up. For Derrida, the extra-correlation
instance refers to a structure, for Meillassoux it refers to an event.

Extra-Correlational Instances
Written texts and natural events are very different sorts of things.
One cannot easily generalise from one to the other. If the goal is
to push the correlationist interlocutor to admit to a necessarily
transcendental instance, written texts seem a uniquely challenging
choice of object. Unlike natural events, which our spontaneous,
naive realism already conceives non-correlationally – as unfold-
ing in blind indifference to human concerns – we are not at all
accustomed to thinking of written texts as radically independent

GOLDGABER 9781474438339 PRINT.indd 43 01/10/2020 09:30

 44 Speculative Grammatology

of humans. On the contrary, texts seem like exemplary technical

objects uniquely reliant on human makers.
It is, Meillassoux underscores, the dubious virtue of correlation-
ism to make belief in extra-correlational events and objects seem
incredible. It seems incredible, on the contrary, to think of texts
in extra-correlational terms. Yet, as we shall see below, Derrida
defines texts – exemplary linguistic artefacts – as ‘absolute’ in
Meillassoux’s sense, that is, as indifferent to human survival. Thus,
Meillassoux’s critique of correlationism on the basis of the ancestral
premise vindicates (rather than challenges) everyday ideas about
natural events. On the other hand, Derrida’s notion of absolute
textuality will entail a radical revision of our standard view of texts.
All things being equal – assuming both instances do the same
kind of critical work – we might be tempted to adopt Meillassoux’s
approach. Ancestral events promise to generalise to any and all
(material) events, processes and objects. It is not at all clear how
truths about texts would generalise to non-textual items. But of

FINAL
course, all things are not equal, and the ‘object’ choice is anything
but philosophically arbitrary; it defines their respective approaches
to the problem of correlationism. Meillassoux’s anti-correlationist
project leads to Alain Badiou and his ontology of the event, while
Derrida’s project leads through grammatology, the project of a
general writing.
Must we choose between Meillassoux’s critique of correla-
tionism and Derrida’s? Does it matter if our escape from cor-
relationism involves reflecting on the non-relative or absolute
status of events or texts? In this chapter, I shall argue that it does
matter. Reflecting on the difference between Meillassoux’s and
Derrida’s choice of object will allow us to diagnose a problem with
Meillassoux’s argument, or what he understands to be the entail-
ments of his argument. The distance between the two approaches
narrows considerably once we correct for this problem.
Meillassoux believes that correlationists face a characteristic
dilemma: patterns of reasoning that normally warrant belief in sci-
entific statements do not warrant such belief in the case of ances-
tral events. Therefore, according to him, they must decide between
their epistemic commitments to scientific evidence and their cor-
relationist commitments to the inconceivability of uncorrelated
events. However, on my reading, Meillassoux misunderstands the
entailments of his own argument. As he formulates it, the problem
of ancestrality actually challenges correlationists’ beliefs about the

GOLDGABER 9781474438339 PRINT.indd 44 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 45

nature of evidence and not the nature of events. If there is evi-

dence of ancestral events, this evidence must itself be ancestral. If
there is evidence of the non-correlational then there is also non-
correlational evidence. But this violates the correlational account
of evidence, that is, that all evidence is a form of givenness, which
Meillassoux’s argument assumes rather than challenges.
If my reading is right, Meillassoux’s thought-experiment
motivates not a non-epistemic conception of objects and events
but a non-epistemic – or recognition-transcendent – conception
of evidence. This difference has non-trivial consequences for
Meillassoux’s critique. By contrast, Derridean textuality just is,
under a different description, recognition-transcendent, non-
correlational evidence.
Non-correlational or recognition-transcendent evidence will
sound as odd as recognition-transcendent texts. Indeed, as I will
suggest in the conclusion of this chapter, the structure of textuality
and what I shall call the structure of indicative evidence are the

FINAL
same. On the standard view of texts and indicative signs, which
Derrida challenges, both are defined as correlates of transcenden-
tal activity. Texts and signs are only such when seen as indicating
the meaning of something to a subject who recognises them as
playing this functional role. If, on the other hand, this standard
(correlational) view is wrong, then evidentiary traces, including
those traces of ancestral events, and written texts both require
a non-correlational account. Once we reinterpret Meillassoux’s
argument in a way that correctly identifies its pertinence to evi-
dence rather than events, I argue that it entails revisions to our
everyday notion of events of the same order and scope as those
prompted by Derrida’s notion of textuality.
In the remainder of this chapter, I reconstruct Meillassoux’s
account of correlationism, emphasising the realist variant, which
is the target of his critique in After Finitude. As he argues, the cor-
relationist believes herself to be a realist, roughly someone who
can affirm the meaning of scientific statements, despite affirming
the unsurpassability of the correlation.6 Meillassoux’s argument
depends upon defeating the belief that correlationism is compat-
ible with realism. I argue that Meillassoux does defeat the cor-
relationist on the grounds he lays out, just not in the way that
he thinks he does. A more granular account of correlationism’s
philosophy of evidence allows us to identify a critical equivocation
in his argument.

GOLDGABER 9781474438339 PRINT.indd 45 01/10/2020 09:30

 46 Speculative Grammatology

I then reconstruct Derrida’s critique of correlationism from tex-

tuality. Derrida argues that texts are necessarily extra-correlational
– they retain their function autonomously from transcendental
subjects or readers. This relates essentially to their structure. Non-
correlational texts are what I call ‘in-formed’ or parasitic forms.
In the chapter’s final section, I turn to the significance of
Derridean textuality for Meillassoux’s critique from ancestrality.
Despite the fact that Meillassoux’s account would deny any mean-
ingful distinction between ancestral evidence and ancestral events,
his argument from ancestrality implicitly preserves the opposition
between indirect or indicative evidence and directly given events
that is essential to the correlationist position. Deconstructing the
opposition makes clear that a sufficiently powerful critique of cor-
relationism will require, as a matter of priority, a non-correlational
account of evidence. Derridean textuality provides a schema for
just such an account.

FINAL
Meillassoux and the Critique of Correlationism from
‘Ancestrality’
In After Finitude, Meillassoux argues provocatively that the truth
of the correlationist can be found in the bad-faith science sceptic.7
Consider the biblical creationist who believes in the literal truth of
Genesis. When presented with radiometric evidence (radioactive
dating) for the antiquity of the Earth, the biblical literalist can
either argue that she is not interested in scientific evidence, or, if
she wishes to avoid charges of irrationality, she can argue that
such evidence does not compel belief.
Belief in an ancient Earth seems to be well supported by radio-
active dating. But according to the sceptic, this belief does not
warrant the credence that scientists assign to it. We must believe
that the presence of the stable argon ‘daughters’ that we test for
today, and that we use to calculate the age of rocks, were not
present at the rock’s accretion, but are exclusively effects of the
very long process of radioactive decay (the half-life of potassium-
40 is around 1,250 billion years). The sophistical sceptic will dis-
armingly agree that this is a reasonable assumption on the basis of
the conditions we find today. But how can we be sure that these
same conditions were present then, at the origin?
It is true: we cannot be certain of initial conditions at the Earth’s
genesis, a time before any possible witnesses. Scientific inquiry

GOLDGABER 9781474438339 PRINT.indd 46 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 47

attempts to gain knowledge of these conditions by examining

traces of these events as they persist in the present, for example
in the geological record. The sceptic reminds us, however, that
evidence is a relational game between inquirer and world; the
game does not extend to a world that does not correlate to the
inquirer. Scientists must assume (nature’s) continuity and uni-
formity, that the conditions we find and observe today can be
retrojected to the ancient past. But, the sceptic argues, this belief
is unwarranted by scientific standards of evidence. We cannot
validate that the same conditions that we observe today prevailed
then, at the time of the Great Unwitnessable. Therefore, science
too rests on faith.
The bad-faith sceptic emphasises the central role of the witness
for establishing not only the ‘initial conditions’ but also the per-
sisting uniformity of natural laws. The presence of corroborated
evidence of these conditions validates the chain of inferences from
initial conditions to the present inquirer. However, where there is

FINAL
a manifest break in this chain, the epistemic warrant for scientific
inferences fails. The sceptic’s argument assumes that the (ideal)
standard of evidence is that which is seen with one’s own eyes – or
otherwise given to the senses – and that such givenness is also the
ground for all inference. For evidence of any process or event to
be sufficient, it must in some sense be contemporaneous with an
observer, or at least a possible observer.
Meillassoux argues that bad-faith science sceptics avail them-
selves of all the best resources that correlationist philosophy has
to offer. It follows that resisting the present wave of irrational-
ism will not require doing battle with biblical literalists in the
name of scientific rationality. It will require doing battle with the
seemingly responsible and institutionally respectable correlation-
ist philosopher who unwittingly provides the science sceptic with
so much sustenance. The sophisticated biblical literalist allows us
to make explicit the correlationist belief that evidence divorced
from witnesses, from a possible mode of givenness, is no evidence
at all.
If the science sceptic depicted above is in bad faith, this is because
their primary interest is not to consider the evidence of science from
an indifferent point of view, but to defend their own dogmatic
views as equally rational and warranted. Meillassoux does not
think the correlationist in general is a bad-faith sceptic; indeed, he
thinks that correlationists are usually not even aware that they are

GOLDGABER 9781474438339 PRINT.indd 47 01/10/2020 09:30

 48 Speculative Grammatology

science sceptics. Thus, pointing out that the correlationist shares

her epistemic bed with the biblical literalist is meant to serve as a
wake-up call.
At this point it is worth emphasising that Meillassoux’s argu-
ment does not aim at so-called post-structural correlationists who
would doubt that even our best (witness-corroborated) evidence
could ever (literally) be about the world. The correlationist who
believes that what we call ‘evidence’ is really experience as it
conforms to contingent conceptual or linguistic schemes is not
likely to be moved by charges that she undermines scientific claims
to objectivity. Often, her arguments call for radical revisions to
accounts of scientific objectivity. Meillassoux aims rather at those
correlationists who believe that their epistemic commitments do
nothing to impugn the validity of scientific claims. In other words,
he aims at realist-compatible varieties of correlationism and wagers
that making explicit the link between correlationism and forms of
contemporary irrationality will galvanise these correlationists to

FINAL
revise their views in the direction of a realism that will radically
decentre the role of the finite observer.
Meillassoux’s strategy assumes, then, that at least some
correlationists – the realist ones – are in error about their own
beliefs. But how can correlationists be so mistaken about their
views – in particular the epistemic consequences of their views?
For Meillassoux, the correlationist is mistaken about her ability to
endorse the literal meaning of scientific statements. This misappre-
hension is clarified when we consider scientific statements about
the distant – pre-human, pre-carbon-based life – past. Exemplary
of such statements is ‘the Earth accreted 4.5 billion years ago’.
He calls these sorts of statements ancestral statements. In turn,
they are grounded in statements about ‘arche-fossils’ and ‘fossil-
matter’. As Levi Bryant glosses:

Meillassoux refers to ‘arche-fossils’ and ‘fossil-matter’ as not simply

materials indicating the traces of past life in the sense of our familiar
understanding of ‘fossil’, but also as materials indicating the existence
of ancestral realities or events anterior to all life. An arche-fossil is
thus not an ancestral being like the big bang ‘in the flesh’, but is rather
something like the radioactive decay of isotopes that allows us to infer
the ancestral or that which precedes all life. Meillassoux’s question is
thus two-fold: On the one hand, he asks, under what conditions are
these statements meaningful? That is, what must be the case for ances-

GOLDGABER 9781474438339 PRINT.indd 48 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 49

tral statements to have any sense? On the other hand, and more fun-
damentally, how must the correlationist interpret these statements?8

The first question Bryant attributes to Meillassoux – ‘what must be

the case for ancestral statements to have any sense?’ – seems easy
to answer. According to a certain view, one knows the meaning of
‘the Earth accreted 4.5 billion years ago’ when one knows its truth
conditions. ‘The Earth accreted 4.5 billion years ago’ is true if and
only if the Earth accreted 4.5 billion years ago.
The second question, about how the correlationist must inter-
pret the statement, is less clear. As Bryant glosses, the correlation-
ist can unproblematically agree that fossils, in general, indicate
the existence of past life. They can affirm this because the past life
these fossils indicate is (presumably) correlatable, something that
could in principle be given ‘in the flesh’ to a witness. However,
arche-fossils are material indications (in the present) that allow us
to refer to that which cannot, in principle, be given.

FINAL
Herein lies the rub for the correlationist. For her there can
be no evidence, indicative or otherwise, of what cannot in prin-
ciple be given. The correlationist must insist that arche-fossils,
unlike fossils contemporaneous with life on Earth, do not evidence
anything. They cannot be indications, they cannot function as
evidence. The correlationist reasons that we take the indicating
evidence to be evidence of an event which, were there witnesses,
would have appeared ‘in the flesh’ in such and such a way.
The conditions for givenness are retrojected backwards. Indeed,
as Meillassoux emphasises, this is just how the correlationist makes
sense of any and all appeals to unwitnessed events. However,
the difference between the unwitnessed and the ancestral – the
reason why the latter poses a problem to the correlationist that
the former does not – is that, in the case of the ancestral, this ret-
rojection violates the core correlationist view that an object that
cannot in principle be given ‘in the flesh’ cannot be thought. Thus,
Meillassoux argues that any statement equivalent to ‘we have
indicative evidence of ancestral events’ cannot be meaningful to
the correlationist. Or else, the correlationist must admit that there
is evidence of the uncorrelated.

GOLDGABER 9781474438339 PRINT.indd 49 01/10/2020 09:30

 50 Speculative Grammatology

Correlationism as a Theory of Evidence

Correlationism is, I argue, best construed as a theory of evidence. A
brief detour through this theory will clarify the issues surrounding
‘evidence of the uncorrelated’. For the realist correlationist – for
example, the Husserlian phenomenologist – the (ideal) standard of
evidence is that of self-evidence or givenness. As Husserl writes in
the fourth Logical Investigation, ‘the appearing object announces
itself as self-given’ in an ‘originary giving intuition’ – this object
appears, in his formulation, ‘in the flesh’.9
The object as it gives itself in perception testifies to its presence
and, we might add, even to its ontological independence from its
own appearance. For Husserl, part of the meaning of a worldly
object’s appearance is that it transcends its appearance. However,
as with any other knowable feature of the world, this ontological
independence is evidenced through the ways it is given. A phenom-
enological analysis makes the mode and manner of its givenness

FINAL
explicit.
Givenness, or self-evidencing, is not limited to perception, which
remains nonetheless exemplary for the phenomenologist. Many
things self-evidence beyond perceptual objects, including math-
ematical idealities, the meaning of proofs and theoretical entities.
What all forms of givenness have in common is that they involve
ideal content or meaning. In the case of a present perception,
meaning (formulable as a logical statement) is intrinsic to its appear-
ing. Meanings can, however, float free of the perceptions that make
them meaningful. To understand the meaning of a statement such
as ‘the cat is on the mat’ requires, for Husserl, knowing what sort
of perceptual state of affairs would make that sentence self-evident.
Husserl calls this movement from language to meaning ‘reacti-
vation’. To get at the meanings of our statements, we must be able,
at least in principle, to go back to the primary self-evidences upon
which these statements are built. For example, we reactivate the
meaning of Euclidean geometry when we go through the proofs
and reawaken the sort of cognitive self-evidences that animated
Euclid when he wrote his proofs. Reactivation is important for
scientific activity, in particular, because of the long chains of infer-
ential reasoning and judgements involved in scientific work and
communication.10
Just as givenness is not limited to a present perception, evidence is
not, strictly speaking, limited to forms of direct givenness. The cor-

GOLDGABER 9781474438339 PRINT.indd 50 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 51

relationist admits a secondary form of evidence: indeed, one that is

closer to the everyday use of that term. Evidence is either self-giving
or it is what Husserl calls ‘indicative’. However, indicative evidence
is evidence only insofar as it leads back to a form of possible given-
ness. That is, indication points the observer to an absent or past
form of presence. It is, as it were, a placeholder for meaning. The
meaning of any indicative sign is never in its primary givenness –
how it directly appears – and to take it as it is given is to miss that
it is evidence of something, in short that it is a sign of something.11
Indication supposes an inferential leap from absence to pres-
ence. Non-presence is part of the way that indicative evidence
‘gives’ itself. Husserl’s example: when we see smoke indicatively,
we see it as indicating the ‘presence’ of (a perceptually absent) fire.
The difference between seeing smoke directly (as it gives itself) and
seeing smoke indicatively is not ‘intrinsic’ to the appearing object
(that is, smoke); the difference is in how the perceiver ‘takes’ or
‘intends’ the perceptual object (as a given, or as a sign). Thus, for

FINAL
example, when the dendochronologist (who analyses and dates
the growth rings of trees) discovers a spike in carbon-14 in a
tree’s rings, she recognises this presence as a marker or trace of
a past event, say a solar flare. She can then date the event of the
solar flare, knowing that each tree ring marks a terrestrial year.
The meaning of any indicative sign is whatever more original,
primary form of presence it points to. This meaning is established
not by the sign, but by the knower’s tutored inferences. Indeed, as
Husserl writes in Ideas I, ‘Nature is only in being constituted in
regular concatenations of consciousness.’12
Indicative evidence, then, is no more recognition-transcendent
than other primary forms of givenness. It requires seeing something
as evidence, or as a sign of something else, which in turn requires
making the sort of associations that will lead back to (a form of)
possible givenness. The difference between indicative evidence and
givenness is the sort of evidentiary warrant each gives. Givenness,
Husserl specifies, is apodictic or self-evident.13 Indication, by con-
trast, is never self-evident. We need a web of associations and
empirical correlations to grasp something as indicative.
The detour through the correlationist theory of evidence is com-
plete. We needed the distinction between indicative evidence and
evidence ‘in the flesh’ or givenness in order to get a precise sense
for why a correlationist cannot affirm indicative evidence of the
ancestral.

GOLDGABER 9781474438339 PRINT.indd 51 01/10/2020 09:30

 52 Speculative Grammatology

Evidence of the Absolute as ‘Absolute Evidence’

As we saw above, Meillassoux seeks to weaken correlationists’
intuitions that they can vindicate the realist meaning of scientific
statements. He argues that the weakness of correlationism as a
satisfactory interpretation of scientific evidence is manifest in cor-
relationists’ inability to affirm ancestral events. Ancestral events
are prior to the possibility of self-evidence, thus they cannot be
equated with any form of (direct) givenness. But neither can ances-
tral events be indicated, properly speaking, since as we just saw,
indication must refer to a possible presence or form of givenness.
Hence, the correlationist cannot affirm, after all, what the scientist
affirms when affirming evidence for an ancient Earth – not because
the events referenced are uncorrelated, but because there cannot
be evidence of the uncorrelated.
Meillassoux focuses on the difference between the meaning of
the scientist’s statement for the scientist and its meaning for the

FINAL
correlationist. Unfortunately, he does not discuss whether the sci-
entist would endorse the non recognition-transcendent status of
their evidence. It is true that for the scientist the arche-fossil will
count as evidence (a valid ground of inference for the ancestral
event) whereas it cannot count as such for the correlationist. But
why does arche-evidence remain a valid ground of inference for
the scientist? It must be because the scientist, at least implicitly,
has a different view of indicative evidence than the correlationist.
The scientist does not (or need not) believe that indicative evidence
is essentially tied to givenness. There can be evidence of the non-
correlatable because evidence is uncorrelated.
Curiously, Meillassoux seems to retain the very correlational
account of evidence that he otherwise puts into question. To para-
phrase Meillassoux, what the correlationist cannot accept is that an
‘arche-fossil can manifest [evidence] an entity’s anteriority vis-à-vis
manifestation [givenness]’.14 But can the arche-fossil also manifest
this anteriority without manifesting to someone? Apparently not,
since Meillassoux writes just before the line quoted that ‘fossil-
matter is the givenness in the present of a being that is anterior
to givenness’. Strictly speaking, he ought to have written, ‘fossil-
matter is evidence (in the present) of a being anterior to givenness’.
This is what the correlationist cannot believe but what the scientist
must believe, since, for both, the fossil matter will always have
been traces of anterior being and not givenness qua self-evidence.

GOLDGABER 9781474438339 PRINT.indd 52 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 53

The question for Meillassoux is why he accepts the correlationist’s

description of evidentiary traces in the first place?
Given that the arche-fossil dates back billions of years – we are
speaking here, again, of things such as radioactive decay, of the
1.2-billion-year half-life of potassium-40 – what does its ‘givenness
in the present’ matter to the evidentiary worth of the arche-fossil?
Unless we are inclined to agree with our biblical literalist, we
would like to say that the decay ‘clock’ of potassium-40 continues
to tick whether it is manifesting (to someone) in the present or not,
which is what makes it such good evidence of the remote past. We
would want to say, simply – in order to take the scientist literally –
that we have evidence of the ancestral because the evidence is itself
ancestral. The evidence evidences the ancestral, it indicates this
event, even in the absence of any possible correlation.
Put another way, if evidence can be of the ancestral, if this
evidence can itself be ancestral, then indicative evidence ought no
longer to be defined by the inferential activity of the scientist any

FINAL
more than the ancestral event ought to be defined by the enabling
activity of a knower. Evidence itself must be thought in non-
correlational terms.
What I have suggested is that Meillassoux’s argument from
ancestrality motivates, in the first instance, a non-correlational or
recognition-transcendent account of evidence (or, as he puts it,
‘traces’) rather than a non-correlational account of the event. This
sets a rather different path and agenda for the anti-correlationist
than the one he anticipates – since he is interested in disentan-
gling both the thought of the event from the thought of evidence,
and thought of the ontological from that of the epistemological.
However, if evidence were ‘absolute’ in this sense, then these dis-
tinctions no longer mean what they mean for the correlationist.
Absolute evidence deconstructs these distinctions because it entails
thinking of evidence as independent of transcendental activity. For
such a view we find surprising resources in Derrida’s notion of
textuality, to which we now turn.

The Critique of Correlationism from

‘Posthumous Texts’
The philosopher John Searle, addressing a meeting of literary
scholars at the height of deconstruction’s influence in the United
States, attempted to correct what he saw as literary theory’s

GOLDGABER 9781474438339 PRINT.indd 53 01/10/2020 09:30

 54 Speculative Grammatology

increasingly unhinged views about texts and meanings.15 There

are, he insisted, exactly two options to review when considering
theories of textual (and by extension, linguistic) meaning. The first
is to cash out textual meaning in terms of authorial intention. The
second is to think of meaning as assigned by the recipients of the
text. Meaning, then, is a matter of what is present at the text’s
inception or reception, through the activity of either the author
or the reader. In either case, meaning implies the reduction of the
material signifier, an exit from the text towards meaning.
For Searle, it seems obvious that Derrida’s dictum ‘there is no
outside-the-text’ would imply not only that texts have no assign-
able or determinant meaning, but simply no meaning at all:

For Jacques Derrida meaning is a matter of, well, what? Meanings are
‘undecidable’ and have ‘relative indeterminacy’, according to Derrida.
Instead of fully determinate meaning, there is rather the free play of
signifiers and the grafting of texts onto texts is within the textuality

FINAL
and intertextuality of the text.16

Searle’s critique of the ‘generalised’ intertextuality that he attrib-

utes to Derrida is an auspicious starting point for elucidating
Derrida’s concept of textuality. This is not because it offers us the
right account of Derridean textuality (it does not), but because
it presents, with admirable economy, how Derridean textuality
appears from the correlationist’s point of view. Since for the cor-
relationist a text only has meaning when and if it is being read by
a reader or written by an author, to affirm ‘no outside-the-text’
means that there is no reader/writer whose transcendental activity
can bring the text ‘home’, returning meaning to the written words.
It reduces the text to a heap of insignificance.17
As Searle argues, philosophers have typically understood written
texts in terms of the sort of subjective activities they make possible:
reading and writing, communication at a distance. Just like speech,
written texts communicate meanings. However, following Searle’s
descriptions literally, texts (and speech) do not have meaning;
speakers and readers have meanings. Texts signify or indicate ‘tran-
scendental’ or correlational Meanings through a certain economy
of the (artefactual) linguistic mark. This picture of texts as vehi-
cles for meanings assumes, Derrida argues, a radical distinction
between textual signifiers and their (transcendental) meanings or
signifieds. From this it follows that texts can neither be construed

GOLDGABER 9781474438339 PRINT.indd 54 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 55

in non-relational terms nor considered exclusively from the side of

their objective materiality. The idea of an extra-correlational text
or sign is a transcendental illusion. To be a text is to be readable,
an unreadable text is no text at all, and texts without any possible
readers are unreadable. The possibility of reading refers to the pos-
sible presence of a reader who can restore their meaning. Meaning,
on the other hand, is essentially non-textual.
Contrary to this tradition, Derrida argues that the readability of
a text – the essential possibility defining it – is not dependent upon
the existence of any reader. Asking us to consider the status of a
radically orphaned text, outside the horizon of any transcendental
subject, Derrida makes the case for the non-correlational status
of texts. ‘A writing that is not structurally readable – iterable –
beyond the death of the addressee – would not be writing.’18 Here,
as will become clear, Derrida is not referring to the death of this
or that potential addressee – as is often supposed in the secondary
literature – but to the radical absence of all possible addressees.

FINAL
Hence, the readability to which Derrida refers – and which con-
nects with the more unfamiliar term ‘iterability’ – would redefine
the nature of texts, and by extension the nature of meaning.
Let us consider in some detail the moment in Derrida’s 1972
essay ‘Signature, Event, Context’ when he defines texts and writing
in terms of this posthumous readability. After arguing, unobjec-
tionably, that writing distinctively functions in the absence of its
author or intended reader – a text may always be intercepted, its
recipient lost or confused for another – Derrida argues that the
same possibility that ensures that texts survive their authors and
intended audiences also entails that texts function in the radical
absence of all readers:

In order for my ‘written communication’ to retain its function as

writing, i.e., its readability, it must remain readable despite the abso-
lute disappearance of any receiver, determined in general. My com-
munication must be repeatable – iterable – in the absolute absence of
the receiver or of any empirically determinable collectivity of receivers.
Such iterability – (iter, again, probably comes from itara, other in
Sanskrit, and everything that follows can be read as the working
out of the logic that ties repetition to alterity) structures the mark of
writing itself, no matter what particular type of writing is involved . . .
a writing that is not structurally readable – iterable – beyond the death
of the addressee would not be writing.19

GOLDGABER 9781474438339 PRINT.indd 55 01/10/2020 09:30

 56 Speculative Grammatology

While the classic definition of texts that Searle’s analysis calls upon
assumes that a transcendental, meaning-giving activity defines
texts, Derrida insists that the absence of such activity defines
writing of ‘no matter what particular type’ from the start. If
written messages are readable (‘retain [their] function as writing’)
at the two standard termini of the communicative circuit (writer
and intended audience), then it follows that the message remains
readable (and also interceptable) at any point in its itinerary. If a
text is readable at any point in the itinerary, even when it is not
being read, is it not strange to think that it would lose its read-
ability in the case of the sudden, apocalyptic and, let us further
stipulate, terminal disappearance of any and all possible readers?
The structure of texts accounts for the possibility of reading.
Yet this is precisely what the classic definition of a text – the one
to which Searle subscribes – denies: the possibility that readability
names is located on the side of the transcendental subject that is
capable of restoring a text’s meaning through the mnemonic asso-

FINAL
ciations of exteriorised textual marks with ‘internal meanings’.20
On the other hand, if a text is readable in the absence of all pos-
sible readers, then what, now, is reading?
One may object that Derrida’s argument, here, depends upon
conflating ‘legibility’ and ‘readability’. It seems possible to deflate
Derrida’s radical-sounding claim about readability to a seem-
ingly less radical claim relating to the formal features of any text.
Readability above translates the French lisabilité, which signifies
equivocally between two distinctive English terms, ‘legibility’ and
‘readability’. Is Derrida, then, making the less interesting claim
that the text retains its legible form after the radical disappearance
of all reading life: that a text’s structural features are recognition-
transcendent? Or is he making the incredible-sounding claim
that the text remains readable (and hence meaningful) outside
of reading life? In fact, as I will now argue, Derrida means both;
indeed, a proper understanding of textual form will forbid any
clear distinction between structure and signification.
At the conclusion of the text cited above, Derrida writes that the
text remains ‘structurally readable [lisible]’ beyond the death of the
addressee. This death is qualified first as the ‘absolute disappear-
ance’, then as the ‘absolute absence’ of any receiver ‘determined
in general’. The notion of ‘structural readability’ that Derrida is
after must be interpreted in terms of the text’s radical ‘sur-vival’
– or sur-vivance, literally (in French) its ‘living on’ and/or ‘living

GOLDGABER 9781474438339 PRINT.indd 56 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 57

beyond’. What is the meaning of the ‘structure’ that qualifies read-

ability? The subsequent line appears to offer a response. This read-
ability should be understood in terms of the structure of iterability.
But iterability is a neologism – it designates a kind of ‘future
concept’ (should we discover the need for one) – for a relation
logically linking repetition and alterity. Why should the structure
of readability require recourse to such an unheard of relation?
Iterability, it is seldom enough emphasised, is not identical
with the concept of a repeatable structure in general. It is not a
synonym for repetition. Iterability is not a novel way to designate
the old idea or ‘paleonym’ of a linguistic mark (e.g. the letter ‘e’)
or a pattern of elements (e.g. a word or sentence). Indeed, it is
because the available concept of repetition does not specify the
nature of the textual (readable) structure that Derrida justifies the
introduction of iterability, which will now designate the distinc-
tive sort of repeatable structure that a text is. If Derrida selects
the term iterability to designate this structure, it is because, on the

FINAL
basis of his speculative etymology, the particle ‘itera-’ captures
economically the idea of a repetition ‘always already’ entailing
alterity. The alterity in question is not the ‘once again’ of an ideal
repetition (of the same). It is the idea of a heterogeneity implicit in
each such repetition, a heterogeneity that remains obscure, unre-
marked in the classic concept. ‘Such iterability . . . structures the
mark of writing itself, no matter what particular type of writing
is involved.’21
What does it mean to say that iterability structures the mark of
writing ‘itself’? Iterability both produces the mark itself and makes
the mark something other than itself. The ‘othering’ in question is
not the one implicit in the difference between tokens of an ideal
type. Attempting to distinguish the structure of iterability from
the latter, Derrida introduces the image of parasitism: ‘iterability
alters, contaminating parasitically what it identifies and enables
to repeat “itself”’.22 The mark is never simply identical with itself
because it always already ‘hosts’ another mark that structures it.
The infrastructure of a mark is a heterogeneous mark. The mark is
‘in-formed’ by another, in-formed form.
Texts are parasitic structures; the relation between textual signi-
fier and its presumed ‘transcendental signified’ is something like
the relation between a parasite and its host, where the parasite
exists in and through the body of its host.23 If we follow this meta-
phor, the signifying stratum, composed of its pattern of differential

GOLDGABER 9781474438339 PRINT.indd 57 01/10/2020 09:30

 58 Speculative Grammatology

marks, would be the host or medium in and through which the

parasitic signified elements are expressed. A nested structure, the
parasitic pattern in-forms the host pattern. The form of the host
is always already due to the form of a heterogeneous other. While
this description of a text may sound unfamiliar and even fantasti-
cal, Derrida suggests that such structures are, in fact, perfectly
ordinary and common. They describe, for example, the written
texts all around us. The extra-correlational ‘life’ of texts is a para-
sitic life.
I have just argued that Derrida’s neologism, ‘iterability’, speci-
fies the sort of in-formed structure that permits writing to function
both within and beyond the horizon of any subject. This structure,
‘foreign to the order of presence’, binds repetition to alterity. It
repeats a heterogeneous difference (signified) in repeating itself.
Textually speaking, differences live on or sur-vive only in the
‘flesh’ of the heterogeneous differences that they alter or trans-
form. Whatever is structured like a writing – and this includes,

FINAL
minimally, language – will involve differences or systems of differ-
ences characterised by the modifiability or in-formability of their
elements. The repetition characteristic of language is not limited to,
nor can it be exemplified by, the story of a conventionally adopted,
arbitrary mark, which would be indefinitely repeatable. Such an
account of linguistic repetition suffers from a one-sidedness that
the account of iterability corrects by drawing our attention to the
potentially incalculable manifold of encoded differences repeated
along with any repetition of the surface or host text.
On the basis of Derrida’s non-correlational account of the struc-
ture of texts, we can say that one experiences something as a signi-
fier, or understands the meaning of a signifier, when one performs
– however spontaneously or unconsciously – something like a
reverse translation, retrieving the parasitic or nested pattern from
the manifest, host pattern. However, this possibility – the pos-
sibility of reading or translation – is not a function of the reader.
Translatability is a function of the structure of the text. Therefore,
as Derrida points out, a text remains structurally readable without
anyone actually performing such an act of translation – or even
recognising it as a text. An empirical example will help illustrate
the point.
James Gleick, in his book The Information (2011), retells the
famous story of the drum language of the Kele people in Central
West Africa.24 Gleick describes the ‘talking drums’ as patterned

GOLDGABER 9781474438339 PRINT.indd 58 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 59

after spoken Kele, a tonal language with two sharply distinct tones:
each syllable is either low or high. The Kele language is ‘spoken’
– or more properly, ‘written’ – by a pair of drums that, isomorphi-
cally, produce two distinct tones. Kele words are ‘pronounced’ on
the drums in a sequence of high and low tones that repeat the tonal
pattern in the spoken language.
Kele, like any spoken language, is rich in contrastive opposi-
tions. Tonal differences are just some of these ‘marked’ contrastive
magnitudes. But tonal differences are important enough contrasts
to allow other marked differences to be dropped without too
much loss of information – jst lk w cn drp vwls frm phntc nglsh –
while still allowing the drum-writing to convey differences marked
in speech. Gleick notes that European travellers and missionaries
did not know how to ‘read’ the drum language and were always
surprised that news of their impending arrival preceded them.
They simply could not hear the ‘readable’ messages that were
everywhere around them.25

FINAL
It is true that the textual iterability exemplified by the drum-
writing – the same iterability that we find in English phonetic
writing – may seem a very poor model for understanding the
difference between signifier and signified, generally speaking, pre-
cisely because phonetic writing does not seem to implicate the
signified or conceptual element in language. Indeed, drum-writing
as described makes it a form of phonetic writing, where the sig-
nified (structuring) element is, perhaps non-standardly, another
signifier, differences in speech. We will concern ourselves further
with the question of meaning in the next two chapters. For now,
I want only to note that, if iterability generalises to all linguistic
structures, it must apply to the link between signifier and signified
– or, as Derrida emphasises, ‘to writing of whatever sort’. If this
claim is warranted, then the signified or semantic content can also
be preserved within and/or as the differential patterns that define
speech or the signifying strata of language.

From Posthumous Texts to Ancestral Evidence

As with Meillassoux’s anti-correlationist strategy, Derrida’s
account of writing asks us to think the status of the written text
under conditions that exclude the possibility of any correla-
tion taking place. He defines the text in terms of a recognition-
transcendent structure – iterability – that also suggests a new way

GOLDGABER 9781474438339 PRINT.indd 59 01/10/2020 09:30

 60 Speculative Grammatology

to think about reading. Reading will no longer be understood

in terms of the movement from indicative signs (signifiers) to
conscious meanings (transcendental signifieds). Instead, reading
involves a reverse translation, the reconstruction or reconstitu-
tion of the parasitic text. The iterable text has no vital need of a
transcendental subject to be a text or retain its function as written
communication. Derrida’s non-correlational account of textuality
thus entails challenging the standard, correlationist view of texts
as indicative signs that point to transcendental signifieds.
What I want to argue now, however, is that Derrida’s rejection
of the view that texts are structurally explicable in terms of cor-
relational, indicative signs is not limited to linguistic texts. Indeed,
the claim is not simply that written texts do not function indica-
tively, that is, that they cannot be understood as the correlate
of a transcendental activity connecting them to a signified. The
claim is also that there are no purely indicative signs. The mark in
general, in whatever domain it is found, will not be indicative in

FINAL
the correlationist sense. This has repercussions for Meillassoux’s
anti-correlationism and his notion of the absolute, that is, non-
subject-relative structure or being.
As we saw above from Levi Bryant’s gloss, ‘Meillassoux refers
to “arche-fossils” and “fossil-matter” as not simply materials indi-
cating the traces of past life in the sense of our familiar under-
standing of “fossil”, but also as materials indicating the existence
of ancestral realities or events anterior to all life.’26 Arche-fossils,
though they refer to ancestral events, are still interpreted correla-
tionally; they are signs grounding inferences.
How does Derrida’s notion of textuality permit an understand-
ing of arche-matter in non-indicative, non-correlational terms?
The arche-matter is not an indication of ancestral realities, if this
means that it ceases to be readable outside of inferential activity.
Arche-matter, its evidence and traces, must remain legible outside
any and all transcendental activity. If we were to understand arche-
matter textually, in Derrida’s sense, how should we now describe
it? If we use the schema of textuality, we would be looking at the
present indicative marks of ancestral events as structured by a
heterogeneous pattern or mark.
Let us return to the example of the dendochronologist, who
finds in one of the central (that is, oldest) rings of a tree a ‘spike’
of carbon-14. How shall we describe the pattern of discovery?
Shall we say she sees the spike as an indicative sign, by linking the

GOLDGABER 9781474438339 PRINT.indd 60 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 61

presence of carbon-14 to the known efflux of solar flares and then

inferring that a solar flare happened, say, 250 years ago? Such a
description seems perfectly plausible. Is another description pos-
sible? Can we also say that she ‘reads’ the solar flare in the carbon-
14 markers? What is the difference in these two descriptions? Is
one more apt than the other? What is seeing and what is reading?
The carbon-14 can function as a marker or indication of some-
thing because it is differentially present in the tree rings. Each
distinct tree ring, we have learned, marks a terrestrial year of
tree-life. This clear demarcation makes the tree a sort of almanac.
The differential pattern of carbon absorption in each ring records
the differential quantities of carbon in the atmosphere each
year. The differential marks of carbon are already spatially and
temporally structured, without which they could not be read.
These differences, or carbon marks, we might suggest, form a text
within the text of the tree rings. The ‘burst’ in carbon, which we
find spaced let us say in ring 12 – that is, the difference between it

FINAL
and the surrounding carbon marks – is the trace of the solar flare
event. Indeed, the distinction between the trace (of the event) and
the event now seems problematic. Surely the absorption and reten-
tion of the carbon-14 is part of the event. Minimally, this textual
description motivates more theoretical attention to the structure of
evidentiary traces, while the competing description assumes that
the structuring activity happens in the associational activity of the
scientist.
There is, of course, much more in this living record or archive
of the tree. For example, differences in the width of the tree rings
‘signify’ differences in the length of the growing season. What
makes something a significant mark or indication is first of all the
presence of other marks, a differential pattern. What makes these
marks textual, in Derrida’s sense, is that the form or pattern is in-
formed by a heterogeneous pattern. Scientific discovery does not
amount to adding to these marks their transcendent meaning, but
learning to read what in-forms these marks. The different widths
of the tree rings indicate the fluctuating length of growing seasons;
the density of the rings indicates different levels of rainfall. We can
continue to say that the dendochronologist’s activity consists in
inferring past events from these present traces, but this inferential
activity is possible because of the textual structure of the evidence,
which is neither defined by nor dependent on it. The point would
be that not just ancestral events but all events are best understood

GOLDGABER 9781474438339 PRINT.indd 61 01/10/2020 09:30

 62 Speculative Grammatology

not as correlated with a transcendental subject but as correlated to

textual evidence or ‘written’ traces.
Texts, then, are not inherently linguistic; they describe, as we
demonstrated above, an iterable structure. Differential marks are
the material ‘substrate’ of the signified. The signified is, as it were,
inscribed in the signifier, or vice versa. This structure of entangle-
ment characterises textuality, writing ‘of whatever kind’. Derrida’s
notion of ultra-transcendental or post-human texts transforms our
understanding of the structure and materiality of language and
writing. In doing so, it also provides a general or speculative
schema for any kind of textual or grammatological structure.

Conclusion
Why has Derrida’s critique of correlationism been so difficult to
recognise as such? Unlike Meillassoux, Derrida’s aim is not to
dispel the spectre of correlationism by pointing to the sort of

FINAL
object – a readable text – that cannot be described or, indeed, can
only be essentially mis-described by correlationism. His aim, or so
I shall argue in the following chapters, is to dismantle correlation-
ism from within. Not only language but phenomena ‘internal’ to
the correlation will turn out to be textual.
Derrida does not move from an anti-correlationist argument
with respect to texts to a general argument about the viability of a
materialist programme where nothing would be outside ‘the for-
mation of form’. Nor does he explicitly investigate how thoughts
about the absolute (non-relative) are possible, as does Meillassoux.
Rather, Derrida moves from an anti-correlationist argument about
the status of texts to an anti-correlationist argument about the
textual status of consciousness or lived experience.
Written texts and their purported meanings (signifieds) are not
only independent of subjects and subjective life. They are the broad
possibility of what Husserl called ‘reactivation’. Consciousness
and perception, understood as the very essence of ‘the correlation’,
are thus equally mis-described by correlationism. Indeed, correla-
tionism (or the metaphysics of presence) would be plausible in the
first place only because of correlationist descriptions of texts and
belief in transcendental signifieds.

GOLDGABER 9781474438339 PRINT.indd 62 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 63

Notes
1. Beck, quoted in Zahavi (2016: 306, emphasis mine).
2. Meillassoux (2008: 9).
3. Meillassoux (2008: 5, emphasis in original).
4. Malabou (2016: 41).
5. Derrida (1988: 7).
6. The phenomenologist believes, in other words, that even though
she rejects the naive metaphysical realism implicit in scientific state-
ments, her perceptual realism retains the meaning of these state-
ments. The ‘translation’ from metaphysical to perceptual realism
preserves the sense and validity of the statement.
7. Bad-faith sceptic because the sceptic in question is not interested
in the truth or falsity of scientific statement; they do not have a
genuine concern for the truth of scientific statements. Seeking to
justify their a priori disbelief, they attempt to show that scientists
have no better grounds for their beliefs than biblical literalists do

FINAL
for theirs. Correlationism – and hence unwittingly contemporary
philosophy, which is overwhelmingly correlationist – gives resources
to the sceptic, precisely by linking credence, on the one hand, to
proximity to testimonial evidence and, on the other, to the possibil-
ity of, as Husserl described it, reactivating each link of an unbroken
chain of inference to the primary self-evidences that ground belief.
8. Bryant (2009: n.p.).
9. Husserl (2000: 252).
10. Husserl (2000: 376).
11. Husserl (2000: 103).
12. Husserl (1983: 116).
13. Seebohm (1995).
14. Meillassoux (2008: 14).
15. Searle (1994: 634).
16. Searle (1994: 637).
17. In Introduction to the Origin of Geometry, Derrida writes:

Writing, as the place of absolutely permanent ideal objectivities and

therefore of absolute Objectivity, certainly constitutes such a [subjectless]
transcendental field. And likewise, to be sure, transcendental subjectivity
can be fully announced and appear on the basis of this field or its pos-
sibility. Thus a subjectless transcendental field is one of the ‘conditions’ of
transcendental subjectivity. But all this can be said only on the basis of an
intentional analysis which retains from writing nothing but writing’s pure

GOLDGABER 9781474438339 PRINT.indd 63 01/10/2020 09:30

 64 Speculative Grammatology

relation to a consciousness which grounds it as such, and not its factuality

which, left to itself, is totally without signification [insignifiante]. (1989:
89)

Transcendental writing, in other words, can dispense with the pos-

sibility of actual (existent) readers, and think of writing as among the
transcendental conditions of objectivity. However, precisely because
it is conceived correlationally in terms of a pure (possible) rela-
tion to a consciousness, such an analysis must determine factual
writing (equally deprived of actual readers) as essentially insignifi-
cant (without Meaning). Thus, when we look at empirical phenom-
ena such as hieroglyphics, we get a sense of what Derrida calls a
‘transcendental sense of death’, the appearance to consciousness of
its own (impossible) radical absence.

But if the text does not announce its own pure dependence on a writer

FINAL
or reader in general (i.e., if it is not haunted by a virtual intentionality),
and if there is no purely juridical possibility of it being intelligible for a
transcendental subject in general, then there is no more in the vacuity of
its soul than a chaotic literalness or the sensible opacity of a defunct des-
ignation, a designation deprived of its transcendental function. (Derrida
1989: 90)

In short, the text must appear as both unreadable and a priori read-
able. In The Introduction to the Origin of Geometry, Derrida does
not resolve the ‘paradox’ of transcendental death, he announces it.
On my reading, in this announcement we can find a central key to
his project.
18. Derrida (1988: 7).
19. Derrida (1988: 7).
20. Indeed, as Peter Bornedal argues, Derrida ought to have come to
exactly the opposite conclusion from the one he did: rather than tes-
tifying to the possibility of their function under conditions of radical
absence, texts testify to the need for a reader’s radical presence
(1997: 201–5). Texts need some – any – possible readers (together
with whatever empirical conditions are required to decode them)
– at least when we assume that written texts are supplementary to
some original (form of) living presence. If a text’s inner meaning or
content is radically exterior to the text, it follows directly that the
survival of a text is radically dependent upon the life of its possible
readers. Bornedal’s interpretation is interesting, because unlike some

GOLDGABER 9781474438339 PRINT.indd 64 01/10/2020 09:30

 From Ancestral Events to Posthumous Texts 65

of Derrida’s interpreters who see him as arguing that the definition

of a text is to transcend any particular reader or writer, Bornedal
reads him correctly as claiming that the radical absence of all readers
defines all texts. This pattern of interpretation is unfortunately wide-
spread. Derrida’s detractors tend, on the whole, to be more likely to
take his claims literally and seriously. On the other hand, Derrida’s
defenders tend to deflate his claims, even where these deflations lead
either to distorting the text or to implausible positions, or both.
21. Derrida (1988: 62).
22. Derrida (1988: 62).
23. ‘Transcendental signified’ economically captures the twin assump-
tions of a correlational interpretation of texts: 1) a text is a pattern of
signifiers whose signified is presumed to be transcendent to the text
and 2) the signified element is the effect of a transcendental activity.
24. Gleick (2011: 19).
25. The story of the talking drums and the way that they ‘speak’ Kele is
even more interesting if we consider that the distinctive sounds of the

FINAL
talking drums account, according to some musicologists, for some
of the distinctive structure of American popular music and dance
forms, including jazz, rock ’n’ roll and tap. One might speculate
that the musical structures descended from Kele drumming are still
replete with readable messages, though such messages would be pro-
duced without any sort of intention on the part of the drummer or
tap dancer. The readability of such messages is entirely unaffected by
the presence or absence of Kele speakers. In the drum-writing, Kele
spoken language is found in and as the pattern of differences that the
drums repeat – in much the same way, we might add, that spoken
language is found in phonetic writing.
26. Bryant (2009: n.p., my emphasis).

GOLDGABER 9781474438339 PRINT.indd 65 01/10/2020 09:30

 4

Rewriting the
Course in General Linguistics:
From Sign to Spacing

The name ‘my ink pot’ seems to overlay the perceived object, to belong
sensibly to it. This belonging is of a peculiar kind . . . not word and
inkpot.
Edmund Husserl, Logical Investigations

To think the ‘holding together’ of the disparate itself. Not to maintain

FINAL
together the disparate, but to put ourselves there where the disparate
itself holds together, without wounding the dis-jointure, the disper-
sion, or the difference, without effacing the heterogeneity of the other.
Jacques Derrida, Spectres of Marx

Multiple heterogeneous iterations all: past, present, and future, not in

a relation of linear unfolding, but threaded through one another in a
nonlinear enfolding of spacetimemattering, a topology that defies any
suggestion of a smooth continuous manifold.
Karen Barad, ‘Quantum Entanglements’

The Most General Concept of the Gramme

The grammatologist, founder of an autonomous science of writing,
must first clarify its central concept, writing. In order to do so,
they must clarify grammatology’s position vis-à-vis linguistics. If
Saussure indicated that, ‘by rights’, the field of linguistics belonged
to the (more general) field of semiotics, so would semiotics be
covered by the field of grammatology.1 To free grammatology
from linguistics requires grammatology also to declare its inde-
pendence from the concept of writing generated by linguistics.
Who, then, is the grammatologist for whom, in particular,
arche-writing would be so pertinent? Of Grammatology appeared
in 1967, and the expansion of writing, Derrida argues in the first

GOLDGABER 9781474438339 PRINT.indd 94 01/10/2020 09:30

 Rewriting the Course in General Linguistics 95

chapter of the book, was in the cybernetic air du temps. In popular

and scientific thought, ‘writing’ no longer meant what it once did.
It was no longer a technical appendage to human language. It was
already, to all appearances, becoming general, extending from the
code of life (genetics) to computer programs (informatics).
The expansion of writing in scientific, technical and aesthetic
discourses was essentially speculative, both in the sense of infla-
tionary, demonstrating a certain willingness to capitalise on a
metaphor, and in the sense of using the model of writing to think
or explain that which underwrites and supersedes experience. This
change in the concept of writing, Derrida insists, calls for philo-
sophical explanation and explication, a critical justification of this
speculative discourse. Of Grammatology answers these calls.2
The grammatologist aims to think writing in the broadest pos-
sible terms – that is, in terms that both subtend and extend beyond
the human. The grammatologist requires the most general concept
of the gramme. How then to get at this general concept? Does the

FINAL
generalisation of writing imply that graphic writing serves as the
model for what has not traditionally been thought of as writing?
There is no doubt that the model or metaphor of writing had been
productive – especially for linguistics. But the generality of writing
will not refer to the fecundity and productivity – the general appeal
– of this metaphor.
Certainly no science of general writing can be founded on a
metaphor. Minimally, grammatology will require principles for
determining what makes something an instance of writing. In the
context of grammatology, phonetic and graphic writing would
constitute instances of writing that have been mistaken for the
whole. But how to get from the example or model of writing, in
the narrow sense of phonetic or graphic writing, to the general
concept of writing? Is it a matter of taking into account the struc-
tural features of phonetic writing or of linguistic signs generally?
If so, how would grammatology differ from structural linguis-
tics, particularly the formalist analyses of a linguist such as Louis
Hjelmslev, who resolved to think of systems of signification exclu-
sively in terms of form and irrespective of the ‘substance’ of expres-
sion (OG, 61)? More generally, in what sense has the project of
general linguistics failed at precisely this project? Do quantitative
models of message, information and communication – such as
those provided by Claude Shannon – recommend themselves as
substitutes for more traditional linguistic models?

GOLDGABER 9781474438339 PRINT.indd 95 01/10/2020 09:30

 96 Speculative Grammatology

To specify the domain of writing, the grammatologist must gen-

erate a general concept of writing – to make explicit what we talk
about when we talk about ‘writing’. Curiously, as just mentioned,
Saussure’s successors in the field of structural linguistics appear to
offer a more grammatologically apt model of writing than Saussure
did himself, ones, moreover, that are more apt for quantification.
Curiously because, in the second chapter of Of Grammatology,
entitled quite simply, ‘Linguistics and Grammatology’, Derrida
directs the grammatologist not to follow in the path of Saussure’s
formalist successors, who appear to transcend the master’s teach-
ings by taking rather more seriously than did Saussure his own
dictum that ‘language is a form and not a substance’.3
Indeed, Derrida insists that the grammatologist must follow
Saussure precisely where the grammatologist might be expected to
part ways with him. We ought, he counsels, to resist the tempta-
tion to break with Saussure too soon, for then we will fall short of
the most general concept of writing. This strategy seems paradoxi-

FINAL
cal because, in the Course in General Linguistics, Saussure begins
his investigation of the linguistic sign by excluding writing. It is
as if writing needed first to be excluded or bracketed in order to
attain to an account of its generality.
To generalise writing, then, we must first exclude its ‘popular’
and even its technical conception. Saussure was right to claim
that writing misleads the linguists by providing a false image of
language and the linguistic sign. To let empirical writing guide our
intuition, then, will lead us, every time, to the wrong account of
the sign. Unfortunately, Derrida notes, Saussure does not consider
excluding only a certain image of writing – in particular phonetic
writing – while leaving open the possibility that excluding this
model will permit a better account of writing and the linguistic
sign to emerge. This is certainly a problem. Nonetheless, exclud-
ing a certain model of writing permits Saussure to generate an
account of the linguistic sign that is much closer, Derrida argues,
to the account of general writing that the grammatologist requires.
Furthermore, unlike many of those who followed him on the ques-
tion of the structure of language, Saussure’s inquiry into general
linguistics was also a radical investigation into the origin and
genesis of the linguistic sign. Indeed, he arguably subordinated the
question of structure to that of genesis. It will be the problem of
genesis, rather than structure, that permits the general concept of
writing, or arche-writing, to appear.

GOLDGABER 9781474438339 PRINT.indd 96 01/10/2020 09:30

 Rewriting the Course in General Linguistics 97

It seems, then, that to generate the most general concept of

the gramme, to get to arche-writing, we have only to follow in
Saussure’s footsteps and see where Saussure errs in his determina-
tion of writing. The grammatologist must demonstrate not only
that everything Saussure says about language is also true of the
writing he excludes, but also that the most general concept of
writing accounts for both language (speech) and writing. Language
and writing cannot be thought in terms of their radical heteroge-
neity but rather as sharing the same conditions: namely, arche-
writing. Indeed, as we will see, this is just what Derrida does in
Of Grammatology, and more specifically in the second chapter
entitled ‘Linguistics and Grammatology’. There, we can find the
passage between narrow writing and arche-writing via Derrida’s
reading of Saussure. Not, however, without difficulty.
Indeed, this passage has eluded (as we will see in a moment) many
of Derrida’s best readers. And for good reason. It is a tortuous path,
and one easily loses the trail. This is because, perhaps somewhat

FINAL
contrary to readers’ expectations, Derrida connects arche-writing
to the apparently transcendental problem of experience, which
seems orthogonal both to the goals of the linguist, who seeks a
scientific account of language, and those of the grammatologist,
who wishes to show writing’s generality and autonomy rather than
its dependence upon consciousness and transcendental subjectivity.
It seemed that the generalisation of writing would require excis-
ing it from the phenomenological and experiential horizon rather
than insisting on its pertinence. Yet Derrida unmistakably con-
nects writing to transcendental experience, and he does so on the
basis of Saussure’s analysis of the linguistic sign:

Origin of the experience of space and time, this writing of difference,

this fabric of the trace, permits the difference between space and time
to be articulated, to appear as such, in the unity of an experience. (OG,
66)

Even if, as Derrida writes, writing has always marked off a position
of technical exteriority (‘supplementarity’) vis-à-vis the human life
it purportedly served, arche-writing appears to lead writing back
to the interiority of (transcendental) life. This move seems, as
well, to erase the sort of conceptual advances that structuralism
achieves – namely of thinking about language and its structure
objectively, in particular in terms of linguistic value.

GOLDGABER 9781474438339 PRINT.indd 97 01/10/2020 09:30

 98 Speculative Grammatology

If arche-writing will not be limited to experience, but will be

absolutely general, as Derrida claims, the passage from narrow
writing to arche-writing requires a passage through the problem
of lived experience. To get to generalised writing, even one that
will eventually lose touch with the category of experience, we must
pass through the transcendental ‘text’.
What has been difficult to explain is, on the one hand, what
necessitates, justifies or prompts the intervention of the transcen-
dental problematic in Saussure’s account of language. Why exactly
does Derrida think that we must situate the transcendental ques-
tion as a moment in the grammatological inquiry? If it turns out
that Saussure’s analysis of the sign leads to these transcendental
questions, would this not count as a reason for the grammatologist
to reject Saussure’s analysis?
Certainly, the necessity of this ‘transcendental’ passage will not
always be obvious to the grammatologist, or it will only be justi-
fied in retrospect. Without this passage through the transcenden-

FINAL
tal, Derrida writes, we will likely confuse the ultra-transcendental
notion of ‘writing’ or ‘text’ (arche-writing) with a pre-critical
(‘objectivist’) sense of these terms (OG, 61). In any case, we
cannot, without traversing the transcendental text, understand
what Derrida means by ‘spacing’ and why this term is a cognate
for arche-writing. Spacing is, indeed, so far apparently from the
narrow notion of writing as to lose touch with it entirely.
Let us grant, then, that the path to the most general notion of the
gramme – arche-writing – requires a passage through Saussure’s
text and through the problem of the relation of the linguistic
sign to conscious experience more generally. ‘Requires’ may be
too strong a term. This suggests a grounding relation, such that
arche-writing depends upon the truth of Saussurean linguistics (its
being right or wrong). This cannot be right in any simple sense,
since we know in advance that Saussure is wrong – especially
about the nature and status of writing. It may be, however, that
he is crucially right about something else – and that this will
permit us to forge not only a notion of general writing but also to
demonstrate its absolute generality. Minimally, Saussure is right
that the popular conception of writing is wrong, and that this
conception of writing, insofar as it determines the meaning of
language, leads us to misunderstand language. He is also right,
according to Derrida – and herein lies his relative uniqueness in the
linguistic tradition – that the linguistic bond is constitutive and,

GOLDGABER 9781474438339 PRINT.indd 98 01/10/2020 09:30

 Rewriting the Course in General Linguistics 99

hence, that the two faces of the sign, the signifier and signified, are
indissociable.
If Saussure is useful to the grammatologist, it is, then, to extent
that his work is already deconstructive, seeking to oust traditional
concepts that have made it impossible to think about the nature
of linguistic signs as anything but secondary and derivative with
respect to an original (non-linguistic) consciousness. In insisting
on the originality of the linguistic sign vis-à-vis consciousness,
Saussure’s approach is proto-grammatological, soliciting the
inherited concepts of a tradition in order to conceive what they, in
their present formation, exclude. In this respect, finally, Saussure’s
treatment of writing will be an index of the extent to which this
solicitation (of the tradition) has failed – of the extent to which the
tradition still solicits him and therefore limits his analysis.

Rereading ‘Linguistics and Grammatology’

FINAL
In the second chapter of Of Grammatology, Derrida invites the
grammatologist to turn to Ferdinand de Saussure’s Course in
General Linguistics (Cours de linguistique générale) to develop
a general concept of writing. The choice seems obvious at first,
but becomes increasingly puzzling. It is true that the founders of
grammatology are apparently in the same position with respect
to writing as Saussure was with respect to language. Founding
a general science of writing will entail an inquiry into the nature
and limits of writing – and hence, the grammatologist may find
Saussure’s foundational text instructive. Yet, prima facie, the
grammatologist is sure to find Saussure’s treatment of writing not
just unsatisfactory, but deplorable.
Derrida is clear about the fact that the founding gesture of the
father of modern linguistics was to exclude writing. Saussure,
Derrida emphasises, begins by uncritically defining writing as
‘exterior’ to language proper, but bases this exclusion on the most
cursory, uncritical consideration of writing. Indeed, he limits his
attention exclusively to phonetic writing. For Saussure, ‘writing’
always designates the ‘sign of the [linguistic] sign’, the ‘represen-
tation’ or ‘image’ of speech. Had he but considered examples of
non-phonetic writing, from so-called ideographic writing to math-
ematical systems of notation, Saussure could not have so easily
characterised writing as a derivative, inert supplement to speech.
If we look just to those places where Saussure attempts to sharply

GOLDGABER 9781474438339 PRINT.indd 99 01/10/2020 09:30

 100 Speculative Grammatology

distinguish language (proper) from writing, we find so much con-

tradiction that one can easily doubt his value as a linguist.
Saussure’s exclusion of writing is not only a problem for gram-
matology but for general linguistics. He risks the latter’s claims to
generality. Even if phonetic writing were the only and universal
instance of writing (which it is not), why, Derrida asks,

[d]oes a project of general linguistics, concerning the internal system in

general of language in general, outline the limits of its field by exclud-
ing, as exteriority in general, a particular system of writing, however
important it might be, even if it were to be in fact universal? (OG, 39)

If one were trying to find the limits of language – that is, to mark
off the regional boundaries of a science – why not take better
care to ensure that systems of ‘graphic writing’ were not in fact
a subset of linguistic systems? Even if Saussure’s intuitions were
that language had a distinctive form and function, which it did

FINAL
not share with phonetic writing, why not wait to confirm these
intuitions by first establishing this form and function and then
clarifying the status and nature of writing? Why not consider
that if a certain model of writing has given us the wrong image
of language, this model also gives us the wrong model of writing?
No doubt this uncritical and unsupported exclusion of writing
evidences what Derrida calls Saussure’s ‘phonocentrism’. But the
latter is not simply another word for blind prejudice. Saussure,
who was otherwise so alert to the problem of inherited preju-
dices, is not motivated to resist a philosophical tradition that has
systematically elevated speech over writing. The linguistic sign is
phonocentric not because it excludes writing, but because, as we
shall see, it positively affirms the view that sound and thought are
essentially, constitutively entangled.
Based on the rationale that Saussure offers, one might well
conclude that the Cours de linguistique générale will not be pro-
ductive for grammatology, and for similar reasons that it will
be of limited use to the linguist. Would it not, then, be more
productive to begin again? And with a more promising text?
For example, we might start with Hjelmslev’s Prolegomena to a
Theory of Language (1953). The latter’s theory of glossematics,
as a general semiotics, aimed to abstract the form common to all
linguistic systems. Hence, it included rather than excluded writing,
and introduced the generic term glosseme to designate any linguis-

GOLDGABER 9781474438339 PRINT.indd 100 01/10/2020 09:30

 Rewriting the Course in General Linguistics 101

tic signifier. Indeed, glossematics seems, in every respect, closer to

grammatology than Saussurean linguistics. If Saussure gave us the
tools to think of all systems of signification in formal terms (as an
open system of values or differences) – and if, in terms of structure
and function, systems of writing are formally indistinguishable
from systems of signification – then why does Derrida tarry so
long with the archaic Cours de linguistique générale? Why does
he stage partings and breaks, only to return again to Saussure’s
account of the sign?
Derrida is hardly unaware of Hjelmslev or other formalist
approaches. Indeed, he indicates that, with respect to Saussure,
these approaches, in ‘trying to go beyond the master’, fall short.
Indeed, Saussure tells us why. They work with and implicitly rely
upon the model of popular writing. Therefore, they miss the con-
stitutive relation between the signifier and signified. Instead of
thinking the genesis of linguistic form, they abstract from the
constituted form.

FINAL
Saussure insists upon the entangled, mutually constitutive rela-
tion of the signifier and the signified, while the formalists treat
language like a system of writing (where this constitutive relation
is, according to Saussure, absent). If Saussure is wrong about the
nature of writing – if it too is constitutively entangled with the
signified – then Saussure’s analysis of the structure and genesis
of the linguistic sign will help the grammatologist understand
the nature of writing. But to see how, we must think with and
against Saussure; we must stay with him longer than seems reason-
able or promising.
Saussure’s analysis of the sign is radical in the sense that – unlike
the glossematics approach – it does not subordinate the question of
genesis to that of structure; his analysis of the sign’s form is always
articulated with the question of the genesis of this form. Saussure
grants speech primacy over writing not because of the purported
order of linguistic acquisition – this would ground writing’s deriv-
ativeness in a contingent fact – but on the basis of his analysis of
the origin of the dual-sided sign. For Saussure, as we shall see,
the constitutive role of sound is decisive. If Saussure is wrong to
centre sound – that is, if the grammatologist should break with his
phonocentrism – they should continue, with Saussure, to think the
genesis and structure of writing together.
In the remainder of this chapter, I will track Derrida’s writing
of arche-writing into the founding text of linguistics. In my

GOLDGABER 9781474438339 PRINT.indd 101 01/10/2020 09:30

 102 Speculative Grammatology

reconstructive reading, I will focus on the grammatological signifi-

cance of Saussure’s phonocentrism. Why did Saussure think that
sound was originally constitutive of the sign and what happens
to our account of ‘language’ when sound is decentred? How does
rejecting Saussure’s exclusion of writing modify or transform the
project of a general linguistics? What is the extent or scope of
general writing? What do terms such as arche-writing, trace and
spacing do with or make of the narrow, standard understanding of
writing and of Saussure’s account of the linguistic sign?

The Enigma of Arche-writing

I have read and re-read these passages many times. I still find them
enigmatic.
J. Hillis Miller

I want to begin this proposed rereading by remarking the enig-

FINAL
matic appearance of the unfamiliar term arche-writing in the text.
‘Arche-writing’ describes the ‘gramme’ of grammatology and, like
its cognates, trace and spacing, ‘communicates’ with the ‘vulgar
concept of writing’.4 This communication enables the regulated
extension and generalisation of writing, while assuring that,
however much arche-writing displaces everyday intuitions about
what writing is, and however general it becomes, it will account
for writing in the ‘narrow’ sense. Yet arche-writing often appears
to lose touch with narrow writing entirely. This is nowhere more
palpable then in the connection of arche-writing to what Derrida
calls spacing.
In addition to being the possibility of ‘the graphie, a possi-
ble unit of graphic expression’ (OG, 65), arche-writing, we read,
is also the articulation of time and space, which Derrida calls
spacing. Spacing, in turn, is also the ‘articulation of the living on
the non-living’ – which seems to be a different, but related sort
of articulation (OG, 65). Why should we use ‘writing’ to speak
generically about time and space and the relation of life to its
others? What justifies these dramatic leaps away from the narrow
linguistic context?
Of course, it is true, as Derrida writes in Of Grammatology’s
opening pages, that the inflation of writing has been, for some
time now, well underway. But the work of the grammatologist is
not to ‘give in to this movement’ but to justify it. The point is not

GOLDGABER 9781474438339 PRINT.indd 102 01/10/2020 09:30

 Rewriting the Course in General Linguistics 103

to assume that writing is pertinent beyond its narrow linguistic

domain, but to demonstrate its pertinence, to keep the lines of
communication between narrow and general writing open and
clear.
The puzzle for Of Grammatology’s readers, however, is not just
the question of what terms such as arche-writing mean but how
Derrida gets the words to mean what they mean. As J. Hillis Miller
notes in a close reading of several key pages in the chapter of Of
Grammatology under consideration, ‘trace’, like ‘writing’, is a
perfectly ordinary English word, with a more or less determinable
meaning. But it comes to mean something else, something seem-
ingly extraordinary in Derrida’s hands. ‘Just what does Derrida
make the word “trace” mean?’, Miller asks:

Why does he twist just this word? Here my difficulties begin. I do not
think what Derrida means by the word is at all self-evident or clear, in
spite of the fact that everyone knows it is a key Derridean word. The

FINAL
word ‘trace’ can be traced all through Chapter 2, like a red thread in
a tapestry. Read Chapter 2 again for yourself, dear reader. I’ll bet you
will still find what Derrida means by ‘trace’ obscure, occulted. Derrida
says as much . . . I have read and re-read these passages many times. I
still find them enigmatic.5

Now Miller, I should underscore, is not just one reader of Derrida

among others; he is one of Derrida’s best readers. Perhaps because
he does not doubt his reading abilities, and nobody else does
either, Miller can afford to say what often goes unsaid among
Derrida’s readers. We, the readers of Of Grammatology, are often
unsure what these words mean.6
What Miller says of ‘trace’ can be said of ‘writing’. Again, the
terms are cognates, in the way that sign and speech are cognates
in Saussure’s texts. Writing, of course, is a term familiar enough –
and certainly indispensable to grammatology, a science of writing.
But arche-writing is no longer familiar; Derrida admits that the
term cannot be found in the mundane text of any science, or even
in the text of metaphysics. Indeed, ‘no metaphysics can describe it’
(OG, 244). None of this should be taken to mean that these terms
have no meaning. The point it to remark that these terms remain
enigmatic, despite over fifty years of critical readership.
Miller formulates a critical interpretive question: ‘what does
Derrida make [these words] mean?’ To which I would like to

GOLDGABER 9781474438339 PRINT.indd 103 01/10/2020 09:30

 104 Speculative Grammatology

add: how does Derrida make these words mean what they mean?
How are they wrought – weaned from their ‘narrow’, ‘famil-
iar’ designations and ‘generalised’? I will begin by considering a
complex passage in which arche-writing appears, in order to take
the measure of the interpretive task I have set. The one with which
I begin offers, apparently, a definition of arche-writing:

Arche-writing, at first the possibility of the spoken word, then of the

‘graphie’ in the narrow sense . . . this trace is the opening of the first
exteriority in general, the enigmatic relationship of the living to its
other and of an inside to an outside: spacing. The outside, ‘spatial’ and
‘objective’ exteriority which we believe we know as the most familiar
thing in the world . . . would not appear without the gramme, without
différance as temporalization, without the nonpresence of the other
inscribed within the sense of the present. (OG, 71)

We note at the outset that to understand arche-writing we will have

FINAL
to understand it in terms of a network of other, more or less unfa-
miliar terms, including gramme, graphie, spacing and temporalisa-
tion. What, then, according to this definition, is arche-writing?
Arche-writing names a common ‘possibility’; first of speech, then
of writing in the narrow sense of the graphie – a possible unit of
graphic expression – and also perhaps of language in general. But,
moving on, arche-writing is equally the possibility of appearance
and of general relationality (‘of an inside to an inside’). Arche-
writing is ‘spacing’, which Derrida defines a bit further on in the
text as ‘the becoming-space of time and the becoming-time of
space’. But what is the logic that binds all these disparate relations?
Arguably, what is most surprising in this passage (which is
merely exemplary) is the claim that time and space, language and
writing refer to a common possibility, namely arche-writing. The
latter seems to function as the name of a hitherto undiscovered
principle, gathering together seemingly disparate phenomena not
previously held together by a single logic.7 Because we are missing
a sense for what logic ‘hold[s] together . . . the disparate’, this
definition of arche-writing delivers something like the shock to
thought that Michel Foucault described experiencing while reading
the taxonomic entry for ‘animal’ in Borges’s fictional ‘Chinese
encyclopedia’, the Celestial Emporium of Benevolent Knowledge.8
Echoing Foucault, we might ask of Of Grammatology’s entry for
arche-writing: How is it possible to think these things together?

GOLDGABER 9781474438339 PRINT.indd 104 01/10/2020 09:30

 Rewriting the Course in General Linguistics 105

Derrida presents these disparate attributes of arche-writing as

if summing up what has been previously established. Yet, as with
the logical connections between all that arche-writing is said to
make possible, the connection between arche-writing’s debut in
Of Grammatology and the discussion preceding it is unclear and
the reader may be unprepared for it.
The proximate context to the quote above is a discussion
of French linguist Roman Jakobson and his description of the
structure of speech. According to Jakobson, speech cannot be
understood – as it usually is – in terms of a continuous (temporal)
stream; speech is made up of ‘bundled’, ‘distinctive features’ – it
has a ‘manifestly granular structure and is subject to quantile
analysis’ (OG, 69–70). These distinctive phonological elements
or bundles are no more spatial and simultaneous than temporal
and successive. If this is right, Derrida argues, then phonologists
should renounce any sharp distinction between the spoken and
written word; succession and simultaneity, temporality and spati-

FINAL
ality characterise both speech and writing.
Jakobson claims to be describing speech, but he may just as
well have been describing writing. Having apparently established
just this much, namely, that speech and writing are structurally
indistinguishable but that phonologists are reluctant nonetheless
to relinquish a distinction that their own analysis undermines,
Derrida concludes – somehow – that ‘signification is a priori
written . . . in a “spatial” element that is called “exterior”’ (OG,
70). He then reserves the term ‘arche-writing’ for the ‘opening
of the first exteriority in general’. If speech and writing share
a common structure, why call both of them ‘writing’ and why
designate exteriority and its opening as ‘writing’ (rather than
‘speech’)? What or where is the ‘first exteriority’ and why should
the ‘first exteriority’ be bound up with questions or problems of
signification?
It is true that if we think of writing as a concrete possibility for
a subject, the possibility of writing down and hence exterioris-
ing and materialising particular intentions or thoughts, it might
make a kind of analogical sense to denote the relation in which
any inside is expressed in any outside with the name ‘writing’. On
this interpretation, arche-writing would name the generic possi-
bility of this particular kind of exteriorisation or expression. But
one should also note that speech apparently performs the same
sort of expressive function, exteriorising thought. What, then,

GOLDGABER 9781474438339 PRINT.indd 105 01/10/2020 09:30

 106 Speculative Grammatology

do we gain from the ‘grammatological’ substitution that Derrida

proposes?
Perhaps the substitution of ‘writing’ for ‘speech’ is strategic or
rhetorical. It decentres the latter and thereby loosens the doxa that
leads us, wrongly, to think of writing as always exterior and deriv-
ative to speech. Indifferent between speech and writing, arche-
writing would name the ‘opening of the first exteriority in general’.
Countering uncritical habits is certainly all to the good. But ‘first
exteriority in general’ seems a poor translation for ‘expression’.
Matters are further complicated by the fact that both Saussure and
Derrida explicitly reject the view that language can be thought in
expressive terms. Language does not express ideas, thoughts or
meanings – it is constitutive of these.
Indeed, Derrida emphasises that even as Saussure insists –
against the tradition he inherits – that expression is the wrong
model for linguistic function, he accepts, without expending any
critical energy, that it is the right model to understand writing

FINAL
(OG, 11). For Saussure, signifier and signified, word and thought,
are indissociable. The linguistic bond is constitutive of what it
relates. From this perspective, two things seem to be at stake in
Derrida’s substitution of writing for Saussure’s speech: 1) cor-
recting Saussure’s uncritical exclusion of writing from language,
and 2) thinking the relationship of linguistic ‘inside’ and ‘outside’
differently, otherwise than as expression.
Saussure’s critique of expressivism (as a theory of linguistic
function) is primarily motivated by his unorthodox view of the
linguistic sign’s structure. He believed that the two aspects of
the sign, the signifying and signified elements, are not mutually
exterior and independent. If form follows function, then the func-
tion (of the linguistic sign) cannot be the expression of a signified
element that is already constituted elsewhere. The function of the
sign must rather be, Saussure speculated, the individuation and
articulation of the signified element. That the two faces of the
sign are indissociable means that thought’s exteriority to language
is an illusion. ‘In language, one can isolate neither sound from
thought, nor thought from sound; one arrives at neither except by
abstraction.’9
Yet as Derrida points out, despite Saussure’s best efforts to
explain the structural entanglement of signifier and signified,
or what Saussure calls thought-sound (pensée-son), via a novel
account of the sign-form, Saussure’s account ends up reasserting

GOLDGABER 9781474438339 PRINT.indd 106 01/10/2020 09:30

 Rewriting the Course in General Linguistics 107

the priority and independence of the signified element. In other

words, his theoretical efforts fall short of their aim: theorising the
constitutive entanglement of signifier and signified. This failure,
Derrida suggests, is related to Saussure’s failure to adequately
interrogate the relation between speech and writing. The entangle-
ment of signifier and signified, essential to the linguistic sign, may
also characterise the relations of speech and writing. In any case,
Derrida introduces arche-writing as the name for this other rela-
tion: the entangled, indissociable relation of thought to language,
and of speech to writing.
If arche-writing registers the need for a novel schema of entan-
glement capable of specifying the relations between speech and
writing, between signifier and signified, and between language
and thought, Derrida’s definition of arche-writing still reads as
too inflationary, over-generalising and hyperbolic. Rather than
speak specifically about the relation of thought to language –
the problem of the ‘articulation’ of thought-sound in Saussure

FINAL
– Derrida describes arche-writing as ‘the enigmatic relation of the
living to its outside’, and then generalises this relation to describe
the relation ‘of an[y] inside to an[y] outside: spacing’.
What justifies this apparent leap from the linguistic context,
and the limited problem of whether or not writing is expressive
or constitutive with respect to what it signifies, to the general
problem of the relation of life to its ‘outside’? The latter sounds
like a problem facing a systems theorist attempting to think the
relation of the organism to its environment, which seems different
from Saussure’s problem. It may be fascinating to speculate about
the extent to which the problems are the same. More interesting
still is to consider how these disparate problems may be subsumed
under the heading of grammatology, but again the logic guiding
Derrida’s analysis is far from explicit. Does arche-writing say that
all things are entangled (with their outside) in the way that thought
is entangled with language and speech is entangled with writing?
But why should we think that?
These questions are not meant to impugn Derrida’s reading
of Saussure, nor to imply that these are so many unanswerable
questions. Rather, these questions underscore that the fame of
Derrida’s reading of Saussure has not resulted in the sort of insight
into the key concept of Derrida’s early work, arche-writing, that
we might have expected. If the meaning of arche-writing were well
known or well attested, the sort of reconstructive reading I propose

GOLDGABER 9781474438339 PRINT.indd 107 01/10/2020 09:30

 108 Speculative Grammatology

here would be by now redundant. It would be possible to answer

the question of what the project of grammatology consists of, and
to understand its pertinence or obsolescence for today. However, I
am aware of no reading that satisfactorily makes explicit the logic
that leads from the seemingly limited problem of writing’s linguis-
tic inclusion or exclusion to an enlarged notion of ‘writing’ that
encompasses seemingly fundamental ontological relations.
The central interpretive problem, I think, can be stated in the
following way: what connects arche-writing both to language and
to what Derrida calls spacing? The meaning of the grammatologi-
cal project, and in particular its claims about the absolute general-
ity of writing, hinge upon the answer to this question.
I am, of course, not the first person to pose some version of this
question. Indeed, it has taken on some urgency recently with the
materialist and realist turns in continental philosophy. In Telling
Flesh (1997) Vicki Kirby offers a reading of Saussure’s account of
the sign as evincing the indissociably entangled relations between

FINAL
language and world. But here, Kirby’s reading departs from
Derrida’s reading in Of Grammatology by attempting another
reading of Saussure and his successors, one that will make clear, in
the way that Derrida apparently does not, how we can move from
an account of the Saussurean sign to an account of language’s
entanglement with its ‘others’, and, finally, to the generality of
entanglement. The supplement to grammatology that Kirby’s
innovative reading offers, however, seems to attest to the incom-
pleteness of the original, as if Derrida’s claim about the generality
of writing was a speculative promissory note rather than the con-
clusion of a philosophical argument.
Kirby’s reading has influenced Karen Barad, among others, who
generates, via a reading of Niels Bohr and contemporary experi-
ments in quantum physics, an account of generalised entanglement
or ‘space-time mattering’ – without, for that matter, connecting
this ‘space-time mattering’ with the ‘vulgar’ concept of writing.
Both Barad’s and Kirby’s accounts of generalised entanglement
are arguably in the spirit of grammatology, while giving up on
the letter of the text. It seems that arche-writing falls short, as
Saussure’s account of the linguistic sign did before, of thinking both
entanglement and its generality, which would support Catherine
Malabou’s influential argument about the limits of grammatology.
Malabou has argued that supplementary readings such as
Kirby’s cannot vindicate Derrida’s claims about general writing.

GOLDGABER 9781474438339 PRINT.indd 108 01/10/2020 09:30

 Rewriting the Course in General Linguistics 109

The latter was ‘programmed to fail’.10 The warrant for Derrida’s

generality claims is missing because this generalisation is
impossible – on account of the limitations of the graphic model of
writing with which Derrida begins. His starting point in empiri-
cal writing blocks the route to writing’s generalisation projected
in Of Grammatology.11 Kirby, in a recent response, argues that
Malabou’s account of Derrida’s failure to generalise writing is
based on the (false) assumption that Derrida’s ‘graphematic struc-
ture’ was intended to answer the demands of representational
accuracy.12 That is, on Kirby’s reading, we should not expect
arche-writing to look like (empirical) writing. It is rather that the
‘riddles of writing’ (not least of all the relation between repeatable
form and material substrate) – as with the ‘riddles’ of the two
‘faces’ of Saussure’s linguistic sign – are generalised, and hence
gesture towards the thought of generalised entanglement. While I
agree with Kirby’s contention that arche-writing does not ‘repre-
sent’ graphic writing, this does not release us from the obligation

FINAL
to account for the difference between narrow and general writing.
Without such an account, Derrida’s claims about the generality
of writing appear as ungrounded speculative gestures in the guise
of rigorous philosophical claims. Despite his philosophical claims
about the impossibility of strict literality, Derrida everywhere
expends remarkable efforts to take a text at its word. It is therefore
the letter – rather than the spirit – of Of Grammatology to which
we must attend, and this is a text that insists on demonstrating the
necessary generality of writing.
Martin Hägglund, in Radical Atheism (2008), has perhaps gone
the furthest in defending Derrida’s generality claim. He argues
that the generality claim gets its necessity from the form of arche-
writing. Hägglund’s reconstruction of Derrida’s argument moves
from a stipulated generalised ‘finitude’ to the form that generalised
finitude logically imposes: the material trace. For anything to be,
it must persist, and persistence entails material supports. The trace
is necessarily materially inscribed; it is what Malabou would call
graphic. The trace does not persist forever, however: this is because
the ‘arche-materiality’ of the trace entails its future erasure.
Whether or not Hägglund’s argument succeeds in justifying the
generality of the trace, it is anything but a generalisation of writing.
His argument moves from the stipulated generality of finitude to
the arche-materiality of the trace – exemplified by the image of
inscription. That is, it reverses the trajectory of Derrida’s own

GOLDGABER 9781474438339 PRINT.indd 109 01/10/2020 09:30

 110 Speculative Grammatology

argument, which moves from narrow writing to arche-writing.13

However, even if Hägglund accurately reconstructs Derrida’s
argument for the generality of writing, this argument seems to
entail precisely the limits on writing’s generality that Malabou
diagnoses.
As soon as we think the trace in terms of graphic inscription, it
cannot be generalised or enlarged, and it cannot support a specula-
tive argument about the ultra-transcendental necessity of finitude.
This is because matter will always be the condition of the trace,
and as condition, it will be logically exterior to it – as the appeal
to a material substrate suggests. We will return to Malabou’s
diagnosis of arche-writing’s ‘graphism’ in the following chapter.
For the moment, I take it that this sort of debate, between some of
the most interesting and innovative materialist readers of Derrida,
would not be possible if the meaning of arche-writing and its rela-
tion to narrow writing were clear and explicit.
In the next sections, I will challenge Malabou’s claim that no

FINAL
account of writing’s generalisation exists because no passage from
narrow to general writing (or arche-writing) is possible. Or rather,
I will to take her claim as a challenge to hazard this passage.
Malabou’s argument assumes that arche-writing, the most general
concept of the gramme, does not transform our common-sense
understanding of writing as graphie, a ‘possible unit of graphic
inscription’, in a way that would make it generalisable. Her argu-
ment, I believe, is persuasive in the absence of an alternative, non-
graphic account of arche-writing. The aim of this chapter, then,
is to offer just such an alternative account, one that distinguishes
between arche-writing (as generalised writing) and the graphie
(OG, 46).
How does arche-writing compel us to revise what Derrida calls
a ‘popular’ – or, as Malabou suggests, ‘graphic’ – conception of
writing? The fact that Derrida sides with Saussure in insisting that
the popular conception of writing does not offer the right model
for the linguistic sign gives us at least one reason in favour of
thinking that arche-writing may entail a substantial revision to this
popular, graphic conception of writing (OG, 60).

GOLDGABER 9781474438339 PRINT.indd 110 01/10/2020 09:30

 Rewriting the Course in General Linguistics 111

Language and Appearance: Saussure’s ‘Transcendental

Semiotics’
For the grammatologist, Saussure’s exclusion of writing from lan-
guage is problematic. The reasons he gives are puzzling. On the
one hand, Saussure seems to open up the possibility of gramma-
tology by insisting, famously, that ‘language is a form and not a
substance’.14 On the other, he (fore)closes this possibility by insist-
ing that this form is indissociable from the phonetic substance of
the signifier. The essentially phonic nature of the sign, in turn, is
related to its entanglement with the signified element, which has a
psychic provenance.
Saussure’s structural analysis would seem to obviate any refer-
ence to experience, or its conditions of possibility. The Saussurean
sign, according to Thomas Pavel, is that of an open system of dif-
ferential marks.

FINAL
Saussurean signs . . . constitute an open network where each position
is defined and valued only by its non-identity with the surrounding
positions . . . The linguists have identified differential (contrastive)
networks precisely there where language, disposing of limited means,
must arrive at a maximal rendering.15

This description of the Saussurean sign is, however, almost cer-

tainly wrong – at least as a reading of Saussure.16 Pavel con-
flates Saussure’s definition of writing (or a system of writing) with
his definition of a linguistic sign, unwittingly erasing the differ-
ence between linguistics and grammatology, between speech and
writing, which Saussure insisted upon.
Wherein lies the difference? For Saussure, a system of writing is
solely defined in terms of its contrastive value. The terms of such a
system are purely arbitrary and purely quantitative. This is not the
case with linguistic signs, which are dual-sided, composed of both
signifier and signified. Saussure likens the dual-sided linguistic sign
to two sides of a sheet of paper. This means that linguistic signs
cannot be defined in terms of linguistic value, grounded only in
relations of difference; signs are rather the condition for linguistic
value. The contrastive relations between signs produce linguistic
value. Insofar as the function of language has to do with contras-
tive elements, it behaves like a system of writing. But such contras-
tive relations presuppose the existence of signs.

GOLDGABER 9781474438339 PRINT.indd 111 01/10/2020 09:30

 112 Speculative Grammatology

Saussure offers an image that he hopes will help us understand

the distinction he is drawing between the relations that constitute
linguistic value and those that originally constitute the linguistic
sign. If language were like a sheet of paper (with its two sides),
dividing this paper into strips would produce linguistic value
through the contrastive relations between the strips. However,
this possibility is dependent on the more originary relationship
that constitutes the indissociable linguistic sign.

Language is a system of interdependent terms in which the value of

each term results solely from the simultaneous presence of the others
. . . Putting it another way – and again taking up the example of the
sheet of paper that is cut in four, it is clear that the observable relation
between the different pieces A, B, C, D, is distinct from the relation
between the front and back of the same piece as in A/A’.17

What can this account of language tell us about writing and the

FINAL
relation between language and writing? For Saussure, systems of
writing are like language (or perhaps language is like writing)
insofar as the ‘value of each term results solely from the simulta-
neous presence of the others’. But writing differs from language
insofar as its terms (unlike language) may be analysed solely in
terms of linguistic value. This is what makes systems of writing
relatively simple and characteristically substitutable, one for the
other. Language is different; we cannot understand linguistic form
solely by considering linguistic value. This would be to mistake,
as Pavel does, language for writing, and it misses the fact that
the linguistic sign is constitutively dual-sided. It will turn out,
according to Saussure, that we cannot account for the form of the
linguistic sign without including the perspective or point of view
of the speaker.
In his classic text, ‘Saussure and the Apparition of Language’
(1976), Samuel Weber connects Saussure’s account of linguistic
form to temporalisation and spatialisation. On Weber’s reading,
the latter rejects inherited views that define language as ‘a rep-
resentation or expression of thought by sound’, thinking of lan-
guage instead as an intermedium. Language moves between the
two spheres of sound and thought, delimiting each sphere through
decomposition – before reorganising these intervals, primarily
through relations of opposition.18
Linguistic intermediation articulates thought by also articulat-

GOLDGABER 9781474438339 PRINT.indd 112 01/10/2020 09:30

 Rewriting the Course in General Linguistics 113

ing sound and then binding the two. Weber specifies, in a valuable
footnote, that:

Among the meanings of intermedium, the O.E.D. lists: ‘1. Something

intermediate in position: an interval of space. 2. Something intermedi-
ate in time; an interlude; an interval of time’ . . . The first two mean-
ings are considered to be obsolete. The confusion of the spatial and
the temporal interval, combined with the fact that these meanings are
historically obsolescent renders the word an appropriate sign for the
ambiguous position [in Saussure] of ‘la langue’.19

Here, Weber makes explicit how the ‘confusion’ of the spatial and
the temporal interval, or what Derrida calls ‘spacing’, emerges as
central to Saussure’s account of language. That is, it is Saussure
(and not Derrida) who demands that we think la langue in terms
of intermediation, temporalisation and spatialisation or spacing.
For Saussure, the effects of linguistic intermediation are spatial

FINAL
intervals (in thought) and temporal intervals (in sound).

[A]gainst the floating realm of thought, would sounds by themselves

yield pre-delimited entities? No more so than ideas [thoughts]. Phonic
substance is neither more fixed nor more rigid than thought; it is not a
mold into which thought must of necessity fit, but a plastic substance
divided in turn into distinct parts to furnish the signifiers needed by
thought. The linguistic fact can therefore be pictured in its totality –
i.e. language – as a series of contiguous subdivisions marked off on
both the indefinite plane of jumbled ideas and the equally vague plane
of sound.20

The intermediation of language does not just produce intervals of

sound (time) and intervals of thought (space). It produces these
intervals, Weber emphasises, as entangled – or, as ‘the confusion
of the spatial and the temporal interval’.
The dual-sided linguistic sign is the reciprocal articulation of
sound and thought. These entangled intervals are then rearticu-
lated through relations of opposition. Though it is often said that
Saussurean signs are constituted through relations of opposition,
Patrice Maniglier argues that this interpretation mistakes what
Saussure calls the ‘post-elaboration’ of the sign for the genesis
of the sign. Saussure’s account makes clear that ‘the game of
opposition is played between terms already given . . . rather than

GOLDGABER 9781474438339 PRINT.indd 113 01/10/2020 09:30

 114 Speculative Grammatology

supposing that [this game] constitutes signs . . . it redetermines

them’.21
The game of ‘redetermination’, which produces contrastive
terms or values, is not limited to speech. Indeed, Saussure defines
writing as a system of purely contrastive terms. This means, accord-
ing to Maniglier, that most of the features that we associate with
Saussure’s account of linguistic signs – differential elements, terms
constituted through oppositional relations – are actually attributes
of what he calls ‘writing’. What makes Maniglier’s and Weber’s
readings of Saussure different from more familiar accounts, such
as Pavel’s, is that both recognise Saussure’s rigorous distinction
between the constitution of the sign and its redetermination in the
game of linguistic value.
Language, for Saussure, cannot be defined by its expressive
or communicative function with respect to thought because it is
constitutive of that which is expressed or communicated. Indeed,
Saussure argues that what individuates a thought is constitutive of

FINAL
the thought as such.

Psychologically our thought – apart from its ‘expression’ in words – is

only an amorphous and indistinct mass. Philosophers and linguists
have always agreed in recognizing that without the help of signs we
would be unable to distinguish ideas in a clear and consistent fashion.22

Interestingly, Weber remarks, Saussure appeals to tradition at the

precise moment he breaks with it:

If it is doubtless true, as [Saussure] asserts, that ‘philosophers and

linguists’ have almost always concurred in attributing to language
an indispensable function ‘in distinguishing two ideas in a clear and
constant fashion,’ [philosophers] have still distinguished between the
process by which ideas are distinguished from one another, involving
language, and the process by which they are constituted in themselves,
which has been construed as transcending language, de jure if not de
facto.23

In other words, ‘the tradition’ may grant that, without the right
sort of identity conditions – one of which is language – we could
not distinguish thoughts clearly and distinctly. However, Weber
points out, ‘the tradition’ is not likely to agree that the condi-
tions for distinguishing thoughts are the same as those constituting

GOLDGABER 9781474438339 PRINT.indd 114 01/10/2020 09:30

 Rewriting the Course in General Linguistics 115

thoughts. If there are thoughts to be distinguished, the assumption

is that there are differences according to which they may be distin-
guished. Despite what Saussure claims, the idea that articulation,
differentiation and constitution are the same seems to be uniquely
Saussure’s.
For Saussure, intermediation ‘decomposes’ or articulates thought
and sound into intervals. ‘[W]ithout language, thought is a vague
uncharted nebula. There are no pre-existing ideas, and nothing is
distinct before the appearance of language.’24 Indeed, for Saussure,
language – coeval with its ‘apparition’ – is the condition of possibility
for thought’s appearance. Language describes the reciprocal demar-
cation of units. Thought, for Saussure, is not yet the elaborated, sig-
nified element that will emerge through what Maniglier describes as
‘the game of opposition’. Prior to its individuation, thought can only
refer to something like inchoate sense variations, albeit variations not
yet spaced or intermediated by language. We must imagine, it seems,
something like a language without syntax. If we grant Saussure this

FINAL
much, however, it is still unclear why sound – as opposed to another
sense modality or signifying ‘substance’ – should do the work of
individuation? Is sound exemplary, for Saussure, of the signifying
substance in general, or does it have a necessary constitutive role?
By all indications, it is the latter. ‘The original, indissociable bond is
between thought and sound [pensée-son].’25
Weber notes that a philosophical tradition, which extends back
to Aristotle, takes natural languages to be grounded in shared
(non-linguistic) ideas. This tradition does not give language a
radically constitutive role vis-à-vis thought. Indeed, it assumes, as
Weber writes, that the processes constituting thought are de jure
language-transcendent. However, as we will now see, even if this
view is dominant in the Western philosophical and linguistic tradi-
tion, there are other, ‘minority’ traditions that make Saussure’s
position appear far less radical. That is, Saussure’s position may
not be as novel as Weber’s reading suggests – something that
Derrida also seemed to recognise. Recalling this other philosophi-
cal tradition will help explain why Saussure assigns sound the role
that he does.
* * *
Thinking of language in radically constitutive terms was not
foreign to Saussure’s philosophical milieu. Indeed, a lively post-
Kantian lineage assigned to the form of the linguistic sign the

GOLDGABER 9781474438339 PRINT.indd 115 01/10/2020 09:30

 116 Speculative Grammatology

same transcendental role that Saussure assigns it. This milieu was
apparently familiar to Derrida as well, who reminds us of it in the
course of his reading of Saussure, via a quote from Main de Biran:
‘the word is intention “sensibilized”’ (OG, 73).26 For this tradi-
tion, sensibilisation, and not expression, is the original function
of language.
For Herder, von Humboldt and other figures within post-
Kantian German philosophy, language had a distinctive transcen-
dental function. As Michael Forster writes, this tradition shared
the view that ‘[1] one cannot think unless one has a language and
[2] one can only think what one can express linguistically’.27 The
second, ‘expressibility’ thesis is grounded on the first, ‘constitut-
ing’ thesis. Language, or linguistic form more precisely, makes
thought possible in making it sensible. Language renders thought
sensible by provisioning or shaping it with the right kind of form.
‘Transcendental semiotics’ is how Markus Messling and Kurt
Mueller-Volmer (2003) refer to the view that language provisions

FINAL
thought with the form of sensibility. Humboldt’s account of the
relation between language and thought On Thinking and Speaking
(Ueber Denken und Sprechen) (1795/6), they argue, develops a ‘tran-
scendental semiotics’, which anticipates by over one hundred years
the claims made in Saussure’s Course. Humboldt took language to
be central to what he called ‘reflection’. ‘Thinking consists in reflect-
ing, that is, in the act by which the thinking subject differentiates
itself from its thought.’28 Language and linguistic form, in particu-
lar, are the transcendental conditions for this reflective activity.
Language, Humboldt writes, consists in:

arresting the continuous flow of impressions in order to . . . com-

prehend this something as a separate ‘unit’ (Einheit), and set it as an
object over against our thinking activity . . .
[T]he mind can now proceed to compare several of these ‘units’,
divide and combine them in different ways . . . In segmenting its own
process, it thereby forms whole units out of certain portions of its
activity, and in setting these formations separately in opposition to one
another, collectively allows these units to stand in opposition to the
thinking subject . . .
No thinking, not even the purest, can occur without the aid from
the general forms of our sensibility (allgemeinen Formen unsrer
Sinnlichkeit) . . . [t]he sensory designations of those units, into which
certain portions of our thinking are united, in order to be opposed

GOLDGABER 9781474438339 PRINT.indd 116 01/10/2020 09:30

 Rewriting the Course in General Linguistics 117

as parts to other parts of a greater whole as objects to the subject, is

called in the broadest sense of the word: language (Sprache).29

The resonances of this Humboldtian account of language with

Saussure’s account of linguistic intermediation are unmistakable.
In Humboldt’s account, in reflection the mind creates ‘objects’
out of its own activity. Units (Einheit) are formed through seg-
mentation, which are then opposed to one another, as parts ‘of a
greater whole’. This segmentation, however, requires the help of
‘the general form of our sensibility’. Thought cannot become an
object for itself, unless it first takes the form of sense, and the form
of sense it takes is sound. As we saw above, Saussure writes that
sound ‘is plastic substance divided in turn into distinct parts to
furnish the signifiers needed by thought’.30
I quoted Humboldt at length, above, because his account makes
explicit (in a way that Saussure’s does not) the philosophical
rationale for the claims that language is radically constitutive of

FINAL
thought, and that speech, or sound, plays a necessary role in
producing the units of thought.31 Derrida’s critique of Saussure’s
phonocentrism refers to the constitutive role that Saussure assigns
sound. When Derrida writes of Saussure’s phonocentrism, he is
referring, not generically to Saussure’s preference for speech over
writing, but specifically to the role he assigns sound in his ‘tran-
scendental semiotics’. This has not been more apparent, perhaps,
because the philosophical tradition in question deviates from the
mainstream (Aristotelian) logocentrism, which takes Meaning and
thought to be outside and prior to language.32
Comparing Saussure’s phonocentrism to Hegel’s, Derrida
argues that this deviation in the philosophical tradition does not
amount to much of a challenge to Aristotelian logocentrism; the
element of the phoné or sound, which plays the ‘sensibilising’ role,
remains distinct from empirical language, distinct even from the
phonic signifier. In this respect, phoné designates not the empirical
voice, nor the phonic signifier, and least of all ‘the written signi-
fier’. These are all necessarily non-ideal and ‘derivative with regard
to what would wed the [phoné] indissolubly to the mind or to
the thought of the signified sense’ (OG, 11). For the same reason
‘the written signifier . . . has no constitutive meaning’ (OG, 11).
The materiality of both sound and writing excludes either from
playing the transcendental role that Saussure assigns to the phoné,
which designates sound as a pure (temporal) form of sensibility.

GOLDGABER 9781474438339 PRINT.indd 117 01/10/2020 09:30

 118 Speculative Grammatology

Saussure’s signifier, wed indissolubly to the signified or conceptual

element, is not, in this sense, a mundane or empirical signifier.
I have suggested that there is something in Saussure’s tran-
scendental semiotics – or ‘aesthetics’ in the Kantian sense – that
Derrida believes is indispensable to the grammatologist and to a
generalised account of writing. Indeed, the dialectic of the second
chapter of Of Grammatology, which redesignates ‘speech’ as
‘arche-writing’, entails rejecting not the transcendental role of
language, but the view that sound uniquely and originally plays
this sensibilising role. Substituting arche-writing for sound permits
questioning the role that sound or speech plays.
Does Derrida’s substitution of arche-writing break with
Saussure’s phonocentrism but affirm his transcendental semiotics,
or does arche-writing imply a modification or break with Saussure’s
transcendental semiotics? Is arche-writing to be thought according
to the conditions that make experience possible – that is, in cor-
relationist terms – or does arche-writing break with Saussure’s

FINAL
correlationist understanding of the sign as a form of sensibility?
Or again, is Derrida’s notion of spacing a modification or break
with the notion of an a priori form of sensibility? My aim in the
last sections of this chapter is to answer these questions.

Genesis and Structure in Saussure’s Semiology

Derrida is clear that Saussure’s phonocentrism is problematic and
cannot be sustained. The consequences of his critique of phono-
centrism for the project of a transcendental semiotics are less clear.
What is clear is that the philosophical meaning of arche-writing
– the sort of transformation or modification of Saussure’s phono-
centrism that it entails – is legible only in the passage through this
transcendental semiotics.
In a section of The Course appropriately entitled ‘Language as
Organized Thought Coupled with Sound’, Saussure writes:

The characteristic role of language with respect to thought is not to

create a material phonic means for expressing ideas but to serve as a
link between thought and sound, under conditions that of necessity
bring about the reciprocal delimitations of its units.33

Here, Saussure refers to the necessity of the link between thought

and sound. But what are these conditions that ‘of necessity’ bring

GOLDGABER 9781474438339 PRINT.indd 118 01/10/2020 09:30

 Rewriting the Course in General Linguistics 119

about the ‘reciprocal delimitation of units’ – and why are these

units reciprocally delimited, rather than just delimited?
The sign articulates and organises sense experience, in a way
that is radically constitutive of sense. The articulation, Saussure
writes, proceeds by ‘decomposition’.

The linguistic fact can therefore be pictured in its totality – i.e. language
– as a series of contiguous subdivisions marked off on both the indefi-
nite plane of jumbled ideas (A) and the equally vague plane of sounds
(B). The following diagram gives an idea of it:

Figure 1.34
FINAL
In Figure 1 – reproduced from the pages of the Course – it looks
as though linguistic articulation consists in an ‘arbitrary’ correla-
tion of plane A (undifferentiated, ‘jumbled’ thoughts) and plane B
(undifferentiated, ‘vague’ sounds). The result of the articulation of
these two heterogeneous planes, Saussure supposes, are intervals
of thought-sound. If sound is distinct or separate from thought,
it can only be because Saussure conceives of sound as something
like ‘pure’ time and thought as a kind of jumbled simultaneity of
unformed cross-modal sensations and impressions (which do not
include sound). In order for there to be one element ‘discrimi-
nated’, there must be two.
Saussure is asking us to imagine something like the following:
a temporal difference or interval of sound (it matters not what) is
marked off or discriminated by a spatial interval, of whatever sort,
but let us say a visual impression such as ‘brightness’. The tem-
poral interval is placed, located or ‘spatialised’ by its correlation
with the spatial interval, and the spatial interval is differentiated

GOLDGABER 9781474438339 PRINT.indd 119 01/10/2020 09:30

 120 Speculative Grammatology

and made determinate by its correlation with a temporal interval.

Sound as temporal provides for succession, while space provides
for duration. Sensibility is produced as the intermediation, ‘confu-
sion’ or entanglement of space-time, or as the very form of the
linguistic sign.
Saussure offers another image for the process of discrimination
or decomposition that he has in mind:

Visualize the air in contact with a sheet of water; if the atmospheric

pressure changes, the surface of the water will be broken up into a
series of divisions, waves; the waves resemble the union or coupling of
thought with phonic substance.35

The first analogy presents language as the decomposition of het-

erogeneous fields of sense; in this second analogy, the sign-form
is described as a kind of ‘interference pattern’. Saussure struggles
for the right analogy to describe a constitutive relation that works

FINAL
by differentiation, an articulation constitutive of differences. Both
sound and thought are originally constituted as such by their
differentiation – prior to their ‘interference’ there is ‘not the least
positive term’; the sign-form is produced as the original différance
of each term.
In both images, the sign is a form that permits sound and
thought to appear as such, where ‘sound’ and ‘thought’ refer to
pure (‘unmixed’) forms of sensibility. What is perhaps clearer in
the second, ‘wave’ image is that the relation Saussure describes
implies the exchange of temporal and spatial forms. ‘Sound’ lends
‘thought’ its temporal structure and ‘thought’ lends ‘sound’ its
spatial extension. This interference, according to Saussure, consti-
tutes the form of a present intuition. Yet this account must affirm
what it denies, and assume what it claims to account for.
Saussure argues that the differences in question are effects of
articulation and, he insists, absolutely indissociable. However, to
recur to Saussure’s second image, if changes or differences in air
pressure produce the ‘waves’ or differences at the surface of the
water, then the ‘air’ is already differentiated. What appears as the
spatially differentiated surface of the water are the temporalised
differences in the ‘atmospheric substance’. Of course, one might
counter that Saussure only intends the wave as an analogy or
illustration, and we should not expect it to perfectly comport
with the sign structure that he seeks to elucidate. This is perfectly

GOLDGABER 9781474438339 PRINT.indd 120 01/10/2020 09:30

 Rewriting the Course in General Linguistics 121

true. But the problem is that Saussure can provide neither a

logical nor analogical account of the constitutive relation he is
trying to describe. In each of his attempts, it will turn out that
the structure he is trying to account for is not originally produced
by the linguistic bond, but reproduced. To return to the wave
analogy, the spatialising ‘inscription’ of sound cannot be radi-
cally constitutive; sound must be always already differentiated
or spaced for the differences in sound to be inscribed-spatialised.
The ‘appearance’ or ‘sensibilisation’ of sound is already its rein-
scription. The differences in sound appear as traces in a field of
heterogeneous sense or differences; and this trace was always
already a trace.
Differences in sound cannot be spatially inscribed if we do not
assume them to begin with. Thought cannot be differentiated by
its temporalisation, if it were not already internally differentiated.
Sense is always already a trace structure, which entails that there is
neither a pure temporality nor pure spatiality, but rather spacing,

FINAL
their ‘originary synthesis’ or entanglement.
Derrida writes:

On the one hand, the phonic element, the term, the plenitude that is
called sensible, would not appear as such without the difference or
opposition which gives them form . . . Here the appearing and func-
tioning of difference presupposes an originary synthesis not preceded
by any absolute simplicity. Such would be the originary trace. Without
a retention in the minimal unit of temporal experience, without a trace
retaining the other as other in the same, no difference would do its
work and no meaning would appear . . . The (pure) trace is différance.
(OG, 62)

Here, in rewriting the sign as ‘originary trace’, Derrida ‘corrects’

Saussure’s problematic analogical descriptions. If temporal dif-
ferences appear ‘originally’ as spatial differences, while spatial
differences appear ‘originally’ as temporalised, then we should say
that spacing produces the difference between time and space as
différance.

The temporalization of sense is from the very beginning ‘spacing’. As

soon as we admit spacing at once as an ‘interval’ or difference and as
openness to the outside, there is no absolute interiority. The ‘outside’
insinuates itself into the movement by means of which the inside of

GOLDGABER 9781474438339 PRINT.indd 121 01/10/2020 09:30

 122 Speculative Grammatology

non-space, which bears the name ‘time’, appears to itself, constitutes

itself, and ‘presents’ itself. Space is ‘in’ time.36

While Derrida’s reading has shown that the sign is an ‘originary

trace, a synthesis without original simplicity’, Saussure interprets
it as a delayed form of presence. Moreover, if sound can be spaced,
or spatially inscribed, it is always already a trace. This is not due
to the conditions of possibility of sound’s appearance, but the con-
ditions of possibility of the retention of differences that produce
sound as a ‘text’ of differential elements. Sound implies differ-
ences, and there can be no differences without spacing.
But Derrida goes further than saying that language – linguistic
form – entails the general entanglement of sense. He argues that
the trace structure or arche-writing warrants us to say that lan-
guage is a form or instance of the general entanglement of sense,
which, itself, is a form of a still more general entanglement. What
the linguistic sign shows is that there is no original entangle-

FINAL
ment, but an iterative, differantial structure: generalised writing.
If the Saussurean sign is an ‘originary’ trace, then the movement
of retention-inscription which accounts for appearance implies
the retention of differences in heterogeneous differences, where the
differences in question are already traces, already spaced.
Derrida’s rewriting of Saussure’s transcendental semiotics has
the important consequence of establishing the ultra-transcendental
status of arche-writing. Arche-writing is more general than con-
sciousness and its forms. The activity of thought or reflection does
not constitute differentiated objects, but reconstitutes, rewrites,
or respaces. Texts are the condition of linguistic conscious-
ness without being definable by that which they make possible.
Consequently, the contrasts in speech or the constituent elements
of texts do not depend upon a linguistic consciousness to recognise
them. The contrastive structure of texts is rather the condition of
linguistic consciousness. Arche-writing defines texts from the start,
and if Saussure is right to say that the signified element informs
the signifier such that the latter is indissociable from it, this entails
that the signified element is not ‘in’ thought but ‘in’ the text – that
‘thought’ is always already a text.
If the materiality or textuality of the signifier implies its irre-
ducibility to consciousness, the entanglement of the signifier and
signified implies the materiality of the signified element as well.
The deconstructive and grammatological significance of Saussure’s

GOLDGABER 9781474438339 PRINT.indd 122 01/10/2020 09:30

 Rewriting the Course in General Linguistics 123

phonocentrism can be found here. Despite calling sound the

phonic substance, Saussure emphasises that the sound in question
is ‘ideal’ and interiorised – the image or impression that sound
makes on the senses. Sound allows for the dematerialisation of the
signifier. The refusal of the materiality of the signifier is motivated
by the refusal of the materiality of the signified. Yet spacing, as
the indissociability of signifier and signified, requires us to affirm
that the contrastive elements in speech as in writing are readable,
whether or not they are read, whether or not they are enacted or
accompanied by ‘acts of speech’ or ‘acts of writing’.
This arche-writing, then, cannot designate the production of
sense as interiority in and for a subject. Arche-writing breaks up
the interiority of the transcendental subject by demonstrating that
this interiority depends upon a non-correlational, textual struc-
ture. It is only on this condition that writing can both be general-
ised and continue to communicate with narrow, empirical writing.
Derrida is clear that spacing radically displaces the transcenden-

FINAL
tal status of Saussure’s account. The trace must be understood in
terms of what I called, in the second chapter, an extra-correlational
instance that cannot, in principle, be correlated with a subject or
point of view. This is the meaning of the ‘exteriority’ or ‘space’
that always already interrupts subjective interiority.

[I]t should be recognized that it is in the specific zone of this imprint

and this trace, in the temporalization of a lived experience which
is neither in the world nor in ‘another world,’ which is not more
sonorous than luminous, not more in time than in space, that differ-
ences appear among the elements or rather produce them, make them
emerge as such and constitute the texts, the chains, and the systems
of traces. These chains and systems cannot be outlined except in the
fabric of this trace or imprint. (OG, 65)

Arche-writing or spacing deconstructs the distinction between

the real and the ideal. Recognising the necessity that sense must
first be inscribed, Saussure ought to have defined the entanglement
of sound and sense, signifier and signified, in terms of a general-
ised spacing. Indeed, linguistic forms are produced not as original
inscriptions, but as reinscriptions. Language respaces – or rewrites
– sense. Saussure’s error was to think the sign both in terms of the
transcendental form of sensibility and as the object of a possible
experience. The trace is demonstrably neither. As arche-writing, it

GOLDGABER 9781474438339 PRINT.indd 123 01/10/2020 09:30

 124 Speculative Grammatology

names the condition of possibility for the retention of differences.

It is describable not in terms of the conditions of experience but in
terms of the conditions of retention. Experience must be thought
‘in return’, according to the possibility of retention, or spacing.
‘The pure phonic chain’, Derrida writes, ‘to the extent that it
implies differences, is itself not a pure continuum or flow of time.
Difference always already implies the articulation of space and
time.’37 These differences in the phonic chain necessarily reflect
and are informed by heterogeneous differences. Each system of
writing is ghost-written by another. Or, in the particular spacing
of any system of writing, another system of writing is readable.
If phonetic writing can represent the temporal flow of speech, this
is because the spaces in graphic writing are the ‘becoming-space’
of time, and time is itself spaced or the becoming-time of space.
Because speech is already spaced, graphic writing can never be
purely linear. In the graphic element, the phonic elements become
space and the spatial (graphic) elements become time. The relation-

FINAL
ship or articulation between two chains of writing is itself spacing.
This does not mean that time and space wait for phonetic writing;
it is rather that the possibility of phonetic writing, of the becoming-
space of speech, requires us to think the ‘space-time of the trace’,
which is also ‘a priori the space-time that we inhabit’ (OG, 316).
Spacing, or the ‘space-time of the trace’, breaks decisively with
Saussure’s transcendental semiotics and with the problem of a
transcendental aesthetics that would establish the ‘unitary and
universal ground’ for subjectivity and subjective experience in the
form of presence. Indeed, any ‘new transcendental aesthetic must
let itself be guided . . . by the possibility of inscriptions in general,
not befalling an already constituted space as a contingent accident
but producing the spatiality of space’ (OG, 290).

Encore: What is Arche-writing?

In this chapter, my aim has been to unfold the meaning of arche-
writing and its philosophical significance, particularly in Derrida’s
rewriting of Saussure’s ‘transcendental semiology’. I began by
arguing that the term is not well understood despite its absolute
centrality to the grammatological project. Attending to Derrida’s
‘regulated transformation’ of Saussure’s Course – in particular, the
substitution of ‘writing’ for Saussure’s ‘speech’ or ‘linguistic sign’
– reveals the movement by which a narrow conception of writing

GOLDGABER 9781474438339 PRINT.indd 124 01/10/2020 09:30

 Rewriting the Course in General Linguistics 125

is generalised. Arche-writing, I have argued, is not limited to ques-

tions of signification and meaning but finally to what Derrida calls
spacing, or what Barad has called space-time-mattering and dif-
ferentiation, which ‘is no more real than ideal’, no more ‘material’
than psychic, and which ‘no concept of metaphysics can describe’
(OG, 65).
Now, by way of conclusion, I want to return to the citation from
Of Grammatology with which we began, and which (I argued)
confronted the reader with a number of interpretive challenges: in
particular, discerning the logic that links the question of narrow
writing to spacing.

Arche-writing, at first the possibility of the spoken word, then of the

‘graphie’ in the narrow sense . . . this trace is the opening of the first
exteriority in general, the enigmatic relationship of the living to its
other and of an inside to an outside: spacing. The outside, ‘spatial’ and
‘objective’ exteriority which we believe we know as the most familiar

FINAL
thing in the world . . . would not appear without the gramme, without
différance as temporalization, without the nonpresence of the other
inscribed within the sense of the present. (OG, 71)

In the foregoing pages, I hope to have shown what justifies the

transformation of a narrow conception of graphic writing to arche-
writing, or spacing. In these last pages, I would like to demonstrate
how arche-writing permits us to revise, to think ‘in return’, some
of the familiar concepts that it radically displaces.

1) Arche-writing as original translatability. Experience implies

structured elements, which appear to us neither as purely spatial,
‘convenient’ differences, nor as purely temporal sequences of dif-
ferences. The structure of speech, as Roman Jakobson argued, is
neither purely successive (time) nor, more obviously, purely syn-
chronous. For this reason, sound cannot be represented in purely
linear terms. Therefore, according to Jakobson, phonetic writing,
a purely linear sequence of letters, cannot represent the elements of
speech. The musical chord, which represents simultaneity, would
better represent the elements of speech (OG, 53–5). Of course, for
these same reasons, we might have expected Jakobson to puzzle
over how phonetic writing could ever have come – so widely – to
‘represent’ the elements of speech. The same reasoning, that is,
ought to have led him to question the pure linearity of writing.

GOLDGABER 9781474438339 PRINT.indd 125 01/10/2020 09:30

 126 Speculative Grammatology

Phonetic writing ‘represents’ speech through differential marks

and spaces, punctuation, and other graphic markers. ‘Represents’
is in scare quotes because what is at stake in Derrida’s analysis is
demonstrating that the relation between speech and writing is not
‘representational’. The differences in speech inform and structure
phonetic writing. We could not account for the genesis, struc-
ture and evolution of systems of writing or other representational
notation without accounting for the ways in which, historically
and technically, these systems came to be in-formed or out-fitted
by another system of marks. Arche-writing is the name for this
mutual adaptation and usurpation.
Though phonocentric, Jakobson’s descriptions of speech empha-
sise that the spacing characteristic of speech implies diverse repre-
sentational possibilities. The spacing of the differential elements
in speech will not appear in the same form in systems of writing
adapted to signify speech; the differantial form in which these
differences will appear in a heterogeneous text cannot be known

FINAL
a priori. One can ‘represent’ speech graphically through phonetic
marks or musical notations, but also in another phonic medium
– as the tonal drum-writing of the Kele attests (see Chapter 2) –
and in other indeterminately varied ways. Arche-writing, or the
differantial structure and spacing of differences, entails the ‘origi-
nal’ translatability and substitutability of texts. Spaced elements
appear each time different; the differences in speech (which came
to be represented, for better or worse, through phonetic writing)
are themselves not original to speech; speech is rather the transla-
tion of heterogeneous differantial patterns.
The sensory substitution device described in the previous
chapter exemplifies this principle of sensory spacing and origi-
nal translatability. With EyeMusic, the differantial structures
‘spacing’ vision ‘appear’ as spatialised and temporalised aural
marks. EyeMusic succeeds if the reader can discern and ‘trans-
late’ the aural structures in a way that reveals or discloses the
visual elements. The EyeMusic interface also exemplifies what
any skilled reader of musical notation knows: that musical struc-
tures cannot be mapped on to or represented in purely spatial or
temporal ‘planes’.

2) Arche-writing as the (non-transcendental) condition of sense

experience. When Saussure broaches the question of the genesis
of the sign, he is no longer describing speech or the linguistic sign

GOLDGABER 9781474438339 PRINT.indd 126 01/10/2020 09:30

 Rewriting the Course in General Linguistics 127

per se, but rather addressing the necessity of the entanglement

of sound and thought. He speculates that the sign’s entangled
form – the constitutive indissociability of signifier and signified
– should be understood in terms of the conditions for sensibility.
As we saw, however, he identified the conditions of sensibility as
transcendental conditions of experience, thereby defining the sign
in correlational terms.
Derrida’s grammatological reading argues that Saussure’s tran-
scendental account misinterprets the entanglement or spacing
characteristic of the linguistic sign. Saussure was right to think
that the form of the sign was related to the condition of experi-
ence, but wrong to interpret the sign as a form of sensibility. What
a sufficiently rigorous phenomenology of experience will reveal is
its ‘textual’ structure. Textual elements are not self-transparent
or given apodictically. Something other than a phenomenologi-
cal analysis, then, must be used to understand and explain these
structures: a grammatological analysis.

FINAL
One of the ways to capture the difference between a gram-
matological and phenomenological analysis of experience is to
consider their respective descriptions of the phenomenology of
(everyday) texts. The phenomenologist would enact the phenom-
enological reduction in order to focus on the mode and manner
of the text’s appearance or givenness. The phenomenologist is
prone to ‘aestheticise’ the structure of texts, and hence will focus
on ‘present’ forms (‘calligraphy’) rather than the structure of
the text, which does not appear. This will lead the phenom-
enologist to take the letters and words to be like indicative signs,
present givens that the reader has learned to associate with absent
meanings.
The grammatologist, by contrast, will be attuned not to the
text’s manifest but its spaced structure. To get at the structure of
the text we would have to recognise its constituents not indica-
tively but as differential elements; and to understand the relation
of these differential elements to the apparently absent (signified
elements), we have to understand the way the latter invisibly struc-
ture the text.

Spacing . . . is always the unperceived, the nonpresent, and the

nonconscious. As such, if one can still use that expression in a
non-phenomenological way; for here we pass the very limits of phe-
nomenology. Arche-writing as spacing cannot occur as such within

GOLDGABER 9781474438339 PRINT.indd 127 01/10/2020 09:30

 128 Speculative Grammatology

the phenomenological experience of a presence. (OG, 68, emphasis in

original)

The case of (everyday) texts is merely exemplary. The aestheticisa-

tion problem generalises to all phenomenological description. The
phenomenologist will treat all experience aesthetically – as cal-
ligraphy rather than text – in terms of the form of appearance; by
contrast, the grammatologist is attuned to what does not appear
in the form of appearance.
Sense experience sensibilises heterogeneous differences. These
differences ‘appear’ in the texture or the fabric of sense, experi-
enced as a unified sense modality. The constitutively heterogeneous
structure of the linguistic sign is continuous with the constitutively
heterogeneous structure of sensory perception. What perception
gives it does not give as such; what is given is not given as itself,
but in the guise of the other.
Though decisively breaking with phenomenology and tran-

FINAL
scendental discourse, arche-writing must still be understood as
an ultra-phenomenological, ultra-transcendental structure. If not,
grammatological terms will inherit the problems of naive objectiv-
ism and perceptual realism which critical philosophy dispels. The
speculative structures that grammatology describes pertain as much
to ‘objective’ written texts as they do to texts as they are heard or
read, to experience in the most general sense. However, insofar as
these structures pertain, generally, to the structure of experience,
grammatology transforms the philosophical meaning of experi-
ence, at least the meaning it has had since Descartes. Experience
is not inherently first-personal, suffused with subjective presence.
Elements of experience are, as David Roden puts it, phenomenolog-
ically ‘dark’.38 Their phenomenological description, in other words,
will not give us any insight into their underlying or fundamental
meaning. Insofar as the meaning of ‘our’ experience can escape us,
experience is, as Derrida puts it, ‘no more ideal than real’ (OG, 65).
We can better grasp the meaning and implications of what
Roden refers to as phenomenologically ‘dark’ elements if we con-
sider phenomenological descriptions of apparently non-textual,
perceptual experience involving indication. Phenomenologically
speaking, indication contrasts with whatever is directly given.
Indicative elements in perception necessarily involve past experi-
ence, depending upon the associational correlation of a present
element with a past presence. Classically, seeing smoke as a sign

GOLDGABER 9781474438339 PRINT.indd 128 01/10/2020 09:30

 Rewriting the Course in General Linguistics 129

for fire is exemplary of indication. Indication would be used to

explain, for example, how it is that we can hear a sound and say,
‘You dropped a coin’, or even more specifically, ‘You dropped a
two-euro coin.’ How is that kind of perception possible?39
The phenomenologist might say that the experienced perceiver
hears the sound indicatively, as a ‘sign’ representing what is not
perceptually present. Just as we can see smoke as indicating fire, we
can hear some clinking sound as indicating a coin, or some other
small metal object, falling. Perceiving indicatively works by bring-
ing to mind absent (possible) perceptions. This sort of perceptual
possibility, however, does not alter in the least the sound as it is
given in perception. For the phenomenologist, hearing indicatively
involves additional associative acts, which bring intuitively absent
content to mind on the basis of a present given. Particularly indis-
pensable is the mnemonic and recognitive activity of the perceiver,
which takes a given as a sign.
In what sense will the grammatological description differ from

FINAL
the phenomenological accounts of indication – and what does this
tell us about the structure of givenness? On the grammatological
description, the ‘indicated’ content is objectively present ‘in’ the
sound of the coin falling; it is always already ‘given’, whether or
not it is heard or recognised. This entails, paradoxically, that there
is indefinitely more content/structure informing what we perceive
than what we (actually) perceive. Reciprocally, what we do (suc-
cessfully) perceive is made possible by structures that both condi-
tion and exceed our recognitive abilities.
Grammatologically speaking, perceptual indication refers, in
the first instance, not to subjective, cognitive activity, but to objec-
tive (non-correlational), structural features of the sound heard.
We could refer to such recognisable structures as ‘signatures’.
Here, recognisability does not imply that the signature depends
upon or is indissociable from its recognition (as the correlational,
phenomenological account assumes). On the contrary, (success-
ful) recognition depends upon there being recognisable structures.
Thus, what makes it possible to hear a sound and say ‘I think you
just dropped a two-euro coin’ is that objective features structure
the sound. The weight of the coin, its material composition, the
hardness (or softness) of the floor have distinctive, isolable sound
‘signatures’. Hearing – and perceiving more generally – is a matter
of recognising the structural features or signatures in-forming
sound as heterogeneous differences.

GOLDGABER 9781474438339 PRINT.indd 129 01/10/2020 09:30

 130 Speculative Grammatology

Again, while the formal structure in question is the condition

for hearing, it is not defined by any perceptual or cognitive acts.
Hearing is possible because we can abstract differences or signa-
tures in sound that correspond to visual, haptic or propriocep-
tive signatures in heterogeneous sense modalities. Thus the sonic
signature of weight corresponds to certain visual and propriocep-
tive signatures, none of which are prior, original or more proper.
Instead of any original propriety of sense, we might speak instead
of an original translatability among sense modalities, a translat-
ability in which forms of sensibility are rooted or take place.

3) Arche-writing is the general form of memory. Perception, gram-

matologically speaking, is the recognition and reactivation of what
is present only as a trace. Like all forms of memory, the trace is not
dependent upon being perceived or upon any present perception.
Arche-writing refers the possibility of present experience to the
general mnemonics of the trace, to the generalised retention and

FINAL
circulation of differences in modifiable chains or texts.
‘The trace, arche-phenomenon of “memory” . . . must be
thought before the opposition of nature and culture, animality
and humanity etc.’ (OG, 70). To which we must add that it must
be thought prior to the opposition between materiality and ideal-
ity. This is because the form of arche-writing cannot be confused
with graphic form, a mark on a material surface of inscription. It
entails thinking the entanglement of modifiable, differential ele-
ments; the preservation of texts in a heterogeneous set of marks.
Hence, arche-writing allows us to describe language and writing
as forms of cultural or human memory or archivisation, but it also
describes the general movement of archivisation, which precedes
any distinctions between forms of memory (genetic or phyloge-
netic, technical or natural, organic or inorganic). Grammatology
demands that each element of these oppositions – culture and
nature, matter and ideality – be thought ‘in return’, or in terms
of arche-writing. To think, for example, matter in terms of arche-
writing is to describe and investigate the essentially retentive and
modifiable – mnemonic – aspects of form.
Arche-writing is both ‘the formation of form and the being
imprinted of the imprint’ (OG, 63). The form of a written text is
not the structuration of a swarm of amorphous marks; it is the
in-forming of a modifiable structure of marks by a heterogene-
ous structure. In order for there to be form, there must already

GOLDGABER 9781474438339 PRINT.indd 130 01/10/2020 09:30

 Rewriting the Course in General Linguistics 131

be form. But this means that contrastive elements or marks must

afford themselves to modification and restructuration. Arche-
writing entails the essential modifiability and retentiveness of the
trace. There is no form without modifiability and no modifiability
without form. It is the task of a speculative grammatology to think
the a priori relations of form and memory, retention and inscrip-
tion, but also of modification and erasure.

Notes
1. Derrida writes, ‘By a substitution which would be anything but
verbal, one may replace semiology by grammatology in the program
of the Course in General Linguistics: “I shall call it [grammatology]
. . . Since the science does not yet exist, no one can say what it would
be; but it has a right to existence, a place staked out in advance.
Linguistics is only a part of [that] general science . . . the laws discov-
ered by [grammatology] will be applicable to linguistics”’ (OG, 51).

FINAL
2. Francisco Vitale’s recent work on biodeconstruction and the recent
publication of Derrida’s La Vie/La Mort are absolutely central in
this regard. While I will not deal here directly with Vitale’s work
– which was written prior to La Vie/La Mort’s publication – it is
important to note that the account of arche-writing I present here
would be most pertinent to biodeconstruction, or an account of the
biological, whether the notion of the biological be that of a code or
not.
3. Saussure (2011: 122).
4. ‘An arche-writing whose necessity and new concept I wish to indicate
and outline here; and which I continue to call writing only because
it essentially communicates with the vulgar concept of writing. The
latter could not have imposed itself historically except by the dis-
simulation of the arche-writing, by the desire for a speech displacing
its other and its double and working to reduce its difference’ (OG,
56).
5. Miller (2011: 47–51).
6. This lack of certainty came out in an interesting admonition
by the influential Derrida scholar Geoffrey Bennington, in a
London Review of Books essay entitled ‘Embarrassing Ourselves’
(Bennington 2016). Bennington laments that the new translation of
Of Grammatology, far from dispelling certain interpretive problems
arising from translational issues, makes things worse, leaving known
errors from Gayatri Spivak’s version in place while introducing new

GOLDGABER 9781474438339 PRINT.indd 131 01/10/2020 09:30

 132 Speculative Grammatology

ones. Moreover, Bennington argues that Judith Butler’s introduction

manages to confuse even the most basic terms. What is presumably
embarrassing is that Derrideans ought to know what they are talking
about when they are talking about terms such as trace and writing,
and when they demonstrate that they do not they also seem to
demonstrate what Derrida’s detractors have always suspected: that
Derrida’s texts make no sense, have no determinate or determinable
meaning; and that, where sense can be made of them, Derrida’s
arguments are trivial or easily defeated. I believe, however, that the
interpretive problems with respect to grammatology are rooted in
a fundamental lack of clarity on the meaning of arche-writing, and
that this is due in no small part to difficulties in reading the second
chapter, on Saussure. In my view, the main issue is that we, Of
Grammatology’s contemporary readers, have inherited a very differ-
ent Saussure than the one Derrida inherited as a student of philoso-
phy in Paris in the 1950s and 1960s. This Saussure, however, is not
Derrida’s invention. Indeed, his reading is compatible with several

FINAL
recent, innovative readers, including aspects of Beata Stawarska’s
‘phenomenological Saussure’ (Stawarska 2015).
7. We can recall, from the discussion in Chapter 2, that Derrida has
already employed a similar strategy with the term iterability. He
coins ‘iterability’, based on a speculative etymology (‘iter, again,
probably comes from itara, other in Sanskrit)’, which would name
the necessity that binds together repetition and alterity. The ‘form’
of the trace, in turn, demonstrates the pertinence of this speculative
relation.
8. In the preface to The Order of Things, Foucault writes,

This book first arose out of a passage in [Jorge Luis] Borges, out of the
laughter that shattered, as I read the passage, all the familiar landmarks
of my thought – our thought that bears the stamp of our age and our
geography – breaking up all the ordered surfaces and all the planes with
which we are accustomed to tame the wild profusion of existing things,
and continuing long afterwards to disturb and threaten with collapse
our age-old distinction between the Same and the Other. This passage
quotes a ‘certain Chinese encyclopedia’ in which it is written that ‘animals
are divided into: (a) belonging to the Emperor, (b) embalmed, (c) tame,
(d) suckling pigs, (e) sirens, (f) fabulous, (g) stray dogs, (h) included in
the present classification, (i) frenzied, (j) innumerable, (k) drawn with a
very fine camelhair brush, (l) et cetera, (m) having just broken the water
pitcher, (n) that from a long way off look like flies’. In the wonderment of

GOLDGABER 9781474438339 PRINT.indd 132 01/10/2020 09:30

 Rewriting the Course in General Linguistics 133

this taxonomy, the thing we apprehend in one great leap, the thing that,
by means of the fable, is demonstrated as the exotic charm of another
system of thought, is the limitation of our own, the stark impossibility of
thinking that. (Foucault 1970: xv)

9. ‘[D]ans la langue, on ne saurait isoler ni le son de la pensée, ni la

pensée du son; on n’y arriverait que par une abstraction’ (Saussure
1967: 158).
10. Malabou (2007a: 432).
11. Malabou (2007a).
12. Kirby (2016: 48).
13. Critics, including me, have argued that Hägglund’s account of the
generality of the trace departs in critical respects from Derrida’s
generalisation of the trace structure. See Goldgaber (2017); Johnston
(2009); Staten (2009).
14. Saussure (2011: 122).
15. Pavel (1988: 108), as quoted in Maniglier (2007: n.p., my translation).

FINAL
16. For an excellent account of this sort of typical confusion, which
conflates Saussure’s account of a system of writing with that of the
linguistic sign, see Maniglier (2007) or his magisterial rereading of
Saussure, La Vie énigmatique des signes (Maniglier 2006).
17. Saussure (2011: 114–15, my emphasis).
18. Maniglier (2007) offers a robust defence of the reading that I have
just attributed to Weber. Maniglier argues, correctly in my view,
that the inherited readings of Saussure have conflated relations of
opposition and contrast from the sort of relations that account for
the genesis of the linguistic sign.
19. Weber (1976: 924).
20. Saussure (2011: 112, emphasis added).
21. Maniglier (2007).
22. Saussure (2011: 111–12), as quoted in Weber (1976: 922).
23. Weber (1976: 922).
24. Saussure (2011: 112).
25. Saussure (2011: 111).
26. In de la Grammatologie, Derrida uses the expression ‘vouloir sensi-
bilisé’ to refer to Main de Biran’s ‘intention’. Spivak translates this as
‘wish sensibilized’, but this translation is arguably less ‘relevant’, as
Derrida would write, than ‘intention’, given that Derrida translated
Husserlian intentionality as ‘vouloir-dire’, that is, as Meaning or
‘wanting to say’.
27. Forster (2010: 56).

GOLDGABER 9781474438339 PRINT.indd 133 01/10/2020 09:30

 134 Speculative Grammatology

28. Mueller-Vollmer and Messling (2017: n.p.).

29. Mueller-Vollmer and Messling (2017: n.p., translation modified).
30. Saussure (2011: 112).
31. I should point out that while scholars of this period, including those
just cited, are aware of the resonances between Humboldt and
Saussure, there is considerably less awareness on the Saussurean
side. I owe the first suggestion of this connection to Christina Lafont,
in private conversation.
32. Interestingly, scholars of German philosophy of language are well
aware of the extent to which Humboldt’s account anticipates and
overlaps with Saussure’s. Derrideans and Saussureans, by contrast,
seem unaware of the extent to which Saussure is continuous with the
German Idealist tradition.
33. Saussure (2011: 111–12, my emphasis).
34. Saussure (2011: 112). Figure found at <http://www2.sims.berkeley.
edu/courses/is296a-3/s06/Saussure.pdf> (last accessed 15 July 2018);
this reproduces an earlier edition [1959] of the same translation.

FINAL
35. Saussure (2011: 112).
36. Derrida (1973: 73).
37. Derrida (1978: 248).
38. Roden (2013).
39. It would be perfectly plausible to say that one did not perceive a
two-euro coin falling, but instead that one heard a sound, and then
inferred that it was a coin, and even a certain kind of coin – based
on past perceptual experiences of similar events with similar objects.
In this case, one might say that the perception is of a certain sound,
and the propositional content the non-perceptual work of inferences
(associations, memories) subsequent to hearing the sound. This view
can easily explain why experienced hearers will have more success
in identifying the sound by, say, comparing it to the memory of the
last time they dropped a two-euro coin. The account I develop above
defends a non-correlational account where the sound ‘signature’ is
an essentially perceivable, structural feature of the sound.

GOLDGABER 9781474438339 PRINT.indd 134 01/10/2020 09:30