structure. Following an argument analogical to the argument on the sign, Derrida puts the
word “experience” under erasure:
As for the concept of experience, it is most unwieldy here. Like all the notions I am using, it
belongs to the history of metaphysics and we can only use it under erasure. “Experience” has
always designated the relationship with a presence, whether that relationship had the form of
consciousness or not. Yet we must, by means of the sort of contortion and contention that
discourse is obliged to undergo, exhaust the resources of the concept of experience before
attaining and in order to attain, by deconstruction, its ultimate foundation. It is the only way to
escape “empiricism” and the “naive” critiques of experience at the same time (89, 6o).
Now we begin to see how Derrida’s notion of “sous rature” differs from that of Heidegger’s.
Heidegger’s [Being med kryss] might point at an inarticulable presence. Derrida’s [trace med
kryss]is the mark of the absence of a presence, an always already absent present, of the lack at
the origin that is the condition of thought and experience. For somewhat different yet similar
((xviii))
contingencies, both Heidegger and Derrida teach us to use language in terms of a trace-
structure, effacing it even as it presents its legibility. We must remember this when we wish to
attack Derrida or, for that matter, Heidegger, on certain sorts of straightforward logical
grounds; for, one can always forget the invisible erasure, “act as though this makes no
difference.” (MP 3, SP 131) 13
Derrida writes thus on the strategy of philosophizing about the trace:
The value of the transcendental arche [origin] must make its necessity felt before letting itself
be erased. The concept of the arche-trace must comply with both that necessity and that
erasure. It is in fact contradictory and not accepta-ble within the logic of identity. The trace is
not only the disappearance of origin, ... it means that the origin did not even disappear, that it
was never constituted except reciprocally by a non-origin, the trace, which thus becomes the
origin of the origin. From then on, to wrench the concept of the trace from the classical
scheme which would derive it from a presence or from an originary non-trace and which
would make of it an empirical mark, one must indeed speak of an originary trace or arche-
trace (90, 61) .
At once inside and outside a certain Hegelian and Heideggerian tradition, Derrida, then, is
asking us to change certain habits of mind: the authority of the text is provisional, the origin is
a trace; contradicting logic, we must learn to use and erase our language at the same time.
In the last few pages, we have seen Heidegger and Derrida engaged in the process of this
curious practice. Derrida in particular is acutely aware that it is a question of strategy. It is the
strategy of using the only avail-able language while not subscribing to its premises, or
“operat[ing] accord-ing to the vocabulary of the very thing that one delimits.” (MP 18, SP
147) For Hegel, as Hyppolite remarks, “philosophical discourse” contains “its own criticism
within itself.” (SC 336, 158) And Derrida, describing the strategy “of a discourse which
borrows from a heritage the resources necessary for the deconstruction of that heritage itself,”
remarks similarly, “language bears within itself the necessity of its own critique.” (ED 416,
SC 254) The remark becomes clearer in the light of writing “sous rature”: “At each step I was
obliged to proceed by ellipses, corrections and cor-rections of corrections, letting go of each
concept at the very moment that I needed to use it, etc.”14
There is some similarity between this strategy and what Lévi-Strauss calls bricolage in La
pensée sauvage.15 Derrida himself remarks:
Lévi-Strauss will always remain faithful to this double intention: to preserve as an instrument
that whose truth-value he criticizes, conserving . . . all these old concepts, while at the same
time exposing . . . their limits, treating them as tools which can still be of use. No longer is
any truth-value [or rigorous meaning] attributed to them; there is a readiness to abandon them
if necessary if other
((xix))
instruments should appear more useful. In the meantime, their relative efficacy is exploited,
and they are employed to destroy the old machinery to which they belong and of which they
themselves are pieces. Thus it is that the language of the human sciences criticizes itself. (ED
417; SC 255, 254)
One distinction between Lévi-Strauss and Derrida is clear enough. Lévi-Strauss’s
anthropologist seems free to pick his tool; Derrida’s philosopher knows that there is no tool
that does not belong to the metaphysical box, and proceeds from there. But there is yet
another difference, a difference that we must mark as we outline Derridean strategy.
Lévi-Strauss contrasts the bricoleur to the engineer. (“The ‘bricoleur’ has no precise
equivalent in English. He is a man who undertakes odd jobs and is a Jack of all trades or is a
kind of professional do-it-yourself man, but . . he is of a different standing from, for instance,
the English ‘odd job man’ or handyman.” 16) The discourse of anthropology and the other
sciences of man must be bricolage: the discourses of formal logic, and the pure sciences, one
presumes, can be those of engineering. The engineer’s “instrument” is “specially adapted to a
specific technical need”; the bricoleur makes do with things that were meant perhaps for other
ends. 17 The anthropologist must tinker because, at least as Lévi-Strauss argues in Le cru et le
cuit, it is in fact impossible for him to master the whole field. Derrida, by an important
contrast, suggests that the field is theoretically, not merely empirically, unknowable. (ED 419
f., SC 259 f.) Not even in an ideal universe of an empirically reduced number of possibilities
would the projected “end” of knowledge ever coincide with its “means.” Such a coincidence
—“engineering”—is an impossible dream of plenitude. The reason for bricolage is that there
can be nothing else. No engineer can make the “means”—the sign—and the “end”—meaning
—become self-identical. Sign will always lead to sign, one substituting the other (play-fully,
since “sign” is “under erasure”) as signifier and signified in turn. Indeed, the notion of play is
important here. Knowledge is not a systematic tracking down of a truth that is hidden but may
be found. It is rather the field “of freeplay, that is to say, a field of infinite substitutions in the
closure of a finite ensemble.” (ED 423, SC 260)
For Derrida, then, the concept of the “engineer” “questioning the uni-verse” is, like Hegel’s
father-text encompassing the son-preface, or Heidegger’s Being as transcendental signified, “a
theological idea,” an idea that we need to fulfill our desire for plenitude and authority. He