I. Introduction: structural explication vs. linear explanation

There is an interesting parallel between the ways in which G. W. F. Hegel and Ferdinand de Saussure are criticized. They are, respectively, said to deny or devalue the domain of extra-conceptual ‘being’ and of extra-linguistic ‘reference’. In short, the charge brought against them is that the former believes that one should ignore the empirical in the exposition of truth, and the latter that one should ignore the referential in the exposition of language. For Hegel, truth does not primarily pertain to representation (die Vorstellung). And, for Saussure, language is not primarily a representational lexicon of things in the world (a nomenclature). The parallelism can be made to converge in the following way: Hegel denigrates the philosophical importance of non-conceptual ‘being’, and Saussure denigrates the linguistic importance of reference to non-linguistic ‘being’. Generally speaking, they disregard the ‘real’, positive, or external content of concepts and words. I will call this the ‘standard criticism’ (SC).Footnote ^1, Footnote ²

In the present article I argue that the SC stems from a conflation of two distinct levels of theorization: the SC understands the structural in terms of the linear (thereby effectively obfuscating the structural altogether). To clarify the difference between these two levels, I have opted to introduce a distinction between structural ‘explication’ and linear ‘explanation’. Much of the present article will be devoted to clarifying this distinction through the works of Hegel and Saussure, both of whom are major exponents of structural explication.

In order to keep focus on the issue at hand, namely these two general levels of theorization, I have had to omit any discussion of what could be termed ‘Saussure's Hegelianism’ and, following Derrida, ‘Hegel's semiology’. The (historical) question of Hegel's influence on Saussure will therefore not be pursued here, though this deserves more scholarly attention.Footnote ³ Derrida's essay ‘The Pit and the Pyramid: Introduction to Hegel's Semiology’ was one of the first to bring attention to the similarity between Saussure's semiology and Hegel's account of the sign in the Philosophy of Spirit (third volume of the Encyclopaedia) and in the Aesthetics lectures.Footnote ⁴ Though I do not here discuss this specific similarity, I do think that the present article supports it, while suggesting an even deeper and more general affinity between Hegel and Saussure in their scientific approaches. Furthermore, I think that the importance of what I have here termed ‘structural explication’ reaches far beyond such a question of influence. Indeed, it could be argued that the distinction between the structural and the linear cuts through all scientific and theoretical fields.

II. Preliminary definitions of the ‘linear’ and the ‘structural’

The paradigm of a ‘linear explanation’ is causality. This includes both causality of the mechanistic cause–effect type as well as quasi-causality of a non-mechanistic antecedence–consequence type, whether of a temporal, conditional or inferential order. (The ‘quasi-’ is meant to indicate that the relations are not strictly reducible to mechanistic causality.)

The other level of theorization is ‘structural explication’, and it is somewhat more cumbersome to state in positive terms. It is characterized by the non-prioritizability, non-separability, and non-reducibility (whether temporally, conditionally, or logico-inferentially) of the parts of the explication. This means that the explication aims to show the primacy of a whole that is not straightforwardly reducible to its parts. (Though this definition is still mostly negative, part of the present article will consist in providing a positive and fuller definition of ‘structural explication’.) It is ‘deep’ in the sense that the individual parts do not exist prior to the structure (there is structure ‘all the way down’). The deep structural explication corresponds to Hegel's conceptual exposition and to Saussure's synchronic approach to language.

III. Antecedence and consequence in ‘linear explanations’

I think it is safe to say that most, if not all, explanations tend towards the linear form. ‘Explanation’ is virtually synonymous with ‘causal explanation’. An explanation is ‘linear’ by virtue of a spatiotemporal or logical unidirectionality of the antecedent and consequent, and by virtue of a clear separation of the antecedent and consequent. Let us look at some examples. Classical physics offers the most obvious paradigm of ‘linearity’ as it is used here (not to be confused with ‘linear’ in the sense of pertaining to linear differential equations—of which classical physics is, however, also a paradigm). Cause and effect are here distinctly separable in space and time, and a cause (= antecedent) serves as the explanation of an effect (= consequent). Though an explanation in classical physics could in principle work equally well both backwards and forwards in time, it always has a direction and a separation of antecedent and consequent. However, the unidirectionality of this linearity has been firmly consolidated in non-classical physics that accentuate ‘the arrow of time’ (= the irreversibility of antecedence and consequence).Footnote ⁵ Moreover, so-called non-linear (chaotic) systems are defined by a high sensitivity to small variations in initial conditions (= antecedent) with respect to very large divergences in outcome (= consequent), thus affirming the explanatory linearity (in the sense used in the present article) within a ‘non-linear’ (chaotic) context. And in the theoretical field concerned with so-called ‘emergent properties’ there is always a lower-order antecedent level to a higher-order consequent level. To say that, for example, mental events supervene on physical events is to say that the former are a consequent of the latter (and not the other way around), even if not straightforwardly mechanico-causally related. Though emergent properties generate seemingly autonomous levels of explanations (from particle physics to, say, psychology) they are all prioritized vis-à-vis each other, meaning that a lower-order is always prior to a relative higher-order (the latter supervenes upon the former). And if we move to the realm of practical philosophy it is common to distinguish between causes (for natural events) and reasons (for human actions). Yet, despite their distinctness they still have the same kind of relative separable and directed ordering in time. Reasons are defined and explained as the prior motivations (= antecedents) that initiate, and explain, an action (= consequent).Footnote ⁶ Even when seen as metaphysically distinct from mechanical causality, practical reasons are still quasi-causal (or, in Kant's terms, a kind of ‘spontaneous’ causation). Furthermore, if we consider the idea of a logical space of reasons—as famously evoked by Wilfrid Sellars—it goes without saying that the relations of giving and asking for reasons are structured according to a priority of premises (inferential antecedents) with respect to conclusions (inferential consequents), despite not necessarily being strictly temporally ordered. Again, Leibniz's principle of sufficient reason is based on a relation of antecedence (the sufficient reason, cause, or ground) and consequence (the fact). Generally speaking, the unidirectional linearity of the conditional (if p, then q) in modus ponens (if p, then q; p; thus q) is evident from the logical fallacy called ‘affirming the consequent’ (i.e., an argument with the form if p, then q; q; thus p is invalid; only the premise that affirms the antecedent yields the consequent, not the converse). Other species of explanations that are based on antecedence and consequence include historical, narrative, probabilistic and evolutionary explanations. Axiomatic-deductive systems, like Euclidean geometry, are also based on linearity, in the sense that there is a hierarchical priority between, on the one hand, the ‘axioms’, ‘common notions’, and ‘definitions’ and, on the other hand, the theorems that can be deduced from these, and again between theorems (or corollaries) that are presupposed to deduce further theorems. When raised to the level of utmost generality, most explanations can be comfortably decomposed into an antecedence-side and a consequence-side. And with this decomposition there is a relative priority and a relative separation of these two sides in the explanation. A consequent may function as an antecedent in another context, and vice versa, but not in the same context. In general, the order of a linear explanation also tends to be, directly and indirectly, mappable onto an analogous spatiotemporal order (though not necessarily onto causal relations in the strict mechanical sense).

This list of various kinds of ‘linear explanations’ is not intended to be in any way exhaustive, nor is it intended to demonstrate anything. Instead, it is simply put forth to suggest the sheer pervasiveness of linear explanation to such a degree that it is virtually synonymous with explanation as such. On this background it is not surprising that there can be a tendency to understand all theorizations as being of this kind. But, as I hope to show, there is at least one general and indispensable mode of theorization that cannot be reduced to the linear form of explanation.

The thesis proposed in the present article is the following: it is the (probably unconscious) expectation to find a similar kind of linear explanation in Hegel and Saussure that is the source of the confusion of the SC. I will focus especially on why the basic claims of the generalized SC are incompatible with the structural explications of Hegel and Saussure.

IV. Philosophical science and linguistic science

It is interesting to discern a more positive parallel between Hegel and Saussure in the formulation of their scientific programmes. They both share an ambition to give their respective disciplines a proper scientific form, based on defining the scientific object, which they both take to be fundamentally different in kind from that of any of the other established sciences, because it (the object) bridges domains that are usually seen as sharply distinct, and because these objects cannot be clearly separated out from the cognizing subject.Footnote ⁷ For Hegel, the scientific ‘object’ is the concept (der Begriff) and the conceptual system (die Logik), and for Saussure it is the sign (la signe) and the language system (la langue).

Here I will present a brief parallel three-point overview of the ‘scientific programmes’ of Hegel and Saussure. In this way we can detect the inherent inadequacy of the linear form of explanation when faced with the (strange) objects of these sciences.Footnote ⁸

V. Hegel

(1) Hegel rejects the idea that philosophy should be grounded in first principles that can play the role of absolute primary antecedents from which to draw consequences (see e.g., PhG: 15/27).Footnote ⁹ In Hegel's time ‘science’ basically meant knowledge grounded on some first principle(s) (in conformity with Aristotelian episteme). Hegel rejects this for the following reasons: first principles are by definition dogmatic, unproven or unprovable, excluded from rational accounts, and thus only grounded in a dubiously indubitable ‘certainty’ or a private ‘intuition’. The felt need for a first principle is exclusively based on the lack or the paucity of any conceivable alternative. It would surely be convenient to be in possession of a primary antecedent that could serve as an absolute fixed point from whence to derive all positive knowledge as its consequence (a philosophical equivalent of a mathematical axiom). This would yield perfect linearity. But, for Hegel, a first principle, as an immediate piece of knowledge, is a vague and inherently incoherent notion. For Hegel, all knowledge entails mediation (even the Cartesian cogito ergo sum is deemed to be a syllogism).Footnote ¹⁰ The problem is that mediation is usually seen as linear and thus cannot constitute a first principle (because it involves parts that, when arranged linearly, have a priority, and the first in this line is then the actual first principle, which is ‘immediate’). Hegel's solution is a logical theory in which immediacy (self-identity) and mediation (self-differentiation) are not mutually exclusive.Footnote ¹¹ (And it is precisely here that structural explication becomes pertinent, as we shall see later.) This also involves showing that any claim to positive knowledge without mediation can only be a mere assurance of knowledge (Versicherung) or a private certainty (Gewissheit), meaning that this ‘knowledge’ cannot be demonstrated. And as such it is not really knowledge.

(2) The representational model of cognition of a subject standing over against an object cannot yield true knowledge for it will always be vulnerable to scepticism. There is no way to step outside the subject–object relation and into an epistemologically ‘neutral’ zone, some external Archimedean standpoint, whence to assess if our subjective representations ‘match up’ with the object. Hegel's solution to this problem is to unearth a kind of knowledge (which he calls absolute knowledge) whose truth does not depend on representational correspondence. (For Hegel, representational ‘truth’ would more aptly be called ‘correctness’, but cannot be called truth in an emphatic sense.) To show this he must show that all forms of knowing that separate knowledge (i.e., the subject) and truth (i.e., the object or substance) end up in internal conflict when confronted with the requirement of correspondence. However, it is also this internal conflict (or dialectic) that provides the solution of how to achieve absolute (non-representational) correspondence.

In the Introduction to the Phenomenology Hegel gives the general name Gestalten des Bewußtseins to all these internally conflicted kinds of subjective knowledge (see especially PhG: 52ff./72ff.). Gestalt is usually translated as ‘shape’, ‘form’, or ‘figure’ (of consciousness), but could here perhaps more fittingly be translated as ‘structure’ (of consciousness). For, as Hegel repeatedly stresses, the subject-pole and the object-pole are both internal to the consciousness-Gestalt. It is these two poles that together constitute a structure of consciousness. The consciousness-Gestalt is not over against the object, but is the totality of the knowledge-truth structure.Footnote ¹² Thus, as Hegel notes, the criterion of truth is not external to the consciousness that has (or aspires to have) true knowledge. Said in a slightly different way: what constitutes the criterion of objectivity is not given externally to the subject (even if this criterion states that objectivity is ‘subject-independence’). When reconfigured in this way we are no longer faced with an unbridgeable epistemological dualism. We simply have to compare the two sides of the duality of ‘knowledge’ and ‘truth’. In the end, the conclusion Hegel draws from this is that the disintegrated dual representational structure, initially presented as divided into a separate subject-pole and object-pole, is actually an integrated total logical structure between a differentiation-pole (für sich) and an identity- or unity-pole (an sich). The conflict of representation is resolved in general, once we come to see that there can be no pure difference (for itself) and no undifferentiated identity or unity (in itself).

It is, roughly speaking, at this point that the Science of Logic begins (with the conceptual self-differentiation of pure undifferentiated identity, or ‘pure being’). This means that the fundamental mistake—for which the Phenomenology serves as the propaedeutic antidote—is the separation and prioritization of one pole over the other. In using very abstract terms for these poles, the ‘an sich’ and the ‘für sich’, Hegel can move effortlessly from the representational to the logical. These terms do not refer to substantial entities or distinct realms of existence, but irreducible structural moments. Together, as a unified structure, this in-and-for-itself constitutes the object of a philosophical science proper (= a science of logic).Footnote ¹³

(3) Science has to be internally systematic and cannot be based on what Hegel calls ‘external reflection’ (see e.g., WL: 348–51ff./II.28–32ff.). This means that the two structure-poles, the an sich and the für sich mentioned just above, cannot be unified except ‘by their own accord’. I will try to sketch what this means here. The negative point is that the subject cannot intervene and ‘externally’ manipulate the parts to synthetize them into a whole. The parts, or moments, as Hegel calls them, in some sense, have to structure themselves, ‘internally’. Though this may initially sound like a linear temporal or successive process, importantly it is not. If it were so, then the moments would exist perfectly well outside the structure, and the structure would just be an ‘external’ relation of parts. And if this were the case, there would be nothing to prove, no necessary ‘internal’ connection. There can be no necessary relation between parts that can subsist outside the relation. The necessity, therefore, relies on the inseparability of the parts, i.e., that the parts are actually not separate parts but in fact structural ‘moments’. Hegel's argument thus depends on showing that there are structures that cannot be decomposed or reduced to more primitive structureless parts. Hegel's method proceeds by demonstrating that one ‘part’ is only rendered possible in absolute unison with the other ‘part’, and that the same is true for the reverse. They are parts (different), and yet they are inseparable (identical). Hegel calls these inseparable parts ‘moments’. This unison between the moments is so absolutely irreducible that Hegel refers to it as an ‘identity’, as in the famous formulation: the identity of identity (= an sich) and difference (= für sich). This then forms the An-und-für-sich-sein structure. The radical and strange nature of this kind of structure must not be overlooked. Under a normal linear interpretation it is a contradiction. As Hegel himself acknowledges, we here find ourselves on the razor edge between the logical and the paralogical.

In not shying away from what is prima facie contradictory, Hegel believes that it is possible to rationally work through this critical ‘limit’ to show its positive structural truth (see e.g., WL: 742f./II.558f.). For Kant, this critical limit was rationally insurmountable. However, Kant was too ‘linear’ in his approach. For Hegel, the extra-critical or extra-antinomial truth is only accessible if we break with linearity. In Kant's dialectics, extra-critical thinking is presented as two irreconcilable yet equipollent parallel lines of reasoning. The way Hegel executes the break with linearity is by showing that there is no separation between the two lines (they form a total structure). The reason for this is that they mutually render each other possible (they ‘transition into’ or ‘become’ each other in a non-linear way). It is for this reason that the antecedent and the consequent are interchangeable in Hegel's philosophy. The so-called ‘speculative sentence’ expresses this chiastic inter-changeability (PhG: 39f./59f.). In other words, the linear forward progression in a logical argument must, in Hegel's exposition, also be a ‘folding back’ upon itself or a ‘return’ to the starting point. It is this double movement—where each move in one direction is also a move in the opposite direction—that is the sublation (Aufhebung) of linearity. Instead of a succession, it affirms absolute logical simultaneity. For this reason, Hegel describes the logical progressions as a circle (Kreis). If it were a linear progression this circle would be vicious. But since it is not linear but structural—non-prioritized, non-separable, non-reducible—it is a non-vicious circle. The circle is evoked to indicate the break with linearity. And Hegel's name for this kind of irreducible structure is der Begriff.

VI. Saussure

With this sketch of Hegel's project of a philosophical science laid out, I will now proceed to sketch Saussure's project for a linguistic science.Footnote ¹⁴ Here I will also select three central points that accentuate his break with linearity.Footnote ¹⁵

(1) What is the nature of the scientific object of linguistics? Saussure holds that most sciences have an object given in advance, that can be considered from several points of view, and in relative independence of these. But the linguistic object is, it seems, completely dependent on the point of view that one takes. Saussure identifies two primary but heterogeneous, even antinomial, ‘sides’ of the linguistic object: it consists both of a ‘sound’-pole (‘acoustic image’) and a ‘thought’-pole (‘concept’).Footnote ¹⁶ The dilemma is that it is not possible to study one side without neglecting the other side. Furthermore, it is not possible to study either side individually without theoretically neglecting the essential duality of the linguistic object. The problem with the linguistic object is that it is not an integrated whole. Its inherent duality seems to preclude the unity of the object of study. It is dismembered by being studied from many points of views, e.g., as sounds in phonology and as development of worldviews in philology (and in other aspects in anthropology, grammar, psychology, etc.). But there does not seem to be a unique scientific object of an autonomous linguistic science. Saussure's answer to this is to propose a strange kind of object. The object of linguistics is an irreducibly dual unity that he gives the deceptively simple name le signe (or la langue, the language system, to designate the system of signs). The sign is a unity that is inseparably composed of the duality of signifier (signifiant) and signified (signifié). It is important to note that the ‘signified’-pole is not an external object of reference, but is a so-called mental concept. Both of the poles (in short: communicable sounds and their meanings) are internal to human psychology. But linguistics cannot be reduced to psychology, or to any other science for that matter, because of the unique nature of the linguistic object that combines seemingly antinomial and dual aspects into one unified object of study (i.e., the sign). Therefore, linguistics must become an autonomous science of the sign (semiology).

(2) Even if Saussure speaks of the sign as though it were an isolable entity, i.e., as the object of study, we must not forget that there can be no lonely sign outside a system of signs. In his linguistics this system is called la langue, usually translated as the ‘language system’. The system is not made up of the sum of individual connections of an independent ‘signifier’ with an independent ‘signified’. This erroneous view is what Saussure describes as a nomenclature (CLG: 16, 65, 114, 122/34, 97, 155, 158). The two sides of the sign (signifier/signified) are inseparable and irreducible to each other: ‘The linguistic entity exists only through the associating of the signifier with the signified. Whenever only one element is retained, the entity vanishes’ (CLG: 102–3/144). But we can still talk about the two sides of the sign in abstracto for analytic purposes, although we cannot present a single ‘pure’ signifier or single ‘pure’ signified. The sign is both inherently dual and inseparable. Saussure has to show that the one side is only rendered possible because of the other side, and vice versa. But he cannot do this with the nomenclature model of language, for it presupposes a given signifier and a given signified that can be individually identified, externally manipulated and combined. Furthermore, in the duality there can be no part that has priority over the other, in so far as each part can only exist together with the other part.

But there is, as noted earlier, another irreducible aspect besides the unity of the two distinct ‘sides’, signifier and signified. This aspect is the irreducible correlation of two distinct orders of language: the system of signifiers and the system of signifieds. Saussure describes the language system as a ‘system of equivalence of things of different orders’ (CLG: 79/115, translation modified). The ticklish problem here is that these two different orders, between which there are systematic equivalences, are themselves each orders of immanent differences.

I think the most concise way to formulate what this means is the following. There is never any given individual signifier,Footnote ¹⁷ but only a system of signifiers that are individuated by their differential relations to others. Conversely, there is no given individual signified,Footnote ¹⁸ but only a system of signifieds that are individuated by their differential relations to others. But, as was just pointed out, there also cannot be any signifier and signified in isolation from each other. This also goes for the systems of these two sides. What generates (in a strict non-linear sense) a language-system is the correlation between the signifier-system and the signified-system. It is a difference in signifiers that picks out and individuates a difference in signifieds, and, conversely, it is a difference in signifieds that picks out and individuates a difference in signifiers. If we remove this correlation language as such disappears. Instead of a system of signifiers we are left with (at best) an undifferentiated sound-spectrum, or even less: a random series of meaningless sounds with no clear divisions, no individual sounds, and no clear beginning or end (CLG: 103–2, 111–12/145–46, 155–56). And instead of a system of signifieds we are (at best) left with an amorphous and indistinct mass of inexpressible meaning (112/155). I would add that meaning (the signified) in isolation from its expression (the signifier) is a fictitious analytic construct.Footnote ¹⁹ To sum up: it is not merely the case that each sign only is what it is in relation to what it is not. Things are somewhat more complicated: any individual sign is only possible as a product of differences in the correlations or equivalences between the two levels of difference-systems (of signifiers and signifieds).

(3) This brings us to the third and final point, which is probably also the least acceptable aspect of Saussure's theory, according to the SC. It is summed up in the infamous formula: ‘in language there are only differences without positive terms’ (CLG: 120/166). How can there be only differences? Is it not like adding up zeros where, no matter how many zeros one adds up, the final sum is still zero? Furthermore, the critic will ask, differences between what? What are the given terms of the differences? However, these two points harbour two dogmatic assumptions: (i) that language is the result of some sort of numerical addition of atomistic pre-systematic building-block-like language-parts, and (ii) that the terms of the difference pre-exist the difference (either temporally or logically). We can clearly see that these critical points implicitly, yet unmistakably, expect Saussure's formulation to be of a linear sort.

To clarify this mistake let us first try to comprehend what is meant by a signifier being purely differentially constituted (and afterwards take on the signified). The differential constitution is intimately correlated, as Saussure himself says (CLG: 118/163), to the thesis of the arbitrary nature of the signifier. I have, to not complicate matters unduly, spoken of the signifier as if it were only made of sounds. But this is just one kind of material ‘element’ by which signifiers can exist. It can also be made of written letters, colours, Braille-signs, hand gestures, Morse code, binary codes, eye movements, smoke signals, light beams, or anything for which it is possible to mark a sufficiently practicable degree of differentiation. Whatever these differences are the differences of is irrelevant for the general constitution of language. What makes a signifier is only its difference from other signifiers; it is the reciprocal opposition of signifiers that counts, and not what these opposing elements are made of (CLG: 117–20/163–66). The positive terms are thus always only an effect of their differential constitution.

Even if one accepts this, it remains hard to accept the thesis that the signified (the level of semantic content) of language is only differentially constituted. This domain seems much too rich to be captured by mere exclusions. First of all, the obvious problem (if we stay within Saussure's theory) is that there is no articulated signifier-less signified-content. So whereas we could imagine, or at least imagine imagining, some signifier-differences independently of their relation to any signified (meaning), it seems much less likely that we can imagine, or even imagine imagining, some unmarked signified-hood as a kind of self-radiating isolated identity-point of ‘meaning’. But it is also, of course, precisely this impossibility that Saussure is underscoring: these two sides of the sign do not exist in themselves. It is merely easier to imagine signifiers in isolation, because they have a material support, to which they are, however, irreducible.Footnote ²⁰ As soon as we imagine some signified it will always already be tied up with the level of signifiers (some possible form of expression). Thus, in so far as signifiers are only differentially constituted, the signifieds are, ipso facto, also differentially constituted. In so far as nothing means anything in itself, in isolation—there can be no punctual semantics—the nature of the signified is, ceteris paribus, based on a system of differential relations. (It should be noted that, as such, this does not exclude that language also has a referential role.) Saussure is nowhere denying that there are positive terms in language, but only that language is not linearly constructed from the addition of simple positive terms (individual language ‘substances’). His point about the negative differential constitution of language only holds for the orders of signifier and signified in separation (see CLG: 120–21/166–67). But when taken in their totality they yield a positive result. However, this positivity is not the result of a linear process. The fact that there can be a positive, meaningful, and simple term is only a result of the complex systems out of which it arose. It is never the case that there can be a simply positive, primitive, punctual sign prior to a system.

VII. Comparison (Hegel and Saussure)

Let us compare these two surveys. Their respective scientific objects are both characterized as inherently composed of (prima facie) incommensurable sides. This is what makes it difficult for them to pinpoint the object. For Hegel, the ‘object’ is the relation between a substantial and self-identical ‘truth’ (the an sich) and the differentiated and subjective ‘knowledge’ (the für sich) of this ‘truth’. For Saussure, the ‘object’ is the systematic relation between differential signifiers and differential signifieds. If we assume a linear mode of explanation we would have to prioritize one of the sides, where one explains the other. But as soon as we do this we lose the integral scientific ‘object’. For both Hegel and Saussure this would entail the disintegration into an irresolvable dualism. The result of this would be that their sciences are lost, because they lose their proper (albeit strange) objects. It would be lost either in the dualistic opposition of ‘natural consciousness’ that divides the integrity of the concept into a subject-extreme and a substance-extreme (Hegel), or in the ‘natural assumption’ of the nomenclature that externally joins two individual ‘substances’ (sound and meaning) in a linear-accumulative way (Saussure). None of these linear explanations can capture the inherently double and extra-antinomial nature of the philosophical concept or the linguistic sign.

VIII. Succession and simultaneity (Saussure)

The main challenge faced by any structural explication is that it cannot be made all at once. It must necessarily show a strict simultaneity, but the discursive performance of the explication will necessarily follow a successive order. As such, the linear explanation has an advantage over the structural explication in so far as its discursive performance mirrors its content: if x explains y, then x will be prior to y, both in the order of explanation and in the order of time. It may be conjectured that the syntactic order or the grammatical subject–object ordering implicitly favours linear explanations. This would help account for the all-too-easy conflation of the linear with the structural. But even if any subjective discursive performance is necessarily successive, it does not ipso facto entail that the objective order is successive. Nevertheless, expressing a non-successive order in a successive way is inherently more difficult than expressing a successive order in a successive way. But, generally speaking, the discursive antecedent does not necessarily have to be fixed as the objective antecedent (and similarly for the consequent). In a structural explication the parallel between the order of discursive performance and the order of objects cannot be sustained.

To clarify this problem Saussure draws the all-important distinction between diachronic (or evolutionary) linguistics and synchronic (or static) linguistics (see CLG: 79–100/114–40). This distinction is drawn with a view to study the proper scientific object of linguistics, i.e., la langue. The failure to draw this methodological distinction results in confused works of linguistics. At the outset it must be emphasized that it is a methodological distinction and not a distinction de re.Footnote ²¹ Saussure's special emphasis on the synchronic aspect of language is due to the fact that it is here that the object of linguistics can be determined. The methodological opposition between the two modes of doing linguistics is absolute, but the results of the two modes are complementary, and they do not divide the world into two disparate ‘objectivities’. Language in itself can contain what for us constitutes ‘irreconcilable’ perspectives that cannot be studied simultaneously. Any (actual) language is a unity of these dual aspects. It is always both in continual variations in the linear flow of time and determined by systematic lateral relations. (Or more precisely: the diachronic changes in language are changes in the synchronic lateral value-relations.)

Despite the fact that any language always contains both axes, the science of language must prioritize the synchronic axis. The reason behind this is that the diachronic axis focuses on the changes within the sign system, whereas the synchronic focuses on the sign system itself (as constitutive of the nature of the sign). Thus, the former ‘presupposes’ the latter. (The scare quotes indicate the proviso that they are only separable for analytic purposes. The object of linguistics (la langue) is only accessible from the ‘axis of simultaneity’ but not from the ‘axis of succession’. Note also that the priority is not of a temporal nature. There is not first a static state and then the evolution of that state. Language is always both at the same time. But methodologically the two axes must first be distinguished, before their actual unity can be appreciated.)

If we therefore bracket the diachronic developments of language and turn solely to the synchronic perspective, the first methodological problem we face is that the medium (discursive language, parole) by which to explicate the synchronic axis is itself successively structured. Yet, it is not successive in the historical, evolutionary, large-scale diachronic sense, but in the small-scale syntactical sense that any single expression in language is linearly and temporally ordered (CLG: 70/103). Syntax is a successive order with formation-rules arranging expressions in a linear chain. The medium itself (as a linear performance) thus seems to run counter to its own theorization in its non-linear systematic aspect. This system is never disclosed as such in any actual expressions. It is for this reason that Saussure distinguishes between language as a system (la langue) and language as it is performed in speech or writing (parole). ‘Synchronic’ and ‘diachronic’ is the subdivision of langue, and not of parole (see the figure in CLG: 98/139). Parole is simply the mass of actual and explicitly expressive language-performances (like, for example, the text you are reading now).

The Saussurean hypothesis—that there exists a language system (langue)—cannot be verified by any direct observation, because we only ever see its indirect effects in writing or speech (parole). It is for this reason that the methodological perspective or the ‘point of view’ is so important for Saussure. He has to create a methodological space wherein he can write successively about an aspect of language that is strictly non-successive (and thus non-observational). In contrast to this, both parole (syntactically ordered) and the diachronic axis (historically ordered) relate to ‘objects’ that are inherently successive. This explains why he devotes so much effort to clarifying the distinction as (methodologically) absolute.

To sum up: synchronic linguistics studies a static state of a language system (langue) at a given time, whereas diachronic linguistics studies the evolution of the language system (langue) over a span of time; and speech or actually performed language (parole) consists of any and all manifest utterances ever produced by speaking subjects. The apparent irony is that we have to explicate the static state of language (synchronic langue) with its own axial opposite, namely the successive parole that is the mode of all discursive expression. But, of course, parole also presupposes langue. So the ‘irony’ is more apparent than real, and is caused by (mistakenly) separating de re what is only a methodological distinction.

Saussure is acutely aware that the structural explications on the synchronic level present a much greater challenge than the linear explanations on the diachronic level:

In other words, a structural explication is more difficult because it cannot rely on the support of the more familiar causal or quasi-causal notion of succession that speaks to the imagination and to the normal (linear) order of representation. It is difficult to determine the synchronic object of linguistics because of its synchronicity, and there is no linear theoretical shortcut to the determination of the structural.

The normal linguistic ‘way out’ of this difficulty is to turn to reference to stabilize semantics with respect to things in the world. This strategy can be called nomenclaturism. In, for example, the act of dubbing some referent we re-inscribe language into a causal (or quasi-causal) order.Footnote ²² With referential dubbing we can, it is believed, causally stabilize and semantically disambiguate the individual terms of language, so that each refers back to some precise spatiotemporal point. Nomenclaturism thus holds that language is fundamentally a collection of names (both proper names and names for natural kinds), each semantically self-sufficient and fixed in space and time by Adamic acts of dubbing.

Now, there is no indication that Saussure would deny the referential aspects of language. What he would deny is that any language can be reduced to these kinds of simple relations (dubbing, pointing, indexing, direct reference, ostension, etc.). This is not the kind of ordering that stabilizes language. If we want to explain how language is acquired it is, of course, necessary to talk of causal or quasi-causal relations. However, language acquisition already presupposes a language. And, strictly speaking, we do not need to know what language is in order to begin to describe how it is acquired. Analogously, we do not need to know what numbers are to begin to describe how children acquire the use of numbers. Polemically put: to say that language boils down to simple referential relations would be equivalent to saying that number theory boils down to counting oranges and apples.Footnote ²³

When Saussure, allegedly, claims that a simple word, such as ‘horse’, is not positive and simple, but only negatively determined by means of what it is not—i.e., that it is nothing ‘in itself’ but only a difference from ‘giraffe’, ‘cow’, ‘donkey’, etc.—it will always seem less intuitively appealing than explaining its meaning directly, by way of reference to something (an exemplary horse) that really positively exists. But would Saussure say that the meaning of the word ‘horse’ is nothing but non-giraffe, non-cow, non-donkey, etc.? Surely not; and I think that the question misses Saussure's point. In this question we already see the shortcomings of the linear mode of explanation. For ‘horse’—as well as any other linguistic unit—is always already embedded within a language system. It is a gross simplification to merely say that a given sign is positively individuated by not being any other sign within a language system.Footnote ²⁴ The individual sign is a sign because it has a linguistic value, and this means that it relates differently to different terms. ‘Value’ is a much more finely modulated term than bivalent ‘exclusion’. Furthermore, we cannot isolate a sign as though it were some atomic and simple unit and then ask how this unit became entangled in a system. Saussure's point is that linguistic units are always system-immanent. The point is not that reference is ‘expendable’, as if we could acquire language in a state of total sense-deprivation. His point is instead that language is never built from simple linguistic units that are pinned to external objects, as the nomenclaturist would claim. For Saussure, it does not really matter how language originated or how we acquire a system of language. What matters is what language is in its most general aspects. But this cannot be explained by way of a summation of simple bottom-up reference relations. Reference is always system-based, whereas not all aspects of the linguistic system are reference-based.

IX. Brandom's linearization of structure

A similar problem to the one described above has occupied some prominent recent commentators on Hegel (though with no mention of Saussure). To keep focus I will only draw on the commentator most relevant for the present context.Footnote ²⁵

In Robert Brandom's essay, ‘Holism and Idealism in Hegel's Phenomenology’ (Brandom Reference Brandom2002: 178–209) the discussion is especially focused on Hegel's idea of ‘determinate negation’. Brandom credits Hegel with being the first thinker of ‘semantic holism’. This holism entails that any determinate individual conceptual content is based on relations of exclusion to other such concepts (183). He (Brandom) subdivides this into a ‘weak individuational holism’ and a ‘strong individuational holism’ (183). The question here is whether negative relations of exclusion on their own are sufficient to individuate semantic content. Can one, for example, understand the content of ‘horse’ solely on the basis of all the semantic relations that ‘horse’ excludes? Can a term be positively individuated only as a result of systematic (holistic) negative relations (i.e., relations of material incompatibility)? Taken out of context this sounds like the central problems invoked by the SC of Saussure, but here it is invoked entirely within a discussion of Hegel.

According to Brandom, the strong version of semantic holism, where the ‘holistic relational structure’ is entirely based on exclusion, is incoherent. Again, his criticism of this position sounds like the SC against Saussure:

For Brandom, it is this basic incoherence of a system based solely on simultaneous relations of exclusion (without any antecedent relata) that requires us to turn to a kind of linear explanation of how semantic holism could ‘get off the ground’. This explanation must begin with the ‘antecedent relata’ as desiderata. For, as he says, ‘without antecedent relata, we cannot really understand the relations (and so the relata)’ (200–1). Brandom's solution is a ‘two-level model’ (201–2, 204–6), and, at times, this model sounds uncannily like a linearized version of Saussure's structural theory of the two-level model of the sign (but it is used to provide an interpretation of Hegel's ‘determinate negation’). Here follows a sketch of this model.

According to Brandom it is (what he calls) ‘signs’ that give us the pre-relational ‘antecedent relata’ that we need to get holism ‘off the ground’. These constitute the ‘first stage’ of his three-staged two-level model. He describes them as ‘primitively individuated signs’ (204): ‘These are things we can immediately distinguish, that is, noninferentially discriminate or tell apart’ (201).Footnote ²⁶ This point is very important for Brandom, because if the ‘signs’ were only relationally distinguished, then they would not provide us with the pre-relational relata. They constitute the ‘given’ antecedents that allow us to start an explanatory line. In a footnote, he adds the following qualification to understanding ‘signs’ as immediately differentiated:

The last line of this note comes very close to resembling Saussure's dictum that the materiality of signs (signifiers) is purely differentially constituted (see CLG: 117–20/163–66). But this is not the point of the note. It addresses another issue, namely that there cannot be a difference in abstracto, i.e., without some actual material ‘content’ to this difference. Immediate difference is not ‘autonomously intelligible’ because we cannot imagine pure difference; it must be a difference between something. However, when Saussure says, regarding signifiers (or ‘sign designs’), that it is only difference that matters, he is not saying that there is some pure difference before there are actual differences between actual signifiers. It is not a linear relation. However, for Brandom, it is linear (but reverse): the ‘mere difference’ is a consequence (abstracted after the fact) of given individual signs. But, as we shall see, the essential importance of the ‘mere difference of signs’ cannot be a consequence of already individuated signs, because it constitutes the necessary structure of possibility of their individuation.

Let us, for now, grant Brandom such ‘already identified and individuated signs’. The next stage is to ‘look at relations among them [the signs]’. In order to explain such relations Brandom invokes some generic ‘community’. We are asked to ‘consider the relation two sign kinds p and q stand in if, in some community, tokening both of them is subjected to a distinctive sanction.’ (Reference Brandom2002: 201).Footnote ²⁸ In a note Brandom expounds a bit: ‘we can imagine the sanction as consisting in the person being (normatively) obliged to alter her disposition to token p and q in some systematic way’ (383, n. 36). In other words, tokening the ‘sign kinds’ p and q would be met with disapproval and, in some unspecified way, corrected (by cultivating the indisposition to tokening these together).

We have now seen the two initial stages (which together constitute the first level) of Brandom's (linear) model: first, we have ‘signs’ (as a given); second, we have the (social normative) practice of sanctioning the simultaneous tokening of two kinds of ‘signs’. This practice provides us with a relation between the pre-relational ‘signs’ (the ‘signs’ in themselves do not stand in any relation to each other). The relation is that of ‘practical incompatibility’. In the third stage (and second level) we ‘define the roles played by signs with respect to that relation’ (201). We can, for example, define new cases of sanction-inducing tokening of ‘sign kinds’ p and q by them being equivalent (though not strictly identical) to the first case. And, furthermore, we can build a whole network by associating one ‘sign’ with all the other ‘signs’ to which it stands in an equivalent relation to the first case of ‘practical incompatibility’. This can be done without needing any explicit sanctioning (practical incompatibility). The ‘sign’ in this way becomes expressive of a content. Thus, we do not just have a level of practical incompatibility (first level), but also a content incompatibility (second level). The content is the set of ‘signs’ to which the ‘sign’ stands in an equivalent negative relation as in the first case of ‘practical incompatibility’. The ‘sign’ is now no longer just an immediate given, but is also expressive of a mediate role (defined, primarily, by its incompatibility content relations to other signs). However—and this is crucial for Brandom's account—such abstract roles are defined by virtue of an underlying structure of normative practical relations among non-relational ‘signs’. In this way, as he tells us, we can escape the ‘paradoxical formulation of strong holism’ (i.e., relations without relata), because the higher level of ‘an abstract relational structure of roles’ is built from the lower level of normative practical relations between ‘signs’ (202).

Without entering the labyrinthine details of Brandom's normative pragmatist inferentialism (which clearly underlies the above account) I want to draw attention to two problematic points that would critically interfere with the linearity (beginning with ‘signs’ as the primitive antecedents) of his account of how holism is supposed to ‘get off the ground’.

The first problem is with ‘signs’. Brandom says that ‘signs’ are immediately differentiated. However, the nature of being differentiated (distinguished, discriminated) is not immediate, but only possible against a differential background. Unfortunately Brandom only provides one example, namely the sentence letters p and q, and these are themselves differentially structured in the most direct way (they are identical, except for the one being the mirror-reverse of the other). Moreover, both are inherently tied to an alphabet system, and thus not merely sensuously differentiated.Footnote ²⁹ I think the issue here is that Brandom's account only leaves us with two alternatives: either ‘signs’ are inferential or they are noninferential. But what if signs are neither? Brandom clearly wants ‘signs’ (in his sense) to be noninferentially given as immediately differentiated, but this givenness is based on a structure. There are indications of this structure when Brandom talks about ‘the mere difference of the signs’ (383 n35). As we saw, this refers to a consequence; but it cannot be a consequence if ‘signs’ are only rendered possible because of this ‘consequence’. This would constitute a vicious circle (and the whole point of Brandom's model is to escape such a circle with the help of some given antecedent). If we acknowledged this circularity, we would instead need to explicate the aforementioned structure. And in so doing we would have to come to terms with the necessary structure of possibility of differentiation, which is an ontological structure that Brandom would most likely want to avoid. (However, I do not think that it can be avoided, even on Brandom's own terms.)

A related question is the following: why does Brandom talk of ‘signs’? He never states this explicitly, but two reasons are discernible: (i) ‘signs’ are, as we have seen, supposed to provide him with something given by which to get a holistic structure off the ground, and (ii)—more in line with the usual meaning of the word ‘sign’—‘signs’ are expressive of a content. For example, when introducing his ‘holistic role abstraction’ model he characterizes it as ‘beginning with signs, and ending with roles played by those signs, or contents expressed by them’ (51). But are (i) and (ii) even compatible? For ‘signs’ to be given means that they cannot have an expressive content, but for ‘signs’ to be signs means that they have an expressive content. How do (immediate) ‘signs’ with no role become (mediate) ‘signs’ with a role? Is this even possible? Is it merely the repetition of the word ‘signs’ in both instances that carries all the explanatory weight?

This brings us to the second problem: that social normative practice is supposed to provide the transition from (i) to (ii). The immediate ‘signs’ are attached to normative practical relations out of which a system of incompatibility relations can be constructed and wherein each ‘sign’ plays a holistic role. But I think it can be shown that such a normative practice already presupposes the transition Brandom wants to explain (and is therefore also viciously circular). To be able to perform such complex actions as administering a distinctive sanction, and to identify a simultaneous tokening of p and q (whatever this involves), as well as follow this up with corrective measures, and to express such authority, the community must already have a developed system of signs. They must, at the very least, be able to signify a case of such tokening as sanction-worthy. And how (and indeed why) would it be sanctioned, if there was not already some larger meaningful social structure, wherein such-and-such actions could be deemed sanction-worthy? Even if we say that this is only embodied in practice, such a practice still seems to require a system of signs (with a degree of expressive power equivalent to those Brandom wants to explain) in order to function.

I will not pursue this discussion further, but only note that I think it is possible to find (semantically essential) irreducible structures ‘all the way down’ on each stage in Brandom's model, even if his aim with this model was to provide a linear explanation of how (holistic semantic) structures emerge in the first place.

As previously mentioned, Brandom's aim for this model is to provide a linearized interpretation of Hegel's ‘determinate negation’, specifically in the dialectical movements from the ‘immediate’ (an sich) to the ‘mediate’ (für sich) to ‘mediated immediacy’ (an und für sich) (202–8). But to critically engage with his reading would require another (long) article. So instead of digressing further into a direct discussion, I will engage it indirectly by presenting an emphatically non-linear interpretation of these movements.

X. The temporal form (succession) and the conceptual form (simultaneity) (Hegel)

Since Brandom mainly refers to the Phenomenology of Spirit in his essay, I will also refer to this work. Let us ask the following question: are the developments of the Phenomenology linear developments? I think the answer is both ‘yes’ and ‘no’—but for very precise reasons (as we shall see shortly). Still, it can already be noted that the philosophical importance of the developments of the Phenomenology is not their linear progressions, but that the developments explicate irreducible structures (i.e., concepts).Footnote ³⁰

In the dense last chapter called ‘Absolute Knowing’ we read that although the developments surveyed in the text appear in time, its aim is to account for something fundamentally atemporal, or conceptual:

There are many things happening in this passage, drawing in many different directions. First, Hegel seems to equate time with the concept, but only the concept that is there (‘der da ist’). What Hegel means by the ‘temporal form’ (Zeitform) is the succession whereby one thing, say A, becomes something else, non-A, i.e., it is the linear movement from A to non-A. But the problem with this temporal-linear form is that the terms exclude or replace each other. It is therefore not a simultaneous structure, but a succession. First there is only A, then A changes into non-A (and there is now only non-A, for A has passed away). But this is not the ‘conceptual form’. In this form there is a strict simultaneity (atemporality) between the two terms. The problem is that when expressed (in time) their simultaneous relation always appears as a transition, where one seems to replace the other.

Any event or performance, including thinking, is a successive process in time, so there is nowhere for thinking to appear except in time. Yet, this does not mean that what is being thought should be understood as being temporal. Or more precisely, just because an explication is necessarily performed successively (one thing after another) or manifested in a ‘linear’ manner, it does not entail that the object of the explication is itself therefore successive or linear.

Hegel absolutely needs the transitional- or becoming-like exposition in order for it to function in a demonstrative way. One part must necessarily develop (become, transition, etc.) into the other. However, it must be a kind of inclusive transition. This means that it is a transition between moments that already necessarily make up a systematic whole. The transition takes place because one part is necessarily simultaneous with another part. When Hegel speaks of the movement of cognition (Erkennen) or of thinking we should not picture it as a temporal linear movement, but as a movement that establishes the static structure that was already there at the outset, and whereupon each moment necessarily depends. Hegel makes this clear already on the first pages of the ‘Preface’ to the Phenomenology, where he presents the famous metaphor of circular organic growth. The successive moments from bud to blossom to fruit should not be seen merely as so many separate independent individuals, where one excludes and supplants the next. It is this one-sidedness that gives rise to a linear successive explanation. Instead, their continuous temporal development is (paradoxically) indicative of their deeper unity (or simultaneity, in the sense that no one part can have priority over the others):

The stages of the plant's development are, despite being temporal, also circularly unified (in so far as the last stage is not just a result of the first stage, but the first stage is also a result of the last stage). The plant's successive stages are in some sense objectively separable (I can nip a bud, pick a flower, pluck a fruit), but this separation provides a distorted picture of the ‘life of the whole’ of the plant (because there can be no flower without the bud, no fruit without the flower, no bud without the fruit (seeds), etc.). The successive exposition of the concept is itself meant to prove the inseparable non-successive nature of the conceptual ‘moments’ as a structure (and the structure is deep, because ‘picking it apart’ means that it, and the parts, totally vanish). The ‘concept’ is, in general, a kind of unity (Einheit) that sublates the one-sidedness (Einseitigkeit) of a separation of the successive moments.

A philosophy of the concept therefore has to be systematic. For a system is an irreducible relation between simultaneous terms, and not a succession or connection of pre-existing terms. The successive nature of an explanatory exposition necessitates a developmental (organic) kind of argument, where the unity of its parts is preserved despite its stage-wise progression:

In his writings Hegel consistently formulates this point in terms of a folding or a turning back upon itself of the linear-successive form of the exposition. In his conceptual exposition a change, a differentiation, or a becoming initially resembles a temporal (successive) development. However, the temporal form is sublated in the logical movement of folding back on itself, whereby it assumes the (true) conceptual form. It deprioritizes the parts into a unified conceptual whole. It is no longer a succession of passing stages, but a structure of simultaneous moments. Hegel calls this structural becoming the ‘simple becoming’, which is meant to capture that it is a becoming (and thus a differentiation) that does not undermine the simple unity of the whole. In other words: the ‘moments’ are necessarily simultaneous, inseparable, and irreducible to the linearity of their mode of exposition.

As already noted, Hegel often describes this simultaneity with the traditional image of a circle, and the whole Logic as a ‘circle of circles’ (WL: 751/II.571). It is this (non-vicious) circular logical movement that breaks with linearity. It is an atemporal becoming of a ‘whole’ out of a ‘part’, because the part is only rendered possible in that whole. It therefore becomes what it already is ‘in itself’: ‘its becoming what it, the substance, is in itself’ (PhG: 462/585, translation modified). And logical ‘absolute’ cognition is the exposition of a non-linear concept: ‘the movement which is cognition … is the circle returning back into itself, which presupposes its beginning and reaches its beginning only at the end’ (PhG: 462/585). The difficulty of reading Hegel can (partly!) be explained by the difficulty of explicating the structural in a medium (i.e., parole) that is inherently linear.

In general, it can be said that the scientific form of Hegel's exposition always begins with some ‘untrue’ point of departure presupposed in a normal linear account. But it is from the impossibility of such an abstract insular point that the underlying structure has to be unearthed. The actual truth of the ‘untrue’ beginning is, as Hegel puts it, the resultant whole (PhG: 13ff./19ff.). The exposition is a ‘static’ movement from immediacy to mediation and back, recollecting the parts by showing that the possibility of each resides only within the total structure:

It is only when the linear mode of explanation has been ‘sublated’ into the structural mode that immediacy and mediation become compatible (and not contradicting opposites). A conceptual structure, in this precise sense, is an immediate mediation.

XI. How to make content out of form (necessary structural possibility in Hegel and Saussure)

To end this article I return to the SC mentioned at the beginning. In the SC of Hegel, his logic is mockingly portrayed as claiming to somehow generate extra-conceptual content out of conceptual form; in the SC of Saussure, his linguistics is likewise mockingly portrayed as claiming to somehow generate extra-linguistic content out of linguistic form. But such portrayals deeply misunderstand both Hegel and Saussure. The core of the misunderstanding lies in the separation of form and content, and the linear relation between the form and content. Neither Hegel nor Saussure holds that ‘form’ can generate ‘content’ in some linear way. They reject the notion that ‘content’ can be given prior to, and separately from, ‘form’. In so far as they argue that ‘form generates content’, it is always in a strictly non-linear way. And any claim to the effect that there is an immediately ‘given’ content can be shown to have a form that is (immediately) mediate. Without this form the purported, given content would not be possible (in a non-linear sense). Neither a representationalist form/content dualism nor a nomenclaturist sign/reference dualism can account for this, for they can only take the sides of the dualism as ‘given’, prior substances. As such, they are unable to non-dogmatically answer questions concerning the ‘given’. It is simply that—given. To answer such questions a structural explication is indispensable, since the linear explanation becomes incoherent (e.g., regressive or viciously circular) when forced to account for the (dogmatic) ‘given’. If content is posited as something ‘in itself’, or, alternatively, if form is posited as something ‘in itself’, the explication of such posits will reveal that there is always already content in form and form in content. The abstraction of the two is a metaphysical fiction. If we abstract one side and give it priority over the other, then what results is the dialectical transition into the opposite (form to content, and content to form). However, this transition is not linear, but structural.

Despite the emphasis on the non-priority of the ‘sides’ or ‘parts’ of the structure, there is a kind of priority of the structural (over the linear). But this priority is itself not linear. It is not that the structural ‘comes before’ the linear in some chicken-or-egg kind of way. It is rather that since the (structural) unity of form and content must be there ‘from the beginning’, the beginning of the linear must itself presuppose the structural. This thwarts the self-sufficiency of any linear explanation. In general: the structural can explicate how the linear is rendered possible, but the linear cannot explain how the structural is possible. In Saussurean terms: the theory of la langue can explicate the possibility of parole, but parole cannot explain the possibility of la langue (because parole must itself necessarily presuppose a language system, however primitive, to be possible). In Hegelian terms: der Begriff can explicate the possibility of ‘given’ determinate individual objects and subjects, but such individual objects and subjects cannot explain the possibility of der Begriff (because they too must necessarily presuppose it to be possible). Nevertheless, la langue and der Begriff are not linearly prior to parole or to determinate individual objects and subjects, and they do not ‘generate’ manifest parole or manifest objects or subjects in some linear (causal or quasi-causal) way. If the term ‘condition’ was not already so linear, la langue and der Begriff could both rightfully be called deep structural conditions. Instead it would, I think, be more fitting to call them by another name: the necessary structures of possibility (they are also ‘deep’ in the sense that there is no deeper condition that is not itself also a necessary structure of possibility). Thus we can characterize Hegel's logic as a science of the ‘necessary structures of possibility’ of anything being determinate at all. And similarly for Saussure, we can characterize semiology as the science of the ‘necessary structure of possibility’ of anything being significant at all.

If we read these necessary structural possibilities as linear (i.e., causal or quasi-causal) accounts of how parole or determinate objects or subjects are ‘produced’, we are bound to end up in all kinds of absurdities. It is exactly this mistake that the SC directed against Hegel and Saussure is guilty of.

Against Saussure the ‘standard critic’ will do the following: he will try to corner Saussure by saying that adding up these indeterminate formal and negative differences (implicitly: in a linear manner) will never yield a positive content. He will, instead, turn to reference for support, and tell a linear narrative that equates meaning with the referential content. But in so doing he will already have presupposed as ‘given’ the meaningful structure that he is attempting to explain. For reference is not primitive or a simple act of establishing the content of signification, but is necessarily bound up with the differential form of language from the outset.

In a similar way, the ‘standard critic’ of Hegel will claim that he (Hegel) cannot explain how concepts ultimately relate to actual beings in the world or to the facticity of being. But, for Hegel, there is no need to prove that the logical structures have such content, because the notion that there is any kind of ‘pure’ unstructured content (‘pure being’) has been debunked. So Hegel's concepts will not have to be ‘applied’ to individual beings (or being ‘as such’), because they always already apply in so far as they account for the necessary structure of possibility of any individual being at all. There is no unformed content-in-itself (or being) that can be given priority. For anything to exist ‘in itself’ is already for it to be structurally determined, i.e., it is differentially self-related, or ‘for itself’. To be merely in itself is structurally insufficient for existence. There is therefore no given fact prior to its necessary structure of possibility.

In short: the structural explication accounts for the deeper non-linear possibility of that which the linear explanation has already (implicitly) presupposed as given.