Language as Vision: The Ocularcentrism of Chomskyan Linguistics by Chris Werry
Note: this paper is based on material in my dissertation, Rhetoric and Reflexivity in Chomskyan and Cognitive Linguistics (Carnegie Mellon University, 2002). Please do not quote from this paper. If you would like to read or quote from my dissertation, email me at firstname.lastname@example.org.
References to vision pervade Chomsky’s work. They are a key component of the figures he uses, the examples he provides, the analogies he makes, and the argumentative warrants supporting his central claims. When dealing with opponents Chomsky repeatedly exploits the rhetorical potential of visual analogies and metaphors in order to construct rebuttals. References to vision and to spatio-visual phenomenon constitute a key component of the most characteristic rhetorical moves Chomsky makes, and are central to the way Chomsky defines the project of linguistics. From Syntactic Structures to his most recently published texts, Chomsky’s writing is permeated by a constellation of terms centered on space, vision, optics and form. This is perhaps not altogether surprising, given that Chomsky is a thinker who identifies so strongly with Descartes, and who describes his theoretical project as “Cartesian”. In Modernity and the Hegemony of VisionDavid Levin argues that Descartes is the modern philosopher most obviously indebted to the metaphor of knowledge as spatio-visual, a writer whose work most clearly exemplifies a discourse that is dominated by an ocular metaphoric. Levin writes:
For Descartes, darkness is a nightmare. There is nothing to be learned from entering its domain. He is the philosopher obsessed with clarity and light. If a discourse in which light, vision, and its metaphorics are constitutive of its very logic may be called ocularcentric, then it would be difficult to deny that Descartes’ philosophy exemplifies ocularcentrism.
Correspondingly, Chomsky is perhaps the post-war linguist most focused on clarity, light, space and vision. While a range of figurative expressions characterize generative discourse, ocular metaphors are assigned a place of particular importance. It might be said that the generativist vision of language has been articulated primarily in the language of vision. It has tended to be heavily “ocularcentric”, to use the term coined by Martin Jay. The assumption that language exists primarily as something visible; that it can be represented in terms of comparisons with various forms of visual phenomenon, and that linguistic analysis can be understood as a kind of visual perception, has been integral to the objects, concepts, data and methodologies that characterize generative inquiry, and is deeply embedded in its conceptual logic. This tendency has shaped the field’s predominant “ways of seeing” language, and in particular its treatment of grammar.
This paper explores the ocular metaphoric that characterizes generative discourse. It demonstrates the degree to which vision constitutes the central analogical figure in Chomsky’s writings. The paper examines the role visual imagery plays in Chomsky’s main arguments, analyzes why it is so important to his theoretical framework, and considers the rhetorical work it is used to carry out. I discuss how ocularcentrism shapes Chomsky’s understanding of what is central to linguistic inquiry and what is peripheral; what is “visible” and “invisible” within his theoretical framework, and what the nature and limits of linguistic inquiry are. The first section of the paper provides examples that show how pervasively Chomsky makes use of visual references.
1.1 Examples of Chomsky’s Vision-Centered Vision of Language
Chomsky’s discussions of linguistic theory often begin with references to the rationalist tradition, and make reference to figures such as Plato, Descartes, Leibniz and Cudsworth. The rationalist framework upon which Chomsky’s work is built is articulated and justified largely in terms of references to vision. When arguing for the basic soundness of rationalism, and by extension his own theories of language, Chomsky regularly cites the nature of the visual faculty as the central piece of evidence for rationalism. For example, in a discussion of rationalism in Knowledge of Language Chomsky writes that “the Platonistic conception would suggest that knowledge of a particular language grows and matures along a course that is in part intrinsically determined, with modifications reflecting observed usage, rather in the manner of the visual system” (Knowledge of Language, page 2). In Reflections on Language Chomsky states that his views of language are based on rationalist presuppositions, and that while certain aspects of rationalism must be updated, the overall approach is valid. As evidence for the plausibility of rationalism, Chomsky cites research indicating that the visual system is “wired in”, and that human infants are equipped from birth with an inbuilt ability to perceive and process three-dimensional space. Chomsky argues that the same is likely true of the linguistic system, and the idea that language can be considered in terms analogous to a bodily organ such as the visual system is “quite natural and plausible” (11). The plausibility of a rationalist theoretical framework is articulated primarily through references to vision and the visual system.
Chomsky’s discussions of rationalism are typically used to introduce one of the central claims of generative linguistics, namely that language is an innate faculty of the human mind. Chomsky’s arguments about the innate character of linguistic knowledge turn repeatedly on references to and analogies with vision. His argument for the plausibility of innateness rests on establishing that vision constitutes a particularly compelling example of an innate, biologically based cognitive faculty, and that important similarities exist between language and vision. From his earliest works to his latest writings, Chomsky uses ocular metaphors and references to support the innateness hypothesis. In Aspects of a Theory of Syntax he writes that “on the basis of the best information now available, it seems reasonable to suppose that a child cannot help constructing a particular sort of transformational grammar to account for the data presented to him, any more than he can control his perception of solid objects or his attention to line and angle.” (Aspects, 59.) In Reflections on Language Chomsky’s arguments for innateness are based on a series of references to the nature of visual perception: the “wired-in” character of the visual system; the inbuilt ability of the infant perceptual system to understand various aspects of 3-dimensional space; the notion that an innate “grammar of vision” may be built into the human nervous system (Reflections, page 8.) In Rules and Representations Chomsky states that the language faculty may be thought of as an innate “mental organ” analogous to the human visual system (39). In Language and Thought he asserts, “the child’s language ‘grows in the mind’ as the visual system develops the capacity for binocular vision.” (page 29). In a recent interview in which Chomsky outlines the main components of his theory of language, most of his major points are made through examples, comparisons, or research having to do with visual perception. On the issue of innateness, Chomsky writes:
If I say, “He thinks that John is intelligent,” you know that “he” is somebody other than John. If I say, “his mother thinks John is intelligent,” “his” may refer to John. Those are things we aren’t taught. The basic features of language structure are built into our biological nature. They’re like the principles that determine why mammals see things in a particular way and insects see things in a different way.
Supporters and expositors of Chomsky’s works frequently use similar language when discussing innateness. For example, throughout Smith’s book on Chomskyan linguistics innateness is supported through appeals to the similarity between language and vision. Comments such as the following are typical: “in vision as in language our innate endowment is such that minimal input can give rise to great richness of knowledge. Arguments from the poverty of the stimulus are comparable in the two domains.” (Smith, page 94.) Smith writes that development of the language faculty does not proceed through learning or interaction, but is instead “more like the development in the child of stereoscopic vision”. (Smith, 117).
Chomsky’s “innateness hypothesis” leaves little room for learning, experience or practice in the development of linguistic abilities (the marginal status of learning is signified by the fact that it is often put in quotation marks.) The acquisition of language is represented in terms of the unfolding of a biological program. The nature of this program, and Chomsky’s explanation of why learning has such a minor role in the overall process, is typically discussed in terms of analogies with visual perception. For example, in Rules and Representations Chomsky writes:
If the general structure of binocular vision is genetically determined, then naturally we must seek to explain its origin in terms of biological (evolutionary) development rather than in terms of learning. Exactly the same is true when we turn to cognitive structures or the (unknown) physical mechanisms that underlie them. If, say, we find extensive evidence that the principles that underlie the wh-island constraint belong to universal grammar and are available to the language learner without experience, then it would only be rational to suppose that these mechanisms are genetically determined and to search for a further account in terms of biological development. (209)
The development of linguistic ability is described as activated through the triggering effects of experience, or, in Chomsky’s later work, as the setting of parameters. This process of development is also described and supported largely in terms of parallels with visual development. For example, in Reflections on Language Chomsky argues that development takes place via the triggering effects of external stimuli on innate faculties. Chomsky’s argument for this position is constructed through references to work on the development of visual perception. He writes that “work of the past years has shown that much of the detailed structure of visual experience is ‘wired in’, though triggering experience is required to set the system in operation”. (Reflections, page 8.) In Aspects of the Theory of Syntax Chomsky uses figures and examples that center on visual perception to describe the role that exposure to the environment has on the triggering of language acquisition. For example, when arguing that exposure to language is entirely “peripheral” to linguistic development, Chomsky draws on parallels between the role learning plays in the acquisition of language, and the role it plays in the “acquisition” of vision in humans and in animals. He writes:
This distinction [that acquisition requires exposure to the environment, yet this exposure plays no determining role in development] is quite familiar outside of the domain of language acquisition. For example, Richard Held has shown in numerous experiments that reafferent stimulation (that is stimulation resulting from voluntary activity) is a prerequisite to the development of a concept of visual space, although it may not determine the character of this concept…or, to take one of innumerable examples from studies of animal learning, it has been observed (Lemmon and Patterson, 1964) that depth perception in lambs is considerably facilitated by mother-neonate contact, although again there is no reason to suppose that the nature of the lamb’s ‘theory of visual space’ depends on this contact. 33-34.
And when Chomsky takes up the issue of the triggering effects of the environment in Rules and Representations, he turns once again to the example of visual development. He writes that the factors that shape the development of language “are on a par with the factors that determine that a child will have binocular vision”, in the sense that both language and vision:
develop in the individual along an intrinsically determined course under the triggering effect of appropriate social interaction and partially shaped by the environment – English is not Japanese, just as the distribution of horizontal and vertical receptors in the visual cortex can be modified by early experience. (Rules, page 44-5.)
The description of language development in terms of visual analogies is common not just in Chomsky’s work, but also in the wider generative literature. Consider, for example, Smith’s argument for the viability of a Chomskyan theory of linguistic development:
We do not ‘learn’ how to see in color or in 3-D…. We obviously need some visual input for the process whereby we become able to see in this way to be triggered: being brought up in total darkness will guarantee that one sees not at all, hence not in 3-D or in colour. Similarly, we need to be exposed to faces for the appropriate development of the face recognition module to be triggered; and we need to be exposed to examples of language in order to acquire normal knowledge of a language. Smith, page 26.
Smith’s discussions of acquisition, linguistic development, and learning are couched in similar terms. Smith argues we no more “learn” language than we “learn” how to see, and uses a series of analogies with visual perception to construct a distinction between the development of innate, internal abilities, and the development of abilities derived from external stimulation. In a statement that typifies Smith’s treatment of language, he writes: “In vision, the contrast between developing stereoscopic vision and learning to identify different kinds of moth is clear; in language the contrast between acquiring the principles of binding theory and learning the vocabulary of biochemistry is similarly not in dispute.” (133)
Visual references and examples are used not just to advance points concerning the object of inquiry, but are also used when talking about the theoretical approach to language adopted by generative linguistics. For example, when explaining how generative theories of language can simultaneously consider physical mechanisms, theoretical abstractions, and several other levels of analysis, Chomsky frequently explains this practice in terms of a hypothetical approach to the study of vision. For example, after describing how generative theories of language refer to both physical mechanisms and theoretical abstractions, Chomsky argues that a similar situation exists in theories of vision. He writes:
In the same way, a theory of vision might be formulated in concrete terms, referring, say, to specific cells in the visual cortex, and their properties; or it might be formulated abstractly in terms of certain modes of representation (say, images or stick sketches), computations on such representations, organizing principles that determine the nature of such representations and rules, and so on. (Rules & Representations, page 5.)
Chomsky frequently supports the theoretical approach taken by generative linguistics via analogies with hypothetical theories of vision. For example, Chomsky states that a generative theory of language is similar to a theory of visual perception in that both do not concern themselves with evidence taken from the intuitions or self-report of particular subjects, but must instead focus on what the subject “actually” knows or sees:
A generative grammar attempts to specify what the speaker actually knows, not what he may report about his knowledge. Similarly, a theory of visual perception would attempt to account for what a person actually sees and the mechanisms that determine this rather than his statements about what he sees and why. Aspects, page 8.
Furthermore, Chomsky supports key aspects of his theoretical framework (idealization, abstraction, “systematic ambiguity” between theoretical and physical objects, the use of small amounts of data taken from introspection) via analogies and examples of how various studies of vision might, in principle, proceed. For example, in Rules and Representations he writes:
There is a familiar morass of problems about just what is meant when we take a theory to be true: what is the status of its theoretical entities, its principles, its idealizations…Consider again the study of vision. Suppose that some series of experiments leads to the conclusion that particular cells are sensitive to lines with certain orientations. In this case, no special problems are held to arise, though of course the conclusion is underdetermined by evidence, the cell is abstracted from its environment, nothing is said (at this level) about the mechanisms that might be responsible for what the cell is alleged to be doing, the results are obtained under highly idealized conditions built into the experimental situation and apparatus, and so on. (p. 105)
Chomsky presents hypothetical studies of vision in order to legitimate the methodological principles adopted by generative linguistics. He assumes that the study of vision is a natural analogue to the study of language, and that methodological choices that are legitimate in the study of vision therefore apply also to language. When defending the practice of abstraction and idealization Chomsky turns to visual analogies particularly often. The quotation below captures this rhetorical move nicely:
Consider the matter of ‘defining’ such notions as ‘human language’ or ‘French’. The former task poses questions of a familiar sort, as when we speak of the ‘human visual system’ as distinct from that of cats or bees. In speaking of these terms, we abstract away from individual differences and from interconnections among systems, focusing on one (idealized) element of a complex integrated whole…We might postulate that some principle for identifying the structure of a body in motion is a property (perhaps innate) of the human visual system, and that some principle (say, opacity or locality) is a property of the human language system. In this respect, it seems to me useful and appropriate to think of the human visual system or human language as analogous to an organ or bodily system, and to try to characterize their properties.
Throughout Rules and Representations, the text from which the above two quotations are taken, issues concerning what is properly scientific, what are adequate theoretical assumptions, and what the proper methodological framework for the study of language ought to be, are discussed in terms of figures, comparisons and hypothetical studies concerning vision.
One of the most characteristic aspects of Chomskyan linguistics is the way it constructs a fundamental isomorphism between language and vision, an alignment that works to superimpose research on vision with generative research on language, and vice versa. This alignment takes many forms. To begin with, Chomsky singles out research on vision as the only other area of the cognitive sciences that has had success that is comparable to generative linguistics’. Thus he writes: “In the specific empirical areas of the so-called cognitive sciences, some have been doing pretty reasonably, like the study of vision and the study of language”, and “there are pretty successful computational theories of, say, vision, and language…they achieve some pretty surprising things”. In considering whether “there other areas of human competence where one might hope to develop a fruitful theory, analogous to generative grammar”, Chomsky writes that one might “consider the problem of how a person comes to acquire a certain concept of three-dimensional space.” (Language & Mind, page 73.) Chomsky argues that visual processing is a prime candidate for being studied in a “similar way” to language, because the visual system operates in a way that is fundamentally akin to language (that is, vision is an innate system that develops “in a more or less uniform way on the basis of restricted data.”) Chomsky often talks about vision as if it is amenable to the kind of analyses carried out by generative linguists. Chomsky is thus particularly interested in the work of researchers such as Richard Gregory, who have argued that language and vision are based on common cognitive ground, and thus must be considered together. Chomsky cites Gregory’s work, and states that this work suggests that:
There may be a ‘grammar of vision’, rather like the grammar of human language, and possibly related to the latter in the evolution of the species. Employing this ‘grammar of vision’ – largely innate – higher animals are able to ‘read from retinal images even hidden features of objects, and predict their immediate future states’, thus ‘to classify objects according to an internal grammar, to read reality from their eyes’. (Reflections, page 8)
One of the most frequent analogies Chomsky’s makes involves drawing parallels between language and the faculty (proposed to exist by some cognitive scientists) that infants have for visually identifying faces. Chomsky proposes that children possess an innate “grammar of faces,” and writes:
Consider recognition of faces. A person can recognize an enormous number of human faces and identify a presentation of a single face with various orientations. This is a remarkable feat that cannot be duplicated with other figures of comparable complexity. It might therefore be interesting to try to develop a ‘grammar of faces’, or even ‘a universal grammar of faces’, to explain these abilities. Perhaps, at some stage of maturation, some part of the brain develops an abstract theory of faces and a system of projection that allows it to determine how an arbitrary human face will appear in a given presentation. Rules and Representations, page 248.
Some of the most explicit and extended comparisons between vision and language occur in Chomsky’s discussions of the computational neurophysiologist David Marr. Chomsky describes Marr’s work on vision as highly compatible with his own research, as evidence for generative linguistics, and argues that generative work on grammar and Marr’s work on vision constitute the most advanced research in cognition. Chomsky reserves his highest praise for Marr, calling his work “interesting”, “real science”, and “analogous to what we are doing.” Chomsky writes:
They [Marr and his group] are interested in developing systems of representation and levels of representation which will on the one hand have a basis in physiology, if they can find it, and on the other hand will account for important perceptual phenomena…what they are doing is, in a sense, artificial intelligence, but what we are doing is in the same sense artificial intelligence…what they point out is that there are a number of different levels of investigation that you might imagine. First there is the level at which you deal with particular elements like diodes and neurons, and that is biochemistry, or physiology. Second, there is a level at which you talk about bigger units, assemblies or groupings of these units, and that is more abstract. Third, there is a level at which you talk about algorithms, actual procedures, parsing procedures, etc. And finally, there is what they call the theory of the computation, where you try to abstractly characterize the general properties of the system and how it functions and what its nature is. They argue, plausibly I think, that the theory of the computation is the most crucial and also the most neglected level of research, and that the most fundamental work will be at the level of the theory of that computation. Chomsky 1982, pp. 8 – 10
Chomsky argues that Marr correctly differentiates between levels of conceptual analysis, just as his own theory does (and in stark contrast to Chomsky’s critics, who Chomsky suggests unfairly criticize him in this regard.) Chomsky notes that Marr’s research proposes that vision is innate, algorithmic, and that the same kinds of restrictions on reflexive access to innate knowledge characterize both vision and language. Both research programs are strongly computational and focus on “real systems and their nature.” (11) Chomsky often expresses a sense of kinship between his work and Marr’s. He applauds Marr’s criticism of competing areas of AI that concentrate on too “concrete a level” of analysis, or on human behavior rather than on more conceptual/systemic considerations; he sympathizes with Marr’s claim that scientific work on vision has neglected more abstract levels of analysis, and in particular theoretical approaches that are computational in character. Chomsky argues that the four levels of analysis described by Marr provide an important methodological model for generative linguistics, one that parallels and validates Chomsky’s own methodological choices. Chomsky writes:
He [Marr] and his colleagues distinguish four levels:
At the lowest level there, there is basic component and circuit analysis – how do transistors (or neurons), diodes (or synapses) work? The second level is the study of particular mechanisms: adders, multipliers and memories, these being assemblies made from basic components. The third level is that of the algorithm, the scheme for a computation; and the top level contains the theory of the computation.
They suggest that the ‘top level is the most neglected [and] also the most important’, and that current research is misguided in constructing algorithms without appropriate prior understanding of the top level. Adopting this framework, we may consider the study of grammar and UG to be at the level of the theory of the computation.
Marr’s research is represented as homologous with Chomsky’s in terms of scientific goals, approach, categories, concepts and methodology. Marr’s work on vision thus functions as key exemplar, analogy and support for generative linguistics.
The references Chomsky makes to Marr, along with the various other examples cited above, help indicate just how central vision is to generative linguistics, and the degree to which Chomsky constructs language in terms of an equivalence with vision. One area of Chomsky’s writing that illustrates particularly well just how powerfully this equivalence functions can be seen in the sections devoted to rebuttals. Chomsky continually draws on references to vision and visual perception in the counterarguments and refutations he constructs (Chomsky is justly famous for his skill in dealing with opponents, and for his ability to construct rhetorically powerful rebuttals and refutations. Yet one of the most distinctive characteristics of these rebuttals – his use of visual analogies, metaphors and examples – appears to have gone unnoticed by both critics and admirers of Chomsky.) This is a standard move, a rhetorical strategy that Chomsky returns to with extraordinary frequency. For example, in Reflections on Language Chomsky engages in a sustained debate with the communication theorist P.F Strawson, in which Strawson’s criticisms are answered with rebuttals based on visual analogies. Chomsky cites Strawson’s argument that linguistic rules are social, conventional, public phenomena, and Strawson’s claim that generative linguistics ignores this. Chomsky considers Strawson’s argument that ‘the function of communication remains secondary, derivative, conceptually inessential’ in generative linguistics, and a proper account of learning is not given. Chomsky responds:
The question whether communication is ‘primary’ and ‘conceptually essential’ is begged throughout Strawson’s counterarguments. Furthermore, the picture that Strawson rejects as perverse and arbitrary seems quite reasonable and probably correct…The organism is so constituted that it acquires a system of language that includes ‘meaning-determining rules’ (again, perhaps, in interaction with other faculties of mind.) These rules are then used by the speaker to express his beliefs (inter alia). The learner has no ‘reason’ for acquiring the language; he does not choose to learn, and cannot fail to learn under normal conditions, any more than he chooses (or can fail) to organize visual space in a certain way…As for the fact that the rules of language are ‘public rules’, this is indeed a contingent fact. Having acquired the system of language, the person can (in principle) choose to use it or not, as he can choose to keep to or disregard his judgments concerning the position of objects in space. He cannot choose to have sentences mean other than what they do, any more than he can choose to have objects distributed in perceptual space otherwise than the way they are…. It is a fact of nature that the cognitive structures developed by people in similar circumstances, within cognitive capacity, are similar, by virtue of their similar innate constitution. Thus we share rules of language with others as we share an organization of visual space with them. (Reflections on Language, page 71.)
As is often the case, the force of Chomsky’s counterargument rests largely on an assumed equivalence between language and vision, and his critics’ failure to appreciate this. He consistently replies to criticism concerning the methodological adequacy of generative linguistics (the focus on abstraction, idealized conditions, data based on introspection, etc.) by comparing the study of linguistic and visual systems. Chomsky argues that the same issues arise in the study of vision, yet this isn’t considered a problem. Since language is the same kind of object as the visual system, such criticisms have no force for generative linguistics. Such a strategy is used throughout the debates Chomsky conducts in Rules and Representations. For example, Chomsky outlines the objections commonly leveled at generative linguistics, and then argues for their irrelevance via an extended comparison with vision. He concludes his discussion with the following statement:
Consider again the study of vision. Suppose that some series of experiments leads to the conclusion that particular cells are sensitive to lines with certain orientations. In this case, no special problems are held to arise, though of course the conclusion is underdetermined by evidence, the cell is abstracted from its environment, nothing is said (at this level) about the mechanisms that might be responsible for what the cell is alleged to be doing, the results are obtained under highly idealized conditions built into the experimental situation and apparatus, and so on. Suppose next that it is proposed that identification of objects involves analysis into stick-figures or geometrical structures, though nothing is said or known about neural mechanisms that might carry out such analysis. Is the situation fundamentally different in some way, apart from the physical and chemical properties of the brain? It is not clear why one should assume so.
Chomsky’s rebuttal is predicated on the notion that the study of language and vision are largely equivalent when it comes to methodological considerations. Similarly, in response to Putnam’s objections to the generative concept of innateness and the marginal role assigned learning, Chomsky writes that ‘He [Putnam] makes the tacit assumption that language is cognitive in a way that vision is not, and hence that discussions of language have to meet additional criteria of adequacy.’ (172) InKnowledge of Language Chomsky engages the critiques of cognitivists who take a connectionist approach to the study of language and mind. Chomsky opposes the connectionist argument that preexisting structure and innate mechanism are not required in order to explain linguistic and cognitive structures. His frequent response to connectionist opponents consists of the counterargument that they cannot explain vision without reference to innate mechanisms and preexisting structure, yet they try to do with language. Since language is a system that is analogous to vision, such an approach must therefore be flawed. Chomsky argues against the relevance of functional approaches to linguistic inquiry with this memorable riposte: ‘The child does not acquire the rule by virtue of its function any more than he learns to have an eye because of the advantage of sight.’ (Rules and Representations, page 231.) And when arguing against the notion that language is something constructed by its users, Chomsky often fashions rebuttals in terms of analogies between language and the visual system. Thus in Rules and Representations Chomsky opposes constructivist theories (as suggested by Vico and Rorty) in the following terms:
Have we, as individuals, ‘made’ our language? That is, have you or I ‘made’ English? That seems either senseless or wrong. We had no choice at all as to the language we acquired…there is no more reason to think of language as ‘made’ than there is to think of the human visual system and the various forms that it assumes as ‘made by us’. (Rules & Representations, page 11.)
For Chomsky, language is self-evidently the same kind of object as the visual system, and since we do not ‘make’ our visual faculties, it makes no sense to imagine language as socially constructed.
In general, vision is a central analogic figure in Chomsky’s writings. His work is full of expressions that suggest fundamental parallels between the visual and verbal domains. Often these parallels are broad (as for example when Chomsky writes that research on vision “is highly suggestive for the study of cognitive structures such as language”, or states that when it comes to the biological development of the visual system, “comparable conclusions seem to hold in the case of human language”.) Sometimes they are specific (as for example when he outlines his computational approach to grammar, then writes that “by parity of argument, a scientist studying vision might develop a theory involving certain types of computation and representation for the identification of objects in motion.”) Ocular metaphors, examples, and analogies pervade Chomsky’s work, and
references to actual or hypothetical research on visual systems function as important sources of support, authority and evidence for generative linguistics. However, Chomsky’s comparisons between language and vision rest on a very distinctive and selective understanding of vision. In the sections that follow I examine Chomsky’s vision of vision, consider where it comes from, and how it is used.
1.2 Chomsky’s Construction of Vision
It is worth looking in some detail at precisely how Chomsky understands vision, for Chomsky’s comparisons between language and vision rest on a very particular, and rather idiosyncratic understanding of vision. He is very selective in the examples and research he cites, and he constructs a model of vision that affords specific rhetorical uses. In this section I look at the particular representation of vision Chomsky constructs, how this relates to Chomsky’s theory of grammar, and consider some of the rhetorical uses this model of vision is put to.
1.2.1 Chomsky’s Understanding of Vision is Cartesian and Computational
Descartes’ theory of vision plays an important role in Chomsky’s linguistics. Chomsky uses it as evidence for innateness, and describes the Cartesian theory of vision as crucial for the development of computational approaches to mind. Describing how his work is prefigured by the “Cartesian Revolution” of the seventeenth century, Chomsky writes:
Inner mechanisms and inner processes appear to be computational systems, mentally representative and, in some unknown manner, physically instantiated. But that again is highly reminiscent of something that took place in the seventeenth century – in particular, Descartes’ theory of vision, which was a crucial breakthrough and developed a kind of representational, computational theory of mind.
Descartes’ theory of vision is presented as a “crucial breakthrough”, and as forerunner to the ‘cognitive turn’ that Chomsky has become synonymous with. For Chomsky, the Cartesian model of vision is the foundation of an internalist, computational approach to language and mind. Chomsky is so interested in Descartes’ theory of vision, and draws on it so frequently in his descriptions of grammar, because it so neatly fits Chomsky’s heavily mathematized, computational account of language. The above quotation nicely illustrates an important chain of equivalences that exist within Chomskyan discourse. The focus of linguistic science is “inner processes”, which are essentially “computational systems”. And computational systems can be likened to the operation of vision, since vision (in its Cartesian form) is itself computational in character. Language, vision, and computational systems constitute a set of foundational concepts in generative linguistics. The generative object of knowledge is evoked and given shape through a range of figurative comparisons, parallels, and substitutions involving language, vision and computation. Throughout Chomsky’s works these three domains are equated with each other to such an extent that Chomsky often forgoes the use of overt verbal cues signaling comparison, and will instead simply juxtapose references to language, vision and computation when constructing a line of argument, as if their equivalence can be taken for granted. Consider, for example, an argument about innateness in Language and Mind. Chomsky introduces the familiar rationalist distinction between inner and outer vision, between the “eye of the mind” and the “eye of the body,” in order to argue that knowledge proceeds from internal principles rather than exposure to external objects. Chomsky endorses Descartes’ notion that when we see a triangle, our knowledge of the triangle proceeds not from the “figure depicted on paper,” but from the internal process of mentally envisioning “the authentic triangle”. Chomsky’s discussion of innateness connects internal vision of a triangle with computational processing and generative grammar:
In this sense the idea of the triangle is innate. Surely the notion is comprehensible; there would be no difficulty, for example, in programming a computer to react to stimuli along these lines…Similarly, there is no difficulty in programming a computer with a schematism that sharply restricts the form of a generative grammar, with an evaluation procedure for grammars of the given form, with a technique for determining whether given data are compatible with a grammar of the given form, with a fixed substructure of entities (such as distinctive features), rules, and principles, and so on – in short, with a universal grammar of the sort that has been proposed in recent years. For reasons that have already been mentioned, I believe that these proposals can be properly regarded as a further development of classical rationalist doctrine, as an elaboration of some of its main ideas regarding language and mind.
Chomsky’s discussion of innateness proceeds through a contrast between ‘outer’ and ‘inner’ vision, and its analogue in a hypothetical computer system. Chomsky suggests that a computer could be made to simulate the relationship between the exterior (peripheral) and interior (innate) aspects of visual perception. Correspondingly, a computer could be programmed to simulate many of the characteristics of universal grammar. Vision, language and computers are comparable in that each consists of an internal, formal-symbolic, computational center. The plausibility of this set of equivalences rests on the assumption that vision should be understood in neo-Cartesian terms as a computational system, and that vision and language share fundamental similarities.
Another good example of the way Chomsky equates language, vision, and computation occurs in Rules and Representations. In the passage below, in which Chomsky distinguishes between different levels of theoretical analysis, a fundamental correspondence is assumed between these three domains. Chomsky writes:
Ultimately, the study of language is a part of human biology. In the study of any organism or machine, we may distinguish between the abstract investigation of the principles by which it operates and the study of the physical realization of the processes and components postulated in the abstract investigation. Thus, the study of visual perception might lead to the hypothetical construction of certain abstract components – for example, feature detectors – that enter in this system. A further inquiry might reveal the physical mechanisms that meet the abstract conditions postulated. In studying some automaton, we might attempt to determine its program at an abstract level, then proceed to inquire into the circuitry or mechanical principles by which this abstract program is realized. We may say that the same program is represented in devices of very different design and constitution. Rules and Representations, page 226.
The passage is organized around the conceptual dualism “abstract/physical.” This dualism is developed through a set of parallel linguistic constructions that align language, vision and computation. Chomsky writes that language can be studied at two different levels (“abstract investigation of principles”, and “physical realization of processes”). He follows this with the assertion that vision can be studied in terms of “abstract components” and “physical mechanisms”, and that an automaton can be studied in terms of the “abstract program”, or in terms of “circuitry or mechanical principles”. The correspondence between language, vision and computation is constructed without any explicit comparison, but is achieved through parataxis. The degree to which Chomsky equates language, vision and computation is signaled by the fact that after introducing the topic of studying language at different levels of analysis, he then describes potentially parallel studies of vision and automata, without giving any explanation as to how these three realms are logically related. The examination of each object is merely juxtaposed, as if it is obvious that they are conceptually equivalent. Passages such as the one above, in which language, vision and computation are equated, can be seen as enthymemes, in both the Isocratean and Aristotelian senses of the word. Fahnstock notes that the Isocratean sense of the word “enthymeme” connotes a compressed recapitulation of assertions already explicit in previously argued material. Chomsky’s references to language, vision and computation often function this way – as a kind of short hand for a much larger, more explicit argument about language, mind, science, and methodology. The passage also functions as an enthymeme in terms of the more traditional meaning associated with the term. An Aristotelian “enthymeme” suggests a line of argument constructed from a truncated syllogism in which a mediating premise is left out. In the passage above, the missing premise involves the notion that language, vision and computational systems can in fact be treated as isomorphic. Many of the central arguments of generative linguistics can be formulated in the following terms: language is equivalent to vision; vision is a form of computation, therefore language can be understood as a computational system and all three objects can be studied in the same way.
1.2.2 Chomsky is Influenced by Marr’s Neo-Cartesian Account of Vision
Another way in which vision is equated with computation, and both are connected to language, occurs in Chomsky’s use of the work of David Marr. As mentioned above, Marr’s work on vision is a central point of reference for Chomsky. Of all the work done in cognitive science and neuroscience, Chomsky focuses particular attention on the study of vision. And of all the work on vision carried out within cognitive science and neuroscience, Marr’s is selected for particular emphasis. Marr’s influence is apparent in the explicit references Chomsky makes to him, but it can also be seen in the many examples and turns of phrase that have their origin in Marr’s work on vision.
Chomsky is attracted to the work on vision by his colleague at M.I.T. in large part because it so closely echoes key aspects of the generative paradigm. Marr’s approach to vision is heavily formal, algorithmic and computational. Marr “embarked on a vigorous research program seeking computational insights into the working of the visual system, putting them to the test of implementation as computer models”. He treats vision as an autonomous cognitive system, and maintains that within this system there are different modules for computing different aspects of visual information (Marr’s arguments about the modularity of cognitive systems have been influential in cognitive science). His approach parallels Chomsky’s, in that it focuses on the level of the theory of computation, on internal representations and inner processing, and seeks an idealized, purified version of the object of inquiry. Gardner writes that “just as Chomsky wished to examine syntax in its pristine form (uncontaminated by semantics or pragmatics), Marr wanted his analysis of visual processing insulated as far as possible from the intrusion of ‘real world’ knowledge.” Marr is heavily Cartesian; he emphasizes the abstract coordinate geometry of visual perception, and focuses more on the computations that underlie vision than their material manifestation in either physical or organic “hardware”. And like Descartes, he equates external perception with mechanical devices, assuming that both embody the same logic and can be studied in similar ways. Marr’s work on vision thus functions as an important exemplar, analogy and support for generative linguistics. Vision is the central analogy, the root metaphor in generative linguistics, however it is vision understood in largely computational terms, based on the example of writers such as Descartes and Marr.
1.2.2 Problems with Chomsky’s Vision of Vision
A problem for Chomsky, and for generative linguistics, is that the computational model of vision that Chomsky spends so much time comparing language to has had a relatively brief shelf life in the cognitive sciences. Marr’s work has not fared well in neurophysiological research and in the cognitive sciences over the last 15 years. More generally, the computational-representational model of vision that Chomsky advances has seen declining support across a range of disciplines. Edelman and Vaina note that Marr’s computational theory of vision has been associated with the idea that “constructing an internal model of the world is a prerequisite for carrying out any visual task. The accumulation of findings to the contrary in neurobiology and in the behavioral sciences gradually brought to the fore the possibility that vision does not require geometric reconstruction. This encouraged researchers to seek alternative theories”. Many of the alternative theories that have since appeared propose that vision is distributed, emergent, and nonrepresentational. These theories reject the notion that vision is an autonomous system that can be understood in terms of traditional information processing models. Myin summarizes some of the main precepts shared by nonrepresentationalist theories of vision:
The heart of the nonrepresentational theory is that vision is seen as a process of active exploration of the environment, rather than the elaboration of an internal replica of the external world. The force of the perceptual poverty of the stimulus argument is denied in this approach, because vision is seen as a property of the whole animal (in its environment) and as extended in time. From the point of view of the whole animal, the stimulus is no longer fragmented. Neither is it ambiguous, in this theory, because movement in time disambiguates stimuli that are ambiguous when seen from a stationary position.
More generally, the view of “the nervous system as a modular, information-processing machine” has come under attack from connectionist and “dynamical systems” approaches to cognition. For example, Thelen and Smith write that “recent neurophysiology and developmental studies have turned this eminently plausible [the modular-information processing paradigm] view on its head…studies show vast and previously unimagined networks of interconnections both within and among anatomically distinct areas. There is the primary sensory organization within modalities, for example for color, form, and motion in the visual system.”  They write that there is
strong evidence of convergence of sensory information…multiple areas where projections from the visual, auditory, and somatosensory cortices converge…Indeed, some of the neural structures showing intersensory integration are subcortical, early maturing, and phylogenetically old. In their elegant book, The Merging of the Sense, Stein and Meredith (1993) build a compelling case for sensory convergence as a fundamental and enduring characteristic of animal nervous systems, universal in all phyla, and occurring at many levels of the neuraxis.
Chomsky has been consistently hostile to connectionist and dynamic accounts of vision, cognition and language, continuing to ally his theory of language with modular, information processing models.
Chomsky has received a certain amount of criticism from linguists, psychologists and cognitive scientists for hitching his theoretical wagon to what might be deemed “outdated” paradigms. As far back as 1980, in a roundtable discussion of Chomsky’s work in The Behavioral and Brain Sciences, critics attacked what they saw as an anachronistic account of biology and cognition. A number of the respondents argued that Chomsky’s focus on separate, modular systems, rather than the more integrated terms of modern biological research, risked resurrecting Gall’s organology. Lakoff’s charges are among the most severe. He argues that the biological analogies and metaphors Chomsky uses, and in particular the analogies with vision he constructs, focus exclusively on “the biology of separate systems”. Lakoff argues for the relevance of modern biological models that foreground interconnection, particularly between vision and language. He writes: “For me, the most exciting question at present is what similarities there are in the various cognitive faculties – language, thought, vision, motor control, and so on.” (page 23) He also argues that Chomsky tends to base his biological analogies on scientific work that is rather antiquated, since such work tends to treat biological systems as isolated (Lakoff writes: “It is particularly striking that the biological analogies Chomsky uses come not from contemporary biology – molecular biology, genetics, and so on – but from earlier biology – the biology of separate systems – the circulatory system, the visual system, and so on.”) While Chomsky’s opponents make some important points, much of their criticism consists of charges that generative linguistics lacks proper scientific legitimacy (ironically, this is a rhetorical strategy used often by generativists against opponents within linguistics.) What is ultimately of most interest to me is not so much whether Chomsky has accurately based his arguments on the most up to date research on vision and cognition. What does seem worth examining is the selectivity with which Chomsky draws on scientific work on vision (along with other areas of biology). Chomsky draws heavily on Marr’s work, ignoring most other work on the neurophysiology of vision. And as research on the neurophysiology of vision has moved away from Marr’s approach, Chomsky’s analogies to the study of vision have not lessened, but have become more general and abstract. References to research on vision are much more likely to be formulated in terms of hypothetical cases and generalized examples. A more relevant criticism of Chomsky’s use of biological and visual analogies can be found in the way he reconstructs research to fit his particular agenda. This brings us to the topic of how vision is used as a rhetorical strategy in generative linguistics.
1.2.4 Constructing Vision as ‘Generative’
When describing how language is comparable to vision, Chomsky constructs vision in terms of the language and categories of generative linguistics. A good example of this can be found in Aspects of the Theory of Syntax, where Chomsky argues that parallels can be drawn between vision and language with respect to the way development proceeds in infants. Citing the work of Held, Lemmon and Patterson on the visual development of children and newborn lambs, Chomsky writes:
Richard Held has shown in numerous experiments that reafferent stimulation (that is stimulation resulting from voluntary activity) is a prerequisite to the development of a concept of visual space, although it may not determine the character of this concept…or, to take one of innumerable examples from studies of animal learning, it has been observed (Lemmon and Patterson, 1964) that depth perception in lambs is considerably facilitated by mother-neonate contact, although again there is no reason to suppose that the nature of the lamb’s “theory of visual space” depends on this contact. 33-34.
Chomsky describes the work of Held, and Lemmon and Patterson as confirming his own hypotheses about the development of language. Yet he does this by extrapolating from their research, and by redescribing it in terms of the concepts and categories of generative linguistics. For example, nowhere do these researchers discuss “the development of a concept of visual space”, or of an animal’s “theory of visual space”. Nor do they argue that environmental stimulation exerts no influence on the development of a ‘concept’ of visual space. In a critique of Chomsky’s use of this research on vision, De Beaugrande argues that the three researchers in fact merely state that they observed the “facilitation” of development by external stimulation, and “not its irrelevance to some theory of visual space.” De Beaugrande argues further that “the scientific, biological finding does not support his [Chomsky’s] thesis except through his own non-biological interpretation,” and he accuses Chomsky of distorting the work he cites. While there is much truth to De Beaugrande’s charge, what is more interesting from a rhetorical perspective is the way Chomsky’s treatment of these researchers illustrates several key argumentative strategies. Chomsky constructs vision and the study of the visual system in terms of the conceptual framework of generative linguistics in order to be able to argue several things: that the study of vision confirms his research on language; that language is a cognitive faculty just like vision, and is amenable to the same kind of analyses; and that linguistics ought to model itself on the study of vision (which just happens to be a mirror-image of the generative approach to language). One can identify such rhetorical moves in a range of Chomsky’s writings. Consider, for example, this passage from Rules and Representations. Chomsky proposes that the study of language be modeled on the study of vision. He writes:
How can we proceed to investigate the properties of language? To clarify the issue, we might think about the less controversial task of studying the physical structure of the body. A rational approach would be to select some reasonably self-contained physical system of the body…Consider the kinds of questions we might ask about an organ of the body – say the eye, or more broadly the visual system regarded as an organ. We might organize our inquiry along the following lines:
(1) (a) function
(c) physical basis
(d) development in the individual
(e) evolutionary development.
Chomsky quickly rejects (1) (a) as a viable topic of study, and proceeds to talk about an analysis of the visual system based on the other questions, spending most of his time on 1 (b). Chomsky writes that we ought to begin with “some characterization of the structure of the visual system at the abstract level”, and might then proceed to a set of questions about the “structure of the visual system”, how physical mechanisms relate to structure, development in the individual, etc. This is all couched in terms of the language of generative linguistics, as can be seen in the passage below:
The organism begins in some genetically determined initial state common to the species with variations we may ignore at the outset. It passes through a sequence of states until it attains a mature final state which then undergoes only marginal further change. This ‘steady state’ is, it seems, attained at some relatively early stage in life. But though the organ of vision is essentially fixed in structure at that time, we may still ‘learn to see’ in new ways throughout our lives, for example, by applying knowledge gained later in life or through exposure to some new form of visual representation in the arts, say cubism.
The development of the visual faculty is discussed in terms of an “initial state”, “steady state,” and “final state,” familiar entries in the generative lexicon. It is said that “variation” in visual processing can be ignored. While the visual faculty is largely fixed, a person may produce novel visualizations and “learn” to see in new ways. Having described the study of vision in terms of the approach, methodology, categories, and questions posed within generative linguistics, Chomsky then proposes that linguistic inquiry ought to be modeled on the study of vision. That is, Chomsky argues that we should model linguistics on the study of visual systems represented in terms of the generative framework. After outlining what might be dubbed a “Chomskyan theory of vision”, Chomsky writes: “Suppose that we attempt to study language on the model of a bodily organ [the visual system], raising the questions (1a) – (1e).” (Rules and Representations, page 229.) Chomsky then outlines how the study of language might be based on the “model” provided by research on visual systems – that is, he describes the project of generative linguistics.
1.3 Vision, Knowledge & the Study of Grammar
One of the main reasons that a visual metaphoric plays such an important role in generative linguistics is that Chomsky conceptualizes grammar largely in terms of knowledge, and he associates knowledge with vision. There is a long tradition in the West of representing knowledge in ocular terms. For Plato, true knowledge is modeled on an act of vision – that of ascending to a realm in which the essence of things can be clearly seen, or of seeing behind the surface of things in order to discover the underlying regularities. (Eidos and idea, the words Plato gives for the Forms, both derive from the verb idein, ‘to see’.) In the Phaedrus Plato describes true knowledge as “seeing that which is well ordered and ever unchangeable”; in the Meno Plato describes knowledge as anamnesis, or the recollection of the soul’s experience of seeing the Forms in their pure state. Aristotelian philosophy is similarly visualist. In the first sentence of the Metaphysics, Aristotle writes: “Of all the senses, trust only the sense of sight”, and he defines vision as the “most noble sense” because it is the faculty that most closely resembles and guarantees true knowledge. Dewey argues that in Western philosophy since Plato and Aristotle “the theory of knowing is modeled after what was supposed to take place in the act of vision”. Rorty writes that the “ocular metaphor seized the imagination of the founders of Western thought”, and a focus on knowledge of universal concepts contributed to make “the Eye of the mind the inescapable model for the better sort of knowledge.” (Rorty 1979, p. 38-39) Arendt echoes these claims, stating that ‘from the very outset, in formal philosophy, thinking has been thought of in terms of seeing…The predominance of sight is so deeply embedded in Greek speech, and therefore in our conceptual language, that we seldom find any consideration bestowed upon it, as though it belonged among things too obvious to be noticed.’ Martin Jay argues that vision is “the master trope of the modern era”. And in Wittgenstein’s later writings he considers how our language may predispose us to think about the world in visualist terms (“a picture held us captive. And we could not get outside it, for it lay in our language and language seemed to repeat it to us inexorably.”)
Nowhere is the representation of knowledge and mind more strongly visualist than in Cartesian philosophy. The metaphor of the mind as inner vision is central to Descartes’ conception of knowledge and cognition, and figures prominently in Chomsky’s writings also. Descartes states, “We shall learn how to employ our mental intuition by comparing it with the way that we employ our eyes”, and “when the mind understands, it in some way turns towards itself and inspects one of the ideas which are within it.” . Descartes represents knowledge as a purified internal vision situated in the mind that stands in contrast to an impure, external vision that operates through the body. Rorty notes that this distinction dates back to classical Greek philosophy, where knowledge of universals and particulars are talked about primarily in terms of two forms of vision. The “eye of the mind” is associated with “knowledge of the highest and purest things: mathematics, philosophy itself, theoretical physics, anything which contemplates universals.” By contrast, “the eye of the body knows particulars by internalizing their individual colors and shapes.”
Chomsky draws often on the work of Descartes and other rationalist writers to explain his understanding of knowledge and language. Chomsky writes that the rationalist account of knowledge, once updated and “purged of the error of preexistence” provides a fruitful and plausible framework for understanding cognition.  Chomsky talks of the rationalist assumption that the ‘natural light’ of common sense can lay bare the basic elements of human reasoning. (Reflections on Language, page 244).
Chomsky’s discussions of knowledge and the mind are strongly Cartesian, and even include the familiar Cartesian distinction between the “Eye of the Mind” and the “Eye of the Body”. Consider, for example, the following statement in which Chomsky describes the mind in strongly visualist terms, and in which he contrasts “inner” and “outer” vision in a way that closely resembles the schema described by Rorty above. Chomsky writes:
The eye perceives, but the mind can compare, analyze, see cause-and-effect relations, symmetries, and so on, giving a comprehensive idea of the whole, with its parts, relations, and proportions.
Chomsky’s concept of knowledge is modeled on the visual perception of conceptual objects in a field of ordered, geometric, Euclidian space. The mind “sees” objects, parts and relations, and is predisposed to view these objects in terms of relationships of symmetry and proportion. Chomsky follows Descartes in assuming that there exists a “natural geometry of the mind”. He endorses Descartes’ notion that “the mind is so constituted that it constructs regular geometric figures as “exemplars” for the interpretation of experience.” (Rules & Representations, page 38). Chomsky writes:
Descartes and Cudworth believed the mind to be endowed with the principles of Euclidian geometry as an a priori property. We see a presented irregular figure as a (possibly distorted) triangle, straight line, circle, and so forth, because our minds produce these figures as ‘exemplars’, just as ‘the intelligible essences of things’ are produced by ‘the innate cognoscitive power’. In Kant’s phrase, objects conform to our ‘modes of cognition.’ Rules and Representations, page 246
There are thus as Chomsky puts it “first principles of geometry” that are internal to the mind – the mind both constructively envisions and makes visible objects of a certain orderly character, and inclines us to understand the objects we perceive with our external senses as ideals. Chomsky sometimes elaborates on this metaphor of the mind as a space of inner vision by drawing on another strongly optical metaphor, that of visually deciphering a piece of writing. For example, citing Cudsworth, Chomsky states in Reflections on Language:
The ‘book of nature’, then, is ‘legible only to an intellectual eye’….just as a man who reads a book in a language that he knows can learn something from the ‘inky scrawls’. 
The ocularcentrism of rationalist philosophy is frequently expressed in terms of analogies with written language, and Chomsky follows figures like Descartes, Cudsworth and Berkeley in this regard (I take up this aspect of Chomsky’s ocularcentrism in more detail in chapter 5 of my dissertation, Rhetoric and Reflexivity in Chomskyan and Cognitive Linguistics.)
In the sections that follow I examine three aspects of the way Chomsky’s epistemological premises are grounded in visual analogies. First, I will consider how Chomsky constructs both the subject and object of knowledge in terms of vision. Second, I will examine how language is defined as a mirror of knowledge, with both language and knowledge represented in terms of visual paradigms. Third, I will address how Chomsky’s visualist representation of knowledge defines some approaches to and aspects of language as knowable/visible, and others as unknowable/invisible.
1.3.1 Chomsky’s Ocularcentric Construction of the Subject and Object of Knowledge
Generative linguistics shares many of the tendencies that characterize ocularcentric disciplinary discourses. This includes the epistemological ideals of objectivity, certainty, transparency, universality, subject-object duality, and corporeal transcendence. These ideals are embedded in the visualist language and imagery Chomsky uses to discuss knowledge, language and mind, and profoundly shape Chomsky’s understanding of the subject and object of linguistic study. In Chomskyan linguistics knowing is seeing, and the object to be known is best understood not only in terms of visual perception, as a visually manifest phenomenon that can be observed with clarity, determinacy and certainty, but ultimately as itself a kind of vision. Chomsky’s ocularcentric construction of the subject and object of knowledge is evident in the key analogies he uses to describe the perspective from which an idealized knower could see/know language in its totality, in its true essential form.
Chomsky frequently argues that while on the surface language seems ‘complex and defective’, mired in ‘difficult and murky contingencies,’ this is only due to the distorting influence of its physical realization in the body, and our inability to adopt a properly scientific and objective perspective. Chomsky writes that ‘this apparent variety and complexity is superficial, reducing to minor parametric differences.’ The reason language appears ‘imperfect’ is that it must be materialized in the ‘sensorimotor system’. Chomsky states:
A large range of imperfections may have to do with the need to “externalize” language. If we could communicate by telepathy, they would not arise. The phonological component is in a certain sense “extrinsic” to language, and the locus of a good part of its imperfection. (Chomsky, 1997b)
Chomsky associates ‘imperfection’, and deviation from the ideal, with the body, and with the requirement that language be voiced. He argues that the imperfections of language caused by its embodiment would disappear if humans communicated by telepathy – that is, without a body. The distorting effects of our physical embodiment mask the unity and singularity of the underlying language system. In ‘Minimalist Explorations’ Chomsky states that the sensorimotor system is ‘extraneous to language,’ a ‘nuisance’ that is ‘imposed by external systems,’ and that ‘the one unique computational process’ would become apparent ‘if we could think and communicate by telepathy.’ He jokes that ‘if you were God’ it would be obvious that the ‘imperfections’ of language result from the vagaries of its physical realization. That is, if one could occupy the perspective of God, forwhom the perfection and design of human cognition is visible in its totality, or if we were unshackled from the distorting influence of the body (a condition often associated with the divine), we would be able to perceive the essence of human language. Later in this same text, Chomsky states that if we could approach language with the objectivity afforded a Martian, a being that has not experienced the cultural conditioning humans undergo, a comparable clarity of vision might be achieved. He writes that ‘Martians looking at humans would say there’s one language with a bunch of lexical exceptions.’ Similarly, in Language and ThoughtChomsky writes that the ‘computational system’ underlying language is invariant, and it follows that ‘there is only one human language, as a rational Martian observing humans would have assumed’. (page 50.) In “Language and Mind: Current Thoughts on Ancient Problems,” Chomsky writes:
In their essential properties, languages are cast to the same mold. The Martian scientist might reasonably conclude that there is a single human language, with differences only at the margins. For our lives, the slight differences are what matter, not the overwhelming similarities which we unconsciously take for granted. No doubt frogs look at other frogs the same way. But if we want to understand what kind of creature we are, we have to adopt a very different point of view, basically that of the Martian studying humans.
And in “Models, Nature and Language” Chomsky states:
There is fairly good reason now to believe that in a certain, rather deep sense, there is only one human language. If a Martian scientist looked at us the way that we look at frogs he might well conclude that with marginal, minor modifications, there is only one language. You and I might say “tree,” and a German would say “baum,” but we’re using basically the same concepts from the same inventory, which is both rich and restrictive. (Chomsky 1994, p. 173.)
Chomsky argues that we are confused by the existence of different languages and by variation in language use because we cannot adopt the proper perspective that will allow us to see the object as it really is. The ‘view from outer space’ presumably enables the hypothetical Martian to see through the surface irregularity and variation of language. A Non-terrestrial Being that is outside human history, culture, and social life, free of earthbound attitudes and prejudices, able to observe human language from a privileged, culturally ‘uncontaminated’ perspective, would see that there is only one underlying language system.
Chomsky also equates the perspective of the Martian with the vantage point afforded the child acquiring language. He writes that ‘from the Martian point of view,’ as well as from ‘a child’s point of view,’ languages are ‘essentially identical’ (Chomsky 1995b). Chomsky argues that “languages must look identical from the child’s point of view”, since ‘otherwise it’s impossible to learn any.’ The speed and precision with which children pick up new words “leaves no real alternative to the conclusion that the child somehow has the concepts available before experience with language and is basically learning labels for concepts that are already part of his or her conceptual apparatus.” Like God and the Martian, the child is represented as encountering language in an idealized, pristine form, his ‘knowledge’ of language untainted by experience and bodily realization. In each case a universal, totalizing vision constitutes idealized knowledge of language, and connects us to the truth as it distances us from the body, history, culture and society. Objectivity, certainty and knowledge of the ideal are associated with the visual apprehension of an object by a transcendental subjectivity. Chomskyan linguistics thus privileges what Nietzsche describes as ‘an eye outside time and history, an eye that no living being can imagine, an eye required to have no direction, to abrogate its active and interpretive powers’. Rorty notes that Western philosophy is characterized by the visualist ‘wish to see the world from above… as spectator of time and eternity,’ and has sought to constitute itself as a ‘discipline which lets us stand over and against the world of everyday practice by seeing it as God sees it, as a limited whole.” Chomskyan linguistics is characterized by similar aspirations, and is animated by a similar set of root metaphors. Chomsky’s descriptions of idealized knowers are permeated with the language of spectatorial epistemology – ‘observation’, ‘perspective,’ ‘point of view,’ ‘look,’ ‘see,’ ‘focus,’ etc. In the examples above, and in much of the rest of his work, knowledge is figuratively characterized as a visual enterprise. The object of inquiry, language, is represented in correspondingly ocular terms, and is assigned many of the same characteristics, including disengagement, objectivity, disembodiment, certainty and universality.
Chomsky’s investment in optical metaphors, analogies, evidence and examples shows up at every level of his work. His philosophical commitments are closely connected to the way vision has been culturally constructed, and to the specific material properties assigned it in the context of its use as a metaphor for knowledge. For example, vision is typically assumed to be the most immaterial of the senses.  Visual interaction with an object is thought not to involve the kind of incorporation associated with touch or taste, taking place as it does via the seemingly immaterial medium of light. Philosophers such as Hans Jonas who study the phenomenology of vision have suggested that sight is preeminently the sense of simultaneity, enabling the subject to instantly survey a wide visual field. Furthermore, while each mode of perception maintains objects at a distinctive distance from the body, vision is typically understood as the most distancing. Echoing Jonas’ arguments about vision, Fabian writes:
Vision requires distance from its objects; the eye maintains its ‘purity’ as long as it is not in close contact with ‘foreign objects’. Visualism, by instituting distance as that which enables us to know, and purity or immateriality as that which characterizes true knowledge, aimed to remove all the other senses and thereby the body from knowledge production. Fabian 1993, p. 99.
Ong makes a congruent argument, stating that “of all the senses sight is the most distancing sense: it requires always that eye and object be removed to a considerable extent from one another.’  Sight is also considered the most disengaged sense, the least attuned to temporality and the body. The externality of sight allows the observer to avoid contact with the object of perception. Vision is thus the sense most associated with objectivity and disengagement. Keller and Grontkowski state:
Although itself one of the senses, by virtue of its apparent incorporeality, it [vision] is that sense which most readily promotes the illusion of disengagement and objectification. At the same time it provides a compelling model for intangible communication. (p. 213)
Houlgate notes that Dewey’s critique of the spectator theory of knowledge centers precisely on the assumption of disengagement, on the idea that ‘the aim of philosophical or scientific inquiry is to come to know reality – as we seem to do in vision – without in any way interfering with it or modifying it through practical activity.’ In Chomskyan linguistics the ideal knowing subject is represented as an all-seeing, all-knowing observer who occupies a ‘view from nowhere,’ and above all, who is disengaged and objective. The perspective of God, a Martian, and the child is deemed perfect precisely because it is knowledge that is removed from the world.
Vision has been constructed as the sense which registers objects of perception instantaneously. Visual metaphors for knowing suggest immediacy, unity, and unmediated access to reality. Rorty states that ocular theories of knowledge seem to offer the hope of an ‘immediacy which would make discourse and description superfluous.’ (Rorty 1993, p. 375) Visual metaphors for knowing are frequently used to repress the role of language in knowledge production and suggest that understanding can proceed in an essentially unmediated way. Ocular metaphors efface the importance of rhetoric and reflexivity by suggesting that knowledge occurs instantly, at a glance, unmediated by language. Ocularcentric discourses tend to background the metaphorical status of vision (Derrida writes that with ‘photocentric’ discourses description is conflated with, or projected onto, the object it is supposed to understand, and that as a result ‘structure becomes the object itself.’) Correspondingly, Chomskyan linguistics is characterized by a constant drive to establish immediacy and efface signs of constructedness and mediation. It seeks to escape the need for language to talk about language. This rhetorical imperative depends on the conflation of the theoretical object and the ‘natural’ object, on closing the gap between the object and the language used to represent it. Chomsky represents knowledge in terms of inner vision and in terms of the act of observation by a transcendental subjectivity. In each case knowledge is immediate, disengaged, disembodied and unmediated. Furthermore, the subject and object are once more aligned by making the act of knowing and the object of knowledge similar in many key areas. Knowledge and the object of knowledge are described primarily in terms of vision, and both in their ideal form are dematerialized, immediate, unmediated and noncorporeal. This tendency is also identifiable in Chomsky’s discussions of the essence of language understood as a form of telepathy – which if not exactly a form of vision, is associated with the same qualities attributed to vision. As telepathy, communication is unmediated, transparent, disembodied and immediate. The properties typically assigned vision are associated with pure communication understood in terms of pure thought. In generative linguistics perfect knowledge thus mirrors perfect communication – a disembodied eye can perceive disembodied communication in its essential form.
Vision has also been associated with certainty, universality, and with what is self evidently true. It has been depicted as atop the hierarchy of senses with respect to reliability, and is often presented as the most universal sense (in the sense that perceiving subjects are assumed to see the same thing). Rorty talks of how Western philosophy has sought to associate knowledge with the certainty that appears to be provided ‘when staring at an object,’ (Rorty 1993, p. 159) and Descartes writes that “sight is the most universal and the most noble of the senses.’ Chomsky consistently associates the certainty attributed to vision with knowledge of language (references to vision in generative linguistics thus play a role somewhat akin to Dr Johnson’s stone.) When confronted with skeptics or dissenters, Chomsky argues that knowledge of language (as well as language as knowledge) is as certain and well-grounded as knowledge of the visual system. For example, when arguing against the notion that linguistic rules are cultural and variable, and that linguistic meaning is conventional and context-dependent, Chomsky writes that a language user
cannot choose to have sentences mean other than what they do, any more than he can choose to have objects distributed in perceptual space otherwise than the way they are…. It is a fact of nature that the cognitive structures developed by people in similar circumstances, within cognitive capacity, are similar, by virtue of their similar innate constitution. Thus we share rules of language with others as we share an organization of visual space with them.
In passages like the one above the universality and certainty commonly attributed to visual perception is associated with language in order to authorize generative linguistics and naturalize the object of inquiry. Chomsky’s frequent appeals to the certainty of visual perception and the visual system thus function as what Potter et al call an “undeniability device.” That is, references to vision constitute rhetorical strategies designed to establish beyond any doubt the existence of a brute reality, something external to talk, discursively unmediated. Like hitting the furniture, kicking a rock, invoking death, power or the Holocaust, Chomsky’s statements about the distribution of objects in ‘perceptual space’ (in other words, the undeniability of what we see) are used to establish that which cannot be denied, a bedrock nonverbal reality that is present, self-evidently true, an ultimately trustable form of experiential realism. Potter et al analyze some of the most common undeniability devices, arguing that each inevitably fails in its attempt to escape representation. For example, they examine the ‘furniture argument’ (“The realist thumps the table. What a loud noise! Much louder than talk. Much more gritty. Much more real. And yet we insist that this noise, being produced in this place, atthis time, in the course of this argument, is an argument, is talk”) sometimes invoked by realists. They write:
The Furniture argument, as the argument of no argument, purports to be the one that ends the rhetoric, is above rhetoric, and demonstrates its limits: it is “the naked truth,” unconstructed, unsupported, unclothed, needing no allies. The counter to this is to name it as a device, a rhetorical construct, occasioned and deployed. For example, we can place it amongst a set of similar devices for accomplishing undeniability. In discourse, the Furniture device shares features with other rhetorical ploys that, difficult to undermine in themselves, are deployed as shields behind which some rather more vulnerable entity is placed; thus positioned, they lend their robustness to some more contentious issue. Page 3.
Chomsky’s appeals to the undeniability of visual perception might similarly be dubbed “the Vision argument.” References to vision function to shield the ‘more vulnerable entity’ of language, to establish its certainty, its status as a bedrock nonverbal reality that is present, self-evidently true, discursively unmediated.
Idealized knowledge of language is depicted in terms of inner vision that delivers universal understanding, or in terms of idealized observers who occupy a vantage point that enables universal understanding. The linguistic examples that accompany Chomsky’s description of ideal knowledge, and which Chomsky uses to establish the universality and certainty of the object of inquiry, also depend on visualist assumptions about language and knowledge. Consider, for example, the passage cited earlier describing a Martian view of language:
If a Martian scientist looked at us the way that we look at frogs he might well conclude that with marginal, minor modifications, there is only one language. You and I might say “tree,” and a German would say “baum,” but we’re using basically the same concepts from the same inventory, which is both rich and restrictive. (Chomsky 1994, p. 173.)
Chomsky asserts that objective knowledge would reveal a single, universal conceptual system that underlies and is reflected in language. The universality of the conceptual and linguistic system is suggested through the example of German and English words for the concept “tree.” What is interesting about the example is that like many other arguments for conceptual universals, it focuses on an individual noun, one denoting a physical object with concrete physical characteristics, and which in many contexts of use connotes availability to visual perception. The plausibility of Chomsky’s example rests on an implicit appeal to the universality of thought and language understood in terms of visual perception. Chomsky does not consider a word expressing process, sensation, or some culturally specific practice or expression. His example presupposes that when an English speaker hears the word “tree,” s/he has in mind a mental image that can be compared and found equivalent to the German word “baum”. Chomsky thus assumes that words and sentences function primarily as pictures, rather than as the later Wittgenstein would argue, as tools integrated into the fabric of human action and life. His example presupposes Aristotle’s visualist dictum that “words are the image of thought”, and Wittgenstein’s remark that “the proposition is a picture of reality.”  Such a model of language clearly depends on an assumed equivalence between thought, knowledge and vision – if, for example, we imagine thought and knowledge in terms of an alternative set of modal metaphors, the plausibility of such a universal linguistic and conceptual system appears less convincing.
Generally speaking, philosophical and linguistic appeals to universality depend at some level on the epistemological certainty and universality that has come to surround visual metaphors. Much of the suspicion toward visual figures identifiable in poststructural and postmodern theories centers on the claims of universality that often accompany visualist language. Becker has produced an interesting interrogation of arguments for conceptual and linguistic universals, one that can also be read as a critique of visualist ontology. Becker examines the sentence “the sky is blue,” a declarative statement that appears to have a transparent, universal meaning. He writes:
To say, as some philosophers have, that everyone in the world agrees that the sky is blue, is not just wrong because of exotic differences in the connotations of ‘sky’, and ‘blue’ (important as these differences are) but more importantly because not everyone has a ‘the’ or an ‘is’. The language games English (and other languages) play because we have a figure ‘The ____ is ____’ (and similar figures) are not played everywhere. The ‘copula figures’ as we might call them (the ones used in syllogisms and definitions and identifications, to cite only some of the common uses of these figures) have no counterparts in some other languages but are, from the point of view of English, silences…It is impossible to recreate in English the experience of speaking a tenseless, article-less, be-less kind of languaging: every example will undermine itself. One simply has to learn to get along in another way of languaging.’ Becker, 227-231.
The universality and certainty that appears attributable to a sentence like “the sky is blue” proceeds via assumptions about what we all ‘see’ (consider how less persuasive this appears if based on an appeal to what we all taste, feel or smell.) Becker’s interrogation of semantic and conceptual universality foregrounds the weaknesses of Chomsky’s visualist faith in a single, universal conceptual system underlying and reflected in language. Becker’s use of terms such as ‘silence,’ ‘languaging,’ and ‘language games’ also suggests the availability of a more dialogic, process-oriented, and less visualist idiom for describing language. 
1.3.2 Language as Mirror of Mind
Rorty has famously argued that “the picture which holds traditional philosophy captive” is that of the mind as a great mirror in which language, thought and reality correspond. Rorty argues that the notion of the mind as mirror is a function of Western philosophy’s tendency to represent knowledge as analogous to ‘looking at something (rather than, say, rubbing up against it, or crushing it underfoot, or having sexual intercourse with it).” Chomsky’s definition of language as a mirror of mind can thus be seen as a central indicator of the visual bias of generative linguistics. In this section I explore Chomsky’s representation of language as a mirror, and explore the way this proceeds from and contributes to a visualist concept of language.
The essential function Chomsky assigns language is for it to be reflective, to operate as a mirror of cognition. Chomsky frequently associates language with specular imagery, with the notion that language is a “mirror” or “reflection” of the mind. Chomsky begins Reflections on Language by posing the question “Why study language?” He states that the most “compelling” answer to this question is “that it is tempting to regard language, in the traditional phrase, as ‘a mirror of the mind’”. This “temptation” is replaced almost immediately by conviction, when Chomsky writes that the innate character of linguistic knowledge, the poverty of stimulus available to activate this knowledge, and the unconscious nature of linguistic knowledge constitute evidence which leads him to the conclusion that “language is a mirror of the mind in a deep and significant sense.” (Reflections on Language,page 4.) And in Language and Thought Chomsky asserts that if, as he believes, the rationalist view of cognition is essentially correct, then “the structure of language can truly serve as a mirror of the mind, in both its particular and universal aspects.” As Gasche notes, the language of reflection and mirroring has strongly visualist connotations, suggesting it does the action by which reflective surfaces throw back light, and display reproductions of objects in the form of images (Gasche, page 16). To represent language as a mirror of the mind is to constitute it as something knowable primarily in terms of vision, as involving relations of isomorphism, correspondence, transmission and reflection. Because language is defined as a reflection of Chomsky’s neo-rationalist, deeply ocularcentric model of knowledge and mind, it shares many of the same properties assigned to the mind.
There is a more complex and far-reaching sense in which relations of reflection and mirroring contribute to the visualist character of generative linguistics. This involves the profoundly specular relationship that Chomsky constructs between language, knowledge and cognition. Generative linguistics is predicated on the notion that language, mind, and knowledge all exist as mirror images of each other in a very strong sense. The faculty of human knowledge in its most general form is represented as reflected in the object of linguistic inquiry (knowledge of language), which in turn mirrors the condition of linguistic knowledge production by the generative linguist. That is, the acquisition and development of knowledge of language in a child, the acquisition and development of linguistic knowledge by the analyst, and the acquisition and development of human knowledge in general are all held to operate according to the same logic, and to mirror each other. Chomsky argues that we are biologically constituted in such a way that our minds see the true nature of reality because our minds partake of, and reflect (if used properly) this same reality. Chomsky argues for what he calls “epistemic naturalism”. This position, which Chomsky attributes to early modern rationalist philosophers, holds that epistemology is a mirror of biology, or more specifically, the structure of cognition. In support of this position Chomsky quotes Peirce’s statement that “man is provided with certain natural beliefs that are true because certain uniformities…prevail throughout the universe, and the reasoning mind is itself a product of this universe. These same laws are thus, by logical necessity, incorporated in his own being.” Chomsky closely follows Plato in this regard. Plato held that the mind is naturally inclined to mirror the nature of the ideal because both partake of the same substance. (Ballow writes that in Platonic philosophy “the soul’s cognitive processes reflect an extreme expression of the like-knows-like principle: it is able to recognize same and different sorts of being because it is also composed of them.” Ballow, p. 85.) In Chomsky’s writings we find an updated version of this position. He states: “our systems of belief are those that the mind, as a biological structure, is designed to construct. We interpret experience as we do because of our special mental design. We attain knowledge when the ‘inward ideas of the mind itself’ and the structures it creates conform to the nature of things.” Theory construction thus reflects the underlying structure of both the mind and the world. Quoting Peirce again, Chomsky writes:
Continuing with Peirce: ‘Man’s mind has a natural adaptation to imagining correct theories of some kinds…If man had not the gift of a mind adapted to his requirements, he could not have acquired any knowledge’. Correspondingly, in our present case, it seems that knowledge of a language – a grammar – can be acquired only by an organism that is ‘preset’ with a severe restriction on the form of grammar.” Language and Mind, p. 91.
Chomsky continues this line of argument by endorsing Descartes’ statement that theoretical inquiry is based on principles “imprinted on the soul by the dictates of Nature itself.” When it comes to constructing theories of all kinds, we are aided by the fact that our minds are naturally adapted to the task. With respect to scientific knowledge, Chomsky states “we might think of the natural sciences as a kind of chance convergence between aspects of the natural world and properties of human mind/brain, which has allowed some rays of light to penetrate the general obscurity”(Language and Thought, page 45.) Chomsky argues further that because scientists construct theories based on “degenerate” evidence, develop theories so quickly, are in general agreement as to the truth of theories, and often arrive simultaneously at particular theories, we must possess a “science forming capacity” that enables us to recognize “theories as intelligible and natural”. He writes:
Some such science forming capacity must be an innate capacity of the mind. That is not to say that all scientific knowledge is ‘preformed’ at birth. Rather, the human mind is endowed with some set of principles that can be put to work when certain questions are posed, a certain level of understanding has been achieved, and certain evidence is available, to select a narrow class of possible theories. Perhaps, these principles, too, might fruitfully be regarded as a general schematism that characterizes the class of intelligible theories, thus permitting us to develop systems of belief and knowledge of great scope and power on limited evidence….Analogously, a rich set of principles of universal grammar permits us to attain our extensive knowledge of language on limited evidence…It is conceivable that we might discover the principles that underlie the construction of intelligible theories, thus arriving at a kind of ‘universal grammar’ of scientific theories. Rules and Representations, page 250-251.
The above quotation makes clear the extent to which Chomsky sets generative linguistics up as an ideal that is mirrored in a whole series of other domains. Generative linguistics is constructed as a grand model of how language, the mind, and the world exist in a relationship of correspondence, and this same model is argued to inhere in everything from scientific knowledge, to moral systems, to common sense (while generative linguistics is described as “analogous” to domains such as scientific theory production, suggesting confirmation of generative linguistics, it is in fact the case that Chomsky constructs these other domains in the likeness of generative linguistics, and then uses this “discovered” similarity as evidence for generative linguistics.) Learning, general human intelligence, and scientific knowledge are described in terms of the vocabulary and concepts of generative grammar. For example, the construction of scientific theory is represented as proceeding on the basis of “degenerate evidence”, the same term Chomsky applies to the acquisition of language. Chomsky proposes that theory construction is characterized by “poverty of the stimulus” (implying that knowledge of science must be innate), and that a “universal grammar” of scientific theories may exist. The Chomskyan system is summed up succinctly by Wittgenstein when he writes: “These concepts: proposition, language, thought, world, stand in line, one behind the other, each equivalent to each other” (PI, ∫ 96). True knowledge occurs when we can see that these realms are properly aligned, or as Chomsky puts it, when we can apprehend that “the inward ideas of the mind itself and the structures it creates conform to the nature of things.” Knowledge stands at the center of this hall of mirrors, reflected in language, the world and cognition, and assumed to function as a kind of vision.
Chomsky is led to this seemingly preposterous position in large part because of the epistemological dilemma opened up by his joint commitment to both rationalism and scientific realism, and his unwillingness to think about knowledge (particularly scientific knowledge) in terms of its existence in the social world. Chomsky’s rationalism invites the problem raised by Kant, that knowledge may be a projection of the structure of the mind, rather than a reflection of ‘Things-in-themselves’, and thus an unbridgeable gulf may exist between reality and our understanding of it. At the same time, Chomsky’s adherence to scientific realism requires that there be a secure correspondence between knowledge and the world. Chomsky’s solution is to argue that a miraculous ‘chance convergence’ exists between the structure of reality and the physical properties of the mind, such that scientific knowledge is naturally equipped to yield the truth. We can be confident that our scientific theories reveal the truth because they just happen to share and reflect the deep structure of reality. Language is assigned the job of doubling or guaranteeing knowledge, of occupying its proper place ‘in line’, as Wittgenstein puts it. This contributes to constructing language as something modeled on, and knowable primarily in terms of, a visual paradigm. And it leads Chomsky to imagine the limits of linguistic knowledge – and even all knowledge in general – in terms of what is and is not ‘visible.’
1.3.3 If Knowledge Is Vision, Then Understanding Is Limited To Those Objects Amenable To Visual Perception
Chomsky’s representation of the relationship between language, the mind and the world, leads him to the conclusion that some types of phenomenon are naturally “visible” to the mind’s eye, and are thus amenable to scientific analysis, while others are largely invisible, and cannot be examined with any rigor. Thus Chomsky writes:
The same properties of mind that provide admissible hypotheses may well exclude other successful theories as unintelligible to humans. Some theories might simply not be among the admissible hypotheses determined by the specific properties of mind that adapt us ‘to imagining correct theories of some kinds’. (Reflections, page 156).
Chomsky states that he agrees with Kant’s argument that the “schematism of our understanding”, when applied to “appearances” and “mere form”, is something that “nature is hardly ever likely to allow us to discover, and to have open to our gaze”. (156) Chomsky divides the field of linguistic inquiry into what is knowable, which he associates with light, vision, mechanism, and geometric form, and what is unknowable, which he associates with darkness, blindness, and mutability. The knowable he calls “problems”, the unknowable, “mysteries”. He writes:
Our minds are fixed biological systems with their intrinsic scope and limits. We can distinguish between ‘problems’, which lie within these limits and can be approached by human science with some hope of success, and what we might call ‘mysteries’, questions that simply lie beyond the reach of our minds, structured and organized as they are.’Rules & Representations, page 6.
Problems exist in a realm in which “the issues seem rather clear and straightforward”. Problems are ‘amenable to rational inquiry,’ whereas mysteries are not. Mysteries exist in a realm where “fundamental insights are lacking… [and] we are as much in the dark as to how to proceed as in the past.” (Reflections, page 138-9) While as human beings we might approach mysteries with “intuition and insight”, there is nothing we can say “as scientists”. (138)
Chomsky defines as mysteries such questions as “how do people succeed in acting appropriately and creatively in linguistic behavior or performance?” (Reflections, page 138). He also argues mysteries arise from the fact that the mind is so designed that it cannot reflect on some of its own processes. Chomsky writes that “the human mind is inherently incapable of developing scientific understanding of the processes by which it itself functions in certain domains.” (The mind cannot, so to speak, view itself in the act of viewing.) Chomsky argues that some aspects of the mind are accessible and can be reflected on, and these are associated with the Cartesian ideals of clarity, distinctiveness, regularity, and form – qualities that appeal primarily to the perception of visual phenomenon. Those that cannot are associated with change, interconnection, and dynamism. In Rules and Representations Chomsky adopts the terms “chemical” and “mechanical” (originally used by Mill) to illustrate the differences that exist between these contrasting aspects of the mind. Quoting Mill on “the laws of the phenomena of mind”, he writes:
When many impressions or ideas are operating in the mind together, there sometimes takes place a process of a similar kind to chemical combination. When impressions have been so often experienced in conjunction that each of them calls up readily and instantaneously the ideas of the whole group, these ideas sometimes melt and coalesce into one another, and appear not as several ideas but as one.
Chomsky associates interpenetration, dynamism, the lack of clear boundaries and fixed units with “mental chemistry”, or that which cannot be known. By contrast, the “mechanical” is associated with clarity, fixity, computation and clear boundaries. Chomsky writes that “where ideas are generated by mental chemistry, as distinct from association on a mechanical model, it is presumably impossible to resolve them into their constituents by introspection.” Chomsky represents the limits of knowledge in terms of the mind’s ability to visually inspect different kinds of conceptual object. Knowledge of the mind is the ultimate goal of linguistics, and Chomsky describes one of the key means of achieving this as “introspection” (the etymological roots of which are “looking into”.)
Furthermore, his description of the mechanical model favors frameworks such as generative linguistics, characterized as it is by a focus on ‘hierarchical constituent structure’. This bias is nicely illustrated in a statement by Pinker, who argues that “the rules of language are discrete combinatorial systems: phonemes snap cleanly into morphemes, morphemes into words, words into phrases. They do not blend or melt or coalesce.
For Chomsky, knowledge is a form of inner vision that is Cartesian and computational, and the focus of linguistics is “knowledge of a language” (and ultimately the structure of the mind). This allows him to define certain aspects of language as “legible only to an intellectual eye,’ and thus as not amenable to vision (proper knowledge.) It also allows Chomsky to define many rival approaches to the study of language as either ‘blind’ or involved in the study of objects that can never be properly ‘seen’ by the light of understanding. Here one can ascertain an example of what Messer-Davidow, Shumway and Sylvan call ‘boundary work,’ or the construction and regulation of a division between what is legitimately inside and outside a field of inquiry (p. 9) Chomsky achieves this by naturalizing knowledge production, and assimilating it to a model of cognition that mirrors generative linguistics. The mind just happens to be constructed so that it can only see/know objects that fit the generative paradigm, and cannot see/know objects that do not fit within the generative paradigm. Because of the way the mind is biologically constituted, if we are to know grammar at all, it will naturally be in terms of the framework outlined by generative linguistics. There is little sense of knowledge as something that must be actively produced within specific institutional and disciplinary conditions. By contrast, knowledge mirrors what ‘the mind, as a biological structure, is designed to construct’.
Chomsky describes not just of the mind, but also disciplinary knowledges in terms of metaphors of vision and optics. The more scientific an area of inquiry is, the more Chomsky associates it with light and vision. For example, in response to the critiques of figures like Quine and Putnam, Chomsky responds that such issues ought to be directed first at established sciences before being leveled at generative linguistics. He writes that ‘questions of a fundamental nature should be raised where the hope of gaining illumination is highest; in this case physics, not psychology.’ (Rules & Representations, page 22.) Speaking in broad terms about the situation of linguistic science, he writes:
Science is a bit like the joke about the drunk who is looking under a lamppost for a key that he has lost on the other side of the street, because that’s where the light is. It has no other choice. 
In an almost parodic version of Plato’s allegory of the cave (complete with a degraded version of anamnesis, in the form of a drunk struggling to find his keys) knowledge production is depicted as an essentially visual affair. Knowledge entails (re)discovering what was already there through a process of visual perception. Written at a moment in which the promise and centrality of the generativist project is in some question, Chomsky’s anecdote functions as an ironic statement of the difficulty, perhaps even the impossibility, of scientific linguistics. However, it also reveals the basic orientation toward the nature of linguistic inquiry found within generativism, the goals it aspires to and the fears it harbors. Scientific knowledge is characterized by the qualities of clarity, illumination and form, while other (nonscientific) fields are characterized by opposing qualities. To the extent that language is knowable, Chomsky assumes that it must be an object that conforms to the “natural geometry of the mind,” is regular, unified, definable primarily in spatio-visual terms, and amenable to being studied by the light of science.
As Chomsky has argued many times, if the underlying structure of language is in fact inextricably connected with social, normative and cultural factors then we are confronted with the opposite of science; with chaos, instability and darkness, and a field unworthy of study. Chomsky writes that if language is best described as a form of ‘bricolage,’ then the questions that drive linguistic inquiry “may have no interesting answers.’ It might be said that language as bricolage raises the specter of that which generative linguistics has systematically excluded or exiled from communication – fragmentation, incommensurability, temporal irreversibility, nonlinearity, practice, speech and the non-visual.
Agre, Philip E. 2000. The Practical Logic of Computer Work. In Matthias Scheutz, ed, Computationalism: New Directions, Cambridge: MIT Press.
Allen, K. 1986. Linguistic Meaning. London and New York: Routledge.
Aristotle. 1984. The Complete Works Of Aristotle. Translated by R. P. Hardie and R. K. Gaye. Reprinted in Jonathan Barnes (ed.), Princeton University Press.
Baigrie, Brian (ed.). 1996. Picturing Knowledge. University of Toronto Press.
Baigrie, Brian. Descarte’s Scientific Illustrations and ‘La Grande Mecanique de la Nature’. In Baigrie, Brian (ed.). Picturing Knowledge.
Baron, Naomi. 2001. ‘Commas and canaries: the role of punctuation in speech and writing. Language Sciences, 23 (2001) 15-67.
Bordo, Susan. 1987. The Flight to Objectivity: Essays on Cartesianism and Culture Albany: State University of New York Press.
Borgmann, Albert. 1984. Technology and the Character of Contemporary life.
Botha, Rudolf P. 1989. Challenging Chomsky: the Generative Garden Game. Blackwell.
Bowker, Geof. 1993. How to be Universal: Some Cybernetic Strategies, 1943-1970. Social Studies of Science, vol. 23, 1993, 107-127.
Boyd, Richard. 1979. Metaphor and Theory Change. In A. Ortony (Ed.),Metaphor and Thought. Cambridge: Cambridge University Press.
Button, Graham (ed.) 1995. Computers, minds, and conduct. Blackwell.
Chomsky, Noam. 1957. Syntactic Structures. Mouton.
Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge: MIT Press.
Chomsky, Noam. 1968. Language and Mind. Harcourt, Brace and World.
Chomsky, Noam. 1972. Phonology and Reading. In Levin, H. (ed.), Basic Processes in Reading, Harper and Row.
Chomsky, Noam. 1975. Reflections on Language. New York: Pantheon Books.
Chomsky, Noam. 1980a. Rules and Representations. New York, Columbia University Press
Chomsky, Noam. 1980b. Rules and Representations. The Behavioral and Brain Sciences, 3:1-21, and 42-61.
Chomsky, Noam. 1982. Noam Chomsky on the generative enterprise: A discussion with Riny Huybregts and Henk van Riemsdijk. Dordrecht, The Netherlands: Foris Publications.
Chomsky, Noam. 1986. Knowledge of language: its nature, origin and use. Praeger, New York.
Chomsky, Noam. 1988. Language and Problems of Knowledge, Cambridge, Mass: MIT Press.
Chomsky, Noam. 1991. Linguistics and Adjacent Fields: A Personal View. In The Chomskyan Turn, Asa Kasher (ed), Blackwell.
Chomsky, Noam. 1991. Linguistics and Cognitive Science: Problems and Mysteries. In The Chomskyan Turn, Asa Kasher (ed), Blackwell.
Chomsky, Noam. 1992. Naomi Chase Interviews Noam Chomsky. Z Magazine,April.
Chomsky, Noam. 1993. Language and Thought. Wakefield, R.I.: Moyer Bell.
Chomsky, Noam. 1994. Models, Nature and Language. Grand Street Fall, 1994, pp. 170-6.
Chomsky, Noam. 1995a. Language and Nature”. Mind 104:1-61.
Chomsky, Noam. (1995b) Minimalist Explorations. Paper delivered at University College, London, 16 June 1995. Transcript is available online at www2.prestel.co.uk/littleton/nc95_min.htm.
Chomsky, Noam. 1997a. Language and Mind: Current Thoughts on Ancient Problems (part 1). Conference paper. Online Conference: the 40th Anniversary of Generativism, December 1-12, 1997. Available online at: http://fccl.ksu.ru/papers/chomsky1.htm
Chomsky, Noam. 1997b. Language and Mind: Current Thoughts on Ancient Problems (part 2). Conference paper. Online Conference: the 40th Anniversary of Generativism, December 1-12, 1997. Available online at: http://fccl.ksu.ru/papers/chomsky2.htm
Chomsky, Noam. 1999. Interview with Brigitte Stemmer: On the Nature of Pragmatics and Related Issues. Brain and Language 68, 393-401.
Classen, Constance. 1993. Worlds Of Sense: Exploring The Senses In History And Across Cultures. Routledge.
Coulter, Jeff. 1995. ‘Logic: Ethnomethodology and the Logic of Language’. InEthnomethodology and the Human Sciences, edited by Graham Button.
Davis, Philip (editor). 1995. Alternative Linguistics: Descriptive and Theoretical Modes. John Benjamins.
De Beaugrande, Robert. 1991. Language and the Facilitation of Authority: The Discourse of Noam Chomsky. In Olson and Gale (eds), (Inter)views: Cross-Disciplinary Perspectives on Rhetoric and Literacy. Carbondale: Southern Illinois UP.
Derrida, Jacques. Of Grammatology. John Hopkins University Press, 1974.
Derrida, Jacques. Writing & Difference, trans. Alan Bass. Routledge, 1978.
Derrida, Jacques. Dissemination. Chicago : University Press, 1981.
Derrida, Jacques. 1983. The Principle of Reason: The University in the Eyes of its Pupils,” Diacritics, vol. 13, no. 3.
Descartes, Renee. The Philosophical Writings of Descartes. J. Cottingham, R Stoothoff, and D. Murdoch (translators). Cambridge University Press, 1984.
Edelman, Shimon, and Lucia M. Vaina. 2001. ‘David Marr’. The International Encyclopaedia of Social and Behavioral Sciences, edited by Paul B. Baltes and Neil J. Smelser. Elsevier.
Fabian, Johannes. 1983. Time and the Other. New York, Columbia University Press.
Fabian, Johannes. 1993. ‘Ethnographic Objectivity Revisited: From Rigor to Vigor.’ In: Allan Megill (ed.), Rethinking Objectivity. Durham and London: Duke University Press. pp. 81-108.
Fromkin, Victoria. 1991. Brain and Language: Redefining the goals and methodology of linguistics. In On the Chomskyan Turn. Kasher, (ed). Basil Blackwell.
Gould, Stephen Jay. 1996. Full House. Harmony Books, New York.
Gardner, Howard. 1985. The Mind’s New Science: A History of the Cognitive Revolution. Basic Books.
Gibbs, Raymond W. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge University Press.
Gibson, J.J. 1979. An Ecological Approach to Visual Perception. Boston: Houghton Mifflin.
Godzich, Wlad. 1994. Language, Images, and the Postmodern Predicament. InMaterialities of Communication, Gumbrecht & Pfeiffer (eds). Stanford UniversityPress.
Grassi, Ernesto. 1980. Rhetoric as Philosophy: The Humanist Tradition. University Park: Pennsylvania State University Press.
Haefner, Joel. 1999. The Politics of the Code. Computers & Composition, volume 16, Issue 3, pages 325-339.
Hanks, William. 1990. Referential Practice: Language and Lived Space Among the Maya. Chicago: University of Chicago Press.
Hanks, William F. 1995. Language and Communicative Practices. Boulder: Westview Press.
Haley, Michael and Ronald F. Lunsford. 1994. Noam Chomsky. New York: Twayne Publishers.
Harris, Roy, 1980. The Language Makers, London: Duckworth.
Harris, Roy. 1986. The Origin of Writing. London: Duckworth.
Harris, Roy. 1987. The Language Machine, London: Duckworth.
Harris, Roy, 1987. Reading Saussure : a critical commentary on the Cours de linguistique generale. Open Court
Harris, Roy, and Taylor, T. J. 1989. Landmarks in Linguistic Thought: the Western Tradition From Socrates to Saussure, London:Routledge.
Harris, Roy. 1990. The Foundations of Linguistic Theory: Selected Writings of RoyHarris, edited by Nigel Love, Routledge:London and New York.
Houlgate, Stephen. 1993. ‘Vision, Reflection, and Openness: The Hegemony of Vision from a Hegelian Point of View’. In Modernity And The Hegemony Of Vision, David Levin (ed). Berkeley: University Of California Press.
Jonas, Hans. 1954. The Nobility of Sight. In Philosophy and Phenomenological Research, 14.
Jonas, Hans. “The Nobility of Sight: A Study in the Phenomenology of the Senses,” in The Phenomenon of Life: Towards a Philosophical Biology. Chicago 1982.
Lakoff, George. 1980. What ever happened to deep structure? The Behavioral Sciences pp 22-23.
Koch, Christof, and Li, Hua (editors) 1995. Vision Chips: Implementing Vision Algorithms With Analog VLSI Circuits. IEEE Computer Society Press.
Lanham, Richard. 1993. The Electronic Word: Democracy, Technology and the Arts. University of Chicago Press.
Levin, David (ed). 1993. Modernity And The Hegemony Of Vision. Berkeley: University Of California Press.
Lindberg, David C. 1976. Theories of Vision from Al-Kindi to Kepler. University of Chicago Press.
Lumsden, C.J. & E.O. Wilson. 1981. Genes, Mind, and Culture. The Coevolutionary Process. Cambridge MA: Harvard University Press.
Lyons, John. 1971. Chomsky. Modern Masters Series. London: Fontana Press.
Nietzsche, F. 1956. The Genealogy of Morals. Doubleday.
Noe, Alva.; Evan Thompson, and Luiz Pessoa. 1999. Perceptual Completion: A Case Study in Phenomenology and Cognitive Science. In J. Petitot, J-M Roy, B. Pachoud & F. J. Varela (eds.) Naturalizing Phenomenology: Issues in Contemporary Phenomenology and Cognitive Science. Stanford University Press.
Ochs, Elinor., Emanuel Schegloff & Sandra Thompson (eds). 1996. Interaction and Grammar. Cambridge University Press.
Olson, Gary and Lester Faigley. 1991. “Language, Politics, and Composition: A Conversation with Noam Chomsky.” In (Inter)views: Cross-Disciplinary Perspectives on Rhetoric and Literacy. Ed. Gary Olson and Irene Gale. Carbondale: Southern Illinois UP.
Ong, Walter J. Orality and Literacy: The Technologizing of the Word. Routledge: London & New York, 1982.
O’Regan, J. Kevin and Noe, Alva. 2000. Perception, Attention and the Grand Illusion. Psyche 6 (15). 2000.
O’Regan, J. Kevin and Noe, Alva. 2001. A Sensorimotor Account of Vision and Visual Consciousness. Behavioral And Brain Sciences 2001 24(5) 2001.
Otero Carlos (ed.). 1994. Noam Chomsky: Critical Assessments. London: Routledge.
Pinker, Steven. 1994. The Language Instinct: The New Science of Language and Mind. Penguin.
Pylyshyn, Z. 1991. Rules and Representations: Chomsky and Representational Realism. In Kasher, A. ed., The Chomskyan Turn. Cambridge, MA, Blackwell.
Smith, Neil. 1999. Chomsky: Ideas & Ideals. Cambridge University Press.
Snell, Bruno. The Discovery of the Mind. New York: Dover Publications, Inc.
Rorty, Richard. 1979. Philosophy and The Mirror Of Nature. Princeton University Press,.
Synnott, Anthony. The Body Social: Symbolism, Self, And Society. Routledge, 1993.
Tyler, Stephen. 1984. The Vision Quest in the West, or What the Mind’s Eye Sees.Journal of Anthropological Research, 40, No. 1: 23-40.
Walker, John A. & Sarah Chaplin. Visual Culture: An Introduction. St. Martin’s Press, 1997.
Werry, Christopher. Rhetoric and Reflexivity in Chomskyan and Cognitive Linguistics. PhD Dissertation, Carnegie Mellon University. 2002.
Wittgenstein, Ludwig, Philosophical investigations (translated by G. E. M. Anscombe). Oxford : Blackwell, 1986.
 As I argue below, Chomsky’s ‘ocularcentrism’ is evident in texts he has written throughout his career. If one wanted to identify a single text where this tendency is most obviously identifiable, one might consider
Chomsky’s 1980 work Rules and Representations. The major points, premises and qualifications are constructed from analogies, figures, evidence, examples or research involving vision. Chomsky couches his discussions of innateness, acquisition of the language faculty, learning, the (largely peripheral) ‘triggering’ effects of the environment, the universality of linguistic knowledge, the modularity of mind, the scientific character of linguistics, and his rebuttals of opponents, in terms of vision. In chapter six, Chomsky describes (and justifies) his approach, methodology, theoretical framework, and the Rationalist tradition he seeks to update, in terms of extended analogies with vision. See in particular pages 5, 44-45, 71, 105-106, 208-209, 217-254.
 Levin, page 9.
 Jay, page 3.
 Mathematical knowledge is the second most frequently used argument for innateness. Comparisons between language and vision often include references to the ‘number faculty’ or ‘mathematical knowledge’, which Chomsky suggests are also innate. Thus in Rules & Representations he writes that like language, ‘the capacity to deal with the number system or with abstract properties of space is surely unlearned in its essentials.’ (p. 39). Note however that innate mathematical understanding is typically represented as having the apprehension of visual space as a central component. In Chomsky (1982) he states that “most of the history of mathematics is based on a very limited class of intuitions, intuitions about number and intuitions about visual space. Those things just seem to be inherent to human beings.” (23) Both faculties are described as part of the computational structure of the mind, or of the “capacity to deal with discrete infinities through recursive rules.”
 Chomsky, 1992.
 For example, Chomsky 1986, p.211: ‘The core…is the essential part of what is ‘learned’, if that is the correct term for this process of fixing knowledge of a particular language’.
 See in particular pages 5, 45, 71, 105, 219-231, 242-251.
 Language and Thought, pages 82 and 89.
 Rules and Representations, page 248.
 Huybregts et al, pages 9–10
 In texts such as Rules and Representations and Reflections on Language, in which Chomsky devotes a lot of space to his interlocutors, it is by far the single most common tactic used in responding to critics.
 The three most common ways in which counterarguments are dealt with in academic discourse are strategic concession (in which parts of an opposing author’s argument are accepted while others are rejected), refutation (outright rejection), and demonstration of irrelevance (in which opposing views are represented as not meeting the criteria of relevance defined by the author). When discussing methodological issues, Chomsky typically uses the last of these strategies. Methodological objections are defined as irrelevant because they do not apply to the study of vision, and since vision is analogous to language, these objections also do not apply to the study of language.
 For example, in Knowledge of Language Chomsky criticizes connectionist accounts of language, stating: “Again the refusal to treat the development of language as parallel in terms of its genetic determinants to the development of vision is left unmotivated.’(173)
 Another good instance of the way Chomsky uses visual analogies to argue against functionalist approaches to language can be found in Chomsky 1997b. Chomsky writes that a biologist ‘would not offer the functional design property as the mechanism of embryological development of the eye,’ and that ‘similarly, we do not want to confound functional motivations for properties of language with the specific mechanisms that implement them.’
 Rules and Representations, pp 229 – 232.
 Ibid, page 71.
 Olson and Faigley. “Language, Politics, and Composition: A Conversation with Noam Chomsky.” Page 70.
 For Descartes, vision is an optical machine characterized by the internal operation of precise mathematical rules and geometric laws that result in the reconstruction of images within the mind. Walker and Chaplin note that “Descartes was the first to assert clearly that light was nothing but a mechanical property, and to mathematize it. Crucially, he displaced perception itself from the surface of the retina to the brain.” (Walker and Chaplin, page 43.)
 In chapter 5 of Werry (2002) I discuss the influence of computational models on Chomsky’s representation of ‘inner processes.’
 Language and Mind, page 83. The full passage from Descartes quoted by Chomsky is as follows:
When first in infancy we see a triangular figure depicted on paper, this figure cannot show us how a real triangle ought to be conceived, in the way in which geometricians consider it, because the true triangle is contained in this figure, just as the statue of Mercury is contained in a rough block of wood. But because we already possess within us the idea of a true triangle, and it can be more easily conceived by our mind than the more complex figure of the triangle drawn on paper, we, therefore, when we see the composite figure, apprehend not it itself, but rather the authentic triangle.
 These two different meanings of ‘enthymeme’ are discussed in Fahnstock, page 29.
 Chomsky’s frequent discussions of the methodological appropriateness of distinguishing between different levels of analysis in linguistic research owe much to Marr. And many examples and analogies involving vision echo remarks made by Marr. For example, when explaining how generative theories of language can simultaneously consider physical mechanisms, theoretical abstractions, and several other levels of analysis, Chomsky explains this in terms of a hypothetical approach to the study of vision. He writes:
‘In the same way, a theory of vision might be formulated in concrete terms, referring, say, to specific cells in the visual cortex, and their properties; or it might be formulated abstractly in terms of certain modes of representation (say, images or stick sketches), computations on such representations, organizing principles that determine the nature of such representations and rules, and so on. (Rules & Representations, page 5.)
This passage strongly echoes statements made by Marr in Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. Even the reference to ‘images’ and ‘stick sketches’ recalls Marr’s work, as these are key terms that feature prominently in Marr’s account of the parsing of visual information. For a discussion of Marr’s work that helps illustrate some of the similarities between Chomsky and Marr, see Gardner, pp. 300-306.
 My discussion of Marr cannot begin to deal with the complex specifics of his work, but rather attempts to show in general terms how and why his work is attractive to Chomsky. My reading of Marr is based mainly on Gardner, Sterelny, Edelman and Vaina.
 Edelman and Vaina.
 Dreyfus puts the matter succinctly when he writes that ‘neurophysiology does not support information processing models’ of cognition. Dreyfus, 25.
 Nonrepresentational theories of vision have been proposed in cognitive science by researchers such as O’Regan, Noë, and Koch and Li. Congruent theories have been developed in the AI tradition of Active Vision by figures such as Gibson and Neisser.
 Myin, “Two theories of perception, meaning and language”.
 Thelen & Smith, A Dynamical Systems Approach to the Development of Cognition and Action, page 187
 For example, in Language and Thought he writes that ‘connectionism is a radical abstraction from what’s known about the brain and the brain sciences…there’s no evidence for it.’
 These comments occur in respondents’ section of issue 3 of The Behavioral Sciences (1980), pp 21-42.
 Lakoff’s critique appears justified. Consider, for example, the description of biology articulated by the Nobel prize winning molecular biologist Max Delbruck. He writes: “there are no ‘absolute phenomena’ in biology. Everything is time-bound and space-bound. The animal or plant or micro-organism he is working with is but a link in an evolutionary chain of changing forms, none of which has any permanent validity.” (Cited in Lanham, page 56.) Clearly such an understanding of biology contrasts significantly with Chomsky’s.
 Vision is a key point of difference between Chomsky and Lakoff theories of language, as I argue in more detail in Werry 2002. Chomsky thinks cognitive systems are separate, and that language as a cognitive system is largely separate from vision (He writes that ‘there seems little reason to suppose that the principles of grammar or universal grammar have any close analogue in other cognitive systems…just as we do not expect the fundamental properties of the visual system to be reflected in language. Confident assertions to the contrary, which are prevalent in recent literature, seem to me rather dogmatic as well as without empirical support or plausible argument.’ Rules and Representations, pp. 245-6.) Cognitive linguistics, by contrast, argues that cognitive systems are interconnected, and that the prime example of this is the way language reflects vision.
 In his more recent work Chomsky has begin to distance himself from Marr’s theories. He writes that Marr ‘was concerned with input-output systems… Language is not an input-output system.’ He also now rejects Marr’s ideas about separating levels of analysis. See Chomsky 1995a page 12; Chomsky 1995b; Chomsky 1999, page 397.
 It should be noted that Chomsky describes many objects and areas of study in terms of generative linguistics. For example, in an interview with Olson and Faigley, Chomsky proposes that a generative substructure underlies morality, mathematical ability, and the ability to read and write (pp. 77-78). The interview makes clear that for Chomsky, almost any regularity in human life seems to imply shared inner structure and innateness. However, this tendency appears most often in his discussions of vision.
 De Beaugrande, page 105.
 In many respects Chomsky’s use of language to describe vision can be seen as a second kind of ‘systematic ambiguity’ or systematic conflation of terms. As described in chapter one of Werry 2002, Chomsky constructs what he calls ‘systematic ambiguity’ between theoretical and biological object. This functions to naturalize language, to efface the manufactured character of linguistic discourse, and the role of reflexive practices in theory construction. The systematic ambiguity that Chomsky establishes between research on language and research on vision is not as explicitly stated, but functions in a similar way.
 Rules and Representations, page 227.
 A similar strategy can be identified in Chomsky’s discussions of face recognition in infants. In Rules and Representations Chomsky suggests that language and the visual perception of faces are analogous in that they both constitute “innate systems that deal with restricted data in a uniform way” (247). He proposes the idea of “an abstract theory of human faces,” and “a universal grammar of faces.” Chomsky describes the ‘innate’ faculty that infants have for visually recognizing faces in terms of generative grammar, and then uses this ‘faculty’ as evidence for generative grammar. See Rules and Representations, page 248.
 The argument that there is a spatio-visual bias in Western epistemology can be found in American pragmatism, hermeneutics, and in much post-structural theory. Some of the writers who have made this claim most strongly include Dewey, Bergson, Wittgenstein, Heidegger, Arendt, Merleau-Ponty, Ong, Derrida, Jonas, Tyler, Jay, and Levin.
 The visual bias in Aristotle’s writings, and its implications for Greek thought is discussed in Hans Jonas’ ‘The Nobility of Sight: A Study in the Phenomenology of the Senses’, in The Phenomenon of Life: Toward a Philosophical Biology, 1982.
 Dewey, The Quest for Certainty, page 23.
 Cited in Levin, page 2.
 Jay 1993, page 114.
 Descartes, Regulae IX. Quoted in Keller & Grontworski, 1983, page 214.
 Descartes, The Philosophical Writings of Descartes. Edited by J. Cottingham, R Stoothoff, and D. Murdoch , page 112.
 Rorty 1979, pages 41 and 43. In Platonic philosophy both ideal and empirical knowledge involve vision. Plato distinguishes between knowledge acquired through visual perception (aisthesis) and knowledge seen by the mind’s eye (noesis). As I argue in Werry 2002, one way of thinking about the difference between generative linguistics and cognitive linguistics is in terms of the ‘Eye of the Mind’ and the ‘Eye of the Body’. Both theories involve strongly visual paradigms, but cognitive linguistics places far more emphasis on the bodily character of vision. Despite this apparent difference, the eye of the body is assigned many of the transcendental characteristics previously assigned the eye of the mind in generative linguistics.
 Reflections on Language, page 6.
 Reflections on Language page 6-7.
 Descartes 1984, page 106.
 Reflections on Language page 7. In passages such as this Chomsky’s conceptualization of knowledge closely resembles a Platonic one, but in ways that he perhaps does not fully anticipate. For as Derrida has argued, writing and vision are powerfully linked in Plato’s logocentric philosophy, and constitute what may be the founding metaphors of Western philosophy. (The best known discussion of this occurs in the chapter of Dissemination entitled ‘Plato’s Pharmacy’.) For Plato, true knowledge is modeled on an act of vision – that of ascending to a realm in which the essence of things can be clearly seen, or of seeing behind the surface of things in order to discover the underlying regularities. In works such as the Phaedrus thePhilebus and the Timaeus, this knowledge is exemplified as a purified form of inscription (See for example the Phadrus 274ff, and the Philebus, 18ff.) The soul is described as coming to know the fundamental elements of the real in terms of the visual apprehension of written form. As Hoskin notes, in Plato’s dialogues the underlying regularity of the world (the “syllables of nature”) is described in terms of “stoicheia”, which refers simultaneously to writing and geometry. (Hoskin, page 39-40.) Vision, writing and geometry are at the center of Platonic philosophy, and they play a similarly important role in Chomsky’s work. Hoskin writes that in Plato we can identify a “whole alphabetic way of seeing”. As I shall argue in some detail later in this study, a “whole computational-scriptist way of seeing” lies at the center of Chomsky’s work, and this is an important part of why a visual paradigm is so strong in Chomsky’s writings. I take up the connections between ocularcentrism, machine metaphors and the influence of written models in generative linguistics in chapter 5 of Werry 2002.
 Chomsky, 1997a.
 Roy Harris argues that the assumption of telepathy, or ‘telementation,’ is a key part of what he terms ‘the language myth,’ a foundational set of assumptions about language that have persisted throughout much of linguistic history. This myth has three main parts: the isolatability of language as an object; its status as a fixed code, and its participation in telementation, or the transfer of thoughts from one mind to another. Harris subjects all three aspects of the language myth to a withering critique in The Language Makers, The Language Myth, and The Language Machine.
 This invocation of a God’s eye view capable of seeing the essence of language and mind contrasts strikingly with Wittgenstein’s statement that “If God had looked into our minds he would not have been able to see there whom we were speaking of.” PI, Part II, p.217e.
 Interestingly, in a seeming parody of Chomsky’s invocation of the Martian view of language, Roy Harris has argued the exact opposite position. That is, he argues that even a telepathic Martian would have a different understanding of language from that of his/her human teacher since “language-using will always be integrated into the Martian’s existence in an ineradicably different way.” Harris 1990, p. 181.
 Chomsky 1988, p.28. The notion that the child is merely learning ‘labels’ for concepts already in place is in line with the ideal of telepathy in generative linguistics.
 In some of his works Chomsky presents another idealized figure able to grasp the essence of language, that of a hypothetical, ideal, “objective” scientist who approaches the study of language de novo. What is striking about this figure is the way that he naturally proceeds to intuit a generative theory of language based in large part on analogies and evidence drawn from the nature of the visual system. For example, in Reflections on Language Chomsky writes:
Imagine a scientist, henceforth S, who is unencumbered by the ideological baggage that forms part of our intellectual tradition and is thus prepared to study humans as organisms in the natural world. Let us consider a course of inquiry that S might undertake, sketching conclusions that he might tentatively reach along the way, and then confront S with some of the questions of methodology and principle that have been raised by a number of philosophers who have discussed the nature and goals of linguistic theory. (p. 139)
Chomsky suggests that S would infer the plausibility of rationalism and innate knowledge based on the evidence of the visual system and the ‘face-recognition module’ in infants. Having mapped the trajectory of S’s development in terms of the lessons learned from studying the visual system, Chomsky states that ‘in the case of grammar, similar remarks apply,’ (147) and proceeds to discuss how S might develop an analogous theory of grammar. Rather like the linguistic-cognitive development of the ideal speaker-hearer, the idealized theorist is described as naturally acquiring a theory of knowledge that is generative.
 Nietzsche, The Genealogy of Morals, Doubleday, 1956, page 255. Cited in Levin (1993) page 4.
 Rorty 1993, page 338 and 346.
 Chomsky’s tendency to treat visual metaphors inscribed in everyday language as a neutral means of talking about knowledge and language, rather than as a particular figurative construction, is another example of the repression of reflexivity in generative linguistics.
 I draw here on the work of writers in many fields – cultural historians of the senses such as Classen and Synnott, theorists of visual culture such as Jay and Levin, analysts of orality and literacy such as Ong and Havelock, and philosophers of phenomenology, poststructuralism and American pragmatism, for whom the significance of visual metaphors in theoretical discourse is a common theme. It should be noted that many of these writers, most obviously Jonas, Ong and Jay, attribute intrinsic characteristics to different sense modalities. That is, they suggest that certain theoretical commitments follow inevitably from the choice of different sensory metaphors for knowledge. While I acknowledge that different senses perhaps provide different metaphorical ‘affordances,’ I do not assume that they have an essence that leads to a corresponding set of epistemological values. Thus while I draw on work that examines the relationship between visualist discourses and particular theoretical frameworks, I assume that vision can and has been culturally constructed in a variety of different ways, and that many different sense metaphors can be used to represent knowledge. For example, Tyler’s work on Dravidian languages and Handelman’s contrastive discussion of Hebrew and Hellenic metaphors for knowledge explore how some cultures use non-visual metaphors for knowledge. Snell’s arguments about early Greek representations of vision also suggest that sight can be constructed in ways that assume a different, less passive relationship between subject and object. And following Nietzsche, a number of poststructuralist writers have described knowledge in terms of a decentered, multi-perspectival, incarnate vision.
 Vesey sums up many of the most common values assigned sight in Greek and Western metaphysics: “Sight does not require our being part of the material world in the way in which feeling by touching does…The directness of seeing when contrasted with hearing, its non-involvement with the object when contrasted with feeling by touching, and its apparent temporal immediacy when contrasted with both feeling and hearing are features that may partly explain the belief that sight is the most excellent of the senses.” Cited in Keller and Grontwowski, page 221.
 Jonas, pp.513-518. Jonas, a student of Heidegger’s, picks up many of the anti-ocular themes expressed in Heidegger’s writings.
 Ong, 1977, page 138.
 The disembodiment and asceticism associated with knowledge as vision can be traced to Plato. Plato writes that in ‘despising the body and avoiding it, and endeavoring to become independent – the philosopher’s soul is ahead of all the rest…If we are ever to have pure knowledge of anything, we must get rid of the body and contemplate things by themselves with the soul by itself.’ (Phaedo 64d-66d) The pure visual contemplation that characterizes philosophical knowledge constitutes a radical flight from the body. Generative linguistics can thus be understood as the linguistic apotheosis of a venerable cultural tradition.
 Stephen Houlgate, ‘Vision, Reflection, and Openness: The Hegemony of Vision from a Hegelian Point of View’. In Levin 1993, page 87.
 Consider, for example, Russell’s description of language as ideal: ‘In a logically perfect language the words in a proposition would correspond one by one with the components of the corresponding fact…in a logically perfect language there will be one word and no more for every simple object, and everything not simple will be expressed by a combination of words…one word for each component. A language of that sort will be completely analytic, and will show at a glance the logical structure of the facts asserted.’ In its reliance on a visualist epistemology it resembles Chomsky’s representation of language as ideal.
 Writing and Difference page 15. Derrida writes that “the metaphor of darkness and light (of self-revelation and self-concealment) is the founding metaphor of Western philosophy as metaphysics . . . In this respect the entire history of our philosophy is a photology, the name given to a history of, or treatise on, light” (p. 27).
 Cited in Keller and Grontowski, p. 213.
 Reflections on Language, p. 71.
 Potter et al, p. 3.
 Consider, for example, the words used in East Cree to express states of mind and feeling. Junker discusses how many of these words are rooted in expressions having to do with the planning and skill required to trap an animal. It is not obvious that English or German speakers ‘reference the same concepts from the same inventory’ in such cases.
 De Interpretatione, 1, 16a.
 Tractatus, 4.01.
 If, for example, knowledge and thought were represented in terms of non-visual metaphors, metaphors suggesting process, embodiment, duration, particularity, and interaction between subject and object, then the universal conceptual/linguistic system Chomsky posits might appear less obviously plausible.
 In the final part of the study I take up the issue of how non-visual metaphorics, as well as different uses of visual imagery, might be used to approach the study of grammar. I explore how alternative theories of grammar which draw on non-visual metaphorics might provide the basis for a more reflexive, rhetorically-oriented understanding of grammar.
 Rorty 1980, p. 39. Italics In the original.
 Reflections, p.77. This is reiterated in the work of other Chomskyans. For example. Smith writes: ‘A major part of Chomsky’s achievement is to have opened up language for inspection and by so doing has opened a mirror onto the human mind’. (48).
 In generative linguistics language could perhaps be more accurately described as a mirror of a mirror, since Chomsky (along with many other writers within traditional cognitive science and artificial intelligence) bases his concept of cognition to a significant extent on models suggested by writing technologies, and in particular computers. Roy Harris’ analysis of Port Royallinguistics thus applies equally well to Chomskyan linguistics. Harris writes that ‘the explanation offered [of mental structures] is simply a reversal of the process by which the analysis itself was produced. What are hypothetically projected back into the mind as observable mental realities are abstractions derived from analysis of the observable verbal structures themselves’. (The Language machine, p. 27)
 This aspect of Chomsky’s work is closely related to the issue of reflexivity. As I argue in chapter 2 of Werry 2002, Chomsky continually seeks to close the gap between the object and language of inquiry in order to escape the contaminating mediation of language. One of the ways in which this is achieved is through the conflation of terms, and the “systematic ambiguity” constructed between different levels of analysis. Another way that Chomsky aligns the object and means of representation is by setting up language, the mind, and knowledge as mirror images of each other.
 Language Thought, page 32. Chomsky writes: “The epistemic naturalism of early modern thought appears to be quite reasonable, and is being rediscovered and given more substance in current empirical research.’ Fodor has made similar arguments, and ‘epistemic naturalism’ is also similar to Quine’s ‘naturalized epistemology.’
 Language and Mind, Page 96.
 See for example Timeaus, 37a-b.
 Keller and Grontowski explain this aspect of Platonic philosophy in the following terms: “the knower and that which is known…are essentially kindred. They are both parts of the whole of Being itself. Thus kinship with the universe and its structures constitutes Plato’s metaphysical presupposition. His epistemological assumption is that we, who were originally part of the lawful divine structure, are thereby in principle able to see (intuit) it fully again.” Page 212.
 The notion of ‘epistemic naturalism’ thus grants enormous authority to those doing cognitive research, since it is presumably they who are in the best position to decide whether a claim conforms to the underlying structure of the mind. Fuller, discussing the ‘epistemological naturalism’ of Quine and Fodor writes: “The conclusion is that only those beliefs are justified that are arrived at in a manner that conforms to the representation producing processes as pictured in cognitive science.” (Fuller, p. 99).
 Language and Thought, page 32.
 This is a common argument in Chomsky. Consider for example his statement in Language and Thought: ‘the typical phenomenon is either we think of nothing, or else everybody more or less thinks of the same thing or at least recognizes it as plausible. That indicates a high degree of structure in the cognitive system.’ (88).
 For example see Language and Thought, pp 32-35. Chomsky states that the mind/brain is ‘a complex system with a highly differentiated structure’ that includes a language faculty, a science faculty, a faculty for moral and aesthetic judgment, and common sense. For related comments seeLanguage and Mind, page 90, and Rules and Representations, page 250.
 This position leads to a rather vicious circle, one very similar to the circularity entailed in Chomsky’s position regarding the underdetermination of theory by data, and to the conflation of terms discussed on page. What unites each of the theoretical moves made by Chomsky is the drive to construct an object that is unified, self-enclosed, removed from culture and society, and sealed from reflexive understanding.
 Reflections, p. 156
 Rules and Representations p. 243.
 Interestingly, the quoted passage from Mill seems to foreshadow connectionist descriptions of cognition, which Chomsky has expressed consistent opposition to.
 Pinker, page 163. Linell also describes how generative linguistics divides “linguistic products into successively smaller segments; the whole discourse (or text) is broken down into sentences, these in their turn into constituent sentences (main and subordinate sentences), clauses, and phrases, and phrases are thought to consist of words, words of morphs, morphs of syllables and/or phonological segments (vowels and consonants), and the latter are finally dissected into the ‘ultimate constituents’, i.e. phonological features.”
 Chomsky does grant that theories that deal with ‘performance’ and the ‘externalities’ of language may produce some interesting research, but he argues that this will always be unscientific knowledge of the ephemeral and epiphenomenal. This echoes a rhetorical strategy first made by Plato against several of his adversaries. In the Theatetus Plato criticizes the sophist Protagorus for his view that perception can deliver true knowledge of Being, but concedes that he is right in his assertion that man is the measure of all things in the world of Becoming. Similarly, Chomsky states that theories that analyze the social and cultural aspects of language may be right about what they say when it comes to the world of performance, however this is a world about which only very limited and retrograde knowledge can exist.
 Robert Barsky, Noam Chomsky: A Life of Dissent, page 95.
 Language & Responsibility, pages 140, 152-3. Chomsky makes some interesting rhetorical moves when discussing the possibility of alternative, competing ways of theorizing language. He admits that language may in fact not be a scientific object, and thus be incompatible with the kind of ‘Galilean’ idealizations he proposes. He allows that language may be multiple and integrated into other areas of social life. However he suggest that to accept this would be to commit a form disciplinary hari kari. If language isn’t systematic and able to be illumined by the light of scientific knowledge, it’s not worth studying – it is in fact not an object at all. Interestingly, part of Chomsky’s concern seems to be that linguistics would not have the right ‘place’ – the space it inhabited would not be bounded, whole, unified and discrete – it would be a distributed place, a ‘practiced’ place. Both the object and the discipline would not have the spatial qualities necessary to be unified, coherent, present and self-identical.
 Chomsky, 1997b.