Published Spring 2007
Why Everyone Should Study Linguistics
All literary critics already do some kind of linguistics. When we make even the most off-hand assertions about the meaning of a literary text, we commit ourselves de facto to assumptions about the nature of language and about specific aspects of linguistic structure. It's no surprise that intricate observations about the linguistic nuances of literary works depend on a theory of language. If we start talking about count nouns, subordinate clauses, or the indicative mood, we are drawing on a technical terminology that only has content by virtue of a specific theory of language (a theory, for example, in which some groups of words count as "clauses" and other groups of words don't). But one need not be a formalist for one's claims to depend on linguistics. Even a mere paraphrase would be unintelligible without unstated linguistic premises, and if its claims were to be justified explicitly these premises would necessarily step into the light. For example, say we are considering the first line of Shakespeare's first sonnet, "From fairest creatures we desire increase." We might plausibly gloss this line as saying "We want the most beautiful things to reproduce." But in asserting this as a paraphrase, we must assume that there is some systematic relationship between the sentence we started with and the sentence we offered as a gloss on it. We would want to say, for example, that "we" remains essentially unchanged between the original and the paraphrase, but to do so we must assume some notion of grammatical subject that explains in what sense "we" remains the same when it is obviously in a different spot. Even the most innocently general summary of what a text says—even the publisher's blurb on the back of a novel—would, if it had to get down to brass tacks and really make all of its claims and assumptions explicit, turn out to depend on premises of this kind. Linguistics is our inevitable hidden premise, just like one cannot infer "I am" from "I think" without assuming that "Everything that thinks, exists" (which is why Descartes goes to such trouble to deny that the cogito is an inference ). The only way one could avoid any implicit dependence on claims about language would be not to talk about the text at all, and a work of literary criticism that did not talk about the text at all could hardly meet even the most minimal standards for evidence.
So no argument need be made that we should draw on some theory of language in our criticism. We all do that already. Unfortunately, the theories that most of us draw on are substantially mistaken and long since superseded. I want to suggest that we should bring our linguistics up to date. We would expect, for example, an avowedly historicist critic to get the history right, and we would be justly critical of someone who produced a clever reading of a text but on the basis of seriously flawed historical research (imagine a historicist reading of Marvell's Horatian Ode that presupposed that Cromwell was unstintingly benevolent towards the Irish). Our expectations for the use of linguistics, on the other hard, are at present astonishingly low. The vast majority of literary scholars expect no familiarity with current linguistics, having no familiarity with it themselves. Surely grammar is the only domain in which virtually the entirety of literary scholarship is willing to accept uncritically the received wisdom of the nineteenth century. One sign of the strange relationship between literary study and linguistics is the fact that structuralism is still seriously discussed by literary scholars (e.g., in theory surveys) when it has been stone dead for nearly half a century in linguistics departments. Most literary scholars would be hard-pressed even to name a single living linguist other than Noam Chomsky—Luigi Rizzi and Norbert Hornstein may be names to conjure with, but not on our side of the quad.
Fortunately, the remedy for this state of affairs is comparatively painless. If we all tacitly employ some linguistic theory already (whatever its limitations) then employing linguistics must not prevent us from also employing approaches informed by history, or by gender studies, or by psychoanalysis, or by whatever else we find compelling. Since language and representation are so often under discussion in literary theory, we might expect a revision of our theory of language to have far-reaching consequences for the basic tenets of our frameworks. But the core disciplines of linguistics are focused on somewhat narrower issues than we might expect, and much of what we would want to say on literary-theoretical grounds about language and representation actually falls well outside of the concerns of linguists. For example, much of our discourse about language depends on the notion that there are different, particular languages, like English, Dutch, Chinese, etc. But this intuitive notion of human language is essentially empty from the standpoint of linguistics. As Chomsky observes:
We speak of Chinese as 'a language,' although the various 'Chinese dialects' are as diverse as the several Romance languages. We speak of Dutch and German as two separate languages, although some dialects of German are very close to dialects that we call 'Dutch' and are not mutually intelligible with others that we call 'German.' A standard remark in introductory linguistics courses is that a language is a dialect with an army and a navy (attributed to Max Weinreich). (Knowledge of Language 15)
The notion that there is such a thing as 'Dutch' or 'Chinese' is, more than anything, a socio-political fact, not a linguistic fact. Linguistically speaking, the variety of the world's languages only seems conspicuous to us because of our extremely delicate sensitivity to linguistic subtleties, but actually human language is much more uniform than it is various. If languages were really dramatically different from one another, then we would expect children learning different languages to learn at different rates and to learn different elements of the language in different orders; but in fact language acquisition is remarkably uniform across the board. There are, for example, no languages that are more 'complex' than others and that therefore take longer for native speakers to learn, and neither are there languages that are more 'primitive' and therefore easier for native speakers to learn. All of this means that in a sense we are wrong even to speak about something like 'the English language.' What we are really talking about when we talk about 'the English language' (or about 'Englishes') is often really socio-political in nature, and insofar as we are talking about a social reality, not a linguistic fact, our observations do not conflict with current linguistics—we are simply sailing in different waters. Here as elsewhere, revising our conception of language doesn't obstruct our inquiries but, if anything, helps us to sharpen them. It allows us to distinguish the properly linguistic from the political far more clearly, and thus to refute such claims as that the English spoken on the BBC or CNN, because it is supposedly 'correct,' should have some kind of special status in our curricula to the exclusion of dialects spoken by urban youths, by rural farmers, or by colonized peoples. Linguistics is not, of course, consistent with everything: there are properly linguistic notions to which, in our ignorance, we have committed ourselves and with which we might have to dispense. Some of the discoveries of current linguistics are exciting precisely because they run contrary to received wisdom. For example, according to current linguistics, the syntactic structure of a sentence is not in fact a matter of linear order (though it eventually has to be "linearized" in order to become a sequence of sounds [see Kayne for one influential account of linearization]). We are more than compensated for the loss of our antiquated notions about language by the benefits that a more accurate theory of language has to offer, and nothing prevents our engagement with linguistics from cohabiting happily with historicism or postcolonial theory or any other approach.
If we accept that we should take up the task of learning current linguistics, we then face the somewhat thorny problem of deciding which linguistics to learn—and it is no simple matter to choose between unknowns. However, there are three criteria which could guide us in making our decision: (1) relevance, (2) coherence, and (3) independent verification. The first of these criteria is the easiest to apply. Some areas of linguistics (such as computational linguistics), however interesting they may be in themselves, don't have any apparent application to literature and thus should not be our focus. We may make discoveries in unexpected quarters, of course, but our priority should clearly be those areas and approaches in linguistics that have the most direct potential bearing on the use of language in literature. The second criterion requires that whatever areas of linguistics we delve into be consistent with one another. We cannot, for example, adopt a functionalist account of syntax alongside a generative account of phonology, since they make mutually exclusive assumptions about how human language works. The third criterion, independent verification, is probably the most difficult to apply in the absence of substantial acquaintance with the scholarly literature in linguistics, but it is vitally important: the theory of language we adopt should be convincing as an account of human language, not just as an account of one particular use of language, like literature or conversation. In other words, a theory such as that proposed by George Lakoff and his adherents might seem appealingly literary (since everything under the sun becomes a metaphor), but if it is not independently convincing as a theory of language then its utility for thinking about literary language will be nil.
In my view, the kind of linguistics that best meets these three criteria (and that is presently the dominant model in linguistics) is what is sometimes known as 'theoretical linguistics' or 'generative linguistics.' This model, associated with Noam Chomsky, is oriented around syntax.1 One of the reasons Chomsky's model is particularly relevant to literary study is that it can help us to get a handle on one of the most persistent problems in literary study: the move from description to interpretation. It's easy enough to notice some conspicuous aspect of the language of a text, but how do we convincingly move from that observation to its basic meaning and thence to its significance? Chomsky's approach is uniquely adapted to grappling with this problem because of its emphasis on syntax. Rather than thinking of syntax as concerned only with form or arrangement (the sequence of words) and regarding questions of meaning as the concern of semantics, Chomsky regards syntax as including much of what has hitherto been regarded as semantics (KL 45). This view is intuitively plausible if we consider a structurally ambiguous sentence like Someone ate every tomato (which could mean either that some one person, Egbert, say, ate all the tomatoes, or it could mean that for every tomato there is some one person who ate it: Egbert ate one, Mathilde ate one, Armand ate one, etc.). This sentence is ambiguous by virtue of its syntactic structure, and it would be impossible either to explain the systematicity of its two meanings without appeal to its syntax or to show that it is syntactically ambiguous without appeal to its meanings. If Chomsky is therefore right that syntax already contains aspects of meaning—that in syntax form and basic meaning are not different things—then syntax allows us a very precise way to talk about how meaning emerges, and thus provides us with a firmer ground on which to erect our claims about significance. It follows from this that what a linguistically-informed criticism amounts to is not at all a resurrection of 'formalism.' One of the things we can learn from linguistics is that 'formalism' is, strictly speaking, impossible, since the basic units of human languages (words, or, more correctly, morphosyntactic features) are intrinsically bound up with meaning.
Theoretical linguistics also provides us with a coherent picture of how the human capacity for language works, thus linking syntax with phonetics and phonology on the one hand and semantics and morphology on the other (not to mention language acquisition). This is not to say that the picture is finished—any more than modern physics is 'finished'—but the basic model Chomsky and others have advocated is consistent with the available evidence and provides a very powerful explanation of why the evidence should take the form it does. Particularly at the beginning, Chomsky's proposals were extremely controversial. His first book, The Logical Structure of Linguistic Theory (completed in 1955), was unpublishable at the time because it was such a dramatic departure from the then-dominant approaches (mainly structuralism and behaviorism)—publishers understandably wondered who would buy it. But Chomsky was able to formulate challenges to the prevailing orthodoxy convincing enough to spur a considerable quantity of research on such issues as the distinctiveness of human language (from artificial languages or animal communication systems) and the richness of the innate language faculty. Part of the reason for the dominance of Chomsky's model in linguistics departments today is that empirical investigation ended up dramatically vindicating his views (see, for example, Crain and Nakayama for a classic experiment testing innateness, and see Piattelli-Palmarini for a discussion of the background issues).
The details of Chomsky's views have undergone many changes over the years, especially with the increase in the number of linguists working along similar lines. His first works, although certainly influenced by linguists like Leonard Bloomfield and Zelig Harris, were comparatively sui generis, but the landmarks of his later career have often synthesized a wide array of research undertaken by others into one coherent picture. The outsider's impression that the single personage of Chomsky utterly dominates the field of linguistics thus has a grain of truth, but it is a rather small grain. More recent developments are particularly important because they have involved significant changes in the basic model of the language faculty. The early forms of generative grammar owed much, as Chomsky often points out, to a long tradition of thinking about language in terms of universal grammar from Descartes through Humboldt and Goethe (see Chomsky's Cartesian Linguistics). This approach is still probably what most people associate with Chomsky, especially in the form presented in his Aspects of the Theory of Syntax: the language faculty uses rewriting rules to produce a "Deep Structure" representation of the sentence that reflects its semantics, then modifies the sentence by way of "Grammatical Transformations" to produce a "Surface Structure" representation, which in turn is converted into a string of sounds by a set of phonological rules. This approach (sometimes referred to as "Standard Theory") continued to be elaborated over a period of decades (eventually becoming "Extended Standard Theory"), which accounts in part for the familiarity at least of its broad outline. Beginning in the early 1980s (and marked by Chomsky's Lectures on Government and Binding), a new model, known as the principles and parameters approach, emerged in work by Chomsky and others that was without parallel in the tradition of thinking about language, namely the view that:
languages have no rules in anything like the familiar sense, and no theoretically significant grammatical constructions except as taxonomic artifacts. There are universal principles and a finite array of options as to how they apply (parameters), but no language-particular rules and no grammatical constructions of the traditional sort within or across languages. (The Minimalist Program 5-6)
This new approach crystallized into a particularly stringent research program in the 1990s, the minimalist program, that focused especially on issues of simplicity, which have long been remarked upon but have only now become a primary focus of inquiry (see Logical Structure 113-28 for an early formulation of the issues; see Rizzi for more recent discussion). These developments are worth noting, in part because they are not well known among literary scholars, but also because we must always remind ourselves that the picture of language we get from current linguistics—although it will be clearer and more true to the facts than the alternatives—is a work in progress. Because of this, we will want to begin learning linguistics through basic textbooks,2 but ideally we will then maintain at least a passing acquaintance with the scholarly literature in the field to keep ourselves abreast of new developments.
Keeping up with the technical literature in linguistics is no small undertaking, but the labor is not merely its own reward. An acquaintance with linguistics can, as we have seen, give us a clearer sense of what human language is and thus how it fits into social life, but it can also help to resolve disputes over linguistic features of literary works. To borrow an example from Timothy Austin (who has written the only serious book-length attempt to use Chomskyan syntax to interpret literature that I have found), consider the following lines from Shelley's elegy for Keats, Adonais:
Thy extreme hope, the loveliest and the last, The bloom, whose petals nipt before they blew Died on the promise of the fruit, is waste[.] (51-3)
At first glance, one might simply regard these lines as ungrammatical, perhaps suggesting the brokenness of the world in which Keats has been lost. But as Austin points out, these lines in fact do not violate any syntactic constraints, but rather a (pragmatic) constraint on processing—they have too many center-embedded clauses. We can add enormous amounts of information to the beginning or the end of a sentence without the sentence becoming difficult to parse: we can begin with Egbert sneezed and expand it indefinitely into Egbert's cousin sneezed or Egbert's cousin's chiropractor's sister's mailman sneezed, and so on. Similarly, we can tack phrases onto the end of a sentence indefinitely: Egbert sneezed in the park, on a bench, under a tree, during an eclipse, and so on. But if we add clauses to the middle of a sentence, it rapidly becomes difficult to parse, as in these examples from Austin, the first of which is quite difficult to parse and the second of which verges on being unintelligible:
? The woman whom the spider whose web broke startled her screamed.
* I'll introduce the woman who when the spider whose web broke startled her screamed to my aunt tomorrow.
Unlike the previous examples of "left-branching" and "right-branching" constructions, these "center-embedded" constructions seem very rapidly to cause problems. But such problems in the Shelley example do not result from the fact that the sentence is ungrammatical, as we can see if we put each of the embedded clauses in brackets:
[Thy extreme hope, the loveliest and the last, The bloom, [whose petals [nipt [before they blew]] Died on the promise of the fruit,] is waste]
This construction is a product of a perfectly legitimate operation: it causes no problems to go from The bloom is waste to The bloom, whose petals died on the promise of the fruit, is waste. The problem is that this legitimate operation has been repeated too many times to be easily processed by the reader (even though each successive application is grammatical). So we would, in fact, be simply wrong to say that the sentence is ungrammatical, and our wrongness would obstruct our understanding of the passage.
An acquaintance with linguistics can in this way save us from errors in our factual descriptions of any artwork made out of language. It can also help us to formulate more clearly our sometimes vague responses to literary language. We might, for example, look at those lines from Shelley and have the vague impression that their syntax is strange or complex. With the tools of linguistics, we can put our finger on exactly how those lines are syntactically complex. This will not, of course, instantly yield an interpretation. The significance of some set of linguistic facts is a product of critical reflection, and is not contained in the linguistic facts themselves (though the plausibility of an interpretation will obviously depend on the linguistic facts to which it appeals as evidence). It would make no sense to try to use linguistics to construct what Stanley Fish calls an "automatic interpretive procedure" (70). But having a clearer picture of what the linguistic facts are can be tremendously helpful for formulating and substantiating an interpretation, not least by making the picture more complex. If we observe that the lines from Shelley are syntactically strange, we might go on to say that the significance of this fact is that it alienates the reader from the language of the poem, thereby suggesting that with Keats's death all poetry has become simply unintelligible. This inference from a vague feeling to a broad generalization would be extremely difficult to justify if it were disputed. Whereas, using the linguistics, we might say something like this: The clause that pushes the sentence over the edge is "nipt before they blew," without which (as noted above) the sentence would produce no parsing difficulties. The addition of that clause is thus not only a quantitative change (one more clause) but also a qualitative change (from intelligibility to unintelligibility). The meaning of the sentence would also be much more straightforward without the additional clause: it would simply be a metaphor comparing Keats to a flower whose petals have withered and is therefore barren, incapable of fulfilling the "promise of the fruit" (OED "waste, a." 1b). But the additional clause complicates matters because if the petals have been "nipt before they blew," then the "bloom" not only hasn't bloomed but will never bloom, and thus isn't a "bloom" at all (OED "bloom, n.1" 1a). The calyx of the sentence, so to speak, thus posits what its gyncium revokes: the outer shell is an image of decrepit old age, whereas the core is an image of youth dead before its time. The merely quantitative addition of a clause to this center-embedded construction thus qualitatively transforms the meaning of the sentence into a "waste" that is not merely the absence of productive potential but both that which has been squandered and sheer emptiness. This interpretation might well be wrong, but its grounds, at least, can be meaningfully disputed, since it relies on the properties of center-embedded constructions (namely, that they have very strict quantitative limits). And it is difficult to imagine how one might arrive at this interpretation without the linguistic analysis, so this example also shows that using linguistics can reveal hitherto unrecognized aspects of a text: new problems and new mysteries, all at a finer level of grain.
An acquaintance with current linguistics can also be very useful in the classroom. In some contexts, directly teaching some linguistic material can make things much clearer for students (meter being one conspicuous example where the less technical traditional account is actually more difficult for students to understand than some forms of generative metrics). Even if one doesn't teach material derived from linguistics, it can be tremendously useful any time one is dealing with language, as when marking student papers. For example, say you come across a sentence like this in a student paper: "He attempts to justify that he is not insane by writing everything down." Just writing "awk." in the margin isn't going to help the student understand what is wrong with this sentence. But some acquaintance with linguistics will tell you that the problem with this sentence is not that it violates some general syntactic principle: if we replace "justify" with another verb like "explain," the sentence sounds just fine. The problem, then, is that the verb justify must be followed by a noun phrase (Egbert justified his statement) and cannot be followed by that+sentence (*Egbert justified that his statement was correct). In other words, the problem here is a result of an idiosyncratic property of the verb justify, namely how it restricts what kinds of words can follow it (a phenomenon that linguists call categorial-selection, or just c-selection), and so one cannot help this student by appeal to a general rule. In order for the student to learn to avoid this error, he or she would have to learn this idiosyncratic property of justify. Once we know this, we have a much clearer idea of how to help such a student.
So employing current linguistics has a wide array of benefits: saving us from error, putting our vague intuitions on a firmer footing, revealing whole new worlds of complexity, and sharpening our pedagogy. In order for us to reap these benefits, we will need to commit, not merely to learning some linguistics ourselves, but to sowing the seeds of a linguistically-informed future for the profession. We can take some initial steps in this direction with relative ease. For example, we could require undergraduate literature majors to take one introductory linguistics course (providing such a course ourselves if it is not already available), and we could require graduate students also to take one advanced linguistics course. This comparatively modest requirement would cultivate, over a period of years, a new generation of literary scholars who share some real acquaintance with the terminology and methods of linguistics. However different each of their theoretical frameworks and approaches might end up being, they would all have linguistics (and thus the text itself) in common, as a lingua franca that could break through some of the barriers that prevent us from talking with one another. Drawing on linguistics is not thereby a kind of interdisciplinarity that would dissolve the distinctiveness of what we do—making us yet another handmaiden to the natural sciences—nor would it bind us together through cherished doctrines, shared values, or a uniform point of view. It would give us new and more precise ways to make our arguments, new modes in which the dialectic of thought might unfold, and new problems to stoke the living fire of critical reflection.
1. One convenient way to distinguish general theories of language is by their orienting subfield: for nineteenth century philology, that subfield was historical linguistics; for structuralism, it was phonetics; for behaviorism, pragmatics; for functionalism, it is again pragmatics, though of a different sort; and for generative linguistics, it is syntax (along with first language acquisition). See O'Grady, et al., for a basic account of the main subfields in linguistics.
2. Speaking of textbooks: for a broad introduction to linguistics as a whole, see O'Grady, et al. (and for a more focused and detailed introduction, see Fromkin). For a good introduction to syntax, see Adger (and for a more advanced textbook, see Hornstein, et al.). For an introduction to phonetics and phonology, see Clark, et al. (and for a more advanced textbook on phonology, see Kenstowicz).
Adger, David. Core Syntax: A Minimalist Approach. Oxford: Oxford UP, 2003.
Austin, Timothy R. Language Crafted: A Linguistic Theory of Poetic Syntax. Bloomington, IN: Indiana UP, 1984.
Chomsky, Noam. The Logical Structure of Linguistic Theory. 1955. New York: Plenum, 1975.
---. Syntactic Structures. 1957. 2nd ed. The Hague: Mouton de Gruyter, 2002.
---. Aspects of the Theory of Syntax. Cambridge, MA: MIT P, 1965.
---. Cartesian Linguistics: A Chapter in the History of Rationalist Thought. New York: Harper and Row, 1966.
---. Studies on Semantics in Generative Grammar. The Hague: Mouton, 1972.
---. Lectures on Government and Binding. Dordrecht, Holland: Foris, 1981.
---. Knowledge of Language: Its Nature, Origin, and Use. Westport, CT: Praeger, 1986.
---. The Minimalist Program. Cambridge, MA: MIT P, 1995.
Clark, John, Colin Yallop, and Janet Fletcher. An Introduction to Phonetics and Phonology. 3rd ed. Oxford: Blackwell, 2007.
Crain, Stephen, and M. Nakayama. "Structure Dependence in Grammar Formation." Language 63 (1987): 522-43.
Descartes, René. Meditations on First Philosophy. Trans. and Ed. John Cottingham. Cambridge: Cambridge UP, 1996.
Fish, Stanley. "What Is Stylistics and Why Are They Saying Such Terrible Things About It?" Is There a Text In this Class?: The Authority of Interpretive Communities. Cambridge, MA: Harvard UP, 1980.
Fromkin, Victoria A., ed. Linguistics: An Introduction to Linguistic Theory. Oxford: Blackwell, 2001.
Hornstein, Norbert, Jairo Nuñes, and Kleanthes K. Grohmann. Understanding Minimalism. Cambridge: Cambridge UP, 2005.
Kayne, Richard S. The Antisymmetry of Syntax. Cambridge, MA: MIT P, 1994.
Kenstowicz, Michael. Phonology in Generative Grammar. Oxford: Blackwell, 1994.
OED Online. 2007. Oxford UP. 12 June 2007 <http://www.dictionary.oed.com>.
O'Grady, William, John Archibald, Mark Aronoff, and Janie Rees-Miller. Contemporary Linguistics: An Introduction. 5th ed. Boston: Bedford/St. Martin's, 2005.
Piattelli-Palmarini, Massimo, ed. Language and Learning: The Debate between Jean Piaget and Noam Chomsky. Cambridge, MA: Harvard UP, 1980.
Rizzi, Luigi. Relativized Minimality. Cambridge, MA: MIT P, 1990.
Shakespeare, William. Shakespeare's Sonnets. Ed. Katherine Duncan-Jones. London: Arden-Thomson, 1997.
Shelley, Percy Bysshe. Shelley's Poetry and Prose. Ed. Donald H. Reiman and Neil Fraistat. 2nd ed. New York: Norton, 2002.