Linguistic Criticism in Poetry

Overview

In many ways, linguistic criticism is a legacy of the work done in the early twentieth century by Ferdinand de Saussure (1857-1913), the great French theorist whose posthumously published work has been the point of departure for all modern structuralism, in linguistics, anthropology, and other disciplines. American structural linguistics, the principles and research program laid down by Leonard Bloomfield (1887-1949) in his Language (1933), was concerned with establishing the structure—phonological, morphological, syntactical, and semantic—of languages conceived as systematic wholes. It dealt with what Saussure’s Cours de linguistique generale (1916; Course in General Linguistics) called the langue, the system of a limited repertoire of sounds on whose differentiation differences of meaning depend and of a limited number of kinds of sentence elements that can be combined in certain orders and hierarchical relationships, as distinguished from parole, particular utterances. To elicit these elements and rules of combination for a given language, linguists depended primarily on speech rather than on written texts. Furthermore, Bloomfieldian linguists tended to concentrate their efforts on the description of “exotic” languages rather than of English. Meanwhile, literary critics focused their attention on particular written texts, deemed literary, and were concerned with the interpretation and evaluation of these works of individual writers.

Poems and other literary works are, of course, works of verbal art whose medium is language, and this fact suggested to some linguists a potential for the application of linguistics to the study of literature. The techniques developed by linguists for the analysis of language could be applied to the language of literary texts. The linguist could bring to bear on these texts an expertise that the literary critic did not have. Much of the work involves the application of linguistics to the study of literary texts falls under the heading of stylistics, the study of literary style. Early contributions in linguistic stylistics tended to carry the animus of bringing a new objectivity to a field that had hitherto been merely impressionistic in its methods. However, later work has generally recognized both the inevitability and the value of a subjective component in stylistic analysis. While the interest of literary criticism and, for the most part, stylistics, is in particular literary texts, modern poetics, which traces its roots to Russian Formalism (a movement in literary scholarship originating in Russia around 1915 and suppressed there about 1930), is instead interested in the question of the nature of literariness or poeticity. The study of poetics is closely related to the study of linguistics.

Besides the application of the specific techniques and categories of one or another kind of linguistics for studying the language of literary texts, another kind of application of linguistics to the study of literature has been of great importance: Linguistics has been taken as a model for the study of literary structures, such as narrative, that are not intrinsically linguistic but are translatable into other media. Literature students have identified constituents of such structures and formalized descriptive rules for their combination by analogy with linguists’ descriptions of the structure of a language (langue). Thus, literary critics have participated in a transdisciplinary movement affecting all the human sciences by adopting the structuralist approach derived from Saussure. However, a discussion of structuralism in literary studies is outside the scope of this survey of linguistic approaches to poetry, which will be limited to applications of the techniques and categories of linguistic analysis to the language of poetic texts. Even within these limits, this survey does not purport to cover all significant works. The contributions discussed below—including general, programmatic pieces and specific, descriptive studies—do not represent all the important figures in the field or even the full range of relevant work by those cited. They are examples of various linguistic approaches to poetry from the early 1950s to the early 1980s.

Beginnings

When, in 1951, the linguists George L. Trager (1906-1992) and Henry Lee Smith, Jr. (1913-1972), published An Outline of English Structure, a description of English phonetics, morphology, and syntax such as had been made of many exotic languages, the linguist Harold Whitehall immediately saw the possibility of application to the study of English literature and went so far as to say that “no criticism can go beyond its linguistics.” Specifically, he saw that Trager and Smith’s account of stress, pitch, and juncture—as each having four functionally distinguishable levels in English—would be valuable to students of meter and rhythm of English verse. The rhythm of lines could presumably be much more precisely described in terms of four levels of stress rather than in the two normally recognized in traditional metrics. Such an application of Trager and Smith’s findings for modern American speech to the study of verse in English was actually made by Edmund L. Epstein and Terence Hawkes, who found in a body of iambic pentameter verse myriad stress patterns according to the four-level system. The question of the relevance of such descriptive analysis to the meaning and aesthetic value of the poetry in question was not raised in these early applications of linguistics to literature.

Archibald A. Hill, who did practical work in the application of linguistics to literature in the 1950s, saw a linguistic analysis of a text in terms of such factors as word order and stress as operating on a preliterary level but considered them to be a useful preliminary to analysis in terms of literary categories such as images. Not unlike the New Critics, he approached a poem as a structured whole. He sought to interpret it with minimal reference to outside knowledge, such as biographical information. Still, he differed from the New Critics in thinking it best to begin with specifically linguistic formal details. In a 1955 discussion of Gerard Manley Hopkins’s “The Windhover” (wr. 1877, pb. 1918), Hill took an analysis of Hopkins’s stress and word order, considered in relation to general English usage, as the basis for resolving ambiguities and determining emphases at particular points—and applied it, as an aid to interpretation of at least local meaning.

Roman Jakobson

In 1958, the Conference on Style was held at Indiana University, in which a group of linguists, literary critics, and psychologists presented and discussed papers on issues relevant to the matter of style in language. This conference, the papers of which were published in 1960 in a collection titled Style in Language, edited by Thomas A. Sebeok, proved to be something of a watershed for the application of linguistics to the study of poetry. Roman Jakobson’s presentation on the relation of linguistics to poetics still remains a point of reference for work on the language of poetry. While the focus of the New Critics and some linguists working on literature was on the individual poem considered as an autonomous structured whole, the concern of poetics, as Jakobson sets it forth, is with the differentiating characteristics of poetic language. He argues that inasmuch as poetics deals with problems of verbal structure, it lies wholly within the field of linguistics, “the global science of verbal structure.” He offers a functional definition of poetic language in terms of the constitutive factors of any act of verbal communication, enumerating six such factors—the addresser, the message, the addressee, the context, the contact between the addresser and addressee, and the code (the rules of the language, also of a certain register, dialect, and so on) in accord with which the message is constructed. In any given utterance or text, focus will be on one of these factors primarily, though to a lesser extent on others, and the predominant function of the utterance or text can be defined accordingly. Jakobson defines the poetic function of language as “focus on the message for its own sake,” stressing that the poetic function is not confined to poetry (appearing also, for example, in political slogans and advertising jingles) and that poetry involves functions of a language other than the poetic; different genres, for example, are partially characterized by the relative importance of the referential, emotive, or conative (focus on the addressee) functions.

Contending that “the verbal structure of a message depends predominantly on [its] predominant function,” Jakobson then studies the effect of the poetic function on the linguistic structure of a text. His famous account of the differentiating feature of language in which the poetic function predominates over the other functions depends on the fact that making an utterance or constructing a text always involves two operations: selection from among a series of items that are syntactically and semantically equivalent and the combination of the selected items into a meaningful sequence of words. “The poetic function,” runs Jakobson’s formulation, “projects the principle of equivalence from the axis of selection into the axis of combination.” In verse, for example, every syllable becomes equivalent to every other syllable, every stress to every other stress, as units of measure. A passage of verse is characterized by the repetition of equivalent units.

Jakobson goes on to cite some of the kinds and operations of equivalence in a broad range of poetry of many different languages, in the process of providing what amounts to a program for research on poetic language. All metrical systems, he says, use “at least one (or more than one) binary contrast of a relatively high and relatively low prominence”; he gives examples of meters “based only on the opposition of syllabic peaks and slopes (syllabic verse),” meters based “on the relative levels of the peaks (accentual verse),” and meters based “on the relative length of the syllabic peaks or entire syllables (quantitative verse).” Besides features invariably present in lines in a given meter, optional features will be likely to occur, and these, Jakobson maintains, form part of the metrical system to be described by the linguist. A full description should not, as traditional descriptions of meters typically do, exclude any linguistic feature of the verse design. Jakobson cites, as an example of a feature that ought not to be ignored, the “constitutive value of intonation in English meters,” “the normal coincidence of syntactic pause and pausal intonation” with line-ending, such that, even when frequent, enjambment is felt as a variation.

Word boundaries and grammatical boundaries may also be among the defining characteristics of a line in a given verse tradition, even if such boundaries are not marked by any distinguishable phonetic features. (Jakobson does not specify a method for ascertaining that the enumeration of relevant features has been exhaustive.) What have generally been treated as deviations from a metrical pattern should, according to Jakobson, form part of the description of the pattern, for they are variations allowed by the rules of the given meter.

For rhyme and meter, Jakobson emphasizes that linguistic analysis should not be limited to sound alone. The similarity of sound between rhyme words throws into relief their grammatical and semantic relations. Whether rhyme words are of the same or different grammatical classes, whether their syntactical functions are the same or different, and whether they have a semantic relationship of similarity or antithesis are all questions relevant to the operation of rhyme in the poetry in question. Poets and schools of poetry that use rhyme differ in favoring or opposing rhyming words of the same grammatical class and function, and grammatical rhymes operate differently from anti-grammatical ones.

While repetition is an important aspect of the sound of poetry, a sound can be important, Jakobson points out, without being repeated: A sound with a single occurrence in a prominent position against a contrasting background and in a thematically important word should not be neglected. Analysis of the sound in poetry must consider the phonological structure of the language in question and which of the distinctive features of phonemes (voiced/voiceless, nasal/oral, and so on) are considered in the particular verse convention. Besides meter, rhyme, alliteration, and other forms of reiteration that are primarily of sound, though also involving syntactic and semantic aspects, poetic language is characterized by other forms of parallelism. For example, lines may be grammatically parallel, inviting semantic comparison between words in corresponding positions, which may be perceived as having a metaphorical or quasi-metaphorical relationship. Concentration on lexical tropes to the exclusion of the syntactical aspect of poetic language is, according to Jakobson, not warranted:

The poetic resources concealed in the morphological and syntactic structure of language, briefly the poetry of grammar, and its literary product, the grammar of poetry, have been seldom known to critics and mostly disregarded by linguists but skillfully mastered by creative writers.

Michael Riffaterre

Jakobson asserted “the right and duty of linguistics to direct the investigation of verbal art in all its compass and extent,” and provided a program for such research. At about the same time, Michael Riffaterre was concerning himself with the problem that a linguistic analysis of a poetic text could provide only a linguistic description of it and could not distinguish which of the features isolated were operative as part of the poem’s style. In a 1959 paper, “Criteria for Style Analysis,” he endeavored to supply a technique for distinguishing stylistic features of a text from merely linguistic features without stylistic function.

Riffaterre’s argument is based on the dual assumption—diametrically opposed to Jakobson’s notion of the nature of poetic language—that the literary artist works to ensure the communication of their meaning, and that this end can be achieved through reduction of the predictability of elements, so that the reader’s natural tendency to interpolate elements that seem predictable from the context will be frustrated, and they will be held up and forced to attend to unpredictable elements. The linguistic elements that are considered in the stylistic analysis of a literary work are precisely these unpredictable elements, which Riffaterre calls stylistic devices or SDs.

While Riffaterre is confident of the efficacy of this means for getting reader attention, at least one investigator of style in poetry (Anne Cluysenaar) cites an instance in which readers failed to notice an unpredictable phrase, simply substituting what they would have expected for what was actually there in the text. Not content that the analyst of style should rely simply on his own subjective impressions as to the location of stylistic devices in a literary text, Riffaterre recommends using informants, readers of the text in question, including critics and editors of the text and lay readers. In using informants’ responses, the analyst of style should, according to Riffaterre, empty them of such content as value judgments and take them as mere indicators of possible sites of stylistic devices. The resultant enumeration should then be verified, by checking whether the points identified coincide with points where a pattern established by a preceding stretch of text has been broken. If so, they are stylistic devices and should be submitted to linguistic analysis by the stylistician.

In choosing to compare parts of a given text with each other, Riffaterre departed from the practice, which became common at the time, of taking ordinary language—a very vague entity—as a norm to compare the text. However, his examples show that he could not dispense concerning language usage outside the text in question, for items can be interpreted as unpredictable in a given context only concerning knowledge or experience of usage in a similar context elsewhere. Besides the departure from a pattern established by a preceding stretch of text, another phenomenon can, according to Riffaterre, help confirm the identification of a stylistic device: This is the presence of a cluster of independent stylistic devices, which together highlight a particular passage.

In a well-known paper published in Yale French Studies in 1966, Riffaterre assesses an analysis by Jakobson and Claude Lévi-Strauss of the poem “Les Chats” (1847; The Cats, 1955) of Charles Baudelaire. Jakobson and Lévi-Strauss had scanned the text on several levels—meter, phonology, grammar, and meaning—discovering equivalences of various kinds, which they took as constitutive of several simultaneous structural divisions of the poem. Riffaterre contends that many of the linguistic equivalences they identify cannot be taken as stylistic features, as elements of the poetic structure, because they would not be perceived in the process of reading. Only such equivalences as would be perceptible should be taken as pertaining to the poetic structure of the text. Equivalences on one level alone generally will not be perceived as correspondences. Grammatical parallelism, for example, will need to be reinforced by correspondence in metrical position.

Transformational-generative grammar

With the development of transformational-generative grammar by Noam Chomsky and others, beginning with Chomsky’s 1957 book Syntactic Structures, came new kinds of linguistic approaches to literature. The theory of transformational-generative grammar is based on the assumption that native speakers of a language internalize grammatical rules for their language such that they are able to produce unlimited numbers of grammatical sentences they have never heard before and to judge a given sentence as grammatical or ungrammatical (or as more or less grammatical, more or less complex).

There is the further assumption that the grammar should reflect native speakers' intuitions of the relation between superficially different sentences, such as between a given declarative sentence in the active voice and its passivization, and of ambiguity as to the construction of certain sentences. In the former case, the superficially different sentences are taken as having the same “deep structure”; in the latter, the surface structure of the given sentence is taken as able to have been reached by two or more different routes from two or more different deep structures. A transformational-generative grammar of a given language ideally consists of an ordered set of rules for generating all possible grammatical sentences in the language. Besides phrase-structure rules, it includes an ordered series of obligatory and optional transformational rules that transform an underlying “kernel sentence” or set of kernel sentences into a surface structure.

The surface structure/deep structure distinction was taken in early efforts at applying transformational-generative grammar to the study of literary texts, such as Richard Ohmann’s, to confirm and clarify the traditional distinction between form and content. A writer’s style could be accounted for by the nature of the optional transformations he chose. Ohmann adheres to the position of early transformational-generative theory that different surface structures produced by the choice of different optional transformations have the same content. Still, he contradicts himself when he says that each writer will make characteristic choices and that these choices correlate with the writer’s way of looking at experience. Likewise, regarding deviance in poetic language from usage restrictions on categories of words, he holds that the kinds of deviance employed by a poet will reflect their vision; the kinds of deviance found in Dylan Thomas’s poetry, Ohmann suggests, reflect his sense of nature as personal, the world as a process.

Samuel Levin was one of the first linguists to apply transformational-generative grammar to the study of style in poetry. Levin was interested in poetry containing sentences and phrases that a native speaker might consider ungrammatical or semigrammatical. In a 1965 paper called “Internal and External Deviation in Poetry,” he takes external deviation in syntax—deviation from a norm of syntactical usage lying outside the text in question—as ungrammaticality. In other words, he assumes that it involves sentences that the grammar of the language in question would not generate. Among such sentences, he recognizes degrees of deviance or ungrammaticality. He does not find the notion of the probability of a given element at a given point in the text (transitional probability) helpful in rationalizing this sense of degrees of deviance because of the unfeasibility of calculating transitional probabilities for the occurrence of a given word after a given sequence of preceding words. His approach is to determine what changes would have to be made in the grammar rules to generate these deviant sentences or phrases. These would be deemed more or less grammatical depending on the number of ungrammatical sentences the changes would generate.

In the case of a phrase such as E. E. Cummings’s “anyone lived in a pretty how town” (1940), the number of ungrammatical sentences would be large, since either “how” would have to be added to the class of adjectives, or adverbs would have to be allowed to occur in the place of the second adjective in the sequence determiner plus adjective plus adjective plus noun, changes involving large classes. In the case of Thomas’s phrase, “Rage me back,” however, “rage” would simply have to be added to the subclass of transitive verbs taking “back” (mostly verbs of motion). Hence, the Cummings example is less grammatical than the example of Thomas. In like manner, the Thomas example produces a sense of richness through the conflation of the verb “rage” with notions of transitivity and motion, while Cummings’s phrase leaves a sense of diffuseness because what is added to “how” is so unspecific. Levin notes that the former kind of ungrammaticality occurs more frequently than the latter in poetry and considers it akin to metaphor in its operation.

J. P. Thorne shares Levin’s interest in the occurrence in poetry of sentences that would not be generated by the grammar; indeed, he says at one point that such sentences form the subject matter of stylistics. Where Levin takes single sentences and phrases from poetry and considers how the grammar would have to be modified to generate each sentence, however, Thorne proposes taking whole poems and constructing a grammar for the language of each poem. The grammar for a poem should be constructed on the same principles as the grammar for the language as a whole, the point being to compare the two and discover how the one differs from the other. A good poet, in Thorne’s view, will invent a new language, differing from the standard language not (or not primarily) in surface structure, but in deep structure, that is, on the level of meaning. The poet will invent a new language to be able to say things that cannot be said in the standard language. In reading a good poem, the reader learns a new language. The grammar that one constructs for this new language will make explicit one’s intuitions about its structure. One must decide which features of the poem’s language are features of that language and which are features merely of the sample, when the sample (the poem) is all of the language there is. One’s assignments of words to categories in this grammar and formulation of selection rules for their co-occurrence will reflect one’s interpretation of the poem.

In a 1965 paper, “Stylistics and Generative Grammars,” Thorne sketches a grammar for Cummings’s poem “anyone lived in a pretty how town.” Thorne also suggests that it may be illuminating to construct grammars for poems in which the language does not seem so manifestly ungrammatical, such as John Donne’s “A Nocturnal upon St. Lucy’s Day” (1633), where animate subjects have verbs normally selected by inanimate subjects, and inanimate subjects have verbs normally selected only by animate subjects. In a 1970 paper, “Generative Grammar and Stylistic Analysis,” Thorne finds a similar deviation from the selection rules for Standard English in Theodore Roethke’s poem “Dolor” (1943), where there are constructions such as “the sadness of pencils,” which attach to concrete, inanimate nouns adjectives that normally select animate nouns or a subcategory of abstract nouns including “experience” and “occasion”; he points out that the decision whether to assimilate “pencils” and the other nouns so used in the poem to the category of animate nouns or to the particular subcategory of abstract nouns will depend on how one reads the poem.

Early efforts at applying transformational-generative grammar to literature have been recognized as limited by (1) the theory’s separation of syntax from semantics (that is, of consideration of sentence structure from questions of meaning); (2) the assumption in early versions of the theory that the transformations that produced different surface structures from a single underlying structure did not affect the meaning; (3) the failure to extend analysis beyond the level of the sentence; and (4) the treatment of literary (especially poetic) language as characterized by ungrammatical constructions, deviations from the supposed norm of everyday speech. With these limitations, transformational-generative grammar could not illuminate the relation between form and content in poetry or relate formal description to interpretation. It did not offer a means of discussing the connections within and cohesiveness of a text as a whole. It had nothing to say about poetry in which the language was not in any sense ungrammatical. Much of the irritation of literary critics with early applications of transformational-generative grammar to literature is attributable to limitations in the applications resulting from limitations in the early versions of the theory.

Category-scale grammar

Besides transformational-generative grammar, other modes of syntactical analysis developed by modern linguistics have been used in approaches to poetry. One of these is category-scale grammar, developed by M. A. K. Halliday and set forth in his 1961 paper “Categories of the Theory of Grammar.” Category-scale grammar analyzes English syntax in terms of a hierarchically ordered enumeration or “rank-scale” of units: sentence, clause, group, word, morpheme. Halliday introduces the notion of rank-shift to refer to cases where a unit operates as a structural member of a unit of the same or lower rank; for example, a clause can be part of another clause or of a group (phrase). Category-scale grammar has been commended to students of literary style as making possible a clear and accurate description of the infrastructuring of the language and helping to discover and specify where the structural complexity of a text resides. Halliday himself has advocated the linguistic study of literary texts, arguing that this should be a comparative study and that it is not enough to discover, say, the kinds of clause structures in a given text, but that their relative frequencies in that text should then be compared with those in other texts, other samples of the language.

In a 1964 paper, Halliday illustrates the sort of treatment he recommends with a discussion of two features of the language of William Butler Yeats’s “Leda and the Swan” (1928)—the use of “the” and the forms and nature of the verbs. “The” is a deictic (a word that points to or identifies); its particular function is to identify a specific subset, by reference to the context (either of the text itself or of the situation of its utterance), to elements of the rest of the modifier or of the qualifier of the noun it modifies, or to the noun modified itself. In nominal groups where there are other modifying elements preceding the noun or a qualifier following it, “the” usually (in samples of modern English prose referred to by Halliday) specifies by reference to the rest of the modifier or the qualifier. While “Leda and the Swan” has a high proportion of such nominal groups in comparison with another Yeats poem, in only one of them does “the” function in the usual way. In all the rest it specifies by reference not to anything else in the nominal group, but to the title. “Leda and the Swan” is also found to differ markedly in the handling of verbs from both another Yeats poem and a poem by Alfred, Lord Tennyson. A high proportion of verbs, including especially the lexically more powerful, are “deverbalized” by occurring as participles in bound or rank-shifted clauses or as modifiers of a noun (rather than as finite verbs in free clauses).

John Sinclair’s 1966 paper “Taking a Poem to Pieces” is the sort of analysis recommended by Halliday. Further, it represents an effort to remedy linguists’ neglect of poetry in which the language is not describable as ungrammatical or deviant. Sinclair hypothesizes that even in poetry, where the language is apparently unremarkable, grammatical and other linguistic patterns are operating in a more complex way than could be described—or even perceived—with traditional terms. He uses the terms of category-scale grammar to describe the language of a short poem by Philip Larkin called “First Sight” (1956).

Beginning with the highest unit of syntax, the sentence, Sinclair sets forth the syntactical structure of Larkin’s text: first the sentence structure (the nature and arrangement of the constituent clauses), then the clause structure (the nature and arrangement of the constituent groups and rank-shifted clauses), then the structure of the groups (the arrangement and relations of the constituent words and rank-shifted higher structures). He shows that the language of Larkin’s poems represents a restricted selection from among the wide range of possibilities afforded by the language. Its stylistic character can be in part accounted for by the persistent selection of certain constructions normally occurring with lower frequency.

It is on the level of the clausal constitution of its sentences that Larkin finds the particular quality of the language of “First Sight” to lie. In what he calls everyday English, in sentences that contain a free clause and a bound clause, the most common arrangement is an uninterrupted free clause followed by a bound clause; discontinuous structure (that is, a free clause interrupted by a bound clause) and the sequence of bound clause followed by free clause are both less common. Of the four sentences in Larkin’s poem, only the last has the generally most common arrangement of free clause followed by bound clause. Thus, while this poem’s language is not deviant in such a way as to require a special grammar to describe it, it is distinguished by the relative frequency of certain otherwise unremarkable structures.

In a 1972 paper, “Lines About ’Lines,’” Sinclair attempts to integrate stylistic description with interpretation and develop theoretical principles and a methodology for stylistic analysis. He assumes that the analyst must begin with a critical understanding of the text; then look for patterns at successive levels; in each case where a pattern is found, relate it to the meaning; and finally synthesize the findings of form-meaning relationships. Sinclair hypothesizes that there will be “intersection points” of form and meaning and adopts the term “focusing categories” or “focats” for such points. The focats found in the analysis of a given text will be initially taken as pertaining to that text alone, but assumed to be general if subsequently encountered in numerous texts.

In this case, Sinclair takes as his example for analysis William Wordsworth’s famous “Tintern Abbey” (1798). He finds two focats operative in the poem. One, the introduction of an optional element in a syntactical structure not yet complete, he calls arrest. The other, which he calls extension, is essentially the continuation of a potentially completed structure by an element not syntactically predictable from any of the preceding elements. Sinclair argues convincingly that arrest and extension do indeed represent instances of significant interrelation of formal structure and meaning in “Tintern Abbey.” These two focats also seem to be likely candidates for generality (occurrence in numerous texts).

Generative metrics

Many of the examples of linguistic approaches to poetry so far considered here have been studies of the language, especially the syntax, of individual poems. Besides focusing on individual texts, however, linguists have also addressed themselves to more general phenomena of poetic language, such as meter and metaphor.

Beginning with a 1966 paper by Morris Halle and Samuel Jay Keyser, a generative metrics was developed, devoted almost exclusively to accentual-syllabic verse, principally iambic pentameter. Halle and Keyser draw an analogy between the native speaker of a language, who has internalized a set of logically ordered rules in accord with which they produce grammatical sentences, and the poet, whom they assume to have similarly internalized a set of rules in accord with which they produce metrical lines. Generative metrics does away with the notion of the metrical foot and replaces the hodgepodge of rules and exceptions of traditional metrics with a brief and ordered sequence of systematically related rules. This sequence of rules governs the realization of an abstract pattern in an actual text. The pattern is represented as consisting of positions rather than feet, each position corresponding to a single syllable. The rules for actualization of the pattern are presented as alternatives arranged in an order from least to greatest metrical “complexity” (greatest to least strictness).

David Crystal: Tone-units

A very different approach to metrics has been taken by the British linguist David Crystal, who, in a 1971 paper, proposed a model for the description of English verse that is supposed to encompass both accentual-syllabic and free verse and to distinguish both from prose. He takes the line, rather than the syllable or the foot, as the basic unit of verse, and hypothesizes that the line normally consists, in performance, of a single complete “tone-unit.” A tone-unit is the basic unit of organization of intonation in an utterance; since intonation functions in part to signal syntactical relations, tone-unit boundaries coincide with syntactical—generally clause—boundaries.

Speech act theory

Transformationalists have approached metaphor as deviance through violation of selection restriction rules (rules formalizing acceptable collocations—for example, that for verbs with certain semantic features the subject must be animate). One interprets metaphors, according to this view (espoused, for example, by Robert J. Matthews in a 1971 paper), by deemphasizing those semantic features entailed in the selection restriction violation.

In the 1970s and 1980s, there was a growing recognition that properties such as metaphoricity, previously assumed to be peculiar to literary language, pertain as well to conversational usage; also, not only the formal features of a text but also the situational context and the speaker-hearer (or writer-reader) relationship were now seen as relevant to its operation.

Growing interest in speech act theory (developed by J. L. Austin and John R. Searle) is a reflection of the concern for context-sensitive analysis of utterances or texts. According to speech act theory, besides performing a locutionary act (that is, producing a grammatical utterance), a speaker will perform one or another kind of illocutionary act (such as asserting, ordering, promising) and possibly a perlocutionary act (that is, bringing about a certain state in the hearer). Besides being grammatical or ungrammatical, an utterance will be appropriate or inappropriate in the given situation; appropriateness conditions—internalized rules of language use that speakers assume to be in force—can be formulated.

While the Austinian treatment of appropriateness conditions is to define them in relation to particular speech acts in particular contexts, H. Paul Grice has generalized the notion of appropriateness, developing rules intended to apply to all discourse. His cooperative principle, which a participant in a speech exchange supposedly will normally assume their interlocutor to know and to be trying to observe and expecting them to observe also, is “Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk-exchange in which you are engaged.” At the same time, linguists have been extending the purview of their discipline in ways that bring it into close accord with speech act philosophy. Since 1968, post-Chomskyan generative semantics (as developed by George Lakoff, Robin Lakoff, Charles Fillmore, and others) has extended the notion of speaker competence to embrace the ability to perform appropriate speech acts in particular situations, as well as phonology and syntax. In addition, sociolinguists are concerned with language use.

How do these developments in linguistics and allied fields relate to the study of literature, particularly poetry? Mary Louise Pratt claims (in Toward a Speech Act Theory of Literary Discourse, 1977) that, with these developments, linguistics is for the first time able to describe literary discourse in terms of the literary speech situation, to define it in terms of use rather than of intrinsic features, and to relate it to other kinds of language use. She cites studies showing that features assumed by poeticians to be exclusively attributes of literary discourse also occur in conversation. Specifically, structural and stylistic features like those found in fiction also occur in “natural narratives.” These formal similarities, she thinks, can be in part explained by the fact that with both natural narratives and literary works, the speech exchange situation is one in which the hearer or reader is a voluntary, nonparticipant audience. She also considers it important to consider the reader normally knows that a literary work was, and was intended to be, published and assumes it was composed in writing with an opportunity for deliberation, hence that it is more likely to be worthwhile than casual utterances; because of these assumptions on the part of the reader, a literary work can get away with being “difficult,” and with making considerable demands on its audience.

Pratt attempts to adapt the categories of speech act theorists to accommodate literary discourse and relate it to other kinds of speech acts. She considers that many, if not all, literary works, together with exclamations and natural narratives, fall into the class of speech acts that have been characterized as thought-producing (as opposed to action-producing), representative (representing a state of affairs), or world-describing (as opposed to world-changing). She sees exclamations, natural narratives, and literary works together as constituting a subclass of representative speech acts that are characterized by “tellability,” that is, by the unlikelihood or problematical nature of the state of affairs represented (whether fictional or not). This characteristic she holds to pertain as much to lyric poetry as to novels and short stories. The subclass is further characterized by (1) detachability from any immediate speech context (this is obvious for literary works, which generally have no immediate relation to the situation in which the reader finds themselves) and (2) a tendency to elaborate on the state of affairs represented. Indeed, Pratt suggests that this elaboration is what literary works chiefly do.

Sociolinguistics

Roger Fowler, a British linguist, advocated and practiced the application of linguistics to the study of literary texts since the 1960s; his early work in this field includes a paper (published in the 1966 collection Essays on Style and Language, which he edited) showing, with a rich variety of examples, that verse of a given meter can have very different rhythmical movements, depending on the relationships of the grammatical units with the lineation. In his 1981 volume Literature as Social Discourse: The Practice of Linguistic Criticism, Fowler argues that linguistic description of literary texts should concern itself with the sociocultural context. He exemplifies the sort of description, essentially sociolinguistic, that he advocates in a treatment (in the same volume) of Wordsworth’s poem “Yew-Tree” (1798).

Fowler’s treatment is in answer to a reading of this poem by critic Riffaterre. Riffaterre concentrates on the lexical aspect of the poem’s language, showing that it consists basically of variations on “yew-tree” through translation of certain of its semantic components from “tree-code” into other codes—for example, “snake-code” in the lines, “Huge trunks! and each particular trunk a growth/ Of intertwisted fibres serpentine/ Upcoiling, and inveterately convolved.” Fowler contends that by neglecting the matters of register (and what it implies of the activity of the speaker of the poem vis-à-vis an addressee) and of the reader’s sequential experience of the text, Riffaterre has failed to give sufficient weight to the shift from a geographical guide register to Miltonic loftiness in these lines. This and other shifts of register in the poem are, he contends, significant, central to its meaning.

New directions

Besides speech act theory and sociolinguistics, later work on developing a linguistic theory of discourse saw applications to literature. Teun A. van Dijk, for example, has done work on grammar by taking the text, rather than the sentence, as the structure to be described and including a “pragmatic” component that would specify appropriateness conditions for discourses. Other works, like those by Isabella Whitney, have been reevaluated by linguistic critics in the twenty-first century, giving the works new perspectives.

Expanding Roger Fowler's application of linguistics to texts, critical discourse analysis (CDA) is increasingly popular among scholars examining modern and historical literature. CDA combines social and linguistic approaches to texts to understand their meaning better and gain new, critical perspectives. Along with Dijk, linguists Norman Fairclough and Paul Chilton are known for using this method in their critical approach to poetry. Additionally, modern research provides insight into language's impact on literary criticism and the role of grammatical and phonological structures in the human perception of literary works.

Bibliography

Carter, Ronald, ed. Language and Literature: An Introductory Reader in Stylistics. 1982. New ed. New York: Routledge, 1991.

Clarke, D. "New Directions in Criticism on Isabella Whitney." Women’s Writing, vol. 31, no. 1, 2024, pp. 1–10. doi.org/10.1080/09699082.2024.2284046.

Dijk, Teun A. van. Discourse Studies. 5 vols. Sage, 2008.

Ellis, John M. The Theory of Literary Criticism: A Logical Analysis. University of California Press, 2022.

Fowler, Roger. "Linguistic Criticism." The Language and Literature Reader. Routledge, 2020, pp. 49-59.

Jeffries, Lesley. The Language of Contemporary Poetry. Palgrave, 2022.

Searle, John R. Expression and Meaning: Studies in the Theory of Speech Acts. 1979. Reprint. Cambridge University Press, 2008.

Searle, John R. Speech Acts: An Essay in the Philosophy of Language. Cambridge University Press, 2011.

Toolan, Michael. Narrative: A Critical Linguistic Introduction. 2nd ed. Taylor and Francis, 2013.

Weber, Jean Jacques, ed. The Stylistics Reader: From Roman Jakobson to the Present. 1996. Reprint. New York: St. Martin’s Press, 1998.