Internal and External Deviation in Poetry

WORD ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: Internal and External Deviation in Poetry Sa...
1 downloads 0 Views 869KB Size

ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage:

Internal and External Deviation in Poetry Samuel R. Levin To cite this article: Samuel R. Levin (1965) Internal and External Deviation in Poetry, WORD, 21:2, 225-237, DOI: 10.1080/00437956.1965.11435425 To link to this article:

Published online: 16 Jun 2015.

Submit your article to this journal

Article views: 3539

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at Download by: []

Date: 27 January 2017, At: 07:38

SAMUEL R. L E V I N - - - - - - - - - - -

Internal and External Deviation in Poetry The recognition that deviation is a distinguishing mark of poetry is as least as old as Aristotle. I Much later, the Prague School estheticians called attention to deviant expressions, referring to them as "foregrounded" or "deautomatized."2 In more recent discussions they are said to be less predictable, or more entropic.3 And we, in everyday parlance, are usually signalizing the same phenomenon when we say of certain expressions that they are striking, heightened, different, or arresting. But however we choose to characterize these expressions, we know that at certain points in a textand especially in a poetic text-the language is used in a way that is not typical, a way which, in particular, constrains us to pause over the expression and reflect upon its form. Two aspects of the phenomenon need to be distinguished: the readerresponse to certain stretches of language occurring in a poem, and the linguistic structure which analysis reveals and which it is claimed stands in some plausible relation to the reader-response. Reader-response is expressed by predicates like "striking," "different," or "original," all of which may be subsumed under the single predicate "novel." Then "deviation" is the term used to describe the way in which the language is deployed.4 The reader-response is held to have intuitional validity, as well as the support of literary criticism; the fact of deviation is given by linguistic analysis. The claim is then made that deviation is a sufficient condition for the response of novelty. It can be shown that most, if not indeed all, of poetry's characteristic devices exemplify deviation in one way or another. Thus, to begin with, the 1 Poetics,

XXII. See Bohuslav Havninek, "The Functional Differentiation of the Standard Language," pp. 9ff. and Jan Mukai'ovsky, "Standard Language and Poetic Language," pp. 21ff. in A Prague School Reader, tr. and ed. by Paul L. Garvin (Wash., D.C., 1958). 3 See Ivan F6nagy, "Communication in Poetry," Word XVII (1961), 201 and passim. 4 This is its fundamental application. In what follows, it is applied somewhat more broadly. 2




use of conventions in poetry obviously lends itself to rationalization in terms of deviation, since the typography of a poem, its rhyme and rhythm, devices like assonance and alliteration, all deviate from the use or appearĀ· ance of these same features in the language as used for ordinary purposes. (This is, of course, not to say that rhyme or alliteration, for example, may not appear in ordinary language discourses; it is simply that the systematic organization of these features in poetry represents a deviation, inasmuch as they occur in the ordinary language in a random, non-systematic way.) Turning to features more strictly linguistic, the phonology,s for example, the syntax, and the lexicon of poetry, it is evident that these features likewise deviate from the way in which they find their expression in the ordinary language. Metaphor, also, regarded by some critics as the central device of poetry, represents a deviation. The proof is that it is only when we see what the metaphor is a substitution for, or, as I should prefer to say, a deviation from, that we understand the metaphor. The same may be said for such staples of poetic language as symbol, image, and myth. It appears, as a matter of fact, that it is only in the themes of poetry that deviation need not occur, since love, hate, passion, and war form the subject-matter of ordinary language as readily as they do that of poetry. All of this leads to the conclusion that deviation in poetry is an attribute ofform. And because deviation is an attribute of form, the interest in responding to it, attaches only secondarily to what the deviant expression is saying, but immediately the expression itself. It is this characteristic of a deviant expression, that of calling attention to itself as object, which gives it its importance in stylistic analysis. And it is this same characteristic that prompts the consistent and sustained use of such expression in poetry. Cases of deviation may be classified into two types. First, there is that type of deviation which takes place against the background of the poem, where the norm is the remainder of the poem in which the deviation occurs. Second, then, is that type where the deviation is to be explicated against some norm which lies outside the limits of the poem in which the deviation occurs. For convenience of reference, we may call these two types internal and external deviation, respectively. 6

s Obviously, phonology is involved in features like rhyme and alliteration. But phonology may also be employed to produce non-"conventional" structures. For an example, see Dell H. Hymes, "Phonological Aspects of Style: Some English Sonnets," in Style in Language, ed. by Thomas A. Sebeok (New York, 1960), pp. 109-131. 6 More precisely, deviation with respect to norms that are internal and external to the poem. The two types of deviation, internal and external, thus differ from the two types of contrast described by Michael Riffaterre, "Stylistic Context," Word XVI (1960), 207218, in that contrasts effected in both the micro- and macrocontexts of Riffaterre's



I shall begin the more specific discussion of deviation with some remarks on typography, which is perhaps not too significant an aspect of poetic language, but which is one that offers some fairly obvious illustrations of the principle at work. Thus, the beginnings and ends of lines of poetry occur not at the margins of the page, but wherever the poet means them to occur, with the result that a poem may have irregular margins. In the same way, breaks in the thematic progress of a poem are indicated not by indentation, but by spacing, and one speaks of stanzas, not paragraphs. Since the norms that are being violated in these two examples are those of the ordinary language typographic practice, these deviations are external. For these two typographic practices in poetry, internal deviation would not be especially expressive, and it does not, as a rule, seem to be implemented. Internal deviation, to be realized here, would require a span of regular margins after a poem had proceeded for a while with irregular ones (or vice versa), or, in the second case, the absence of a stanza-break after a number of stanzas had been set off by spaces (or vice versa). A more interesting typographic convention of poetry, one in which both external and internal deviation may be produced, governs the use of lineinitial letters. The general practice is to begin each line of a poem with a capital letter, producing in this way, of course, an external deviation. In the work of such poets as e. e. cummings and Miss Marianne Moore, however, this particular convention is not observed. Their respective practices in this regard differ interestingly in detail. Cummings' convention is to use lower-case initials, irrespective of whether the beginning of the line begins a new sentence. Miss Moore's practice is to use capitals wherever a new sentence begins-whether at line-internal or line-initial position. Miss Moore's practice, while thus constituting a deviation in respect to the standard poetic convention, actually reverts, as far as capitalization itself is concerned, to the standards of the ordinary language usage. Cummings' practice, on the other hand, represents a deviation not only from the poetic convention, but also from the external norm of the ordinary language usage. Where the real interest in this question lies, however, is in the internal exploitation of these two unconventional standards. Cummings, in breaking with the tradition, goes on to exploit his convention by beginning an occasional (though not a random) line with a capital, thus producing an

analysis would be, equally, instances of internal deviation in the scheme presented here. Also different are Waiter Alfred Koch's "inner" and "outer" styles, "On the Principles of Stylistics," Lingua XII (1963), 418. Style is "inner" according to Koch if the criteria used in describing the particular item are morphological, "outer" if the criteria are syntactic or distributional.



internal deviation. Although there is no reason why Miss Moore's particular convention cannot also be deviated from, such deviation, aside from a few special cases, does not seem to occur. 7 The different tactics of Miss Moore and e. e. cummings in respect to line-initial typography are interesting in the light of information theory, in that Miss Moore, having broken from the tradition, which was meaningless in terms of entropy, has substituted a new convention which is also essentially meaningless. Cummings, on the other hand, in breaking with the tradition, has gone on to exploit the information-theoretic possibilities. In taking up the remaining features of poetic language it is convenient to emphasize the distinction between conventional and purely linguistic features. By "conventional" is intended those features which a poem incorporates as representing a particular literary form, those features, that is, which an author obligates himself to observe-or at least to consider- by the mere election to write in a certain form. The typography of a poem, which has already been discussed, is thus a conventional feature. The two chief conventions of poetry, however, are rhyme and meter. Inasmuch as rhyme and meter occur only accidentally in ordinary language texts, their systematic use in poetry represents an external deviation. But such patterns may also be exploited for purposes of internal deviation. Thus, in a poem where a certain rhyme-scheme has been, or is being, established, the use of a non-rhyme would constitute an internal deviation. Such an occurrence, by frustrating our expectations, would call attention to itself and thus produce a stylistic effect. In the same way a shift in meter at a point up until which the poem had been metrically regular would produce the same result. The fact that one speaks of false rhyme, off-rhyme, inversions, feminine endings, etc. attests to the existence of such deviations. As another example, the notion of metrical counterpoint is predicated on this same dis7 Miss Moore's departure from the typographic conventions of poetry does yield a gain, however. As long as the type of letter in line-initial position is not determined by any conventional conditions, it is possible to manage the line so that words which in the ordinary language orthography are capitalized may be placed initially and thus made prominent. A telling example occurs in her poem "The Steeple-Jack," Collected Poems (New York: The Macmillan Co., 1951). The relevant lines, beginning with stanza 5, are:

A steeple-jack in red, has let a rope down as a spider spins a thread; he might be part of a novel, but on the sidewalk a sign says C. J. Poole, Steeple-Jack, in black and white; and one in red and white says Danger. The church portico has four fluted columns, ....



ruption of the norm, one whose presence must continue to be entertained, however, in order that the counterpoint be effected. I shall not take up specifically the other conventional features of poetry -alliteration, enjambement, caesura, and so forth-but it is quite obvious that they also lend themselves to treatment in terms of both external and internal deviation. Of the purely linguistic features of a poem's composition, the phonology, the lexicon, and the syntax merit discussion. (It does not appear to me that the morphological structure of poetic language is especially interesting from the point of view of deviation.) On the level of phonology, external deviation may be produced in a number of ways-thus, by the presence in a poem of a phoneme which does not occur in the ordinary language, or by the occurrence of a phonological sequence which does not so occur. Such external phonologic deviations have been observed in the songs of certain American Indian languages, s but they are almost never encountered in English poetry. These two types of external phonologic deviation may be called determinate-in the sense that such deviations can be univocally established by comparison with the phonological description of the ordinary language, which we may assume to be either given or obtainable. Another type of external phonologic deviation would be that type which we may call statistical. Here the fact of deviation does not turn on a determinate, yes-or-no decision, i.e., does the phonologic feature occurring in the poem occur in the ordinary language or does it not; it turns on a comparison of the frequencies with which a phoneme or sequence occurs in a poem and in the ordinary language.9 This type of external deviation is fairly marked in a good deal of poetry, in which we find atypical densities of phonological features. Concerning internal phonologic deviation there is not much to be said, except to make the obvious statement that, theoretically, it may take place. In the area of lexicon it is clear that what one refers to as "poetic" diction constitutes an external deviation. As in the case of phonology, external deviation in the area of lexicon may be either determinate or statistical. Purely poetic forms like e'en, o'er, or fain may be determined as deviant by simple reference to a dictionary, which would label them "poetic." The same procedure would serve for poetic nonce words. On the other hand, the incidence of words of a certain type may be externally deviant in the 8 Edward Sapir, "Abnormal Types of Speech in Nootka," in Selected Writings of Edward Sapir in Language, Culture and Personality, ed. by David G. Mandelbaum (Berkeley, Calif., 1949), p. 188. 9 See my article, "Deviation-Statistical and Determinate-in Poetic Language," Lingua XII (1963), 276-290.



statistical sense, i.e., as occurring with a higher relative frequency than those words do in ordinary language texts. Thus, at certain periods archaisms will occur in poetry, both individually and as a class, with a relatively high frequency; in other periods the same will hold true for foreign terms or nonce words. To some extent, then, external deviation in the area oflexicon can be determined by referring to the usage labels in a dictionary. To establish statistical deviation other procedures are of course required. Internal lexical deviation may be occasioned in various ways, depending on the various lexical norms that are established in a poem. To give one example, in a poem where norms of "poetic" diction are being established, the deviation may be effected by the use of a colloquial or dialectal expression, a cliche, a vulgar or obscene term, and so on. In all such cases the deviation clashes against the set of expectations that has been conditioned by the prior establishment in the poem of norms of diction and, in so doing, it arrests and focuses our attention on the deviant form. In speaking of internal deviation, it is necessary to remark on the relative nature of norms and deviant expressions. A priori any given form occurring in a poem may either constitute a deviation or be part of a pattern making up the norm. Thus, if archaisms or foreign terms were used sparingly in a poem, they would be internally deviant. If, on the other hand, they saturated a poem, then they might very well serve to establish a norm against which it would be the standard terms that would be deviant. Before going on to discuss syntax in this paper which is concerned almost exclusively with formal attributes of poetry, I should like t9 make a short comment on content. For my purposes I have selected a poem by Robert Lowell, "Sailing Home from Rapallo." In the first half of this poem Lowell describes how warm and verdant the Mediterranean coastline was as the ship bearing the dead body of his mother embarked from Italy for the trip home and burial. In the second half of the poem, the cold darkness of the family plot in New England is described. The transition can be seen in these the turn of the poem: While the passengers were tanning on the Mediterranean in deck-chairs, our family cemetery in Dunbarton lay under the White Mountains in the sub-zero Like everything else in a poem the contrast here is expressed by formal means, i.e., by language. But the deviation here is not a function of the to Life Studies (New York: Vintage Books, 1956).



language, except trivially; it is a function of the content. The statement made at the beginning of this paper to the effect that deviation is an exclusive function of form thus requires emendation: deviation of content can occur. But it is hard to see how deviation of content can occur except internally. Only on the view which is not seriously maintained today, namely, that only certain subjects are the proper concern of poets, could there be deviation from external norms. But if we are not surprised at any subject that a poet undertakes to write about, then no external norms can be violated. In syntax, as in the other linguistic levels, internal deviation is comparatively easy to explain. Given the response that a feature is deviant, if it is . internally deviant, then, by definition, the norms which condition the response will be found in the poem. Thus, the appearance of an imperative or interrogative sentence in a poem in which a series of declarative statements has been made would constitute an internal deviation. Similarly for a sentence in the subjunctive mood when the poem has been running until that point in tlie indicative. In the same way a sequence of simple sentences would be disrupted by a complex sentence, and conversely. In fact, any of the devices of sentence construction can be used so as to develop a pattern of expectations which the appearance of a counter instance will disrupt. Thus, an instance of inverted word order or of unusual phrase or clause placement in a sentence would be internally deviant in a poem which is otherwise regular in these respects.ll It is when we come to consider external deviation of syntactic constructions that difficulties arise. For in this area it is not immediately clear what might serve as the external norm; that is, there does not seem to be anything in syntax comparable to the dictionary or the phonologic description. It might appear that in the ordinary grammatical description of a language we ought to have something which could be made to serve as an external norm for the explication of syntactic deviation, but unfortunately this is not the case. As a matter of fact, it is only recently, with the development of generative grammar, that this prospect seems possible of implementation. Traditional and even structural, grammars cannot be made to serve for this purpose, inasmuch as they are essentially incomplete. Nor will normative grammars do since, in addition to being incomplete, their norms are unrealistic, at times even whimsical. But before going into a discussion of generative grammar and how it may be used as an external norm, some mention should be made of still another frame of reference which it has u It may frequently happen that a given feature is both internally and externally deviant. This would be true of a word order, for example, that is not an output of the grammar and is at the same time internally deviant.



been frequently maintained can serve as an external norm against which to measure deviation. I refer here to the statistical model. According to this view of norms, all of us, as fluent native speakers of English, have in our memories a body of statistics drawn from our experience with the ordinary language. Among other things, these statistics constitute information concerning the relative frequencies with which individual words occur in the ordinary language, and also information concerning the probabilities with which any word will occur given whatever sequence of words has preceded the text to that point. Actually, it is only the study of transitional probabilities that is relevant for the question of syntactic deviation, frequency counts being relevant to the question of diction. But frequency counts and transitional probabilities are usually considered together in the statistical model, so it seems advisable to treat both here. The statistics thus stored are supposed then to prepare us for the incidence of words in a given text, and they are supposed also to prepare us, on a scale of graduated probability, for the appearance of various words at any given point in a text. Our statistical experience with the language presumably predisposes us, in any given text, to encounter more occurrences of the word the than the word have, and to expect have more frequently than, say, machination. In the same way, given that we have just read or heard the sequence It was a beautiful-, we expect day to follow more readily than, say, elephant, and the latter to occur more readily than must.12 Now since these norms are presumably conditioned by our experience with the ordinary language, we should be able, when we are confronted by a deviant expression in poetry, to rationalize it against this background. There is a certain plausibility in this line of argument. There is little question but what we are conditioned in some such way to react to what we read or hear. When we attempt to make these statistical norms explicit, however, certain difficulties arise.13 These difficulties are not so severe with frequency counts, where it is, as a matter of fact, quite feasible to draw up a table offrequency rank-orders from some representative body of texts (although, naturally, "representativeness" poses problems), and then to use this table as a norm against which to measure deviation from expected word-frequencies. But it is a different matter with transitional probabilities. For this problem it is simply impossible to collect any significant statistics. Where the inventory of units is restricted, as with alphabet letters or phonemes, one may 12 Considering these sequences from the standpoint of external norms. Naturally, a context, setting up internal norms, could affect our relative expectations. 13 Cf. Warren Plath, "Mathematical Linguistics," in Trends in European and American Linguistics 1930-1960 (Utrecht, 1961), ed. by Christine Mohrmann, Alf Sommerfelt and Joshua Whatmough, pp. 27-30.



laboriously accumulate statistics that can be used to plot monogram, digram, trigram, and perhaps larger-context transitional probabilities, since the calculations will have to be carried out for only a small number of units. In syntax, however, where the relevant units are words, we have to deal with a universe in the tens of thousands, and one in which the relevant contexts may be quite extended. Thus, even though the range of existing combinations is finite, the collection of adequate statistics in this area would be a well-nigh impossible task. At a certain point, moreover, even these considerations become academic. This is because for almost any sequence of significant length it turns out that there are no matching sequences anywhere in the millions of books printed to this time.l4 Thus, if the word taste, occurring as the fourteenth word of Paradise Lost, should strike us as novel, we could collect no statistics to explain that reaction, since the sequence "Of man's first disobedience, and the fruit of that forbidden tree, whose mortal-" in all probability does not recur anywhere in the literature (barring citations, of course).15 The use of a generative grammar for purposes of explicating syntactic deviation is certainly not without problems of its own. But such a gra:qunar at least offers the prospect of providing a means for such explication. If a generative grammar is to be adequate to a language, it must generate all and only the grammatical sentences of that language. If we then make the fair assumption that syntactic deviation and ungrammaticalness are the same thing, then the above requirement points directly to the possibility of using a generative grammar to mark externally deviant expressions. For we can test any putatively deviant expression against such a grammar, in the sense that we can ask whether or not the grammar will generate it. Any expression which the grammar will not generate is then ipso facto deviant. The procedure just outlined is somewhat oversimplified. This is because notions like syntactic deviation and ungrammaticalness are too gross as they stand to cope adequately with various kinds of pertinent data. To take two sequences like abysmally winked tornado distraught the and the distraught tornado winked abysmally, and refer to them indifferently as syntactically deviant is unsatisfactory. And as Chomsky and the literature have made plain, to refer to them both simply as ungrammatical is likewise unsatisfactory. The question of syntactic deviation is thus bound up with the question of degrees of grammaticalness. And as now seems to be becoming 14

Cf. Colin Cherry, On Human Communication (New York, 1961), p. 39.

1s To be more precise, statistical investigation would result in assigning a probability

of zero to the sequence, but the assignment would have no significance, since other sequences, including many which might and many which might not strike us as novel, would be the same (zero) probability assignment.



clear, some degrees of "grammaticalness" may turn out to be functions not of syntax, but of semantics. To arrive at a more discriminating view of this general type of deviation, it might thus be advisable to consider incorporating within the general framework of a generative grammar a semantic component like the one proposed by Katz and Fodor.16 Doing so would make possible an immediate and fundamental division into those sequences which are deviant because they are not generated by the grammar, and which are thus syntactically deviant, and those which are generated, but which are marked as deviant (anomalous) by the semantic component. Further, it is possible to make some judgments on the degree of deviation which those sentences manifest that are not generated by the grammar, since this is just the obverse of the degree of grammaticalness. Whether it will also prove to be feasible to define degrees of anomaly is problematical at this point. In general, a sequence may be deviant for one of three reasons: it may instance a wrong word order, it may instance a wrong word selection, or it may instance a combination of the preceding. Put another way, it may violate a phrase-structure or transformation rule, it may violate a wordcategory rule, or it may violate both. Now as Chomsky has pointed out, given a deviant sequence, we attempt to impose some interpretation on it. 17 This attempt takes the form of analogizing the deviant sequence to some well-formed expression. The analogizing process naturally takes different forms depending on what type of deviation we are interpreting. In the remainder of this paper I will take up several distinct types of deviant expression and comment on the different results which the analogizing process yields. Consider the following three sequences, which come from the work of Pound, Thomas, and Emily Dickinson, respectively: Shines in the mind of heaven God My hand unravel/When you sew the deep door The largest fire ever known ... /Discovered is without surprise These represent word order violations; we may speak of such sequences as syntactically deviant. In analogizing these sequences to well-formed expressions we entertain simultaneously the given form and the form it deviates from. Thus, in interpreting "Shines in the mind of heaven God," we 16

Jerrold J. Katz and Jerry A. Fodor, "The Structure of a Semantic Theory,"

Language XXXIX (1963), 170-210; see also Jerrold J. Katz and Paul M. Postal, An Integrated Theory of Linguistic Descriptions (M. I. T. Press, 1964). 17

Noam Chomsky, "Some Methodological Remarks on Generative Grammar,"

Word XVII (1961), 263ff.



entertain the well-formed "God shines in the mind of heaven." 1s For "my hand unravel" we entertain "unravel my hand," and for "discovered is" we entertain "is discovered." As can be seen, the analogue in all these cases is a simple paraphrase of the deviant form. Poetically, the result is comparatively uninteresting. Consider now the following three sequences, by cummings, Thomas, and Auden, which represent a different type of deviation: Anyone lived in a pretty how town Rage me back to the making house Stare the hot sun out of heaven These sequences instance word-category violations; we may say of such sequences that they are paradigmatically deviant.19 In cummings' linedisregarding anyone lived-the deviation is produced by the adverb how occurring where the rules require an adjective. On a priori grounds it is possible to take any deviant sequence and attempt to rationalize it as a case of either word order or word category violation. In actual cases, however, the structure of the deviant sequence will generally dictate the choice. In the present instance, given the amount and type of structure incorporated in the phrase "lived in a pretty how town," we conclude that the structure of pretty how town is Adj., Adj., Noun, i.e. we interpret the sequence as a case of word category violation. The alternative, to construe the sequence as Adj., Adv., Noun, thus making it violate a word order rule, would lead us to conclude that the sequence was nonsense, and our natural inclination to redeem whatever we read rules out this analysis. The result of the interpretation, then, is to conflate how with the category Adjective. This is a paradigmatic conflation and is, in principle, more viable poetically than syntactic conflation, or that type in which two variant word orders are entertained simultaneously. In the particular example it is hard to specify the poetic effect, since the word how is con:flated with the high-order grammatical category Adjective, and these two elements are too disparate for easy fusion. In Thomas' line the sense of novelty arises from the fact that rage occurs where the rules require a transitive verb. In construing the sentence we entertain the notion of transitivity, where this is joined with the word rage, which we must entertain in any case. Thomas' sentence does not seem as We may omit consideration of the metaphor in this line. Syntactic and pamdigmatic violation are types of what Paul Ziff calls variants and inventions, respectively; "On Understanding 'Understanding Utterances'," in The Structure of Language, ed. by Jerry A. Fodor and Jerrold J. Katz (Englewood Cliffs, N.J., 1964), pp. 396ff. 18




novel as cummings'; it is easier to impose an interpretation on it. This reaction can be explained by pointing out that the category of Transitive Verb is lower-order than is the category Adjective, which had to be invoked to interpret cummings' line. Put somewhat differently, rage is a member of a subclass of a class (Verb) of which the category Transitive is also a subclass. There is another factor to be considered. The verb in Thomas' line is actually rage back, not simply rage. The presence of back in the construction limits the class of transitive verbs which can be introduced in the analogizing process to a small subclass of transitives including take, lead, give, and others. In general, as classes of words are analyzed into smaller and smaller subclasses the factor common to them graduates from being syntactic to being semantic. The ability to cooccur with the particle back is common, although not limited, to verbs of motion. The analogizing process in this case thus serves to conflate rage with the factor of motion. The poetic effect induced here approaches metaphor. Auden's line is somewhat similar structurally to Thomas', with the difference that the transitivized verb stare functions with a complement instead of a particle. The complement is the adverbial of place out of heaven. The cooccurrence ofthe locative adverbial limits the analogues for stare to the same set of transitive verbs of motion adduced in the analysis of Thomas' rage, with more or less the same poetic effect. Finally, I will deal with a set of deviations in which the analogizing process entrains the kind of conflation which produces a typical metaphoric effect. Consider the following sequences, from Hart Crane, Miss Marianne Moore, and Eliot: What words can strangle this deaf moonlight [But] why dissect destiny with instruments My self-possession gutters In the line from Crane, words become animate, because strangle regularly cooccurs with animate nouns; the same effect is induced on moonlight by its cooccurring with deaf. In Miss Moore's line the word destiny is con:flated with the notion of animation and the notion of a subtype of abstract noun, since we dissect not only (perhaps dead) animate beings, but also ideas, problems, etc. Destiny is neither an animate noun nor a member of this particular subclass of abstract nouns; it is close enough in its valence to both these classes, however, so that the conflation occurs, with consequent enrichment of the expression. I would suggest that the notion "mankind" is elicited by the analogizing process. In considering Eliot's line, we observe that the intransitive verb gutter is limited in its cooccurrence range to just about the words (burning) candle or flame, so that self-possession



and flame are fused in a metaphor.2o As I see it, the whole question of metaphor is bound up with conflations of the sort discussed here. Metaphors, then, can be described typologically, depending on the degree of remoteness in which the invoked analogue stands to the given form. It may then turn out that a scale of metaphoric effectiveness can be correlated with this typology. Hunter College New York, New York 10021


These would also be inventions in Ziff's terminology.

Suggest Documents