Back to Introductory Page
Back to home page
Back to "Occasional Papers"

Reuven Tsur
Postscript: Cognitive vs. Historic Poetics

I have been lucky to receive detailed comments from a penetrating (anonymous) reviewer on my paper "Rhyme and Cognitive Poetics". As the reviewer professes, s/he comes from a different scholarly tradition, of the so-called Russian school in verse study. "This approach is based not so much on insight and intuition as on wide quantitative analyses of observable facts". Furthermore, his/her approach is historical. "Poetry of earlier epochs was, naturally, written to be read; and not by an abstract reader 'in general,' or by a modern reader, but by the poet's contemporaries. The poet and his audience shared the same poetic culture". "My first disagreement with Professor Tsur lies in his reliance on a reader 'in general,' or on a modern (and sometimes unsophisticated) reader rather than on the reader of a particular period. For example, 'grammatical' rhymes may seem uninteresting and flat, while 'anti-grammatical' rhymes may seem 'vigorous' and clever to the modern reader; but the modern reader is familiar with Butler, Pope, Byron, and Tennyson, while Chaucer's contemporaries (and Chaucer's rhymes are given as an example of grammatical, and therefore not vigorous rhymes) were amazed that rhyming could be used in poetry at all!"

In what follows, I propose to take issues with this criticism. On closer inspection, the "poet's contemporaries" may turn out to be a poetic straw man. First, there is the problem of what E. D. Hirsch (Wimsatt, 1968) has called "the fallacy of homogeneous past". There is no reason on earth to suppose that the poetic tastes and responses to poetry of the "poet's contemporaries" were any more homogeneous than those of our own contemporaries. Second, "wide quantitative analyses of observable facts" can reveal only what the poet wrote, but not how his contemporaries responded to it. Of course, my comments here are meant to legitimise neither paralysing scepticism, nor chaotic idiosyncrasy, nor historical ignorance. It only means that we may not take for granted the responses of the poet's contemporaries, but must make use of every bit of available information to make inferences about the tastes and responses of the poet's contemporaries. However, the only flesh-and-blood respondents to poetry whose responses are accessible for inspection are those of our own contemporaries. These are therefore the responses with which we must begin any attempt to find out how human beings respond to poetry. Then we are obliged to use every available bit of historical information to ascertain, in what respects may be the responses of the poet's contemporaries supposed to differ from those of our own contemporaries.

As for the kinds of rhyme in Mediaeval poetry, I have done some quantitative study on eleventh-century Hebrew poetry in Spain (Tsur, 1968: 137-139) which is, admittedly, a different poetic tradition. One of the prevalent verse forms at this period is the equi-rhyme, in which long series of verse lines (sometimes up to hundreds of lines) rhyme on the same ending. Such poems are very frequently based on suffix rhymes; and indeed, the explicit poetics of the day acknowledges suffix rhymes as legitimate citizens of Mediaeval Hebrew poetics. Notwithstanding, there are quite a few longer poems which insist on rhyming the roots of words.

It is easier to rhyme suffix rhymes than root rhymes. If there are, nonetheless, some longer poems that make a sustained effort to rhyme roots rather than suffixes, it may be an indication of an intuitive preference for such rhymes. If such a preference does exist, one might expect to find a tendency for root rhymes in shorter poems. Such a tendency does indeed exist. I have checked all the five-line-long or shorter equi-rhymed poems in the first two volumes of Schirmann's authoritative anthology (1961), and found the following distribution:

root r.
suffix r.

This table shows a deliberate preference for pure root rhymes to pure suffix rhymes. I expected to find a similar preference for poems in which there is a majority of root rhymes, as compared to poems in which there is a majority of suffix rhymes. The opposite, however, turned out to be the case. This unexpected result can be explained by assuming that some poets insisted in certain of their poems on pure root rhymes, even if they had to overcome difficulties; whereas in poems in which they forsake such an accomplishment, they still inserted, whenever they could, at least one root rhyme, for the sake of diversity. What is perhaps more important from the point of view of homogeneous past, is that there were considerable individual differences among the poets in this respect. I have checked the individual distribution of rhyming in poems by those poets who have at least six poems in the corpus.

root r.
suffix r.
Shmuel Hanagid3420110
Ibn Gabirol221318
Ibn Altaban1.0600
Moshe Ibn Ezra19805
Yehuda Halevy152005
Avraham Ibn Ezra2401

But the most impressive feasts of anti-grammatical rhyme cannot be shown through statistics, only through an analysis of individual cases. And these sometimes achieved a sophistication comparable to the boldest instances of modernism. Moshe Ibn Ezra, who has over twice as many root rhymes as suffix rhymes in our corpus, also resorts to extremely complex semantic structures in the rhymes of some of his shorter poems, as in the following one:


Go out to the yard of death, look at those who sleep there, and be ashamed and abashed
See there bodies covered with stones, whereas the dust is their bedding and [protective] shadow
This became their repose and lot, while the world and their houses [stand] forever

The sounds [lam] constitute the rhyme at the end of the three verses. In nikhlam, the [l] and [m] belong to the root, while the vowel belongs to the conjugation of the verb; silam is composed of two morphemes: the first three phonemes are a transform of sel (shadow), while the last two phonemes [am] constitute an inflection meaning "their". leolam is an adverb meaning "forever". Such semantic and grammatical sophistication is rather rare in eleventh-century Hebrew rhymes, but is by no means accidental.

What can we learn from these results? First, that eleventh-century Hebrew poets made sometimes considerable efforts to rhyme the roots of words and avoid suffix rhymes, in spite of the explicit licence to use them in contemporary poetics. Second, that there were considerable individual differences among the various poets, and that the past is not homogeneous in this respect. And third but not least, that they may have had strong intuitions in respect of the differences between these kinds of rhymes, even though their preferences may have been at variance from each other.

Thus, far from relying on the responses of "an abstract reader 'in general'", Cognitive Poetics discerns fine individual differences between poets of a certain historical period, and, as we shall see, also offers tools to account for individual differences between readers. It avoids a vision of homogeneous past, without allowing for chaotic idiosyncrasies. One of its basic assumptions is that the effects of poetic texts are not perceived in a way in which the effects of, say, touching an electric wire are perceived. The latter are perceived on mere exposure to the wires, whereas the former presuppose a certain kind of cooperation on the perceiver's part. Different perceived effects of the same piece of literature presuppose different kinds of cooperation. Perceptual qualities of texts arise only when they are performed in certain ways, that is, when the reader discriminates certain elements and realises certain relationships between them. When readers discriminate other elements, or realise different relationships even between the same elements, one may expect that different perceived qualities will arise (Tsur, 1992 b: 29). Anti-grammatical rhymes, as described in the foregoing paper, require a different kind of cooperation on the reader's part than grammatical rhymes do. The resulting perceived quality is different too. In the case of grammatical and anti-grammatical rhyme, for instance, we should take into account two kinds of variables: information-processing activities and familiarity. It is the activity of the information-processing mechanism that determines the relatively vigorous or tame character of rhyme. Unexpectedness may heighten the effect of this activity, familiarity may blunt it. Returning now to my reviewer's comment on Chaucer's original audience, it may be construed, in terms of the present paper, as follows: the rich precategorical sensory information reverberating for a very short period of time in short-term memory behind the abstract categories of speech sounds is exploited and enhanced by rhyme; the memory traces of two words considerably apart in time may be fused and perceived as if simultaneously present. In certain conditions, such fusion of auditory information may be perceived as if spread over the intervening section of the poetic passage. Chaucer's contemporaries may have had increased sensitivity to this kind of intense and diffuse sound texture, since it was so innovative to them. This, however, should not have prevented them from perceiving additional tension in rhymes that arouse heightened semantic information-processing activities.

It would appear, however, that the details of my discussion of rhyme allow us to say much more precise things about what is going on in the Mediaeval reader's head when confronted with Chaucer's rhymes. What of the mechanisms described in the foregoing paper can we safely attribute to Mediaeval readers? We may assume that their recourse to phonetic and semantic coding is not greatly different from that of present-day readers. There would be some readers who would easily rely on both semantic and phonetic coding, and some who would feel less at ease with the latter. Perhaps both groups would handle the verse-endings in the ensuing lines in the way we have described, following Wimsatt, the response to homoeoteleuton:

Under his belt he bar full thriftily,
Wel coude he dresse his takel yemanly ...

That is, each has the same meaning, or is the same morpheme, and each supports the logic of the sentence by appearing in a certain place in the structure; they seem to come fairly to the aid of logic; readers or listeners do direct increased attention to the similar sounds as well, but use this similarity merely to reinforce the similar meanings. But, perhaps, readers who make more efficient use of phonetic coding would "take the cue" from the occurrence of the like endings at the end of verse-lines, and use their newly-acquired sensibilities to release the sound information from its attachment to meaning, and direct some attention, however faintly, to the relatively unconstrained reverberation of the rich precategorical auditory information.

The state of twentieth-century literacy and schooling may have, though, affected the relevant cognitive faculties. Longitudinal studies of pre-school children well into the lower grades of elementary school suggest that the use of phonetic coding by some poor readers may improve with the years; researchers disagree, whether it is a matter of slower cognitive maturation, or the result of practice gained in the cognitive skills acquired at school. In the former case, we may assume that differences of schooling would have little affect on the deployment of these cognitive faculties in Chaucer's age and in ours. In the latter case, the differences between present-day and Mediaeval educational methods may have affected the differences between the readers' responses in the two ages in the direction opposite to the one suggested by my reviewer. There will be a smaller proportion of Mediaeval than modern readers who would have acquired in course of their education to literacy the skill or willingness to attend away from semantic to phonetic coding in such like endings as in this quote, and perceive their sound texture as especially vivid.

The historical approach quoted above assumes that when readers are exposed to certain kinds of poetic stimuli, certain kinds of poetic responses may be expected, without asking questions about what is going on in that black box, the reader's head. Cognitive Poetics, by contrast, is much interested precisely in those "mediating structures and processes" that may systematically relate poetic structures to (sometimes conflicting) perceived effects reported by competent readers. By the same token, it provides systematic tools for distinguishing and describing perceived effects, and avoids homogeneity, without allowing for chaos.

Further, the reviewer comments: "I would say, in opposition to Professor Tsur, that the couplet rhyming aabb is not inherently 'witty' or 'clever', but is associated, for the modern reader, with the Classicists' poetic tradition, with the rationalism of the eighteenth-century poetry, while the abab rhyming scheme is associated with the Renaissance and Romantic themes and style". Cognitive Poetics maintains, by contrast, that a rationalist quality is not conferred by readers upon couplets through repeated exposure to couplets in eighteenth-century Classicist poetry. Rather, there is a huge bulk of independently established empirical evidence obtained by gestalt psychologists and the Rorschach ink-blot test, suggesting that predictable good shapes and good form realisation typically go with rational or intellectual qualities; and, according to well-established experimental criteria, the couplet rhyming aabb has a better, more predictable shape than the abab rhyming scheme (Herrnstein Smith, 1968; Arnheim, 1957; Meyer, 1956; Rorschach, 1951). Moreover, an acceptance of my reviewer's position has rather serious consequences: that rather than contributing to the perceived affect of a poetic style, versification receives its perceived affect from the poetic style with which it is regularly associated. This leaves us with an additional troubling question: what are the immediate exponents of perceived affects? Cognitive Poetics assumes, with gestalt psychology, that there is a mutual dependence between part and whole. The parts contribute to the character of the whole, but the whole is a system that determines the character of its parts; perceived affects are characteristics of wholes. As far as I know, the historical approach has no conception concerning this issue.

The same holds true of the reviewer's misgiving: "Why is rigorous iambic verse associated with 'witty, intellectual' style, and a looser form of iambic verse (with numerous accentual 'deviations' from meter) with a more emotional, less 'witty' style? The answer, in my view (based on my experience as a metrist) lies again in the tastes and themes of the epoch that produced such variants of iambic meter in the English tradition. The modern reader associates the first type of rhythm with the Classicism (that valued order), and the second with Romanticism (that valued deviation from order)". Cognitive poetics maintains that the attribution of intellectual or emotional qualities to certain metrical styles based on regular association with "the tastes and themes of the epoch that produced" them is not unlike the way in which Pavlov elicited responses from his dogs. If, on the other hand, this attribution is based on regular association with certain poetic styles as a totality, then we are faced, as I have suggested above, with a mere phantom: these qualities are perceptual qualities of the wholes of which metre is an important part; in other words, the less or more regular metre is one out of several, but very important, perceptual conditions of the respective emotional or intellectual qualities. Predictable metre has a simplifying, or intellectual affect (in certain, highly marked conditions, a hypnotic affect); less predictable, deviant metre has an emotional quality. There appears to be an isomorphism between certain kinds of verse structures and certain kinds of mental processes: convergent metrical structures and mental processes are perceived as typically logical and intellectual, divergent metrical structures and mental processes as typically creative and emotional.

At this stage of my argument, I wish to make a comment concerning the psychological reality of poetic rhythm. Poetic rhythm is not merely a series of observable facts on the paper, amenable to large-scale quantitative investigation; it is also a mental phenomenon, active in the poets' and readers' minds. Cognitive Poetics, with its tools derived from the various branches of cognitive science, has quite a lot to say about these processes. One precondition for poetic rhythm to have psychological reality requires the reader to perceive not only a sequence of more or less regularly alternating stressed and unstressed syllables; but also a higher unit (e.g., line); obtaining a larger unit divided by a sequence of smaller units. This idea is not invented by Cognitive Poetics, and has been familiar for long to such structuralists as Chatman. Thus, when I speak of deviant metre as being perceived as typically emotional, I always imply a constraint: assuming that it is not so deviant as to prevent the reader from perceiving the regularly alternating weak and strong positions and the units above at one and the same time. Now what is the exact limit beyond which deviations prevent the reader from perceiving the unit above is not, and perhaps cannot, be theoretically defined. However, the quantitative range of deviations between the rule-obeying Spenser, Thomson, and Pope, on the one hand, and the rule-violating Milton and Shelley on the other, is rather narrow. I performed a quantitative and qualitative analysis of deviations in these five poets. The overall results were as follows: Spenser: 124 deviations per 100 lines, Thomson: 147, Pope: 156, Milton: 180, and Shelley 191 (Tsur, 1977: 59). This means that the quantitative difference between the most deviant and the least deviant poet in this corpus is no more than 67 deviations per 1000 positions. What is perhaps even more surprising is that the difference between Pope and Milton is no more than 24 deviations per 1000 positions. Nonetheless, readers tend to have very strong intuitions concerning substantial differences between these poets' metrical styles. The reason seems to be that the small quantitative differences are enhanced by considerable qualitative differences. I have established a variety of qualitative scales of markedness (Tsur, 1977: 26-48, and passim).

Concerning these qualitative scales of markedness one should make five observations. First, they have been established on statistical grounds (relative frequency of items), and on cognitive grounds (relative difficulty of processing); the two criteria are significantly related. Second, there is "solidarity" (Roman Jakobson's term) between the items on a scale: poets will not resort to some more marked option, unless they have recourse to less marked ones on the same scale. Third, the poets who have a greater number of deviations are the ones who typically resort to the more marked deviations. Fourth, poets seem to draw the upper limit of their own metricality on each one of these scales; and there appears to be a tendency for consistency between these upper limits: poets tend to aim, intuitively, to similar degrees of markedness on the various scales; but the scales are independent variables. Fifth, most poets have a negligible number of verse lines which violate the utmost limit of their own metricality. Hence, there are marked kinds of deviations which would be counted as metrical in Milton or in Shelley, but not in Pope or Thomson. (An example of how these markedness scales work will be considered below, with reference to the placement of caesura).

Against this background I would like to consider the reviewer's following comment: "But what about the looser verse of the English Baroque of the seventeenth century? What about Donne's 'witty' and unquestionably 'intellectual' Elegies and Satyres written in the loosest iamb in the whole of English poetic tradition?" Indeed, according to my own findings, in the first one hundred lines of "Satyre II", there are 292 deviances, about one hundred more than in Shelley. Donne also has here recourse to an unprecedented number of the most highly marked options on the various qualitative metrical scales (Tsur, 1977: 225). I suggest that this overwhelming deviance may have passed that theoretically undefined limit beyond which deviations prevent the reader from perceiving the regularly alternating weak and strong positions and the units above at one and the same time, resulting in a degree of deviance in which metre cannot be perceived as typically emotional. Without going into details, anybody would agree that the metrical style of Pope's "An Essay on Man" is "witty" and "intellectual" in a different way from that of Donne's Satyre.

The foregoing excursus may perhaps answer the following criticism too: "And what about Surrey's and Sidney's [and Spenser's, we might add] emotional sonnets, written in a very rigorous iambic pentameter?" The point of my argument is, precisely, that one should not try to match "emotional metres" with "emotional contents". Metrical style is a variable independent from, e.g., rhetorical style. In the sonnets in question, metre has a simplifying effect upon the emotional tone generated by their figurative and rhetorical structures; some readers, indeed, perceive a kind of naive charm in some of these poems (I would agree with the reviewer, who would no doubt argue that this naive charm may exist only for a reader who knows Shakespeare's, and Donne's, and Milton's, and Keats's sonnets).

Further, the reviewer comments: "The hemistich form 4 + 6 of the English iambic pentameter has been derived from the pattern of the French decasyllable; it is characteristic of the periods of a more rigorous iambic canon in the English literature. In the works of later Shakespeare, in the verse of Romantic and Post-romantic poets the 'hemistich boundary' moves to the right, positions 6/7. What makes the 4 + 6 form unmarked? The fact that it appears in a more rigorous variant of the English iambic pentameter? I wish Prof. Tsur would at least explain". I am going to do just that, and, in fact, I have so done in Tsur (1977: 74-82).

Consider the comment "The hemistich form 4 + 6 of the English iambic pentameter has been derived from the pattern of the French decasyllable". This statement attempts to refute my argument concerning the perceived affects of the English iambic pentameter by relying on its putative origin. I don't exactly understand what this may mean, or what light this can throw upon an understanding of the perceived affect of the caesura after the fourth position.2 At any rate, I too have found that Villon and Baudelaire overwhelmingly place caesura in their decasyllabic verse lines after the fourth position. I have found that in English iambic pentameter there is an overwhelming tendency to place caesura after the fourth position. I have also found in three Hebrew (Avraham Shlonsky, Leah Goldberg, and Shim'on Halkin), and two Hungarian poets (Árpád Tóth, and Attila József) a similar tendency to place caesura after the fourth rather than sixth position in iambic pentameter lines. This I have taken for a quite interesting finding, that calls for an explanation, namely, that there is an overwhelming tendency to place caesura after the fourth rather than sixth position of decasyllabic lines in languages as remote from each other as English, French, Hungarian, and Hebrew, and in such different versification systems as syllabic and syllabo-tonic meter. I concluded that this is an instance of a wider linguistic principle, mentioned by Jakobson as an example of the poetic function: "'Why do you always say Joan and Margery, yet never Margery and Joan? Do you prefer Joan to her twin sister?' 'Not at all, it just sounds smoother'. In a sequence of two coordinate names, as far as no rank problems interfere, the precedence of the shorter name suits the speaker, unaccountably for him, as a well-ordered shape of the message" (Jakobson, 1960: 356-357). This seems to hold for "precedence of the shorter hemistich" as well. This, in turn, I have taken as an instance of an even wider linguistic principle, that can be accounted for with reference to the limitations of short-term memory, constraining many speech-processing, and rhythm-processing activities. Geoffrey Leech formulated it as follows: "There is a general tendency for the weight of syntactic structure to occur later rather than earlier in the sentence, so as to avoid strain on a person's short-term memory in the course of constructing and interpreting sentences" (Leech, 1974: 197). This enables a person to handle the preliminary processing of the first part of a linguistic unit while new information is still presented to him by the speaker. This principle can also be formulated as "the item hardest to process comes last", and can account for a wide range of phonetic preferences too, among others, the preference of such expressions as sing-song and ding-dong to song-sing and dong-ding, of trick or treat to treat or trick, or walkie-talkie to talkiewalkie, etc. (Tsur, 1992 a: 84-86).

Now, to complicate things, Cooper and Meyer (1960: 61) have found that this principle of the longest item comes last works outside linguistic communication, in music, too. I have already referred to Woodrow's experimental finding that durational differences tend to give rise to end-accented rhythms.3 Among a variety of explanations, Cooper and Meyer (1960: 61) appeal to this finding in order to account for their finding that musical sequences tend to divide into a shorter followed by a longer group. "As lengths of the rhythmic groups increase, the ability of stress to act as an effective organising force diminishes and the role of durational differences in determining grouping necessarily becomes more important". Considering that greater length is usually perceived as greater accent in speech perception as well as in music, it would be but reasonable to expect that whenever a pentameter line is divided into segments of four and six, segmentation into 4 + 6 would be more natural than 6 + 4.

Now let us have a closer look at the numbers characterising the individual poets. I have assigned the caesurae, according to a procedure described elsewhere (Tsur, 1977: 72-74), in the first one hundred lines of Shakespeare's sonnets, Paradise Lost, "An Essay on Criticism", and the first one hundred decasyllabic lines in Faerie Qvene and "Adonais". The following table shows the number of lines in which caesurae occur after the fourth, fifth, or sixth position; "double" indicates lines in which two approximately equal major syntactic boundaries occur within the region of balance (which is not the same as indeterminacy). The difference between "total" and 100, in each poet, indicates the number of lines in which either syntax overrides caesura, or the assignment of caesura would have reflected personal inclination rather than an objectively assigned property of the line.

after IV3325424739
after V3224321815
after VI1423 92718
double 1 1 6 -- 1

All the poets examined have more lines in which the caesura occurs after the fourth position than after the sixth. Pope has almost five times as many, Shakespeare and Shelley only somewhat over twice as many, and Spenser somewhat less than twice as many. The exception in this context is Milton; there is no significant difference in the number of caesurae after positions IV, V and VI. We have to conclude, then, that the unmarked form is the caesura after position IV. Had a similar distribution occurred in Pope, we should have had to conclude that caesurae after positions IV and VI are equally unmarked. Within the present constellation we have to conclude that the unmarked form is, beyond doubt, caesura after position IV. I have distinguished above between convergent and divergent poetry. In Milton's divergent poetry, contrasts are typically blurred (as, for instance, by stressed syllables in weak positions and unstressed ones in strong positions, and by divergent patterns of alliteration). As I have suggested, the divergent effect of Milton's poetry cannot be accounted for merely by the number of deviating stresses, but only by his having recourse to the marked options in a wide range of versification devices, of which stressed syllables in weak positions are only one and marked caesura appears to be another. He has recourse, not infrequently, to such marked forms as strings of stresses that end in weak positions, compounds whose first stress occurs in a weak position, bisyllabics with their stressed syllables in weak positions, and last but not least, stress maxima in weak positions, which are all virtually non-existent in Pope. So we should not be surprised that in respect of caesura, too, Milton resorts to the marked form more frequently than some other poets.

Another feature needs explanation. In other respects, Shelley's style is as divergent as Milton's (in some respects even more). In this particular respect of caesura, however, he deliberately prefers the form which we have decided to be the unmarked one. This would suggest, what is quite reasonable to believe, that the aspects in which there is an opposition marked/unmarked, are independent variables (within the range of solidarity I have established between them). We seem to categorise, intuitively, a certain passage as divergent or convergent according to whether the marked or unmarked forms outweigh the others. The absence or scarcity of any particular unmarked form in a passage may not be criteria for judgment. High occurrence of unmarked forms is to be taken for granted in divergent style too (by definition, the frequency of a marked form is smaller than, or equal to, the frequency of the corresponding unmarked form). Furthermore, I have shown in my book (Tsur, 1977: 200-203), by comparing the use of Spenserian stanza by Spenser and Shelley, that Shelley, who is one of the most deviant poets in my sample, may become more regular than Spenser at certain points--one of the most regular poets of my sample. This is most conspicuous in the sixth position of the dodecasyllabic lines which, not as in Faerie Qvene, in "Adonais" is invariably followed by a caesura and which is the only metrical position in which no deviation occurs throughout this deviant poem. The explanation offered for this remarkable regularity in the sixth position of the dodecasyllabic lines of this poem may probably account for Shelley's preference for the unmarked caesura. Shelley's deviance is anything but chaotic; unlike Donne, he takes deliberate (though inconspicuous) measures to keep his highly divergent verse cohesive.

By essentially the same procedure, I have assigned the caesurae to a random number of iambic pentameter lines by three Hebrew and two Hungarian poets (ibid., 82):

after IV54123132830
after V--25 45134
after VI 2 76 111410
double 4 9 1 3 3
other --5 3 2 3

The same tendency is visible in some decasyllabic verse outside syllabo-tonic meter. In French syllabic poetry, decasyllabic lines are relatively rare. I have checked two of Villon's ballades (65 lines), and all the decasyllabic poems in Baudelaire's Les Fleurs du Mal (94 lines). The overwhelming majority of caesurae in these poems occurs after position IV (49 in Villon, and 64 in Baudelaire).

By now it may be quite clear that both Cognitive Poetics and the "scholarly tradition of the so-called Russian school in verse study" engage in "quantitative analyses of observable facts" (though the latter does it on a much larger scale). They seem to differ rather in the kinds of explanation they offer for poetic affects. Both kinds of explanations are deeply embedded in twentieth century psychological thinking. The historical explanations offered by both Shavit and the reviewer seem to assume that frequent exposure to certain versification forms associated with certain themes (or certain poetic styles) may condition readers (the author's contemporaries, mainly) to respond to the versification forms as if they were responding to the themes (or styles) with which they are regularly associated. What happens within that black box, the reader's head, does not seem to concern them. Cognitive poetics, on the contrary, focuses much interest on the cognitive mechanisms and processes mediating between exposure and response to poetic structures. It assumes that poetic styles and their perceived affects result from an interaction of a wide variety of norms that exist on a variety of strata of the work, and that these norms do have certain inherent (sometimes conflicting) potentials to contribute to the total affect. These potentials may be (and some of them have been) investigated on independent grounds with, e.g., the tools offered by gestalt psychology and psycholinguistics.4

Thus, for instance, there is an enormous bulk of experimental literature indicating that many instances of markedness result from difficulty of processing, or load on short-term memory. What characterises these experiments is, that they tap evasive intuitions, the sources of which are hopelessly subliminal. That is why, though sometimes quite decisive, they look negligible to the casual observer.

Now consider, against this short excursus, the following point of the reviewer. "Probably 'backward grouping' is hard for perception; but dactylic rhymes are scarce in English poetry for phonological and historical reasons"; and, again, "thus, the dactyl (and other ternaries) became scarce, and therefore 'marked', rather than 'marked' and therefore scarce". These comments seem to be quite typical of the conflicting approaches lurking, all along, behind our controversy. The reviewer tends to give priority to statistical and historical reasons, whereas I am trying to rely on a free interaction of a wide variety of reasons of different kinds. Thus, for instance, I am acknowledging the difficulty of finding dactylic rhymes in English; but I mention it as one of several reasons. I am quoting myself: "The dactylic rhyme is highly marked in English poetry, for three different reasons. One reason is that in the English language it is more difficult to produce it than in some other languages, hence its rarity. The other two reasons seem to be less language-dependent. The second reason has to do with the grouping of unstressed syllables with stressed ones. For unstressed syllables to achieve a sense of stability, they must be grouped together with an adjacent stressed syllable in a metrical strong position, and it is easier to group them forward than backward. Backward grouping is always perceived as marked. In the dactylic rhyme, the last two unstressed syllables must be grouped backward since, by definition, they are not followed by a stressed syllable in a strong position. What is more, when the dactylic rhyme occurs in a ternary metre, two weak positions are grouped forward, and two backward, with the same (last) stressed syllable in a strong position (as opposed to the binary metres, in which between any two strong positions there is only one weak position). This may generate a kind of uneasy feeling of perceptual ambiguity and confused directions. The third reason has to do with the acoustic cues for grouping and stress", indicated by Woodrow's and Frye's experiments.

An interesting test case for the assumption that the scarcity of dactylic words in a language is the reason for the markedness of dactylic rhymes would be Hungarian poetry. Hungarian, as it is well known, is a beginning-accented language. Yet, dactylic rhyming is almost non-existent in it. One important reason for this lies outside the scope of the present discussion: for (admittedly) historical reasons, in Hungarian, ternary metres are typically in classical quantitative metre, whereas binary metres are typically syllabotonic.5 This, however, should not have prevented Hungarian poets to write poems in binary metres, with syllabo-tonic dactylic rhymes. When L"orinc Szabó translates "The Bridge of Sighs" into Hungarian, he seems to have no difficulty in reproducing the dactylic rhymes, in quantitative or syllabo-tonic metre. The first two stanzas tend to be distinctly quantitative:

The following two lines from stanza 4 have syllabo-tonic dactylic endings:

But in poetry written originally in Hungarian, it is almost impossible to find dactylic rhymes.6 One of the rare instances, György Faludi's "Mercenary Song", is an ambiguous case. This poem is, in fact, in iambic tetrameter; but owing, precisely, to the beginning-accented nature of the Hungarian language, there is a tendency to override the last stress in the line, and so obtain dactylic rhymes. Here is the first stanza of the poem:

In fact, many readers have reported the perception of a hypnotic affect in this poem. According to the conception propounded above, this affect can be accounted for by the "false security" due to the feeling of security generated by the more than usually regular metre of the poem on the one hand, and the insecurity generated by the dactylic rhyme. This affect is foregrounded, by contrast, precisely in those few cases in which the dactylic construal of the line-ending breaks down, and the rhyme becomes unambiguously iambic. The Hungarian precedent would strongly suggest, then, that the scarcity of beginning-accented words cannot be regarded as the only, or even the main, cause for the paucity (and markedness) of dactylic rhyme in a language.

Finally, I propose briefly to consider a crucial issue raised by the reviewer: "I feel a little uneasy about many instances of impressionistic, subjective statements in Prof. Tsur's work". One way to define Cognitive poetics is to locate it between analytic criticism and impressionist criticism. The former, I suggest, excels in the description of the structure of literary texts, while the latter indulges in their perceived effects. Cognitive Poetics offers cognitive theories that systematically account for the relationship between the two. By the same token, it discriminates between reported effects which may legitimately be related to the structures in question, and those which may not. It should be obvious by now, I hope, that in its speculations on markedness affects, or emotional and witty qualities, Cognitive Poetics heavily relies on systematic knowledge drawn from clinical experience with the Rorschach ink-blot test, and sometimes carefully controlled experiments of the psycholinguists and gestalt psychologists. But I would like to end this Postscript with a few passages, quoted from an analytic philosopher, and a linguist-semiotician. The philosopher Frank Sibley says of "aesthetic concepts":

If we are not following rules and there are no conditions to appeal to, how are we to know when they are applicable? One very natural way to counter this question is to point out that some other sorts of concepts also are not condition-governed. We do not apply simple color-words by following rules or in accordance with principles. We see that the book is red by looking, just as we tell that the tea is sweet by tasting it. So too, it might be said, we just see (or fail to see) that things are delicate, balanced, and the like (Sibley, 1962: 77).

Such a conception is familiar from other areas of poetic theory as well. The linguist Manfred Bierwisch suggests:

Poetics must explain just which structural qualities form the basis for definite effects. It can and must explicate those consciously or unconsciously followed regularities that lead to the understanding of poetic structure and to a judgment of poeticality [...]; it must accept effects as given and determine the rules upon which they are founded (Bierwisch, 1970: 108).

As I have earlier said, Cognitive Poetics proposes to do just that. It uses quantitative investigations to establish regularities, makes a sustained effort to ascertain what "the specific effects of poetry" are, through introspection, in controlled experiments in the psychological laboratory, by considering statements of professional critics in their published writings, or by collecting casual reports of students in the classroom. It offers hypotheses drawn from the various branches of cognitive science to relate, systematically, poetic effects to poetic structures with which they have been regularly associated. By the same token, it helps further to refine the formulation of perceived affects collected in a variety of ways from a variety of respondents. Far from relying "on a reader 'in general,' or on a modern (and sometimes unsophisticated) reader", it helps to break up a homogeneous past into a multiplicity of related attitudes, but avoiding scepticism, idiosyncrasy, or chaos. While the historical approach encountered here "is based not so much on insight and intuition as on wide quantitative analyses of observable facts", Cognitive Poetics assumes that the readers of past ages had insights and intuitions, no less than present-day readers; that to understand such insights and intuitions one must study the responses of living readers, and then try to find out in what respects can the responses of readers in the past be supposed to differ from the responses of present-day readers; and that to abandon such inquiries is too high a price to pay for "scientific objectivity".


1. In Pagis' critical edition of Al-Tabban's extant poetry, there are 12 five-line-long or shorter poems with pure suffix rhymes, 3 with mixed rhymes, and one with pure root rhymes. In addition, there is one six-line-long, and one ten-line-long poem with root rhymes, with distinct anti-grammatic sophistications. Still the number of purely suffix-rhymed poems in this poet's corpus is overwhelming.

2. This is a good example of what Wimsatt and Beardsley called "The Genetic Fallacy". I shall not recapitulate here the arguments for and against it (they have been reviewed and criticised by Wimsatt, 1968).

3. Woodrow's experimental findings seem to be in harmony with some of my other cognitive assumptions governing my argument. That durational differences tend to give rise to end-accented rhythms can be interpreted as 'whenever durational differences arise, the longest segment comes last'. His finding that differences in amplitude tend to give rise to beginning-accented rhythms can be explained, first, by assuming that, other things being equal, greater amplitude at the beginning of a sequence more efficiently arouses the alertness of the perceiving system than at the end; and second, with reference to the finding that greater amplitude does not constitute a greater load on short-term memory.

4. As a matter of fact, the so-called Russian school does not seem to be that homogeneous either. Roman Jakobson, and some other Russian Formalists explicitly embrace mentalistic approaches, and some of their assumptions considerably overlap with the assumptions of gestalt psychologists, and I have learnt from them an enormous lot. One of my favourite quotes is from Victor Erlich:

Works of art are knowable objects, accessible only through individual experience. Consequently, the mechanism of aesthetic response is a legitimate concern of an "objectivist" art theoretician, provided that the emphasis is placed not on the individual reader's idiosyncratic associations, but on the qualities inherent in the work of art and capable of eliciting certain "intersubjective" responses (Erlich, 1965: 178-179).

5. These historical reasons too appear to have linguistic and cognitive reasons. In Hungarian, both length and stress create phonemic differences. This would allow both for quantitative and syllabo-tonic metre. Since, however, phonemic contrasts based on length are relatively rare, quantitative metre is less natural in Hungarian. While syllabo-tonic metre is congenial to Hungarian poetry, the classical quantitative metres were imported to Hungarian by a deliberate classicist effort in the eighteenth century. (In French, by contrast, which is a syllable-timed and not stress-timed language, a deliberate classicist effort to import iambic verse failed).

6. Following the present discussion, I have pursued this issue of Hungarian dactylic rhyme in a Hungarian paper (forthcoming in Literatura)

Reuven Tsur


Arnheim, Rudolf
1957 Art and Visual Perception. London: Faber & Faber.

Bierwisch, Manfred
1970 "Poetics and Linguistics", in Donald C. Freeman (ed.), Linguistics and Literary Style. New York: Holt, Rinehart & Winston. 97-115.

Cooper, C. W. and L. B. Meyer
1960 The Rhythmic Structure of Music. Chicago: Chicago UP.

Erlich, Victor
1965 Russian Formalism, revised ed., London:

Jakobson, Roman
1960 "Closing Statement: Linguistics and Poetics", in Thomas A. Sebeok (ed.), Style in Language. Cambridge, Mass.: MIT. 350-377.

Leech, Geoffrey
1974 Semantics. Harmondsworth: Penguin.

Levy Ibn Al-Tabban
1967 The Poems of. Critical edition by Dan Pagis. Jerusalem: The Israel Academy of Sciences and Humanities (in Hebrew).

Rorschach, Hermann
1951 Psychodiagnostics. Bern: Huber.

Schirmann, Jefim
1961 Hebrew Poetry in Spain and in Provence, 4 vols. Jerusalem and Tel Aviv: The Bialik Institute and Dvir (in Hebrew).

Wimsatt, W. K.
1968 "Genesis, a Fallacy Revisited", in Peter Demetz et al. eds., The Disciplines of Criticism. New Haven and London: Yale UP.

Tsur, Reuven
1968 Studies in Mediaeval Hebrew Poetry. Tel Aviv: Daga (in Hebrew).

Back to Introductory Page
Back to home page
Back to "Occasional Papers"

Original file name: Postscript - converted on Sunday, 30 November 1997, 09:59

This page was created using TextToHTML. TextToHTML is a free software for Macintosh and is (c) 1995,1996 by Kris Coppieters