Reuven Tsur

Back to home page
Back to "Occasional Papers"

Reuven Tsur

Comparing Approaches to Versification Style
in Cyrano de Bergerac


These days I am engaged in the preparation of a second, revised edition of my book Toward a Theory of Cognitive Poetics. Since the first publication of my book many scholars adopted the term "Cognitive Poetics", and it is now widely used—in quite different senses. My prospective publisher suggested that I add a few case studies comparing my analysis to some cognitive linguist's analysis of one issue each, so as to highlight the difference between our approaches.

This paper contains one of those case studies: it focuses on Eve Sweetser's investigation, in the Lakoff tradition, of the contribution of rhyme and metre to a comprehensive interpretation of Edmond Rostand's Cyrano de Bergerac. I will extrapolate to Rostand's play the theoretical machinery I have elaborated during the years for the investigation of rhyme and metre. Sweetser says she is working within "blending theory"—somewhat stretching the concept, so as to apply to versification. At any rate, her approach is strongly meaning-oriented. My approach, by contrast, has gestaltist and phonetic orientation, and when I have recourse to meaning, it is a semantic-oppositions approach. Thus, in the final resort, it is the suitability of two theoretical approaches to handle versification that stands to trial—a meaning-oriented and a gestalt-oriented approach. The purpose of this paper is not to take Sweetser to task for not pursuing a certain kind of criticism. My purpose is—in harmony with my publisher's request—to provide a case study comparing my approach to some other cognitive approach—in this instance, a case study of versification.

Let me state at once the bottom line of this comparison: Cognitive Poetics is not a homogeneous enterprise. Sweetser and I isolate the same text units for attention, but have different methodologies, ask different kinds of questions, and give different answers. Briefly, she concentrates more on meanings; I—on perceptual qualities. It would appear that quite frequently we have similar underlying intuitions; but our different methodologies lead us to ask different questions. I fully agree with the admirable interpretation of the play that emerges from Sweetser's handling of versification; but here I will concern myself only with versification and the perceptual qualities it generates.

The Lakoff school pays curiously little attention to the prosodic dimension of poetry. It is mainly interested in semantics. Two notable exceptions are Masako Hiraga's work, and Eve Sweetser's paper "Whose rhyme is whose reason? Sound and sense in Cyrano de Bergerac".

As a preliminary, I will note that I disagree with Sweetser on one side issue. "Rostand's Cyrano de Bergerac is a particularly complex example of dramatic verse", she writes. "Its form is instantly salient, since its traditional hexameter [sic] (12-syllable lines) rhymed couplets metonymically evoke the frame of mid-17th-century France: the play's setting; and the time when 'classical' French drama was in flower and verse tragedy was considered its highest-prestige form" (p. 32). I claim that this metonymic evocation of the frame of mid-17th-century France, the play's historical setting, is less significant than it would appear. The play's versification is a variable independent from its historical setting. Thus, for instance, Rostand's other verse drama, L' Aiglon (The Eaglet) is placed in the 19th-century (1830–1832), in the Austrian Emperor's court, the "eaglet" being Napoleon's and Marie-Louise's son, with such historical personages for dramatis personae as Marie-Louise and Prince Metternich. The play is "l'histoire d'un pauvre enfant" within that historical setting. The two plays have recourse to exactly the same versification patterns of the alaxandrine couplet. Furthermore, I will argue that Rostand's alexandrine couplets, in both plays, are as different from Racine's as possible within the same versification tradition. In fact, with their occasional enjambments and "effaced" caesurae (see below) they are more similar to the couplets of Romantic than of Classical drama. [1]

I am going to compare my approach to Sweetser's consistently excellent work. As I said, we isolate the same kinds of text units for attention; so, on the descriptive level we would most probably provide quite similar descriptions of the text. The difference would be in the different theoretical frameworks within which we make those descriptive statements. Sweetser adopts the venerable Form–Content dichotomy; I adopt Wellek and Warren's more recent notion (1942) of "Materials" and "Structure". And, as I said, we are working wihin a meaning-oriented and a gestalt-oriented theory, respectively. The different theoretical perspectives generate considerable differences of focus or emphasis on the very same perceptions.

As a point of departure, let me quote Boileau's aphorism: "La rime est une esclave, et ne doit qu' obéir" [Rhyme is a slave, and must do nothing but obey] (L'Art Poétique, I. 30). But who is the master? According to Sweetser's conception it must serve meaning; according to the present conception—it must serve a perceptual whole of which it is a part; and meaning is just another part.

Articulateness and Requiredness

Sweetser introduces her section "The Hôtel de Bourgogne: rhyme, meter, and polyphonic voice" with the following example.

The pickpocket: La dentelle surtout des canons, coupez-la!
(Particularly the canons' lace,
cut it off !)
A spectator:
Tenez, à la première du Cid , j'étais là!
(You know, at the first performance of
le Cid, I was there!)
The pickpocket:
Les montres ...(the watches)
The bourgeois:
                    Vous verrez des acteurs très illustres ...
(You'll see really
illustrious actors...)
The pickpocket:

Les mouchoirs ...(the handkerchiefs)
The bourgeois:

                          Montfleury ....
Someone shouting from above:

                                                Allumez donc les lustres!
(So light the
The bourgeois:...
Bellerose, l'Epy, la Beaupré, Jodelet !
A page:
Ah! Voici la distributrice!...
(Oh, here's the food vendor!)
The vendor (female):
                          Oranges, lait,...

She comments on this excerpt:

In the following example (I. i. 25–30), characters complete not only each other's couplets (25–6, coupez-la and j'étais là), but each other's lines: two speakers' words make up line 27 (illustres), and three speakers (including the preceding two) contribute to the other half of the same couplet, line 28 (lustres). More interestingly, and less typically of verse drama, these characters are not engaged in a single conversation, but in multiple parallel conversations between characters who might not interact directly (given social divisions), and in fact do not even all hear each other. The pickpocket is covertly advising his trainees, the theater buff bragging to his friends, the bourgeois informing his son, the vendor advertising refreshments, and the theater employee calling to colleagues for the lights to be lit.

This paragraph is a description of the excerpt. This, or a similar, description could occur in my discussion as well. What would be most important for me in her description concerns the segmentation of lines and distribution of speech between different speakers and even different conversations. I would focus, on the experiential level, on the perceptual qualities bestowed by them on the couplets. Sweetser, by contrast, says with so many words that she concentrates on meaning: "when multiple characters contribute to a line, or when 'enjambment' gives us salient differences between the two kinds of boundaries, the meaning affordances of metrical line units stand out more clearly" (34). I have nothing against such preoccupation with meaning; except that it misses the perceived qualities contributed by versification. Then she contiues:

By letting these multiple characters' voices 'collaborate' to make hexameter [sic] lines, and rhymed couplets—a structured and unified formal whole—Rostand's blend conveys the broader message that at the content level these parallel interactions between diverse individuals form a unified social whole. Perhaps the message is that Parisian society is somehow an organic whole, despite class and other social divisions; or perhaps a more interesting interpretation is that dramatic artistic performance brings together disparate segments of society into a community. This could well be Rostand's bourgeois late-19th-century message; and if so, it cannot reasonably be put in the mouth of any character within the play's 17th-century setting under the French monarchy. It is appropriately expressed not by those characters' words, but by their interleaving relationship to the metrical structure of the whole (37–38). [and so forth].

This paragraph concerns the interpretation or significance, not the perceived qualities, of the said segmentation and distribution. She offers, à propos versification, a brilliant interpretation of the play. As I said, I embrace most of her interpretation, some of it with great enthusiasm. But in this article I will concern myself only with versification. This paragraph, however, contributes very little to this interpretation. It rather relates the segmentation and distribution to wider issues that are no organic part of the play's action (such as "Rostand's bourgeois late-19th-century message", or "the message that Parisian society is somehow an organic whole"). One thing, however, appears to be quite probable: this discussion is an indication of the author's strong intuition (which I share) that this versification structure is somehow very effective. Sweetser is looking for the key to this effectiveness under the streetlamp of the contents; I claim that versification does something to poetry that cannot be reduced to meaning.

I devoted no professional attention to this play before reading Sweetser's paper. But I have written quite a lot on English versification that, I believe, might fruitfully be extended to Rostand's versification as well. My first English publication (1972) was "Articulateness and Requiredness in Iambic Verse", later included in Toward a Theory of Cognitive Poetics. Articulateness and Requiredness are two sides of the same coin; they are aspects of breaking up a whole into segments. When we speak of articulateness, we imply that a whole has been broken into parts, and that this facilitates perception of the whole. When we speak of requiredness, we imply that each part is essential to the whole: when a part is omitted, there is an acute feeling of incompleteness, of imbalance. Articulateness and requiredness depend on the relative strength of the whole. Requiredness is possible only where the whole is highly organised. If the integrity of the whole is not felt, deficiency cannot be felt either.

According to the gestalt assumption, a perceptual unit tends "to preserve its integrity by resisting interruptions": each of two conflicting units will tend to reassert itself in the reader's perception (Fodor and his colleagues have empirically substantiated this assumption in a linguistic context; Fodor and Bever, 1965; Garrett et al., 1966). The articulation of a perceptual whole suggests its interruption. Thus, articulation enhances requiredness in cases when the whole is strongly organised; at the same time, it also creates a need to reassert the integrity of this whole. The weaker the organisation of the whole, the weaker the impact of requiredness. At the same time, the role played by articulation in perception increases and may become all-important. In a word, the smaller the relative strength of the whole, the more the emphasis shifts from the requiredness aspect of segmentation to its articulateness aspect. As gestalt psychologists have insisted time and again, segmentation may facilitate the perception of a complex whole. Caesura is such a segmentation device in the middle of a verse line, rendering it symmetrical, stable, balanced. A word- or phrase-ending may confirm caesura; in its absence, syntax overrides caesura, generating tension or, alternatively, blurring the line's shape. There may be additional, syntacic breaks in a line. The nearer the break to the caesura, the less it threatens the balance and symmetry, and the less one is likely to mistake the segment for a full but defective line. When it occurs nearer to the end, one is more likely to mistake the segment for a completed unit manqué, and to regard it as a threat on the integrity of the line. Consequently, our relief will be greater when the missing part is supplied. This may generate, in certain cisrcumstances, a sharp, witty effect, turning the last string of syllables into a "punch-phrase", so to speak.

Let us consider a minimal pair, of a couplet from Pope's The Rape of the Lock (a), and an alterated version with a different word order (b):

(a)        Now Lapdogs give themselves the rouzing Shake,
            And sleepless Lovers, just at Twelve, awake.

(b)       Now Lapdogs give themselves the rouzing Shake,
           And, just at Twelve, the sleepless Lovers wake.

The difference between the two versions seems rather slight. [2] Still, (a) appears to be wittier. One of the salient reasons seems to be the difference in the place of the syntactic breaks. In (a), "And sleepless Lovers, just at Twelve" urgently requires completion. A break after the seventh, eighth, or ninth syllable generates increasingly greater tension, so that in (a) "awake" is highly required, and its occurrence is experienced as sudden relief—as wit. In the following couplet from "An Essay on Criticism", there is little that can account for the wit of the second line, except the requiredness of the last word:

Some foreign writers, some our own despise,
The ancients only, or the moderns, prize.

This is a characteristic feature of Pope's wit. And this appears to be one of the characteristics of Rostand's wit as well.

Now consider Sweetser's observation: "characters complete not only each other's couplets (25–6, coupez-la and j'étais là), but each other's lines: two speakers' words make up line 27 (illustres), and three speakers contribute to the other half of the same couplet, line 28 (lustres). More interestingly, and less typically of verse drama, these characters are not engaged in a single conversation". This observation points at something that is intuitively very significant, but Sweetser's theoretical framework cannot account for it. The present theoretical framework, by contrast, is tailor-made for it. The speaker has completed his part, but the verse line remains incomplete. This taking turns enhances the sense of segmentation and, by the same token, the sense of incompleteness, increasing the requiredness of the missing part. The greater the number of interruptions, the more vigorous the verse line's "effort" to reassert itself in the audience's perception (I will return to these two pairs of rhymes). This is, perhaps, the epitome of the argument of the present paper. We both feel that the distribution of one line between several speakers and even several conversations is very significant from the poetic point of view. Blending theory is incapable of accounting for this, whereas gestalt theory is tailor-made for it.

This construal, that "the message is that Parisian society is somehow an organic whole, despite class and other social divisions" is very plausible. But I wish to make four further comments on it. First, in the "OUTSIDE THE CITY GATE" scene in Goethe's Faust there is a similar polyphony of brief conversations of several social groups, and we receive a similar "message" from it. Here, however, the versification strategy is quite different. There is a more complex, less regular rhyme pattern; on the other hand, in the whole 95-line section only one line is distributed between two speakers, with the segmentation not near the end but near the middle of the line. Furthermore, in Goethe's scene rhyme patterns do not "spill over" from one conversation to another. This might suggest, in Rostand, a sharpening of the paradoxical view of creating an organic whole despite class and other social divisions. Second, this structure dominates Rostand's both plays. Cyrano takes place in Parisian society which is said to be "somehow an organic whole, despite class and other social divisions". But we have the same versification structure in L'Aiglon too, which takes place in the Austrian Emperor's court and royal family, whose social structure is quite homogeneous. In fact, in Cyrano too we find the "social polyphony" only at the beginning of Act I, whereas the versification style prevails throughout the play. Third, whatever the social implications, Rostand's versification strategy has considerable "added value": it bestows a witty quality on the dialogues. Finally, a piece of philosophical hairsplitting: to the outright predications "the message is that" and "Parisian society is a" I would prefer a Wittgensteinian conception of "seeing as": "presenting Parisian society as an organic whole, despite class and other social divisions". In other words, this is not the meaning of the scene; merely, Parisian society is concretised in a certain way.

The versification pattern pointed out in Cyrano is no less conspicuous in Rostand's other play. Consider the opening lines of L'Aiglon:

LES DAMES, au clavecin. [...]
Elle manque tous les bémols. — C'est un scandale!
-- Je prends la basse.—Un, deux!—Harpe!—La... la!...—Pédale!

BOMBELLES, à Thérèse .
C'est vous?

                     Bonjour, Monsieur de Bombelles.

UNE DAME, au clavecin.
                                                                                Mi... sol...

J'entre comme lectrice aujourd'hui.

UNE AUTRE DAME, au clavecin.
                                                            Le bémol!

Et grâce à vous. Merci.

                                      C'est tout simple, Thérèse
Vous êtes ma parente et vous êtes Française.

(THE LADIES, at the harpsichord. She misses all the flats. — This is a scandal!
I play the bass. One, two! Harp! La... la!...— Pedal!
BOMBELLES, to Therese It's you?
THERESE Good day, Mr. de Bombelles.
LADY, at the harpsichord Mi... sol...
THERESE I am starting today as a reader.
ANOTHER LADY, at the harpsichord Flat!
THERESE It's thanks to you. Many thanks.
BOMBELLES It's elementary, Therese
You are my kin, and you are French)

Consider the following sequence from the second line of this excerpt: "Un, deux!—Harpe!—La... la!...—Pédale!". All are one- or two-syllable-long chunks, none of them constitutes a clause. The first two chunks are required to complete the first versification segment (that is, up to the caesura); the rest to complete the line. Here, too, the "characters are not engaged in a single conversation, but in multiple parallel conversations". Or consider the following sequence in the same scene: "C'est vous? / Bonjour, Monsieur de Bombelles./ Mi... sol..." All this makes up a single line. The first two chunks are spoken by two different persons, and the conversational turns are completed (if not ended). But, alas, two more syllables are required. These come, eventually, from the harpsichord string of conversation. What is more, these two syllables constitute two semantically related, but syntactically disjunct words.

In the last couplet of the sequence, the first line is divided between two speakers precisely at the caesura. Its second hemistich is further articulated by the vocative "Therese", that interrupts the syntactic flow and is highly required. Finally, the sequence achieves stable closure by ending with a complete line displaying perfect parallelism with a distinct caesura.

This effect of threatening the line's integrity may be further enhanced by a "false" rhyme in mid-line, as in the following excerpt from the same scene:

MARIE-LOUISE, se retournant, à Bombelles.
Je viens d'écrire pour qu'on garde son cheval!

A Thérèse.

Depuis la mort du général...

THERESE, étonnée.
                                                Du général?

(MARIE-LOUISE, turning to Bombelles.
I've just written that they should take care of his horse!
To Therese.
Since the death of the general...
THERESE, astonished. Of the general?)

The audience is fooled into mistaking the first "général" for the expected rhyme word, yielding a grievously deficient verse line. Therese's astonished question "Du général?" provides the genuine rhyme and the missing part of the perceptual field. There are quite a few such instances of false alarm in Rostand's both plays. The next excerpt is from Cyrano:

UN BOURGEOIS, conduisant son fils
Plaçons-nous là, mon fils.
UN JOUEUR                  Brelan d'as!
UN HOMME, tirant une bouteille de sous son manteau et s'asseyant aussi
                                                            Un ivrogne
Doit boire son bourgogne...
Il boit.

                                              ... à l'hôtel de Bourgogne!

(BOURGEOIS, leading his son
Let us sit down here, my son.
CARD PLAYERTriple ace!
A MAN, taking out a bottle from under his coat, sitting down too
A druncard
Must drink his Burgundi ...
... in the hôtel de Bourgogne! )

The unity-in-variety aspect of the Parisian society may be an additional dimension of Rostand's dialogue in Act I. of Cyrano; but cannot account for the witty effect of the lines; this can be traced to the unity-in-variety aspect of the dialogue, in perspective of gestalt theory.

Interruptions will increase the tendency of verse lines to reassert their integrity, up to a point. When this (theoretically undefined) point is passed, the verse line will begin to disintegrate. This effect crucially depends on performance. The actors may take advantage of such segmentation of the dialogue to suppress metre all in all, so as to render their speech as near as possible to realistic prose discourse. In this case, no requiredness will arise. But if the actors preserve the integrity of the verse to some degree, requiredness and the resulting witty effect will be most conspicuous. Paradoxically, the string of segments imitating the cadences of colloquial speech, too, will emphatically reassert itself. In this respect, segmentation is "double-edged".

To compare Rostand's alexandrine couplets to 17th-century verse drama, I have checked the first two acts of Racine's Phèdre (a total of 736 verse lines). There are notably few instances of verse lines distributed between two or more speakers. But the really striking thing is that none of these segmentations occurs in the second hemistich. Consider:

Seigneur, la reine vient, et je l'ai devancée,
Elle vous cherche.


                                    J'ignore sa pensée ...

(THERAMENES Sir, the queen is coming, and I've arrived before her,
She is looking for you.
THERAMENES I don't know what is her intention)

Even such a complex distribution of a line between two speakers as this divides the line into two exceptionally distinct hemistichs with a clear-cut caesura. "Moi!" completes the first hemistich, emphatically articulating the caesura; the second hemistich remains intact. It would appear that in the alexandrine couplet Racine cherishes balance and classical symmetry, whereas Rostand exploits, in the same versification pattern, the requiredness phenomenon, to generate a witty effect.


Not as the iambic pentameter, the twelve-syllable-line (English iambic hexameter and French alexandrine) demands, for good perceptual reasons, a rigidly fixed caesura precisely after the sixth position. The implied author of Cyrano is well aware of this. As Sweetser observes, "Even baking is poetic composition, as we see in Ragueneau's instructions to an apprentice baker (II.i):

Vous avez mal placé la fente de ces miches:
Au milieu la césure, entre les hemistiches!

You misplaced the cut on these loaves;
the cesura should be in the center, between the half-lines!

Notwithstanding this, Rostand the empirical author is sometimes quite "careless" about it. In his plays caesura occurs sometimes in the middle of a word, as in

Depuis la mort du gé/néral... Du général?


Ah! Voici la distri/butrice!.. Oranges, lait,...

It is not enough, however, that caesura should be marked by a word boundary; classical poetics demands that it should be the boundary of a lexical word or even a phrase. In the following example from L'Aiglon, by contrast, it occurs after a preposition, in mid-phrase:

Je viens d'écrire pour / qu'on garde son cheval!

In the first line of the following excerpt from Cyrano we have a caesura right after a conjunction+preposition. The line ends with another stylistic device prohibited in classicist drama: a tense enjambment, introducing an incomplete line, which demands three syllables to complete it. This is eventually provided by another speaker:

On ne commence qu'à / deux heures. Le parterre
Est vide. Exerçons-nous / au fleuret.
Pst... Flanquin...

As I said above, the versification style of Rostand's plays conspicuously differs from, for instance, Racine's classical style in several respects: a predilection for "requiredness" at the line ending; occasional "effaced" caesurae and tense enjambments. Consequently, the style of Rostand's alexandrine couplets is nearer to that of French Romantic Tragedy than that of 17th-century Classicist Tragedy.

Tame and Vigorous Rhyme

Let us turn now to the problem of rhyme. Sweetser comments on the first excerpt quoted above: "Individual rhymes here are meaningful in the usual ways: rhyming lustres 'chandeliers' with illustres ('illustrious') is an amusing change of register from the high-flown to the everyday concrete, as well as a play on the etymological relation between the two words". I would make three comments on this observation. First, we should make clear what it is that we have explained by "the etymological relation between the two words". That there is such an "etymological relation between the two words" is perfectly true; but that this enhances the rhyme's effect is rather doubtful. On the contrary, it appears to tame it. Rhyme is a conspicuous instance of unity-in-variety. The sound structure of its members should be as similar as possible; the meaning structure as different as possible. Its effect—both witty and emotional—crucially depends on this. (I am not speaking of modernist off-rhymes and incomplete rhymes). The etymological relation enhances the similarity of the meaning of the two members. Secondly, the change of register, by contrast, though not very salient, would enhance the opposition of meaning, if perceived at all. Actually, lustres ('chandeliers') need not be of a low register; after all, it is a decorative appliance, typically hung from the ceiling of some large, imposing building. Thirdly, the words lustres and illustres are opposed as parts of speech (noun vs. adjective), reinforcing the dissimilarity of meaning, in spite of the etymological relation.

I have taken the distinction between verse that rhymes similar and different parts of speech from W. K. Wimsatt. In his paper "One Relationship between Rhyme and Reason", (1964) he points out the difference between Chaucer's rhymes and Pope's. Chaucer's are "tame" rhymes, in which the same parts of speech are used in closely parallel functions. Not so Pope's, who achieves his witty effects, among other means, by rhyming, e.g., nouns with verbs, verbs with adverbs, in different syntactic positions. Such rhymes are perceived as vigorous. "We may say that the greater the difference in meaning between rhyme and words the more marked and more appropriate will the binding effect be" (Wimsatt, 1964: 168), or, as Wordsworth would put it, rhyme involves the pleasure of the discovery of similitude in dissimiltude. Roman Jakobson (1956: 82) [3] mentions Wimsatt's distinction between "tame" and "vigorous" rhymes and calls them "grammatical" and "anti-grammatical" rhymes.

Verlaine's notorious attack on rhyme may illuminate certain aspects of the issue:

Prends l'éloquence et tords-lui son cou!
Tu feras bien, en train d'énergie,
De rendre un peu la Rime assagie.
Si l'on n'y veille, elle ira jusqu'où?

Take eloquence and wring its neck! / You will do well while you are about it / to render rhyme a little calmer. If one does not watch, till where will it go?

The phrase "elle ira jusqu'où?" suggests length. If you do not watch, the parallel strings of sound will extend to monstrous lengths. However, the words "cou" and "jusqu'où" share no more than two speech sounds. Still, annotators usually make comments to the effect that this is an example of an acrobatic and funny rhyme. This effect is achieved by opposing the noun cou with jusqu'où. composed of a preposition and an adverb of place. The former is a unitary word, the latter derives its two speech sounds from two different words.

Now let us take a similar but more elaborate case discussed by Sweetser:

In the following passage from Cyrano IV. iii, Cyrano's Gascon regiment is at war, and starving because they are surrounded and their supplies cut off. Cyrano has a musician play a traditional Gascon folk tune on his fife, to distract them from their hunger; they become dreamily homesick instead, for a moment. The Captain fears that Cyrano is making the cadets 'soft' with this sentimental reverie.

Carbon Tu vas les affaiblir en les attendrissant!
(You'll weaken them by softening them too much!)
Laisse donc, les héros qu'ils portent dans le sang
Sont vite réveillés! Il suffit ...
(Don't worry, the heroism in their blood
is speedily awakened! All it takes...)

[He gestures, a drum rolls, and the cadets leap to their feet, arms in hand.]
Les cadets:
What is it ?)
Tu vois, il a suffi d'un roulement de caisse!
Adieu, rêves, regrets, vieille province, amour ...
Ce qui du fifre vient s'en va par le tambour
(You see, all it took was one drum roll!
Farewell, dreams, regrets, old home province, and
love ...
The effects of the fife are chased off by the

The pair of lines rhyming in qu'est-ce / caisse are of course parts of a single statement about the ease of bringing the cadets to attention—and the two rhyming words are closely linked in the meaning frame as well as in sound, since the drum (caisse) is what makes the sound which instantly brings a homonymous watch response ('What is it?') from the cadets. Similarly, the amour / tambour couplet is all about the contrast between sentiment (evoked by the fife) and soldierly courage (evoked by tambour, another word for drum); not only does the couplet encapsulate a complete aphorism on this subject, but the rhyme words come from the two contrasting models—the emotional state involved in one frame, and the musical instrument evoking the other. [...]. The presence of Cyrano's is speedily awakened! All it takes in the same line with the cadets' What is it? is a particularly graphic demonstration of Cyrano's point—indeed, a better completion of his meaning than he could have made by speaking himself. [4]

Rhyme and alliteration and meter all also make specific formal connections between smaller units: the rhyming or alliterating words, or the metrical half-lines, for example.

I've added Carbon's line to Sweetser's excerpt. In this passage we can, again, see most consistently Sweetser's and my contrasting approaches. We both feel that rhyme is particularly significant in this passage. Here, again, Sweetser looks for the key to this significance in the content. I claim, by contrast, that one of the advantages of Cognitive Poetics is that it provides tools to describe rhyme in a way that may account systematically for its perceptual qualities as directly experienced.

Consider Sweetser's following observation: "The pair of lines rhyming in qu'est-ce / caisse are, of course, parts of a single statement about the ease of bringing the cadets to attention—and the two rhyming words are closely linked in the meaning frame as well as in sound, since the drum (caisse) is what makes the sound which instantly brings a homonymous watch response ('What is it?') from the cadets". The assertion that "the pair of lines are of course parts of a single statement" must be understood in a very special sense. Whatever the topic of the utterance, such a virtuoso rhyme will always draw attention to it. Consequently, claiming that "the two rhyming words are closely linked in the meaning frame" is somewhat of a tautology. Actually, a poet like John Donne must make special manipulations to prevent alliteration, for instance, from connecting meanings (see Chatman, 1956; Tsur, 1998: 49–53). This holds true, with the necessary changes, of the amour / tambour couplet as well. The antithesis between a "dreamily homesick" mood and a heroic, soldierly attitude rests in the contents of the couplet, and any pair of rhyme words taken from the description of the respective moods will be "all about the contrast between sentiment and soldierly courage".

The phrase "homonymous watch response" is, again, an accurate description. But what is its significance? Sweetser drops in the word "homonymous", but does not elucidate its relevance. Her theoretical framework is focused on contents, not on verbal structure. The present theoretical framework is, again, tailor-made for it. Homonymity increases the phonetic similarity of the rhyme words; whereas the first member of the rhyme contrasts not one, but three words to the noun in the second member. At the same time, the sequence "Hein!...Quoi?...Qu'est-ce?" forces the line to reassert itself by resisting segmentation—achieving considerable articulateness near the line ending, and generating effective requiredness. We have seen Verlaine's rhyme "cou ~ jusqu'où". Its first member is wholly included in the second; and the two rhyming sounds are derived in the second member from two different words. This was said to be "acrobatic" and "funny". In the present instance, the homonymous rhyme consists of four shared speech sounds, which in the first member of the rhyme are derived from three (!) different words, sharpening its witty effect. The rhyme "attendrissant ~ sang" is vigorous, but far from the vigour of the "qu'est-ce ~ caisse" rhyme pair. It rhymes different parts of speech, a present participle verb with a noun. The rhyming syllables are homonymous; but neither of them contain speech sounds derived from more than one word.

We have already encountered (but not discussed) a similar instance: "Jodelet ~ lait". Here, too, the second member of the rhyme is wholly included in the first one; and the two nouns are contrasted at lower levels: [+/-common noun, +/-animate, +/-human]. In the "amour ~ tambour" rhyme both members are nouns, but are contrasted at a lower level: [+/-concrete]. Or take another instance which we have already encountered:

The pickpocket: La dentelle surtout des canons, coupez-la!
A spectator:
Tenez, à la première du Cid , j'étais là!

We have discussed Sweetser's observations on its versification. As to the rhyme itself, Sweetser merely foregrounds it by typography, without comments. Again, it neatly falls into our model. Both members of this rhyme are composite. The second word in them is the pair of homonyms "la" and "là" respectively, contrasting a pronoun with an adverb of place. The first word in each member is a verb; but the vowel that is part of the rhyme is contributed by different morphemes. Thus, the method proposed here can account not only for the perceptual effect—vigorous or tame—of rhymes, but also for its relative force.

Organising Effect and Immediate Appeal

Sweetser points out in the last act of the play, when Cyrano is about to die, instances of what she calls "intertextuality". I do not deny that in certain cases repeated rhyme pairs may juxtapose, so to speak, two passages far apart in the play, generating some ironic view, as in

ROXANE L'âme, c'était la vôtre!
CYRANO Je ne vous aimais pas.
ROXANE Vous m'aimiez!
CYRANO, se débattant C'était l'autre!
ROXANE Vous m'aimiez!
CYRANO, d'une voix qui faiblit Non !
ROXANE Déjà vous le dites plus bas!

This bas-pas rhyme echoes the final couplet of II.viii, where Le Bret is responding to Cyrano's non, merci!tirade against the world. The audience knows, though Le Bret does not, that Cyrano's intended love declaration to Roxane has just been thwarted by her revelation that she is in love with Christian. Eventually realizing that Cyrano's sudden extreme bitterness is really personal, despite his philosophical justifications, Le Bret says:

Fais tout haut l'orgueilleux et l'amer, mais, tout bas,
Dis-moi tout simplement qu'elle ne t'aime pas!
(Out loud [publicly], act the proud cynic – but
quietly [in confidence],
Just tell me she does
not love you.) (II.viii)

In Act V it is Cyrano and Roxane, using the same rhyming forms, who renegotiate the relationship between low-voiced privacy and denial of love.

I am not suggesting that something is wrong with the investigation of the organising effect of rhyme. I am merely suggesting that if you start straightaway with this effect, you miss very much, most notably the immediate appeal of the rhymes. There are some indications in Sweetser's paper that she did have strong intuitions regarding this immediate appeal. But "blending theory" fails to account for the rhymes' verbal structure or perceived effect, and directs attention away from the verbal structures to the contents. This led Sweetser to a brilliant comprehensive interpretation of the play, relating versification patterns to relatively large chunks of contents. In some of my recent publications I introduced the notion of "relative fine-grainedness" in critical discourse. Sweetser's discussion makes important observations on the play's structure. The critical tools introduced here allow the critic to fill it in with reference to more fine-grained texture. This is one of the great achievements of Cognitive Poetics as I conceive of it.

In her comment on the special issue of Language and Literature (on blending) in which Sweetser's article was published, Margaret Freeman makes a point that may illuminate my foregoing argument in a wider perspective:

As Reuven Tsur (1992, 2003) has cogently argued, blending as defined by Fauconnier and Turner is not the whole story in literary production. In fact, he makes a case for the disruption of the forces that blending purportedly exists to achieve. Cognitive stability and economy, clear-cut categorization, and human-scale reasoning are, he claims, precisely those aspects that poetic devices are designed to unsettle, blur or delay (Freeman, 2006: 111).

Theoretical Conclusion

Sweetser is working within a dichotomic conception based on form and content: "as Hiraga (2005) shows so clearly, blending theory lets us unpack the form–meaning relationships built up by rhyme and meter, giving us a language in which to start talking about poetic units and how they build meaning. This in turn opens the door for unpacking Rostand's particular uses of interaction between dialogue and poetic form". The present paper points out different "particular uses of interaction between dialogue and poetic form".

I am working within a theoretical framework that proposes to do away with the form–content distinction:

Things become even more disastrous for the traditional concepts when we realize that even in the language, commonly considered part of the form, it is necessary to distinguish between words in themselves, aesthetically indifferent, and the manner in which individual words make up units of sound and meaning, aestbetically effective. It would be better to rechristen all the aesthetically indifferent elements "materials," while the manner in which they acquire aesthetic efficacy may be called structure. This distinction is by no means a simple renaming of the old pair, content and form. It cuts right across the old boundary lines. "Materials" include elements formerly considered part of the content, and parts formerly considered formal. "Structure" is a concept including both content and form so far as they are organized for aesthetic purposes. The work of art is, then, considered as a whole system of signs, or structure of signs, serving a specific aesthetic purpose. (Wellek and Warren, 1956: 129)

Sweetser, or Hiraga for that matter, need the form–content dichotomy, because it enables them to talk about "iconicity", that is, to point out that one dimension of the text is similar in some respect to another. I prefer the Wellek and Warren distinction, because it allows for greater flexibility, and makes it possible to account for the emergence of a wide variety of perceptual qualities in a poetic text.

Contents, "projected world", word meanings, phonetic structure, metaphor, metre, rhyme, alliteration, are all materials (Wellek and Warren call them "norms"). Structures are the various combinations of these norms. Poetic effects arise from the subtle interaction of a great variety of norms. Configurations involving stressed syllables in weak positions, alliterative patterns occurring in consecutive syllables, run-on lines, the grouping of lines into asymmetrical or unpredictable structures e.g. by rhyme, and abstract nouns in a landscape defined here and now yield an exceptionally weak gestalt with an exceptionally strong emotional quality (such configurations are typical of Milton and Shelley, said to be the most musical poets in the English language). Configurations involving stressed syllables that occur only in strong positions and in all strong positions, alliterative patterns involving strong positions only (with one or more intervening positions), end-stopped lines, the grouping of lines into symmetrical, predictable structures, and objects with stable characteristic visual shapes in the world stratum of the poem yield an exceptionally strong gestalt, sometimes with an exceptionally strong witty quality. "Hypnotic" poetry typically involves exceptionally regular metre–stress mappings, alliterations of both kinds mentioned above, end-stopped lines, but unpredictable groupings of lines; and so forth. I usually treat these configurations under the heading of convergent and divergent poetry. Anti-grammatical rhymes typically generate a vigorous witty quality in a convergent context, and a vigorous emotional quality in a divergent or hypnotic context (in other words, they typically reinforce, even amplify, the gestalt qualities of the whole). Futhermore, the various configurations need not necessarily comprise homogeneous elements: in some poems, convergent elements serve to prevent a highly divergent poem from disintegrating. This may account for the observation that Milton's and Shelley's highly divergent poetry is typically perceived as musical and emotional, whereas Donne's more divergent poetry is typically perceived as harsh and witty.

This is not "iconicity", but "emergent gestalt qualities". "Iconicity" allows the critic to handle only those instances in which the similarity between form and content exists, or else compels him to read the similarity into them. According to the Wellek and Warren model a wide range of elements (which are independent variables) may occur in any combination, and thus it may serve to describe any unforeseen combination of elements in a poem. Any combination will display some gestalt quality which a poem may utilise or fail to utilise for aesthetic purposes; and gestalt theory may systematically account for the relationship between poetic structure and perceived qualities regularly attributed to them.
Finally, a sympathetic reader of this paper asked a provocative question that gives me the opportunity to end on a more general note.

What is your ultimate goal? Based on results, Sweetser's goal seems to be to produce a "good" interpretation or understanding of the play. Yours remains implicit, and maybe should be made explicit. I have always secretly thought of your work (of, for example, highlighting subtle prosodic effects) as ultimately providing better tools that can be applied, at the end of the day, to interpreting texts in a richer, more comprehensive way. Am I completely wrong?

In my answer I pointed out two ultimate goals. Sweetser and I share, I suppose, both these goals. First, much human research is directed to a better understanding of the world we live in, for its own sake. Just as we want to know how did our universe evolve from the Big Bang, for instance, or how did life emerge from inorganic matter, we also want to know how aesthetic qualities emerge from an aesthetically indifferent string of words. Secondly, a major goal of all aesthetic theory and criticism is to make a crucial recommendation as to what to look for in an aesthetic object, and how to look at it (cf. Weitz, 1962). Interpretation is one kind of activity toward this goal; my research on versification is another.


[1] My student Roi Tartakovsky commented on this paragraph: "Could one not argue, simply, that writing a play in verse in the late 19th century is highly conspicuous, and can potentially 'mean' different things. It has 'signifying potentiality' (my term). In the case of Cyrano, this potentiality is harnessed to evoking the 17th century, and in the other play—it simply does not." His notions "meaning different things" and "signifying potentiality" are well-taken; they have precedent in the theory of metaphor I embrace. The activation of "dormant features" or meaning potentials is the ground for understanding novel metaphors. I had, however, a second point too on this matter: Cyrano's versification is more similar to Romantic than to Classical tragedy. I don't think that putting all alexandrine couplets in one bin is very useful. More subtle distinctions are required. My main objection is, I guess, to jumping to conclusions on the basis of obtrusive but doubtful cues, while ignoring subtle and minimal cues. [back]

[2] In version (a), the caesura occurs after "themselves" in the first line, and after "Lovers" in the second. In the second line of (b), the caesura occurs after "Twelve". (In the iambic pentameter line, not as in the hexameter line, there is no point where it can be divided into two segments of equal length and equal structure; so, there is an "area of caesura", comprising positions 4, 5, 6). [back]

[3] "Anti-grammatical rhyme;s" are rhymes that oppose words belonging to different parts of speech, as sober~October in Poe's "Ulalume", or even sets of morphemes that belong to different parts of speech, such as Pope's in endless error hurl'd~and riddle of the world. [back]

[4] Sweetser makes here an excellent point: the abrupt interruption of the verse line and of Cyrano's fluent speech by the cadets' disjunct interjections is directly experienced as an immediate iconic representation of the disruption of the cadets' "dreamily homesick" mood. [back]


Chatman, Seymour (1960) "Comparing Metrical Styles", in Thomas Sebeok (ed)., Style in Language. Cambridge, Mass.: MIT. 149–172.

Fodor, J., A. and T. Bever (1965) "The Psychological Reality of Linguistic Segments". Journal of Verbal Learning and Verbal Behavior 4: 414–420.

Freeman, Margaret H, (2006) "Blending: A Response" Language and Literature 15(1): 107–117.
Available Online:

Garrett, M., T. Bever and J. A. Fodor (1966) "The Active Use of Grammar in Speech Perception". Perception and Psychophysics 1: 30–32.

Jakobson, Roman (1956) "Two Aspects of Language and Two Types of Aphasic Disturbances", in Roman Jakobson & Morris Halle, Fundamentals of Language. The Hague: Mouton. 55–82.

Rostand, Edmond. L'Aiglon
Cyrano de Bergerac

Available online:

Sweetser, Eve (2006) "Whose Rhyme is whose Reason?: Sound and Sense in Cyrano de Bergerac." Language and Literature. 29–54.

Tsur, Reuven (1972) "Articulateness and Requiredness in Iambic Verse". Style 6: 123–148.

Tsur, Reuven (1992) Toward a Theory of Cognitive Poetics. Amsterdam: Elsevier (North Holland) Science Publishers.

Tsur, Reuven (1998) Poetic Rhythm: Structure and Performance—An Empirical Study in Cognitive Poetics. Bern: Peter Lang. 378 pp.

Weitz, Morris (1962) "The Role of Theory in Aesthetics", in J. Margolis (ed.), Philosophy Looks at the Arts. New York: Scribner. 48–59.

Wimsatt, W. K. (1954) The Verbal Icon. New York: Noonday.

Back to home page
Back to "Occasional Papers"

Original file name: SweetserCompare - converted on Sunday, 17 December 2006, 13:00

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