Download PDF

On Poetry and Mind

[End Page 64]

[0]     What one cannot compute, thereof one must speak.

[1]     The mind happens to think, differently.

[1.0]     “The mind” is, at most, a conceptual index for where ideas, reflections, feelings, calculations, perceptions, or reasoning take place. It could also be understood as a repertoire for such operations, phenomena, or experiences. “The mind,” then, both refers to a set and a site. It seems difficult to take the mental set and site to be completely independent of each other.

[1.1]     One thinks. One is thinking. Does one think because one is thinking? The happening of thinking is both action and virtual inscription.

[1.1.0]     One may be a human, an animal, a machine. And all humans are animals. And all animals may be machines, though this is much less obvious. At any rate, “the mind” is not—or not exclusively—the kind of computing machine we now know, or think, of.

[1.1.1]     One may be more than one. Even minimal mental operations already prepare the bifurcation we witness each time we consider our ways of thinking, or ourselves, or what “each time” could mean, and could refer to.

[1.2]     A mind is more than a brain. It includes an embodied neural system that should never be blindly identified with one specific organ. Moreover, the brain itself is not exclusively nor primarily cortical: the “lower regions,” the more inner and evolutionarily ancient parts, also give us emotions, balance, and memories. Minds, then, are not wholly encapsulated in the heads of individuals. They are routinely distributed through the emergence of group thinking, partially externalized to blocks within their environments (mnemonic objects and loci, papers and pens, books, laptops, and other prostheses), and potentially changed through exchange. The centripetal force of processes leading (back) to neural appropriation cannot conceal the concurrent and centrifugal impetus of minds.

[1.3]     Thinking is not reduced to thought.

[1.3.0]     Let us use the word and category of cognition for mental operations that could be automated and produced according to common rules. While cognitive processes might fail, they seek equilibrium, thus tending to be relatively durable, consistent, and transparent. Instituted, organized knowledge (and primarily “science”) favors the cognitive regime of thought in the fabric of its own theories and practices.

[1.3.1]     Intellection would refer to the differential and variable performance of thinking. Being created ad hoc, it is transient, and often singular, thus resisting automation. The cognitive defects of mental acts—when they lack stability, avoid [End Page 65] standard consistence, arise through opacity, or become “blocked”—strengthenthe advent, and imprint, of intellection.

[1.3.2]     The cognitive regime tries to enfold thought upon itself; the intellective attempts to unfold it through thinking.

[1.3.3]     The intellective stems from the noise of noēsis.

[1.3.4]     The intellective space is where a mind might seek to go beyond the limitations of computability, of decidability.

[1.3.5]     If cognition emerges from the properties of the mind, intellection emerges from emergence and is doomed to be further transcribed at an operational level. But transcription is not an equation.

[1.4]     The mind is multi-dimensional. It is not “wrong” to focus on just one dimension rather than adopting a broader view: it is “poor,” deliberately simplifying, or congruent to specific (pedagogic, strategic, epistemic) goals.

[1.4.0]     A zero-dimensional mind is an ideal, immaterial point. At the extreme tip of methodic doubt, the Cartesian res cogitans (i.e., “the thing that is thinking”) might have no body, no world, no god, and no place to be; it works as the very intent of thought. This contrasts with the “extended thing” (res extensa), that localizable body I may persuade myself to be living in. The image of the mind as a point may have never been other than some intended—transitional—fiction. Similarly, the first cognitive paradigm was essentially downplaying the material role of the “hardware” in the advent of thought.

[1.4.1]     More recent propositions have added one axis and suggested to approach the mind within the extent of its necessary embodiment.

[1.4.2]     The “extended mind hypothesis” is bi-dimensional and addresses the ecological situation of intelligent agents, downloading parts of their noetic functions to a “peripheral” apparatus.

[1.4.3]     The intellective share hints at extension beyond extendedness, through the eventful co-elaboration of transient thinking.

[1.4.4]     The fourth dimension of time deals with the gradual aspect of ideas (being assembled or constructed, appearing and sinking) and also with the noetic ten-ability of what is “the same time” to us. I say: noetic time is tense to us.

[1.5]     Human verbal language is a very powerful organon that makes us think differently.

[1.5.0]     Human verbal language guarantees cognitive routines through syntactic rules. It disseminates knowledge, it emphasizes logical sets, it enforces prescriptions. Words create zones of convergence and attractor basins for ideations that order [End Page 66] our stream of thoughts. Attention disorders in humans accompany weakened verbal abilities; conversely, other “languaged apes” (like members of the Pan species) reach a new mental level that is not confined to added possibilities for expression. As a matter of fact, the bonobos who are endowed with human language do not speak much in general—but they think more, and further.

[1.5.1]     The paradox is this: because human speech has to be granular, open-ended, flexible, and opaque enough to lend itself to possibly unlimited and unexpected usages, because it appears as a proliferating and largely autonomous entity, it is also the source for a lot of “noise” within communication. It operates and it also differentially performs. It “bootstraps” cognition, while betraying the cognitive. It introduces both more and less than we could say or think (or say we think, or think we say, or think we say we think, etc.).

[]     The century we are finally leaving was obsessed with this paradox. The novelty was the interdisciplinary obsession—certainly not the problem itself—as well as the energy deployed to comment on or lament those difficulties, to deepen or erase them.

[]     The practical success of information science—through its major recourse to binary logic and the legacy of founding figures like Alan Turing—constantly invites us to feel the inadequacy of “natural languages” (compared to the code), while promoting an ideal of computability that is neglecting the consequences of the halting problem or of constant Ω. Even from a mathematical viewpoint, the serial identifications of noēsis with thought, of thought with cognition, of cognition with algorithmic processes should lead us to envisage the blind spots, defaults, and incapacities we have to cope with. Thus, computers, as such, do not prove we are Turing machines: they may have some inabilities that are also common to human minds, and more impressive reliability or consistence. But, so far, once they have to deal with the impossible or the unthinkable, they do not know what to do next. Or they select some probable scenario in a sort of prearranged list. Humans often act the same. But they can also bypass their limitations through speculation, intellection, creation.

[1.5.2]     Policing language or vaunting absolute free style will neither completely amend the cognitive pitfalls of speech nor liberate our spontaneous genius once and for all. It is through the iterable performance of cognition that intellection happens. The rules and disciplines of human language may be the unexpected paths leading to the intellective space.

[1.6]     Novel ideas arise through trials and errors, or through invention and recombination. And, one step further, they may be created, when one perceives the abyss and leaps. Creation will need to be cognized, and somehow enfolded, which might even encompass automated or “bureaucratic” mental operations. [End Page 67] However, what made the latter possible derives from a very dramatic move into an unexpected beyond, and back.

[1.7]     Poiēsis is a name I retain to speak of creative mental responses to the uncomputable, especially as they pertain to intellection with language.

[1.8]     Poetry is a verbal art that both extensively explores and shapes intensively the potentialities of poiēsis.

[1.8.0]     Poetry is thinking. Thus, it is not pre-verbal, pre-rational, pre-logical. Percepts and emotions are there, though not as adversaries, or rivals, to the res cogitans.

[1.8.1]     All intellection is not poetic.

[1.8.2]     The whole of poetry is not intellective, but the mental experience of poetic speech events relies on noetic extension. Poetry changes our minds.

[]     If we intend to discuss what a poem allows us to fancy, feel, believe, or formulate, by simply “skipping” the prosodic aspects and directly branching ourselves with some “higher content” (be it conceptual, social-historical, moral, or “theoretical”), we may end up delivering flat or one-sided readings. Conversely, once it is acknowledged, the excess of and to cognition demands from us great care in situating the regulations and calculations a text brings forward—and in describing how those rules are being done, undone, and abandoned. In this we may become capable of reclaiming for ourselves something of the conditions of creation we have just retraced.

[]     A banal misconception is technicism. In his Poetics, Aristotle criticized those who describe poetry as the making of meters.1 Almost two and a half millennia later, we unfortunately have to deal with the same mono-dimensional prejudice (in the generative description of metrics by Nigel Fabb and Morris Halle, with Jacques Roubaud’s arithmetic understanding of the French verse, or in very traditional and “grammatical” taxonomic lists of the dos and don’ts of versification). Here, the technique of the verse helps promote a cognitive reification of the poem: our sight is narrowed, a means becomes an end.

[]     The intersection of cognitive science and literary studies holds promise, if and only if: we do not entrap the mental in the algorithmic (especially the deterministic), we recognize in poiēsis the possibility of disrupting the common order of thoughts, we accept to be touched in return by what we interpret.

[]     The average, standard, historical, or ideal readers are ghosts that literary scholars live with. A typical reader’s brain, as deciphered with fMRI or MEG machines, is just another specter. “Neuro-humanists” who would condemn themselves to overlap the scans of the poem’s readers would not deal with anything more than the common, as long as they lack a theory for understanding differences and idiosyncrasies. [End Page 68]

[]     In the same vein, the “big data” of digital humanities plead in effect for the computational. Oddly enough, many defenders of those new tools do not perceive the connection between their own “mining” and the algorithmic rendition of cognition. Furthermore, “unpredictability, complexity, and abrupt shifts over time”2 are notoriously difficult to handle with vast amounts of information—and those three traits, precisely, are also strongly tied to poetic texts. This epistemic obstacle, once coupled with the discontinuity regarding the status of the algorithmic (from the “data” to “the brain”), logically expels, from research itself, the rich mental experience of poetry reading.

[]     To say it differently, what is happening to us with poems challenges what we believe we know about cognition. It eloquently shows that there are more than rules and operations in thinking, but also that such excess can only derive—and derail—from routines, patterns, and automation.

[1.9]     The literary performance of the form is informed by its mental transform.

[2]     Poetry could be qualified via its mental—cognitive, intellective—experience.

[2.0]     Uniqueness is gauche conceptual framing. I qualify the noetic existence of poetry. I am not defining it. I am not installing an in/out gate. I am not even giving a series of traits that should be reserved for the poetic. Specificity is built through the singular assemblage of non-unique features. It does not require a fixed number of features (all of them, most of them . . .) to crystallize.

[2.1]     Poetry is a noetic and discursive mode that is widespread—and possibly universal—among human societies.

[2.1.0]     Let us consider poetry as discourse, and not as the inherent quality of things that some texts would channel. The “poetry” of a landscape converts the awe into words into awe. As for the fact of language, it is as much before as in us. A poem is a cosa mentale.

[2.1.1]     Cultural relativism will not do, neither will historicism. The local configuration of entities cannot preclude by principle the promise of the universal. True, poetry is not Dichtung, Dichtung is not shi, and shi not ci. But who would believe that poésie in Antonin Artaud and poésie in Nicolas Boileau have the exact same meaning? or that chronologically contemporary authors such as Charles Baudelaire and Victor Hugo are in complete agreement over the definition of a poem? Here, we primarily encounter the banal—though still troubling—issue of signification. The words of poetry allow us to suggest some meaning that will irreducibly be flexible and revisable.

[2.1.2]     It seems that in most—if not all—of the human societies we have a linguistic record of, something like poetry is to be found. Whereas this characterization is [End Page 69] proleptic (the “something like” corresponds to the qualification we are beginning to detail), it does not entail widely divergent practices standardized in advance, under one label, or for the argument’s sake. The plurality of poetry— realized in songs, dances, psalmodies, in thousands of lines or brief forms, acoustic or visual—is a mark of its differential performalism. We shall do justice to the widespread and longstanding existence of poetic modes in human lives, if we are able to reflect scholarly on a scale that would be more than global. Predetermined conceptual unification is inadequate; comparison is not enough. Innovative inquiries on pervasive literary categories will come from both broad and sound examination of a multitude of different traditions and languages.

[2.1.3]     The human mind is not a “blank slate,” it is not unbounded by biological constraints. Nor is it the sole output of a genetic software. It heavily depends on individual development, on the features of the anthropic worlds it evolves into, on the noetic prostheses it can rely upon. Like other animals, our capabilities out-act our capacities. This is why the so-called symbolic revolution in Homo sapiens (probably including the advent of fluent, organized speech) would have appeared dozens of millennia after speciation, and after our ancestors were endowed with “our” “modern” anatomical brain. In the historical timeline, heritable changes in the biological makeup of the central nervous system are either genetic and quite slow (compared to our own life span), or epigenetic and more circumscribed or fragile. Thus, the mind is configured by widely different conditions of existence. “Nature” in us encompasses the given as well as the physical determinations of evolving beings, which ultimately govern “growth,” plasticity, and expression. There are limits to what we can do, achieve, and become. But the phusis of “culture” shows that ordered and meaningful symbolic networks outstretch seemingly “natural” capacities. Hence, arguing in favor of poetry as a biological adaptation really falls short. The omnipresence of poetry among humans is less a “fact of nature” than a second-order consequence of the physical commonalities between our worlds, endowments, and tools. Furthermore, the experience of poetry, in its own turn, alters our minds—or contributes to shift the moving barriers of our noetic capabilities.

[2.2]     Poetry is both internally and externally contrasted with other—ordinary—modes of thought and speech acts.

[2.2.0]     A poem is extra-ordinary and is revealed as such. Even “found poetry,” made of duplicated fragments of heterogeneous texts, stands in contrast with ordinary delivery, through selection, cut, montage; it is an extreme case of what I call “literary response.” More often than not, poetic œuvres are recited or typographically presented in a particular manner. In oral tales from Vanuatu or the Marshall Islands, poetic texts arise as songs, while using another vocabulary, sometimes ancient languages. Stéphane Mallarmé’s groundbreaking conception [End Page 70] of the book page in Un coup de dés exemplifies the same need for obvious contrasts. All of this is convention and artifice, until some sense is made out of it. It remains that the frequent recourse to strong markers of differences—all the more so, if they are readily identifiable—advertises the singular modes of poetry, and “primes” us.

[2.3]     Poetry is enacted ad hoc, through discrete textual objects.

[2.3.0]     Like literature, poetry is both singulative and collective. Each work is able to modify our previous understanding of poetry. Singulative variation is the dual product of writing and interpretation, in a particular time and space. Without the supplementary and collective possibility of recognizing poetry, we would isolate self-defining speech events. Qualifying shared traits without conducting closer readings (as I do in this article) is, by function, a partial endeavor and is intended to be complementarily displaced by ad hoc propositions. It may go without saying that the features I am giving here depend on the examination of (many) discrete texts.

[2.4]     Poetry is differentially performed, with varying degrees of competence.

[2.4.0]     “Performance” is deliberately broad, and includes the use of the voice, prosody, gestures, rhythms, signs, words, cognition. Attention to the delivery is articulated with the scrutiny of language events and care for the emergence of thought.

[2.4.1]     Instead of seeing the differential as a halo, we can, here too, speak of capabilities. What the poem says is not what most readers (even with comparable backgrounds) agree that it tells: it is also everything that could be expressed about it. I am myself a mutable interpreter, whose discourse is conditioned by the extension of the piece I am analyzing.

[2.4.2]     There are gradients of competence in all aspects of poetic performance. Both subjective reports and recent fMRI experiments attest that different manners of reading have vastly different neural signatures.3 Some texts require a high level of practice for each aspect of their performance. For instance, reading aloud— and in Italian—the poems of the Canzoniere demands uncommon verbal agility, due to their plethoric synaloephas. This is in line with the overall Petrarchan poetics and the sophistication of its artifacts.

[2.5]     Poetry maximizes and dynamizes semantics.

[2.5.0]     Once we have been primed by the poetic dispositif, we conventionally enter a maximal semantic world. The slightest element—a comma, an otherwise negligible recurrence of phonemes, a banal ambiguity, an odd turn of phrase—could now make sense to us, in interrelation with the rest of the poem, and other parts of the corpus we select. Meaninglessness is always of second degree: it is not a [End Page 71] semantic lack, it is not “without any sense,” it is the intense subtraction of the maximal potentiality to mean. Asemantic poetry does not exist; only signification of meaningless could be reached, as with glossolalia.

[2.5.1]     In the wake of Russian formalism, Jan Mukařovský rightly elaborated on dynamic semantics and their psychological nature.4 Maxima go with higher velocity. Associations of words, changes of directions happen more, and more frequently, in poems than in ordinary regimes of speech. Our minds may have to refocus several times over the course of a sonnet. Multi-stable meanings are being mobilized, and we oscillate between different strange attractors, never completely finishing course. We do not exhaust the meaning of a poem, we are exhausted by it.

[2.5.2]     Were natural languages unambiguous, transparent, and perfect, then semantic frail extensibility and rapidity would have little appeal.

[2.6]     Poetry is a linguistic construct that also operates in more than “language(s).”

[2.6.0]     In his Rhetoric, Aristotle states that verbal art needs to make language sound both strange and foreign (xenikon).5 Such a duty—later dubbed in Russian ostranenie— appears all the more crucial for poetry, and has often been noticed. Readymade sentences, established connections between semantemes, are either poetically refurbished or rejected. We habituate ourselves to what has been frequently said and thought. Mental path dependency is being challenged by all levels of defamiliarization: metaphorical networks, unusual word order, morphological alterations, vocables . . . There is not exactly a wealth of experiments on the processing of strangeness in language, and observers commonly try to identify very traditional rhetorical objects, with more or less success. Though available results are quite limited, they tend to confirm that stylistic innovation requires more time and effort to be processed than chunks of standard prose, but also— and more interestingly—they confirm the existence of thresholds. Widespread “metaphors” are conventionalized and treated by the brain like other lexical elements, while others, perceived to be more “poetic,” are treated differently, as integrable but exotic objects.6

[2.6.1]     Sometimes the foreign tongue of poetry is indeed foreign, like, for centuries, Chinese in “medieval” Japan, or Sanskrit in vast areas of Southeast Asia. As for juxtaposition, composition, and hybridization of different languages, they are no modernist invention either.

[2.6.2]     The musicality of poetry, concreted through alliteration, rhythm, diction, rhymes, and prosodic reorganization, is often underscored by the adjunction of musical instruments and of the singing voice. To use Aniruddh Patel’s terminology, the “representation networks” of the human mind for language and [End Page 72] for music are distinct, but their “resource networks” do overlap.7 Tecumseh Fitch suggests that this superposition—coupled with other structural commonalities as revealed by poetic practice, whether sung or not—hints at a kind of musical, or primarily prosodic, proto-language in the ancestry of Homo.8 To me, musicality is a high point on the curve. An intrinsically linguistic construct, poetry bypasses the boundaries of language “as such,” through the means of a verbal art.

[2.7]     Poetry uses imagery, especially in deploying the rhetorical armature of speech.

[2.7.0]     The coinage of strong, powerful images may be the most charming and subjective aspect of poetry. I see, I hear, I feel things that a text evokes. Is this suggestion? From Arthur Rimbaud to imagism and surrealism, modern Western poetry certainly recaptured a picturesque and hallucinatory force that predated it. Future experiments will inevitably show that, at least for some of us, perception of alternate realities through poetry is as absorbing and dense (and maybe more) than immersion in movies. For literary criticism, though, images are difficult to grasp. Their qualia is what gives them value; their explanation in theoretical jargon is largely adventitious.

[2.7.1]     Images, then, are epiphenomena tied to word constructs—taken as a differential recording of the poet’s mind in the moment of composition, and as cognitive-linguistic infrastructure for the readers.

[2.7.2]     Tropes pave the way toward imagery. “Blending,” as developed by Gilles Fauconnier and Mark Turner, allows a positive description for the generation of the metaphorical in general. It has been accepted for its simplicity and heuristics. But the theory has many pitfalls. The selection of the traits to be “blended” from each mental space is stunningly arbitrary, and actually far from its scientific objective. And the emphasis on equilibrium tends to conventionalize—and tame—all metaphors or associations. In poems, instability fuels development. When Paul Celan writes “wir schlafen wie Wein in den Muscheln” (we’re sleeping like wine in seashells), complete blending would merely trivialize the adunaton, and would not prepare the reader to interpret the additional connections that the poem is making (through a series of supplementary comparisons and alliterative linkages).9 Furthermore, taking the rhetorical for the literary, as Turner does in The Literary Mind, is a far too exiguous conception for what we seek to establish.

[2.8]     Poetry is a priori regulated through iterative systems.

[2.8.0]     “A priori regulation” does not stand for univocal and absolute norms. Rules could be loosely or strictly enforced. They may be common to practitioners and overly binding, determining the number of counted units, the use of some words, or even elements of thematic content (as in seventeenth-century [End Page 73] Japanese haiku). They may be created or recruited “on the fly,” in free verse, where they constitute evanescent units of organization, whose destiny is to vanish as rapidly as they emerge. Ezra Pound’s Cantos are a masterpiece structured by the unpredictable recourse to ephemeral micro-regulations of all kinds. Incidentally, free verse implies freedom to choose and change rules at will. The pace of those shifts is sometimes so quick and so subtle that liberation from templates is actual, although there is still no absolute liberty. I have a hard time finding poems that would not be even minimally regulated, with the potential exception of some prose poems (especially those by Baudelaire). “Poetic prose” in general is heavily invested in repetition of sounds, words, tropes, and motifs (think of Emily Brontë’s Wuthering Heights, Lu Xun’s Weeds, or François-René de Chateaubriand’s Memoirs).

[2.8.1]     Iterative systems are based on metrical units and/or repetitions.

[]     Measures (or meters) are a staple of many prominent and intercultural systems of poetic regulations like the Arabo-Persian model, whose linguistic, historical, and geographic diversification is impressive. In the Indo-European group, the counting of syllables and/or feet (determined by quantity and/or stress) is so dominant in poetry that most current textbooks, encyclopedias, or “universal” descriptions that are available still suppose metrical laws to be the only, or the best, ones. In our supposedly post- colonial times, the direct conclusion (“without meter, there is no poetry”) is eschewed, though the indirect inference (“poetic lines are metrical”) survives.

Metrics, like rhetoric or performance, will be enrolled in a qualification of poetry, but not as scientia scientiarum. In addition, local revisions are needed. Morae, words, letters, and tones, for instance, could be counted and ordered; lines, rhymes, and stanzas also form super-units. By regarding the variety of items that are being measured, one easily realizes that one solution, in one language and one era, might be favored, but that meters are mutable and additive. A French alexandrine is classically produced by a pattern of six plus six syllables. However, stresses—that, contrary to a tenacious legend, both exist and move, in French—tones—taken at the level of word groups—and vocalic quantities contribute to the fabric of the verse. “Mixed” metrical sets (the accentual-syllabic form of Ancient Aeolic, Renaissance Spanish, and Romantic British lyrics) enforce rules with greater insistence, but, overall, any system of measure is able to use other standards, or even to convert itself. There are deep links between the techniques used by native speakers (from infancy) for speech segmentation in their mother tongue and the favored type of measures. In order to locate the beginning and end of words in a flow of speech most native speakers of Japanese use morae, while speakers of French rely on syllables, and speakers of English use stresses. The traditional poetic meters of those languages are congruent to the “native” linguistic strategies.10 At the same time, under certain—social-historical—conditions, [End Page 74] a foreign (and reputedly “unnatural”) system could be adopted, as when Romans shifted away from their Saturnian stress-based verses and appropriated a Greek quantitative metric, or when Xi Murong adopts classical Mongolian versification (based on alliterative lines with parallelism) in her Chinese writings.

[]     Measures mark time, they come and go, and come again. Repetition (with and without variation) is an important principle for poetic regulation that operates at many possible levels, from individual phonemes to larger ensembles. Parallelisms of tropes, images, themes, phrases, and syntactic structures are, in the absence of detected meters, what constitute for modern readers the most perceptible regularities of Biblical songs or Quechua laments. Almost endless cumulation in the encomiastic genre of Xhosa izibongo looks like progressive parallelism. Line breaks (and the internal presence of caesuras) make silence itself the universal object of poetic repetition. Reiteration of sounds is widespread across languages, but sometimes loosely structured: alliterative groups are “on demand” in most modern European languages, but essential in the old Germanic practice, or scarce in classical Arabic texts; rhymes exist in Vietnamese, Chinese, and Swahili, but not in ancient Greek or Hebrew.

The young Gerard Manley Hopkins stated that “the artificial part of poetry. . . reduces itself to the principle of parallelism,” an idea that Roman Jakobson expounded.11 We can advance that the temporal unfolding of the poem is regulated through recurrences, repetitions, and variations, based on standardized or more idiosyncratic units.

[]     The iterative systems of poetry, in all their artificiality, provide the complex cognitive armature that fosters intellective displacement.

[2.9]     Poetry is a priori regulated through speech reorganization.

[2.9.0]     In many instances, defamiliarization partakes in regulation. It is very rare for taxonomic descriptions of poetry to mention speech reorganization. The phenomenon, far from being absent from the modern European corpus, is perhaps more powerful—or more visibly so—in corpora whose metrics and iterative laws are obscure or relaxed. Here too, a combination of rules is to be expected.

[2.9.1]     High iconicity is sometimes a facultative enhancement. Whereas “imitative harmonies” or rhythmic enactments (speeding up, slowing down, walking, galloping) in Greek or Latin verses have often been disregarded since John Dryden, they resemble a quasi-system. Sound iconicity is essential in the poetry of several American nations; onomatopoeias and ideophones abound in Navajo and Hopi songs. Visual iconicity is the root of modern concrete poetry, and by no means a unicum. Often combined with repetition (from Guillaume Apollinaire’s rainfall in “Il pleut” to Reinhard Döhl’s “Apfel”), visual iconicity is extremely [End Page 75] relevant in traditions using calligraphy or in any performance where body language sustains semantic elaboration—as in Asian traditional theaters, or in Aboriginal and multimedial “sand stories.” Contemporary sign language poetry is another, striking, mobilization of visual iconicity.

[2.9.2]     Requirements in the alteration (oft called “elevation”) of language offer another set of norms. Lending and borrowing from Pali and Sanskrit have long been the rule in Khmer poetry. Many European languages excluded parts of their lexicon from verses, and they reserved portions of their vocabulary for poetic expression. Archaisms are mandatory in most, if not all, songs of the Vanuatu, or those of Brazilian Suyás. Word compounds—kennings in Old Norse—have equivalents in Aztec practices or among the Aranda people. Morphological and syntactic conventions that would not be allowed by native speakers in unmarked use of their tongues are tolerated or even required in poetic speech. Rare terms and neologisms are not incidental for the many writers of free verse: along with repetition, they participate in micro-regulation. Significantly, Aimé Césaire’s A Notebook of a Return to the Native Land ends with “verrition,” an opaque hapax that culminates the constant call for uncommon terms throughout the previous dozens of pages.

[2.10]     Poetry tends to exceed its own algorithms.

[2.10.0]     Poetic regulations (both iteration and reorganization) frame cognition differently. It is through an exceptional re-automation of language—and of its underlying noetic structures—that higher-order de-automation is to be achieved.

[2.10.1]     The “algorithmic” dimension of poetry is dual. One lies at the perceptual level of stimulus processing, when our brains, say, compress alternations of short and long vowels or anticipate the accents of a line. The other deals especially with the “writing program” of fixed forms. The Malay listeners at a pantun recitation expect the sounds of the words at end lines to reappear, and they surmise that the second part of the stanza will echo the first half, both phonetically and thematically. The poetics of “wax and gold” used in Ethiopia encourage Amharic readers to look for the pivotal word that will allow the double entendre of the text. Theoretically, though not presently, algorithmic comprehension could be formalized and computed by machines.

[2.10.2]     This first leaves completely open the question of the algorithm that produces the algorithm (the algorithm that would have “computed” the sonnet at the cosmopolitan court of Frederick II in thirteenth-century Sicily, or the one that would have allowed Gertrude Stein to come up with her own): is it non-deterministic? is there any such algorithm? Secondly, there may be as much joy and satisfaction in feeling that the “template” is realized than in hearing it is not. Jean Racine’s tragedies operate within a mostly regular and melodic distribution of stresses in [End Page 76] each line, but when several verses in a row have the same pattern, they generally give way to other structures, in an unexpected fashion (for how and when). Moreover, strong and abrupt, basically improbable, sequences of stressed and unstressed syllables appear here and there, often due to the expression of passion. Our interest is intact, even higher in those moments. The “regular” algorithm (within the “software” of versification) is done, or half-done, or undone. We like all aspects together. The art is both in the program and in its defection.

[2.11]     Poetry is complex and highly incompressible.

[2.11.0]     While the core unexpectedness of poetic texts needs to be appreciated in contrast with normativity, it defeats in advance any attempt to completely contain it. To borrow a concept from information science and mathematics, we believe poems to be highly complex. Very repetitive lyrics could be “compressed,” though the effects (and affects) of saying almost twenty times “Touch me” over the course of one minute outreaches the notation of the same words followed by the number of utterances.12 Similar remarks could be formed on self-commenting words such as TV On The Radio’s ad lib “My repetition: my repetition is this.”13 Any description that is shorter than the original will simplify the latter or give us a “lossy output” from which we will not be able to satisfyingly retrieve the “input.”

[2.11.1]     Conversely, what makes a poem what it is, with its sounds and rhythms, with its intricate composition and dense allusions, fosters explication. Paraphrase and commentary are noticeably longer and more explicit than what they are applied to. In exposing the artifice, we tacitly admit that any attempt to “unpack things” will still be unable to address the full spectrum of interpretations, and will lag behind the “economy” of the poem.

[2.11.2]     Poetry (from short forms to long epics) is severely compressed, which makes it hardly incompressible. Translation makes this palpable. It was certainly logical for hardcore partisans of artificial intelligence in the 1950s onward to posit the automated translation of poems as a benchmark.

[2.11.3]     The hermetic can only haunt poetry.

[2.12]     Poetry incorporates cognitive dissonance and draws on non-consistent logical reasoning.

[2.12.0]     Cognitive dissonance, or any strong impression of impossibility in the face of what a poem is stating, is induced by the performative disposition of a complex verbal artifact. The movement of de-automation thanks to re-automation only adds to this, as does the status of fictional literary objects. Incompatibilities, contradictions, tensions are otherwise transformed into notions that many individual poems work through. Intellective contradictions or incompossibilities [End Page 77] are neither “explosive” nor “trivial.” They are trans-consistent, beyond what even the symbols of non-standard formalized logics are apt to point at. Poetic contradictions are not always lies, licenses, errors, or episodes of senselessness; they become significant, as we think and embody them.

[2.13]     Poetry is aptly recursive and reflexive.

[2.13.0]     For being a catchword in debates about human and animal cognition, “recursion” is a rather proliferating and multivalent concept. If we recognize in recursion the need for multiple layers, some level of self-similarity, and the drive toward further replications of the current structure, then we see that the metatextuality of literary theory is a recursive process. Poems that comment on themselves, draw attention to their artificiality, include references to writing, speaking and reading, summon characters that versify, sing, or talk all could be found in the traditions of all continents, and across millennia. Texts are embedded within themselves and the depiction of their creation or reception. We are caught in strange loops. A sentiment for artistic achievement is a consequence of the cognitive disorientation we go through, and of the intellective challenge we encounter.

[2.14]     Poetry creates detachable, movable, and livable significations for psychological selves.

[2.14.0]     As our selves are shaped by discourse, we are altered when we accept in us those words that could be ours, and that de facto are and are not ours. To what extent are we altered? This is difficult to say: some more than others, depending on the situation, and on the text. So many poems are known by heart that they live with, in, among us. We neutralize their force by habituating ourselves until, one day, we hear them chanting anew. Fragments of poems are being detached, and rearranged. They enter our memory, they are our memories. They move us. An elegy signifies to me when the no-sense of life and death corrodes me. I may be moved to tears by what is beyond me, and transiently think, perceive, exist in the space of a poem.

[2.15]     Poetry suggests “thinking experiments” for the unthinkable.

[2.15.0]     So much has been written in a century on Gedankenexperiment. We gain ideas by speculating. As Ernst Mach once suggested, literature is filled with “thought experiments,” concerning all the things we do not know yet, or will ever ignore.14

[2.15.1]     Above this, the best poems are thinking experiments for the unthinkable. Along with other noetic excursions outside of the computable range, they suggest to us ways to think beyond the limits of thought.

[3]     What one cannot compute, one must poetize, and think. [End Page 78]

Laurent Dubreuil

Laurent Dubreuil is editor of diacritics. Among his books are The Empire of Language (2013), The Intellective Space (2015), and The Refusal of Politics (2016).


This article is based on the first two-thirds of the University Lecture I gave at Beijing’s Tsinghua University in April 2015. In the spring of 2015, I taught a graduate seminar on “Poetry and Mind” at Cornell University (and, in a condensed form, at Tsinghua). I want to warmly thank all the participants and members of the audience with whom I had the pleasure to exchange on the issues I am discussing here. The weekly discussions in my seminar were essential to the elaboration of this piece. For some specific but key aspects of what I am presenting here, I also greatly benefitted from further conversations with Laurent Ferri, Tom McEnaney, Sue Savage-Rumbaugh, Dagmawi Woubshet, and Yan Haiping, my gracious host at Tsinghua. I also thank Tristan Garcia, Brian Lennon, and Thomas Pavel, who provided important comments on the previous versions of this text. This article is intended to serve as a short introduction to my ongoing—and wider—research project on the mental experience of poetry.

1. Aristotle, Poetics, 1447b.

2. I am quoting Gary Marcus, “Steamrolled by Big Data.”

3. See, for instance, Natalie Phillips, “Literary Neuroscience and History of Mind: An Interdisciplinary fMRI Study of Attention and Jane Austen.”

5. Aristotle, Rhetoric, 3.2.6.

6. See, e.g., N. Mashal et al., “An fMRI Investigation of the Neural Correlates Underlying the Processing of Novel Metaphoric Expressions”; and M. Faust and N. Mashal, “The Role of the Right Cerebral Hemisphere in Processing Novel Metaphoric Expressions Taken from Poetry.”

8. Fitch, “The Biology and Evolution of Rhythm: Unravelling a Paradox.”

9. Celan, “Corona,” in Mohn und Gedächtnis, 33.

12. As Martin Gore sings in “Fly on the Wind-screen” on Depeche Mode’s Black Celebration (Mute Records, 1986).

13. “Repetition,” by TV On The Radio, on Nine Types of Light (Interscope, 2011).

14. See Mach’s brief remark about “the novelist” and Gedankenexperiment in Knowledge and Error, 136. [End Page 79]

Works Cited

Celan, Paul. Mohn und Gedächtnis. Stuttgart and Munich: Deutsche Verlags-Anstalt, 2000.
Cutler, Anne. Native Listening: Language Experience and the Recognition of Spoken Words. Cambridge: MIT Press, 2012.
Faust, M., and N. Mashal. “The Role of the Right Cerebral Hemisphere in Processing Novel Metaphoric Expressions Taken from Poetry: A Divided Visual Field Study.” Neuropsychologia 45, no. 4 (2007): 860–70.
Fitch, W. Tecumseh. “The Biology and Evolution of Rhythm: Unravelling a Paradox.” In Language and Music as Cognitive Systems, edited by Patrick Rebuschat, Martin Rohmeier, John A. Hawkins, and Ian Cross, 73–95. Oxford: Oxford University Press, 2012.
Hopkins, Gerard Manley. The Journals and Papers of Gerard Manley Hopkins. Edited by Humphry House, with Graham Storey. London: Oxford University Press, 1959.
Jakobson, Roman. Selected Writings. 8 vols. Edited by Stephen Rudy. The Hague: Mouton, 1971–87.
Mach, Ernst. Knowledge and Error: Sketches on the Psychology of Enquiry. Translated by Paul Foulkes. Dordrecht: D. Reidel, 1976.
Marcus, Gary. “Steamrolled by Big Data.” The New Yorker. March 29, 2013.
Mashal, N., M. Faust, T. Hendler, and M. Jung-Beeman. “An fMRI Investigation of the Neural Correlates Underlying the Processing of Novel Metaphoric Expressions.” Brain and Language 100, no. 2 (2007): 115–26.
Mukařovský, Jan. On Poetic Language. Translated by John Burbank and Peter Steiner. Lisse, Netherlands: Peter de Ridder, 1976. Patel, Aniruddh D. Music, Language, and the Brain. Oxford: Oxford University Press, 2008.
Phillips, Natalie M. “Literary Neuroscience and History of Mind: An Interdisciplinary fMRI Study of Attention and Jane Austen.” In The Oxford Handbook of Cognitive Literary Studies, edited by Lisa Zunshine, 55–81. Oxford: Oxford University Press, 2015.
Turner, Mark. The Literary Mind: The Origins of Thought and Language. New York: Oxford University Press, 1996. [End Page 80]