Beyond Pure Reason: Ferdinand de Saussure’s Philosophy of Language and Its Early Romantic Antecedents

Saussure’s first brush with the subject apparently came during his stay in Rome in the winter semester of 1906. We remember that when Saussure took a leave from his teaching in Paris in the early 1890s he used the free time for a trip to Lithuania, which proved to be instrumental in shaping his views on the heterogeneity of the historical development of languages. Fifteen years later, Saussure once again took a leave from his duties to travel. This time, however, it was supposed to be a pleasure trip for Saussure and his wife, the only lengthy excursion abroad Saussure made during his Geneva years. And yet this respite from his rather monotonous academic routine resulted once more in a dramatic turn in Saussure’s intellectual interests, with far-reaching consequences.

In a letter to Meillet from Rome in January, Saussure mentioned—perhaps responding to worries about his health—that he was keeping himself busy at the moment with a matter of no particular consequence: just amusing himself by reading archaic inscriptions during strolls in the Roman Forum, with the sole purpose of proving to himself that he had not yet lost the taste for occasionally racking his brains (casser la tête) over a linguistic puzzle. The damaged state of the inscriptions, together with their elliptical writing style, made reading them a rather challenging task. Saussure reported having already established a working rule, according to which wherever an inscription featured the letter i, it usually implied a combination i+x, wherein the x stood for any omitted character, whether a vowel or a consonant, either preceding or following that i.⁴ One can see a curious parallel between this formula and Saussure’s reconstruction of the Indo-European vowel system more than a quarter of a century earlier, whose key “working rule” rested on the hypothesis that any vowel in a protostem should have been originally either preceded or followed by another sound of an indeterminate value. What Saussure was trying to reconstruct this time was not a protolanguage but prototexts—in particular, the proper names that constituted the core of the inscriptions. Having assumed the presence of a coherent message hidden behind the sketchy representation, one then proceeds to retrieve it with the help of “working rules” formulated, if not completely ad hoc, at least, in a way that allows for an adjustment in each particular case. It was this analytical method—which admittedly paved the way for as many abuses as insights—first pursued as a vacation pastime, that would soon emerge in Saussure’s notes as a general strategy of textual analysis whose implementation promised a major breakthrough in understanding the fundamental nature of writing, if not of speech in general.

By the time Saussure resolved to speak out, a few months later—only to his closest friends and colleagues—about the peculiar phenomenon he had stumbled upon, he had already covered scores of notebooks with his analyses. He showed his usual terminological hesitancy in selecting a name for the new phenomenon, using a variety of terms for it, sometimes distinguished as subdivision of the general concept, sometimes thrown in casually as synonyms. The term anagram has become the standard name in later critical literature chiefly because of the prominence given to it in Starobinski’s publication.⁵ Saussure’s starting point was Homer; studies of Virgil and Latin Saturnian poetry (a poetic tradition closely related to tomb inscriptions) followed in close succession.⁶ About the same time, Saussure discovered that the entire Rig Veda was “literally wall-papered” (littéralement tapisée) with anagrams.⁷ Further steps extended the analysis to Germanic epics;⁸ to some examples of French poetry, from Villon to modern times; and finally, to nonartistic prose.⁹ As the circle of texts he studied got wider and wider, the ubiquitous presence of anagrams became absolutely compelling. In Saussure’s words, he “tried in vain, by expanding the scope of the studies in all possible directions, to stumble across a single blank passage” (i.e., one not containing an anagram).¹⁰

The principal mechanism of the anagram as formulated by Saussure was the following.¹¹ At a certain point in a poetic text, persistent repetitions of certain sound combinations occur in a rather conspicuous way. These repetitions are not purely euphonic; rather, they can be construed as latent representations of a theme word—most often, a name—of key importance for this particular passage. On the phenomenal level, the key word is represented only indirectly; all one sees in the text are rudimentary sound combinations whose recurrences show some degree of regularity (they need not be precise, though, allowing for rather wide variations). Each recurring sound fragment as such has no meaning; yet if put together in a certain way, they add up to a word, and, moreover, one that could be understood as the key word of the message carried out by this particular text or textual passage. The effect is as if the signifier of a crucial sign had been broken into splinters and scattered all over the text to be reconstituted—consciously or unconsciously—by the reader. One could say that recurring sound combinations in a text are connected to a key signified of that text not directly and explicitly, as signifiers in a language are, but implicitly and potentially. Such ad hoc semiotic connections are latent signs that are not “given” to the speakers by the language itself but have to be “found” each time anew for a given occasion in the text for which, and only for which, they are relevant. They are transient semiotic phenomena created in the speech act itself.

As far as classical poetry was concerned, the theme word implied in an anagram typically featured the name of a hero or deity, a relevant location, or (in inscriptions) the name of the donor;¹² occasionally, an anagram might also contain the name of a “literary deity,” so to speak, that is, of the author’s literary precursor—the last case being typical particularly for Latin poets, who often employed anagrammatic references to their Greek forerunners. The theme word could be mentioned directly in the text as well, alongside its anagrammatic representation. In many cases, however—Saussure called them “cryptograms”—the theme word figured in the text only “cryptically,” via the anagram, without being directly mentioned. In the latter case, the effect of the anagram was particularly striking: the key word literally lurked beneath the text’s surface, making its presence felt only through overtly meaningless sound repetitions whose presence might be conspicuous but lacked any order. It was this situation that prompted Starobinski to call his pioneering book on the subject Words Under Words.

Saussure’s analysis of a line from book 11 of The Odyssey can serve as a typical example of both the anagram itself and of the analytical methods with which it could be retrieved from the surface of a text. In the line in question, Odysseus is asking Agamemnon, whose shade he has found in Hades, about the cause of his death: was it because Poseidon “aroused the breath of painful hostile winds” against him? The line is repeated twice in quick succession, as Agamemnon replies that no, it was not Poseidon, etc. Agamemnon’s name is mentioned several times in the passage, but not in this particular line, which Saussure cited as ασεν αργαλέων ανέμον αμέγαρτον αϋτμή (áasen argaléôn anémôn amégartôn aütmê). The line contains—in an arbitrary order—the sound clusters a-ga-am-me-em-en-on, which collectively point to the name Agamemnon. It is not particularly important but nonetheless psychologically curious that Saussure apparently misquoted Homer, perhaps citing the line from memory. The initial verb in Homer is not áasen “[he] has exhaled, breathed out” but όρσον (órson) “[he] has incited, aroused.” The difference both in sound and meaning is small, but, in Saussure’s version, the supposed anagram comes out somewhat more strongly. A possible explanation of the psychological mechanism of this substitution could be that Saussure’s memory, prompted by his perception of the anagram, offered him a synonym that fit the anagrammatic pattern better than the word actually used in the text. Ironically, this mechanism was essentially the same as the one Saussure’s theory claimed was directing the unconscious choice of words by the poet as he composed an anagrammatic line.

Another interesting example is Saussure’s analysis of an emotional plea to Venus, the instigator of all life, in the opening passage of Lucretius’s De rerum natura. The name of Venus is, of course, invoked there explicitly. Yet Saussure also observed persistent repetitions of sound clusters that alluded anagrammatically to Aphrodite, i.e., Venus’s Greek antecedent. For instance, in the sentence that stretches over lines 10 to 13, the following sound combinations appear, most of them more than once: a-te-di-af-raf-it-pr-ord. Put in a certain order, they add up to an almost precise rendition of the Greek goddess’ name—not without minor deviations, though, such as ord for rod, or pr in lieu of fr. To justify the latter anomaly, Saussure suggested that Lucretius must have been well-versed in the late classical Greek pronunciation of f as the affricate pf, which could justify the anagrammatic substitution of p for fr (pronounced as pfr) (Starobinski 1979 [1971]:59). In the previous passage (lines 5–9), another less-than-obvious reference appears in the word flores, supposedly presenting fro as a fraction of the goddess’ name. For the representation to be precise, the word in question would have to be froles rather than flores; Saussure justifies this implied substitution by the fact that it often happens in speech practice. Sometimes, Saussure confesses, one needs to repeat a line several times before the convoluted transmutations of its sound motifs conjure up the cryptic name in one’s mind (60).

When Saussure shared his findings with Bally—not without trepidation ¹³—the latter’s reaction, which did not reach us directly but could be inferred from Saussure’s subsequent response, was downright negative. We should keep in mind Bally’s principal scholarly interest was in the stylistic and aesthetic dimensions of language—an area from which Saussure had hitherto stayed aloof—which made him particularly sensitive to matters involved in the issue of the anagram. Not only was Bally doubtful about Saussure’s analytical methodology, but he complained that the vision of poetry as a playground of sound manipulations did a disservice to poetry itself. In Bally’s words, the overall picture that emerges from the anagrammatic analyses makes Homer look like a “strange, frivolous and futile character.”¹⁴

Saussure responded with a remarkable confession that his studies indeed made him experience an “aesthetic cooling” (froissement esthétique), verging on losing all affection, toward the “old bard” (Saussure 1994a:117). As to the studies themselves, he readily admitted his own “moments of doubt” in view of their rather extraordinary consequences, only to reiterate his ultimate conviction that the phenomenon he had discovered was real. His later letters to Meillet on the subject showed the same characteristically Saussurean mixture of timid uncertainties and emphatic, at times almost arrogant, assertions. Responding to Meillet’s “little mountain” (un monticule) of notes pointing out the procedural liberties in Saussure’s analyses, he wrote that all the “obscurities” in Virgil or Lucretius—by which he meant instances where his own analytical principles could not be sustained without ad hoc adjustments—would not “sway” him from believing “with the most absolute certainty in the world” (“je crois absolument certaine pour tout le monde”) that the phenomenon he had discovered extended beyond isolated examples.¹⁵

As to Meillet’s general assessment of his venerable teacher’s idea, it sounds diplomatically ambiguous. Meillet acknowledged that after he himself took a look at some texts they turned out to be permeated with anagrams. For instance, a line in Horace’s ode 2 from book 4: “Nititur pinnis, vitreo daturus Nomina ponto” (Soaring skyward, doomed to give His name to a sea)—contains combinations pin, da, rus, which add up nicely to the name of the poet to whom the ode is dedicated. But why should one limit oneself to the classical languages, Meillet adds: let us open any book at random, say, Chateaubriand’s Mémoires d’outre-tombe (a curious choice, given the atmosphere of a spiritualist séance lurking behind the whole affair). He did, and, lo and behold, in a passage referring to Chateaubriand’s companion Lucile, she is described by a chain of epithets that anagrammatically evoke her name: “Tout lui était souci, chagrin, blessure.” Next comes Baudelaire’s turn: in one of his lines the key word hysterics appears both directly and anagrammatically: “Je sentis ma gorge serrée par la main terrible de l’hystérie.”¹⁶ These offerings could be read as both a reassurance and an implicit doubt. They resounded with Saussure’s own fears about the “superabundance” of his discovery.¹⁷ As examples proliferated, it became increasingly clear that one could find an anagram virtually everywhere—in any text in any language, of any genre, epoch, and literary tradition.

It is quite obvious that, given the rudimentary character of the sound combinations into which the theme word is claimed to be anagrammatically decomposed (they mostly amount to just two sounds, sometimes even a single sound), one could expect any number of words to be potentially reassembled from the same nearly elementary sound material.¹⁸ Saussure was too much of a scholar not to be aware of this problem. His claim to veracity was that the theme word could not merely be composed out of sounds represented in a piece, but that its meaning was of key relevance to that piece. The anagrammatic sound constellations that appeared in a text were not random; they would emerge in a topically relevant passage and fade away as the narrative shifted to another topic.¹⁹ In producing an anagrammatic effect, the sound and the meaning reinforce each other. One can sense the underlying presence of Saussure’s fundamental idea about the indissoluble interconnection between the signifier and the signified in this reasoning: the sound repetitions point to an anagram only insofar as they can be connected to a key word, while the latter reveals its presence only insofar as it is represented through sound repetitions. In a letter to Meillet, Saussure described the anagram as “phonic play” (jeu phonique) in which a sacral name becomes “indissolubly involved” (Saussure 1964:114). Saussure compared the double nature of the anagram with the “famous principle of the leitmotif of Wagner”: in both cases, all the various reiterations of motifs are identified only insofar as they are tied to an idea (Saussure 1994a:118).

Nevertheless, worrisome questions remained, first and foremost among them—why would poets play these games in the first place; why would they all choose to involve themselves in a kind of semiotic tinkering that Bally found, not without reason, “frivolous and futile”? Saussure clearly sounds defensive when he asserts that “preoccupations apparently as childish (puerile) as that of the anagram obsessed Sebastian Bach,” which did not preclude him from writing “extremely expressive music”; a disparaging attitude toward such symbolic games betrays the rationalism of modern art (Starobinski 1979 [1971]:128).

To justify the ubiquitous presence of anagrams in poetic texts, Saussure initially suggested that the phenomenon was of religious, or rather magical, origin, and that it might originally have served as a magic evocation of divine names.²⁰ As the supernatural function of such incantatory texts was gradually supplanted by their aesthetic perception as “poetry,” Saussure argued, the anagram shifted from a magical device into a literary one; it became a matter of literary tradition that subsequent generations of poets acquired and maintained, either consciously or not, through imitation. In a clear reference to his general view of history, Saussure noted that even if there had been a sacral or magic function in the beginning, there was “no mysticism” in the subsequent development of the phenomenon: whatever its origin, it proceeded through (arbitrary) transmutations of the tradition (Starobinski 1979 [1971]:42–45).²¹ For example, while Homer may have borrowed anagrams in his epics from earlier religious practices, later poets (particularly Latin) acquired them by imitating Homer, and poets of still later times by imitating classical poetry at large.²² The only trouble with this argument was that, much as he tried, Saussure could not find convincing evidence from any ancient tradition pointing to the original sacral function of anagrams. Explaining “the embarrassing silence” about the issue in ancient documents by suggesting that secrets of the mystical tradition would have been “meticulously kept” (Starobinski 1979 [1971]:94) was a rather lame excuse. Furthermore, the proliferation of disparate examples, particularly from modern literature, including prose, watered down any claim of even purely literary continuity.

It was Bally’s criticism that helped Saussure find another argument—one born of the general conditions of speech production rather than magic or literary practices. It is well known, Saussure remarked, that euphonic richness is a widespread, indeed universal feature of poetry. To object to the idea of the anagram is not to deny the existence of poetic euphony; rather, it is to deny the presence of a meaning behind euphonic repetitions. Would you be more satisfied, he asked Bally, if it turned out that all that poets do when they employ euphony is pure sound play, a vacuous “waltz of syllables”? Perhaps, Saussure mused, when a poem is being conceived, the poet begins not with an articulated meaning, not even with key words; perhaps the most rudimentary step in a poem’s composition involves the spontaneous emergence in the poet’s mind of fragmentary sound clusters, whose shapes vaguely allude to a nascent poetic idea without yet articulating it. An anagram then constitutes a prototext out of which the text of a poem is born. An anagrammatic effect emerges in a poem not because the poet deliberately sought words that would contain certain sound combinations; on the contrary, it was those primary sound combinations at the core of the poetic idea that “sought” words in the poet’s memory fitting the nascent poem’s prearticulated design (Starobinski 1979 [1971]:96).²³ The sound texture of the poem and its meaning emerge interactively. The original awareness of the ability of sound clusters to attract a potential meaning might have come from ancient rites of mystical glossolalia or else from the heightened awareness of sounds’ combinatory potentials in the Indian poetic tradition resulting from the meticulous phonetic descriptions of Indian grammarians. But whatever their origin, these latent phonosemantic synapses lay stored in a poet’s mind as the primeval linguistic material—semiotic rudiments avant le signe, so to speak—whose coalescence into an articulated meaning proceeded in discourse by means of the anagram.

Whether or not one is convinced by Saussure’s anagrammatic analyses, his general argument reiterates the two properties unique to language: the principle of duality, according to which the material and spiritual elements of a sign have no value of their own, but become a semiotic phenomenon by virtue of their mutual relation; and the principle of arbitrariness that defies any general pattern according to which such relations could be structured and any predictable direction in which they could evolve.²⁴

In this respect the anagram can be seen as a particular facet of Saussure’s exploration of the metaphysics of the sign. The phenomenon of the anagram presents itself as yet another kind of interdependence of a signifier and a signified—one that is not preestablished in language but emerges in the process of speech. What surfaces in a poetic text is, on the one hand, a texture of sound repetitions that by themselves do not signify anything beyond euphony and, on the other, a theme word whose presence in the text remains mute. It is only when these two polarities meet (or, as Novalis would have said, “embrace”) each other in the anagram that their implicit and disparate presence comes into focus as a key element of the poem’s message. The connection between the two poles of the anagram is arbitrary: it exists solely by virtue of being established in this particular way for this particular instance in the text. Nothing in the character of repeated sound combinations itself suggests the name they represent; as a matter of principle, an infinite variety of words could be construed from those combinations. By the same token, looking at the general content of a poem, regardless of its sound texture, one can imagine many words that could serve as its thematic emblem. It is only by projecting our awareness of the highlighted sound material onto our awareness of the poem’s topical content that we grasp their connection in the shape of an anagram. Such moments of semiotic crystallization of the latent textual material emerge as an “imbroglio des formes semi-concordantes,” not bound by any strict rule.²⁵ The manifest lack of rigor in Saussure’s analyses reflects the fundamental nature of the phenomenon. Whatever “working rules” one might form in the process, one should be prepared to bend them at any moment in response to an arbitrary turn in an anagram’s configuration. In the end, no general formula could be found that would cover every single instance of the anagram—nothing beyond simply registering its presence.

The difference between linguistic signs and the anagram is that while the former belong to speakers’ prerequisite knowledge of la langue that exists—at least theoretically—prior to and independent of their speech activity, the latter is entirely a matter of speech itself. The arbitrariness of the anagram is an arbitrariness of the second degree, so to speak. No command of the arbitrary structure of la langue could help one to create or perceive anagrammatic signs. They must be constructed each time anew as a transient connection between an improvised signifier and an ad hoc signified—a connection valid only in a particular instance in a particular text.