The structural properties of the anagram in poetry

1.1 A still unexplored theoretical problem

It will be remembered that the theoretical subject of the anagram in poetry emerged with structuralism in the 1960s–1970s when Starobinski published the notebooks on the anagram, dating from 1908–1910, that Saussure had dedicated to this topic (Starobinski 1964, 1971). At the time, the subject seemed a surprising one since Saussure was the founder of linguistics and this research was evidence of an unexpected theoretical interest. His research method was, in fact, quite unusual. For Saussure, the dissemination of a noun systematically constituted the supporting framework for verse: in the Vieux saltimbanque by Baudelaire, for example, hystérie is, for Saussure, pre-announced by the letters IS, S, TERRI, E which precede this word in the line “je sentIS ma gorge Serrée par la main TERRIblE de l’hystérie.” This discontinuous sequence was given various names by Saussure, paragram, hypogram, anagram, and this latter term took hold attracting many scholars to investigate its dissemination in poetry (Agosti 1972; Wunderli 1972; Zumthor 1975; Johnson 1976; Kristeva 1969; Riffaterre 1978; Jakobson 1984). In 1982, I also published a book on this subject, followed by a longer work in 1993, in addition to some articles in 2001 and 2007 when the subject of the anagram was already in decline. Now that this topic appears to have been practically forgotten, I would like to provide some explanations for the progressive lack of interest shown in the anagram, but also some reasons why, after more than thirty years, I am continuing my research and now propose with this article that the study of the anagram in poetry be resumed.

In fact, the subject of the anagram conceals a difficult methodological problem owing to the absence of knowledge concerning the effective combinatory structure of language, as well as the lack of an adequate software program to facilitate its study for the linguist. I myself became aware of this problem immediately following the publication of my book in 1982 which was, for me, the unexpected result of a very personal path of study: coming from a background in engineering, I was about to graduate in psychology and set out on the journey to become a psychoanalyst. At that time, the theoretical problem I was studying in depth was the associative density of semantic networks which, as well as in dreams, formed the basis of language, subjects that I had just analyzed in detail in my psycholinguistic and neurophysiological studies. It was then that I thought to analyze and compare poetic texts – a literary passion of mine – assuming that their brevity would allow for a better exploration of these themes.

I was not fully aware of the hypotheses concerning anagrams and therefore, unlike Saussure (and other researchers), I had not adopted the hypothesis of disseminated letters, rather that of rigorously consecutive sequences that also straddle word boundaries, but with the advantage of identifying more precise relationships. The anagrams were detected by means of meticulous – but manual – research, however when continuing my studies after the publication of my book in 1982 I started to become uncertain about the possibility of identifying the anagrams correctly. In those years, the first personal computers were just starting to become available and I asked some computer science friends to create a software program that was able to give me an overall picture of the anagrams. To my astonishment I discovered that the number of anagrams in a sonnet could be more than two hundred, and in a canto by Dante – as I later checked – even many thousand.

Such complexity, which could not be detected without the use of a software tool, explained the failure of the research by Saussure as well as the inevitable unreliability of the intuitive study of the anagram and the reasons for its decline. The large number of anagrams was completely unknown: at that time, no studies on the combinatory structure of a text were to be found either in quantitative linguistics or in other language studies, and – as far as I know – are still not available to this day. There was, therefore, no hypothesis on how to evaluate the possible structural significance, which may even have been minimal, of such an enormous combination. However, the poems I had studied retained a certain coherence even in the light of such large numbers of anagrams and this drove me to continue my research. A computer was indispensable, together with a sophisticated software program that could facilitate the research, and it took me many years to develop one: the fact that, still today, it is the only program of its kind in existence is an indication of a theoretical question on the combinatory nature of texts that there seemed no reason to ask.

1.2 The methodology for studying the structural properties of the anagram

Studying a poetic text is not simple given the enormous number of anagrams, and an assessment of their effective structural significance therefore requires a particular methodology. Furthermore, this is counter-intuitive since it explores the text as a succession of words without intervals (also cancelling out the separation into lines), even though this is a determining factor for the research. Both the continuity of the utterance and the comprehensiveness of the analysis are in fact indispensable in order to demonstrate the semantic and formal, as well as statistical significance of certain combinatory phenomena and, above all, to ensure that their study may be shared and discussed by other researchers. ^[1] The principle according to which the anagram operates, once we have become familiar with this method, is however fundamentally simple. The anagram maintains an associative link, between sections of the utterance, that can be interpreted as “a free functor,” i.e., able to perform different structural functions within the text.

The most important of these is, without doubt, the semantic or thematic relationship that its start and end segments maintain between the words (Figure 1(a)). By forming a chain with letters from the start segment of another anagram, the end segment may also make a line that synthesizes a significant semantic or thematic direction (Figure 1(b)).

Figure 1:

(a) Anagrams maintain semantic-thematic relationships between words. (b) Anagrams’ chains synthetize semantic-thematic directions.

The structural function that is most easily comprehensible, perhaps, is the contribution of anagrams and lines to the formal structure – as I shall show immediately in the first examples – and this succinctly explains the importance of formal apparatus in poetry. Closely connected with all these properties is the metaphoric-symbolic function that derives from the semantic-thematic network of anagrams and lines supporting the ordinary grammatical layout.

Some simplification, however, will benefit the undoubted complexity of this research. Anagrams are both graphemic and phonemic, generally working together with each other, enabling their study to be simplified to that of graphemes. Typically, all these properties are hierarchized according to the rank Rn, in which ‘n’ indicates the length of the anagram, and therefore distributed between the longer, rarer, and more important anagrams, as well as the shorter, more numerous anagrams, allowing for further simplification. One useful aspect of this hierarchization, for example, is often revealed in the combinatory coming together at words or sections of the utterance, which often directly highlight the most significant thematic nuclei: of these, generally, the most important are those at the beginning and end of the first line and in the concluding words of the last line, which facilitates a rapid study of the text.

This combination of structural functions is quite extensive and is therefore used in varying ways by each poet, while also revealing similar characteristics: from a comparison of many different texts there gradually emerges evidence of common structural forms, furthering our knowledge of this method. In this article I shall first of all show some simple examples of graphemic anagrams, pointing out in advance that initially they may seem quite extraneous to common sense. ^[2] Subsequently, I shall introduce the theoretical hypotheses which, in my opinion, can explain the structural origin of the anagram. This will be followed by further examples that can elucidate their singular properties.

2.1 The initial formal and thematic generative nucleus

Let us consider the famous poem by Blake The Sick Rose (Erdman 1968), which is only eight lines in length and is included in the 1794 collection Songs of Experience:

1. O rose, thou art sick!

2. The invisible worm

3. That flies in the night,

4. In the howling storm,

5. Has found out thy bed

6. Of crimson joy:

7. And his dark secret love

8. Does thy life destroy.

It is immediately interesting to observe how, even in such a short poem, very many anagrams can already be seen. As many as 89 anagrams, distributed over the various ranks R2 = 71, R3 = 9, R4 = 3, R5 = 5, R6 = 1, overall produce the thick web shown in Figure 2(a). However, several anagrams are formed from simple textual redundancies (the triad “the” and the pair “thy”) allowing for some simplification in the distribution, that can be reduced to R2 = 49 (Figure 2(b)), R3 = 4 (Figure 2(c)), R4 = 3 (Figure 2(d)), R5 = 2 and R6 = 1 (shown together in Figure 2(e)), and enabling us to give a better assessment of their contribution.

Figure 2:

(a) The 89 anagrams. (b) The 49 R2 anagrams. (c) The 4 R3 anagrams. (d) The 2 R4 anagrams. (e) The 2 R5, 1 R6 anagrams. (f) The density diagram.

As can be seen, while the rank R2 anagrams appear to be distributed with no clear order, the longer anagrams are already indicative of some significant aspects. The two rank R5 anagrams connect “rose” at the beginning of the text with the last line, while the rank R6 anagram reinforces the rhyme; the two rank R4 anagrams also involve the last line and the formal structure. Consequently, the beginning and end of the text, together with the formal structure, indicate how the anagrammatic tension tends to be arranged throughout the text making us immediately aware of any textual meaning. In particular, it can be seen how the beginning and end of the text stand out in the density diagram obtained from all the anagrams (Figure 2(f)): the relationship between “rose” and “destroy” is emphasized by the large density of combinatory nuclei that depend on these two words making them clearly evident.

A structural interpretation of this combinatory tension is also immediately given weight by a study of the opening section of the text that shows the important part played by R2 anagrams. We can observe, for example, the simple flow of the first two anagrams that go as far as the first rhyme, w-or-m-st-or-m and to the end of the text, dest-ro-y (Figure 3(a)). Although these are anagrams of only two letters, they emphasize how the textual input is immediately directed to two formal characteristics, namely, the first rhyme and the conclusion. If we consider the next anagram that reaches “d-estro-y,” consisting of five letters (one of the three longest anagrams), it confirms how the flow leads from the beginning to the conclusion (Figure 3(b)).

Figure 3:

(a) The formalization of the first two anagrams. (b) The first flow leads from the start to the end. (c) The structural function of the first line.

Another way of quickly recognizing the structural function of the opening flow is the line produced by the first anagram from “w-or-m” and “st-or-m” that links up with other anagrams (Figure 3(c)). From “w-or-m,” the anagram “orm/tha-torm/ha” extends the linking up of the “worm-storm” rhyme, while from “st-or-m” it is linked with the stor-stro anagram which refers back to “de-stro-y.” The line synthetically indicates the direction of the worm-storm-destroy flow along the formal structure axis, which is maintained not only by the rhyme “worm-storm,” but above all by the particular contribution of the stor-stro anagram of the R = 4 system.

This line, however simple, therefore has the merit of drawing attention to the two most important anagrammatic contributions to the formal structure. The first is the function of the rhyme “worm-storm” extended to the larger combination “orm/tha-torm/ha,” while the second is the supporting function of the stor-stro anagram for the formal structure. Both contributions imply an equivalence of rhyme and anagram, a property that typically characterizes – as I shall show in other examples – formal structure in poetry.

Highlighting these processes is not merely formal, but also indicates how the lexical system identified by the flow should be interpreted thematically. The poet does not limit himself to a brief description of events leading from the rose to destruction, but also establishes the antagonistic relationship between the two subjects of the poem, the rose and the worm and, furthermore, directly entrusts the thematic development of the destructive action, the subject of which is the worm, to the line worm-storm-destroy. The anagram “s-torm – de-stro-y” clearly has a semantic function, while “storm” acts in the flow as an introductory scenario to the destructiveness of “worm,” also serving as a thematic support.

Although the anagram flow “singles out” a limited lexical grouping, it may, however, be interpreted as the nucleus of the grammatical-syntactic development that occurs throughout the utterance. In particular, the “singled out” section indicates the structural significance that can be attributed to the formal structure. Since the formal axis stands out both as regards sound and visual perception, we may consider that its purpose is to favor, in the appreciation of the text, words placed in this position, that is to sensitize the words involved in the flow.

If we adopt this hypothesis, we can understand why, in order to intensify their perception, the poet tends to make use of rhyme and, in this case, also anagrams which contribute to the sensitization of the words in the formal axis.

2.2 The generative nuclei at the end of the first line and at the end of the text

Other characteristics of the anagram may be briefly illustrated by studying this simple poem. Graphemic-phonetic co-operation – an important subject in poetry – can clearly be seen in the other generative apex that is typical of the poetic text, the end of the first line, “art sick,” whose flow lines both converge on “destroy” (Figure 4(a) and 4(b)).

Figure 4:

(a) The graphemic flow. (b) The phonetic flow runs along the formal axis of the first line’s end.

Figure 5:

The graphemic flow of the first line’s R≥3 anagrams.

Of the two flows, the graphemic flow is the more widespread and also extends within the text, while the phonetic flow is directed along the formal axis. Both flows confirm the significance of the concluding word “destroy,” on which the disease of the rose now also converges, implying a particular interpretation that I shall come to shortly. The two flows also reaffirm “storm” as an essential element and, together, clearly define with “invisible” and “his dark secret” the theme of invisibility-secrecy which through the verb “flies” (that belongs to the R = 4 system) marks the start of the destructive action of the worm.

Of equal structural importance is the graphemic flow of the longer anagrams (for R≥3) in the first line, which links up with the last line only (Figure 5). The first and last lines clearly form a privileged pair in the versification, the final line having the purpose of concluding the opening theme.

2.3 Semantic-thematic coherence and interpretation

A more in-depth study of the anagrams also reveals a more general semantic and thematic coherence, which is characteristic of a typical text analysis. For example, “r-os-e” refers with the anagram os-so (the only one in the text) to “crim-so-n,” maintaining the color relationship between the rose and its “bed of joy”; “flie-s” is an anagram of “life,” involving the flight of the “worm” directly in the life that it destroys. This more detailed analysis allows for a more accurate assessment of the general semantic-thematic fabric and its effective contribution to textual understanding.

For example, this poem has given rise to very many interpretations, both of the mystical-allegorical and amorous-sexual type, and in particular with respect to Blake’s conception of woman, and its study provides us with some privileged indications. The entire development demonstrates the priority relationship between the strong opening anagrammatic nucleus, “O rose,” and the strong final anagrammatic nucleus, “destroy.” From a thematic point of view, therefore, this relationship sustains the start-end division together with the life-death division, since the rose evokes beauty and life, but also the fading away foreboded by “sick” at the end of the first line. An anagrammatic analysis, however, highlights certain controversial aspects, for example that the subject of “destroy” is not simply “worm,” but above all “rose.”

In fact, in Figure 5, the two five-letter anagrams of “rose th-ou,” refer in the last line to “d-oes th-y life d-estro-y,” explicitly indicating “rose” as the agent of the verb. Furthermore, the anagram edo-edo which links “b-ed/o-f” to “lov-e/do-es” directly implicates the “bed of joy” of the rose in the secret love of the “worm.” The anagram edo-edo consists of the repetition of the same segment and straddles the two lines in a position that maintains a link between two end-of-line words, thus resembling rhyme: structurally, therefore, it can be considered very important, giving weight to the sexual interpretation of the “bed-love” relationship. Since the rose is “sick” and an anagrammatic examination gives its illness a particular connotation, such analysis lends weight to the interpretation of the rose as a symbol, for Blake, of a dangerous sexual figure.

Although brief, this analysis highlights the general contribution of anagrams to the semantic and formal structure of the text, disregarding only some of the shorter anagrams.

3.1 A study and interpretation of the initial nucleus

The poem A Talisman (1921), by Marianne Moore, is in turn useful for introducing the structural properties of the anagram. Overall this poem (Moore 1984), longer than that by Blake, has 170 anagrams (their quantity increases according to the square of the length of the text), distributed as follows over the various ranks: R2 = 135, R3 = 48, R4 = 8, R5 = 3, R6 = 1. In spite of this large number, we can restrict our study to just the graphemic nucleus at the beginning of the poem, which has the greatest combinatory density (Figure 6) immediately suggesting its structural importance.

1. Under a splintered mast,

2. Torn from the ship and cast

3. Near her hull,

4. A stumbling shepherd found,

5. Embedded in the ground,

6. A sea-gull

7. Of lapislazuli,

8. A scarab of the sea,

9. With wings spread –

10. Curling its coral feet,

11. Parting its beak to greet

12. Men long dead

Figure 6:

The initial nucleus.

For example, we immediately notice the formalizing tension of the first two anagrams which from “unde-r” form the rhyme “fo-und/e–gro-und” (Figure 7(a)). The segment “und” of the rhyme “fo-und – gro-und” is “extracted” from the opening segment of the line “und-er,” and highlights the unconscious attention given by Moore to textual introduction. The extraction “sensitizes” the rhyme, but also the reference to the beginning of the text, with which it forms the lexical combination “under-found-ground.” When interpreted grammatically, this combination expresses the simple concept “I found underground” which summarizes the introductory theme of the poem.

Figure 7:

(a) The formalization of the first two anagrams. (b) The flow of the first four anagrams. (c) The flow of the first six anagrams.

Now let us look at the wider flow comprising the first four anagrams (Figure 7(b)): the longest anagram in the poem, consisting of six letters, is generated which gives formal structure to the entire word “spread,” as well as the less significant two-letter anagram that encompasses “a-nd cast” at the end of the line. Extending the flow to the first six anagram segments (Figure 7(c)) contributes to the important formalization of the concluding word “de-ad,” but also involves “emb-edded” which works semantically with “gro-und.” It also encompasses the subject “sheph-erd” and, above all, the verb “splinte-red,” which with its meaning of “breaking-shattering” introduces the theme of death-decay in the poem that is reflected throughout the entire lexical system.

Consequently, the flow includes three end-of-line words (fo-und/e, gro-und, spread) and the concluding word “de-ad,” but also words forming the end sections of lines (splint-ered mast, a-nd cast) and syntagmatic relationships such as “sheph-erd fo-und/E,” “emb-edded […] gro-und” that reveal the unity of the anagram flow. At the same time, the flow identifies the semantic and thematic development that “under,” in its link with “splintered” (“un-der splinte-red”), has in the text. The meaning of “splinter” is “to shatter, break into pieces”: the broken mast is therefore a symbol of a loss of strength and of life, since the movement of the ship in the sea is dependent on the mast.

This already suggests why the seagull, which also plies the sea, symbolizes death in the poem. Both are representatives of marine life and are now symbols of death on earth, one torn off and broken, the other “embedded,” that is sunk into the ground just as the mast is affixed in the ship, and placed near the ship as a reference to the joint symbolism. Initially, the poet does not reveal the human significance of death, to which the concluding words “men long dead” refer.

At the same time, “splinter” with its meaning of “shatter,” also conveys the idea of “split-open up” and this explains the link from “un-der a sp-lintered” to “spread,” the longest anagram: the wings of the seagull are wide open, but more precisely they are “extended, spread,” in accordance with the meaning of “splinter.” The adjective therefore confirms the symbolic relationship between the mast and the seagull (just like the literary use of “spread” to indicate an “unfurled sail”).

It should also be noted that “shepherd” can be justified by a semantic link with “splintered,” since “shepherd” is a compound word (shep-herd), and “un-der” maintains the “splinte-red – sheph-erd” relationship. From this, the anagram takes “h-erd,” that expresses the ‘sign’ “to group together,” in contrast to the sign “to shatter” of “splintered.” Therefore, while the mast (and also the seagull) symbolize the fragmentation-disintegration of death, the “shepherd” represents the opposite, the cohesion-union portrayed in the symbol of one who gathers together and guides. However, with its derivation from a verb that depicts the breaking of the mast, the shepherd, although a symbol of life, moves along “stumbling,” that is with the risk of falling: so that he, too, is a representation of death, a figure who forewarns, with his uncertain gait, the ending of life portrayed by “men long dead.”

3.2 Combinatory hierarchy of the line flows

Other flows clarify how Moore from the beginning coherently organizes the network of anagrams. Let us observe, for example, the sequence in Figures 8(a), 8(e) and 8(b), showing the line flows of the first anagram as some characteristics in the chain parameters are modified.

For K = 2 (indicating the minimum chain link) and for Q = 1 (indicating the link) of the longest anagram only), the first anagram generates the line that from “found” reaches only to “ground” (Figure 8(a)). However, if we extend the line chain to a second anagram (Q = 2), from “fo-und/e” there is also a link to the segment “de,” whose flow takes in “emb-edded” and then converges on “de-ad” (Figure 8(b)).

Re-written in the sequence “under / ground / found / embedded / dead,” the flow summarizes the entire thematic development of the poem, the sight of something-someone found underground, embedded and dead. When we reduce the chain (K = 1), the flow is extended but converges markedly on the end sections of the last four lines, taking in the whole concluding word “dead” (Figure 8(c)). What we now clearly observe is the strong convergence on “beak to greet” which, together with the confluence on “spread” and “dead,” develops the original nucleus of “splintered,” producing a dramatic image of an open mouth, the first image – we may suppose – that formed in the poet’s mind.

Figure 8:

(a) The first anagram’s line for K = 2 and Q = 1. (b) The first anagram’s line for K = 2 and Q = 2. (c) The first anagram’s line for K = 1 and Q = 2.

Both flows, that of the first six anagrams and the lines of the first anagram, clearly have a structural function: they are directed towards the different semantic-thematic nuclei that have played a part in the conceptual and metaphorical development while, at the same time, contributing to the formal structure.

Lastly, in this poem it is useful to understand the lexical choices of “lapislazuli” and “scarab.” The ninth and tenth anagram segments, ra and aspli, appearing in “unde-r a spli-ntered,” are the generative support for the two words, and from these the flow converges on “coral,” the other material that with “lapislazuli” suggests in its hardness the preciousness of the seagull (Figure 9(a)).

These two symbols help us to understand how a precious material, that is difficult to break, has been evoked as a contrast to fragmentation-death, and assigned to the seagull as a lasting emblem. The five-letter link that reaches “lapis-lazuli” highlights the importance of the main one of these symbols. And finally note the simple flow generated, once more, from “sp-lin-tered” (Figure 9(b)).

Figure 9:

(a) “Unde-r a spli-ntered” generates “lapislazuli,” “scarab” and “coral”. (b) A flow generated from “sp-lin-tered”.

Stumbling, the shepherd moves his feet as the seagull retracts its feet. The contraction of the seagull’s feet in death, and the contrasting opening of its beak, show the common destiny that awaits the unknown explorer while, unsteadily, he approaches the representation of death.

Therefore, the anagram system, although restricting ourselves to the initial anagrams only, enables us to reconstruct the working together of three different linguistic subjects, the mast, the shepherd and the seagull, in order to maintain and diversify the development of a single conceptual nucleus, separation-death. The poet articulates this thematically on a syntactic-grammatical level, while the combinatory flows support and co-ordinate this development at an anagrammatic level.

4.1 The contribution of the anagram and intralexical words to semiotization

The two examples I have shown are fundamentally simple, and their contribution to the interpretation of the text might appear to be – adopting a restrictive hypothesis – a sophisticated version of an ordinary critical evaluation. However, the meaning of the interpretation appears different if we reflect on the semantic-thematic support of the anagrams in the two texts, which implies a structural explanation for the recognized difficulty in clearly defining semiotization in poetry and – a related topic – the function of the formal system.

A short analysis of the sonnet Le Cygne (1885) by Mallarmé (Mondor and Jean-Aubry 1945) may clarify how the anagrammatic contribution to semiotization can be a real determining factor for the interpretation of the text. This example also introduces “words upon words” (to reference the title of the book by Starobinski on the anagrams of Saussure), that is the intralexical contribution of “included or over-included” words in an utterance. In the poem by Blake, for example, “flies” includes the verb “lies,” whose meaning “to lie” is coherent both grammatically with “bed,” the destination of “flies,” and with the anagram of “flie-s,” “life,” the life destroyed by the worm which by flying has reached the “bed.”

In turn, “bed of crimson joy” includes in “crim-son” the root of “crim-e,” thereby implying guilt in “joy,” a word that is already linked by rhyme to “destroy.” “Crim-so-n” refers back to “r-os-e,” and the connotations of “guilt” in “crimson joy,” while the sexual inferences of “flie-life” strengthen the interpretation of the participation of “rose sick” in the destructive love. The contribution of included or over-included words requires particularly careful examination, since it is generally less well defined than that of anagrams.

4.2 Three intralexical generative relationships at the beginning of Le Cygne

In the opening part “Le vierge le vivace” of Le Cygne, however, a much more complex working together of intralexical words is a determining factor for interpreting the obscure first line while, at the same time, giving rise to the formal structure of the first two quatrains and clarifying the difficult symbolism of the sonnet.

1. Le vierge, le vivace et le bel aujourd’hui

2. va-t-il nous déchirer avec un coup d’aile ivre

3. ce lac dur oublié que hante sous le givre

4. le transparent glacier des vols qui n’ont pas fui!

5. Un cygne d’autrefois se souvient que c’est lui

6. magnifique mais qui sans espoir se délivre

7. pour n’avoir pas chanté la region où vivre

8. quand du stérile hiver a resplendi l’ennui.

9. Tout son col secouera cette blanche agonie

10. par l’espace infligée à l’oiseau qui le nie

11. mais non l’horreur du sol où son plumage est pris.

12. Fantôme qu’à ce lieu son pur éclat assigne

13. il s’immobilise au songe froid de mépris

14. que vêt parmi l’exil inutile le Cygne.

This sonnet has numerous anagrams (R2 = 910, R3 = 287, R4 = 73, R5 = 18, R6 = 10, R7 = 3, R8 = 3, R9 = 1), but R2 anagrams are rarely significant and the study can be simplified by restricting ourselves to the other anagrams. First of all, let us consider the complete flow of “Le vierge le vi-vace” (Figure 10(a)), made up of 13 anagrams, that includes the longest anagram in the text, of 9 letters, and two anagrams of 8 and 7 letters. The flow involves the two quatrains only (formalizing the “ivre” rhymes), while the importance of the initial nucleus is clearly seen in the density diagram (Figure 10(b)).

Figure 10:

(a) The complete flow of “Le vierge, le vivace”. (b) The poem’s initial nucleus in the density diagram.

The three longest anagrams reach down to “le givre/le” (frost), immediately indicating the dominant theme of ice-intense cold, to which the anagrams involving “h-iver” and “glac-ier” also refer, while “régi-on” indicates the space in which the swan is held captive. A second correlated theme – the difficulty-impossibility of freeing oneself from the ice-intense cold – can be identified in the flow from the anagrams at “déch-ire-r, ail-e ivre, d-élivre,” to which “v-ivre” is also related.

The word “v-ier-ge” can also immediately be associated with “st-eri-le” on account of the impossibility to procreate, also underlining in “delivre” the meaning of “to give birth,” and so leading to a third theme of the swan being denied birth. At the same time, it can be noted that in the flow the reference to “sou-vie-nt” has no counterpart, and that the two dominant themes, of ice and captivity-freedom, cannot be attributed to the meaning of the opening segment.

The multi-themed origin of the flow can, however, be recognized when tracing the words included in the sequence of letters “le-vierge-le-vivace,” that justify these various developments. In order to trace them, let us once more consider the anagrams produced by the opening section.

The three anagrams which converge on “le givre/le” derive from “le vierge l-”, “viergegele” and “erge le vi-,” so that the “ice” theme is generated by the working together of the entire initial sequence of 12 letters leviergelevi. It can then be seen that the segment leviergele can be broken down to l’evier gelé, in which the different segmentation “gelé” directly expresses the sign “ice” of “givre,” while “l’evier gelé” – “the frozen sink” – anticipates the theme of the frozen lake, whose flow takes in “hiver,” “glacier” and “région.”

This type of re-segmentation is also a feature of the second anagram, which from “le vier” goes to “ail-le ivre” and “d-elivre” and introduces the other important theme of “captivity-freedom.” The sequence le vier can be rewritten as levier – lever-support –, and this word is revealed as the intralexical connotator of the beating wing with which the swan seeks to free himself from the ice. If we consider the re-segmented words l’evier gelé and levier, their appearance can be directly interpreted as the semantic-thematic nucleus of the captivity-freedom dialogue developed in the sonnet.

A third intralexical relationship comes into play in this development. In “v-ier-ge” the segment “ier” corresponds, apart from the letter “h,” to the adverb “hier” (but, phonetically, jɛʁ is included in vjɛʁʒ). In the flow, this justifies the anagram at “souvient” and, above all, explains why “aujourd’hui” is the (apparently inconsistent) grammatical subject of “vierge”: the intralexical opposition “yesterday vs today” is directly personified by the noun syntagma “v-ier-ge aujourd’hui,” forming with “souvient” the other important theme of the sonnet, the memory of the swan. Although the intralexical desegmentation “(h)-ier” is the shortest in this nucleus (and only phonetically complete), this temporal nucleus explains above all the continuity between the initial subject of the sonnet, “le vierge d’aujourd’hui,” and the subject of the second quatrain, the “cygne d’autrefois.” This continuity is maintained grammatically by the opposition “aujourd’hui-d’autrefois,” and intralexically by the identity “hier-d’autrefois.”

4.3 Generative semic opposition within the nucleus

Generative function of this intralexical nucleus in producing the opening anagram flow is therefore explained by the reciprocal penetration of the signs “vitality-movement” of “le vierge le vivace” and the signs “ice-immobility” of “evier-gelé.” Such reciprocal penetration implies the sign “immobility” in the vitality of “vierge le vivace,” and this contrast gives rise to the swan’s attempt to free itself from the ice, which has the intralexical support of levier, while at the same time maintaining the contrast between the present and the past. The internal antinomy of this nucleus therefore allows for ample metaphoric development, with its main link being the impersonal opening “le vierge aujourd’hui” and the subsequent personification “un cygne d’autrefois.”

This link maintains the thematic continuity between the first and second stanzas, pointing specifically to the hidden “anaphoric reference” of “le cygne” to “le vierge” that can only be traced at an intralexical level. In this passage, the temporal axis maintained by the link again becomes more extensive, immediately dramatizing the main theme of the sonnet with the striking difference to the opening scene: “le vierge le vivace” represents “aujourd’hui” and, by contrast, the swan represents “autrefois”.

Through this passage, the past is opposed to an illusory present, enabling us to understand that “cygne” fundamentally expresses a retrospective time of “vierge”: a temporal dimension that contrasts with mature adulthood and suggests a lack of hope for procreation “sans espoir se délivre.” The time structure therefore makes clear that the initial stopping of movement is also that of time, conceptualized as being grounded in a past without a future, as summarized in the concluding words “exil inutile.”

It is important to note that, from the point of view of the semiotization organizing the text, the contrasting co-operation of the two signs “ice-movement” produces two opposing generative processes. The first subordinates the intralexical sign “ice” of “le vier-ge le-vivace” to the grammatical sign “movement” of “vivace,” introducing the opening vitality of the sonnet; while the second subordinates the sign “movement” to the intralexical sign “ice,” producing the symbolism of the swan trapped in the ice, which becomes the main theme of the text. In the nucleus, the inversion of the opposing relationship between the two signs can, however, be interpreted as the main generating force that guides the development from the first to the second quatrain.

This also explains how the anaphoric link between the two quatrains is maintained by a generative type semic inversion. In conscious ideation, the sonnet is first and foremost a triumphalist statement of beauty; however, the intralexical and anagrammatic co-ordination conceals the inversion of the original semic hierarchy that acts as the main generative nucleus, developing the more dramatic theme of the sonnet.

5.1 The problem of anagrammatic and grammatical-syntactic co-operation

These three poems provide some particularly clear examples of the anagrammatic processes that can be detected in poetic texts, while their generative nuclei also have the merit of being able to be statistically validated as shown in paragraph 10 below. If we accept their evident structural function, these processes imply that, during the production of the utterance, a radical control of the combinatory relationships present in the lexical sequence is exercised unconsciously. Since there is a clear co-operation between grammatical-syntactic properties and combinatory properties, this would point to a contemporaneity of the two processes.

I shall now try to give an explanation for such complexity and precision. In my opinion, this concerns the already known associative properties of the sign that are normally present when language is produced, but which evidently include anagrammatic type associative properties whose co-operation is still not known.

If, for the sake of convenience, we keep to the saussurean sign (Saussure 1916), signifieds are characterized by the associative properties of their component parts, the privileged subject of structuralist study (Greimas 1966; Quillian 1968; Katz 1972, etc.) that has sought to determine their semic categorial properties (denotative, connotative, semantic, grammatical, etc. marks); while the signifier is characterized by the associative properties of its morphological components (suffixes, affixes, declensions, etc.), also including the morpho-syntactic properties that are indispensable for working with the grammatical-syntactic functions of “agreement” and “government.”

Ordinary lexical access to a particular word requires the careful selection of both these associative codes: language disorders already studied by Jakobson (1962 [1941]) show the widespread distribution of these codes in the brain, the incorrect selection of which leads to errors in lexical access. When also taking anagrams into account, the situation therefore becomes more complex because it implies that a signifier additionally has associative type anagrammatic properties that are carefully selected from the general co-ordinated semantic and morpho-syntactic elements. Over the years, I have wondered at length about which theoretical model can explain, at the moment of producing language, such a precise selective co-ordination between all these associative properties, of which the anagrammatic element is completely unconscious.

5.2 Theories concerning lexical access in language production

Let us consider, briefly, how language is produced. As is known, the complexity of language generation is a characteristic of theories that are not unique and which, from the structuralist era, have opposed the generative grammar of Chomsky (1965, 1966, 1972) and semantic type generative theories (Bach and Harms 1968; Katz 1970; Lakoff 1971) with various subsequent additions (Hauser et al. 2002). In my opinion, the theoretical key for introducing anagrammatic processes is the controversial passage from the so-called conceptual preparatory phase of the signifieds to that – still keeping to the Saussurean sign for the sake of convenience – of the signifiers.

Traditionally, until the 1980s–1990s (Garret 1976, Garret 1980), it was believed that the first phase predisposed the conceptual-semantic and syntactic-grammatical plan that was simply followed by the selection of the signifiers, while today this selection is better explained by introducing the “lemma” as a distinct level from that of the “lexemes” (Kempen and Huijbers 1983). The lemma does not portray simple signifiers that refer to the signified, but “abstract verbal representations” which combine grammar, semantics, phonology and morphological rules, mediating between conceptual content and superficial utterance (Levelt 1989; Levelt et al. 1999; Levelt 2001).

The process for producing language initiates at a conceptual level involving the lemmas that select the lexemes by activating specific nodes for morpho-phonological encoding. However, such activation is not simply in one direction, but also acts in the opposite direction: that is morpho-phonological encoding can have a backwards influence on the selecting of lemmas and these, in turn, on conceptual-semantic elaboration. The structural nature of such micro-planning implies certain properties in the organization of the brain that can be briefly summarized.

The notion of the lemma, in its current version, is in fact “connectionist,” i.e., determined by modern concepts of how neural elaboration is formed from the system of brain connections which, for language, gravitate towards two main centers: Broca’s area, connected to the frontal lobe and which today is linked to conceptual elaboration, and the occipital-parietal Wernicke’s area, connected to the (acoustic and visual) elaboration of the signifier. The lemma denotes the nodes of the combinatory fabric involved in elaboration leading to the co-operation of these two fundamental areas for language. Consequently, these combinatory properties also give rise to anagrammatic properties.

The model of McClelland and Rumelhart (1981, 1982) provides an example of how a connectionist type network naturally incorporates anagrammatic links that become part of the lexical encoding. In Figure 11, some distinctive primary visual segments categorize the graphemes ‘r’, n’, ‘t’, ‘g’, ‘s’, and the words ‘able’, ‘trap’, ‘trip’, ‘take’, ‘time’, ‘cart’ are dependent on some of these: the anagram ‘art-tra’ is formed between ‘c-art’ and ‘tra-p’, as well as the anagram tr-rt between ‘tr-ip’ and ‘ca-rt’.

Figure 11:

The model of McClelland and Rumelhart shows an architecture with several elaboration levels. The first level analyzes the visual characteristics of the letters, the second level processes individual letters independently of their graphic features, while the third level nodes process words as a unit. The system operates in parallel over the three levels and according to retroactive circuits: the activation patterns associated with a given representation of a word may influence others before the response threshold is reached. According to the theory proposed here, the combinatory structure of these associative patterns leads to the anagrammatic elaboration of lexical selection before the response thresholds for the words are reached.

In a connectionist matrix, words sharing combinatory segments are correlated and so can recall each other helping to explain different types of relationships in lexical access processes (Dell 1986; Dell and O’Seaghdha 1992; Dell et al. 1997). Without entering into specific neurophysiological details, in my opinion it is possible to put forward a model for the development of language, as I have explained in various books (Sasso 1999, 2007, 2011), from which it can be inferred how the lemma incorporates anagrammatic regulation and how this involves a particular concept of the generation of a poetic text.

5.3 The sign: The two combinatory matrices of the signifieds and the signifiers

This type of development implies that the sign is formed by the combinatory selection of the extremely numerous brain pathways distributed between the areas of Broca and Wernicke, a selection process initiated during the development of the infant by the slower maturation of the left hemisphere with respect to the right (Le May 1985; Hopkins and Marino 2000; Bradshaw and Rogers 1993; Andrew and Rogers 2002).

Since the two hemispheres are connected, the information processed in the right hemisphere, for example about an object, is continually transferred to the left hemisphere, but using a system of pathways that are plastic and less mature. Therefore, when the mother says the name of an object, the acoustic area next to Wernicke’s area that codes the name may form associative pathways with those processing the information about the object from the right hemisphere. Broca’s area also gradually forms associative pathways with this information, specializing in its recovery and in verbal articulation.

Although described in an elementary manner, this linguistic re-functionalization explains the “fluid structure” of the sign owing to its roots in the plasticity of the nerve pathways and their categorial elaboration processes, from whose combination each sign is formed and its associability with other signs. In particular, it explains how the development of the signs from a common combinatory matrix entails the emergence, from the maturation of two associative matrices, of the “signifieds” and the “signifiers.” The first matrix concerns the “categorial collection” of elaboration processes that form part of the networks gravitating between the two areas of Broca and Wernicke, indicating the associative development of the signifieds.

The second matrix concerns the associative differentiation of the signifiers which, during maturation, gives rise in areas adjacent to Wernicke’s area to the combinatory matrix of phonetic-graphemic categorial elements. Each signifier corresponds to a particular sequential ordering of such elements, i.e., to a specific succession of links between the elements, so that the differentiation of the ordering gradually produces in the combinatory matrix a dense fabric of pathways.

This means that a given set of categorial elements can be travelled by several order arrangements, which then refer back to sequential segments of different signifiers: a broad associative fabric is then formed in the matrix. The morphemic properties identify the more stable associative properties, while the anagrammatic properties represent an inherently unstable associative combination of categorial elements (without doubt Saussure was aware of this determining influence in poetic language).

5.4 The anagrammatic link and structural cooperation between the two matrices

Such considerations would suggest a fundamentally simple explanation for the anagram. In order to understand the functional origin of the lemma – and its natural role in the regulation of anagrams – the conceptual apex is the contemporaneous differentiation of the two matrices. The differentiation of the signs necessarily produces the structural co-operation of the combinatory matrices of the signifieds and the signifiers, the second of which includes the associative anagrammatic matrix. This process explains what happens in the generation of language.

When the conceptual-semantic preparatory phase selects a particular lemma, this has access to information concerning the signified and, at the same time, to morphological information about the word: this, in turn, activates the phonetic-graphemic combinatory structure. If two or more lemmas share both semic elements and phonetic-graphemic sequences that can be re-arranged (as with “stor-m” and “de-stro-y”), an associative link is established between the two fundamental language areas. With respect to other lemmas that share only semic elements, or only phonetic-graphemic elements, this grounding is more stable and favors its selection in the recursive elaboration phase between the two areas.

The anagrammatic link therefore corresponds to a characteristic of the lemma, that of acting as a mediating element between the conceptual content and the superficial utterance. More precisely, it can be interpreted as a link that “extracts,” and emphasizes, a semic-conceptual content from the combinatory relationship that is established between the two signs or between sections of the utterance straddling several signs. Similarly, in the inclusion and over-inclusion of several signifiers, it extracts a semic-conceptual content from the shared sequential order in the combinatory matrix.

5.5 The anagrammatic link and theories on metaphor

These considerations are therefore accompanied by an important theoretical concept. The anagram, where interpreted as a link between signifiers that links together certain semic properties, re-proposes the model of the metaphor already expressed by the Groupe µ, according to which the intersection between two signifieds leads to their union (Figure 12), but in a structural meaning that is naturally made more complex by the network of combinatory links maintained within the structure of the signifieds.

Figure 12:

For the Groupe µ, metaphor is based on a real identity formed by the intersection of two terms so as to affirm the identity of the whole two terms. It extends to the union of the two terms a property that belongs to their intersection.

This conceptualization shows that the anagrammatic link corresponds to the “interanimation” principle of Richards (1936), taken up by Black (1962, 1979), according to which a primary subject (topic) receives a projected “set” of implications of the secondary subject (vehicle). The implicative contribution of the anagram also translates in a combinatory sense the important theme of the expressive-conceptual limitations of lexical vocabulary. Philosophers of language draw attention to the fact that metaphor is a response to the incompleteness of natural languages (Urban 1939; Ullmann 1962; Alston 1964; Henle 1958) and to the “designative inadequacy” of lexical systems (Weinreich 1964). In poetry, therefore, a network of anagrammatic links “extracts” a conceptual content that cannot be expressed by the ordinary grammatical relationships of the signs of the language while, at the same time, expressing a metaphoric meaning.

The co-operation of the two matrices is evidently made difficult by the complexity of the phonetic-graphemic combinatory control, and this explains why in ordinary language the syntactic-grammatical regulation of the signifieds is normally performed by morphosyntactic regulation only. However, this intrinsic difficulty shows the particular coherence of poetic language. The careful selection of these processes, in fact, reveals the specific nature of semiosis in poetry. Anagrammatic relationships are recursive, that is semantic-thematic characteristics are repeatedly woven. The anagrammatic fabric therefore also assigns a particular meaning to the concept of “infinite interpretant” of Peirce (1931–1966), whose unlimitedness is circumscribed by the “recursive closure” of the combination of interpretants.

9.1 Structural co-operation between the beginning and end nuclei of the first line

As the last example, I shall show, in the first stanza of Il passero solitario (1835) by Leopardi, the extent to which combinatory control of the initial and end nuclei of the first line can act as a coherent support for the formal structure.

1. D’in su la vetta della torre antica,

2. Passero solitario, alla campagna

3. Cantando vai finché non more il giorno;

4. Ed erra l’armonia per questa valle.

5. Primavera dintorno

6. Brilla nell’aria, e per li campi esulta,

7. Sì ch’a mirarla intenerisce il core.

8. Odi greggi belar, muggire armenti;

9. Gli altri augelli contenti, a gara insieme

10. Per lo libero ciel fan mille giri,

11. Pur festeggiando il lor tempo migliore:

12. Tu pensoso in disparte il tutto miri;

13. Non compagni, non voli,

14. Non ti cal d’allegria, schivi gli spassi;

15. Canti, e così trapassi

16. Dell’anno e di tua vita il più bel fiore.

Figure 17:

The nucleus of “torre antica”.

Overall the stanza has 338 anagrams distributed over the ranks as follows R3 = 212, R4 = 70, R5 = 43, R6 = 8, R7 = 2, R9 = 2, R12 = 1, and its highest density nucleus is at the end of the first line (Figure 17): 18 anagrams are produced from this nucleus and support the main development of the stanza, and of these two are of particular semantic-thematic and formal importance. The segment straddling the end of the first line “antica / Passero” gives rise to the two most important anagrams in the stanza, the longest of 12 letters and the other of 9 letters, that produce the long sequence of 22 letters “s-passi cant-i e così trapass-i” forming the basis of the rhyme “spassi-trapassi” (Figure 18(a)). The rhyme “spassi-trapassi” is the most important in the stanza since it summarizes the dramatic contrast between happiness and unhappiness in the life of the rock thrush, and has a counterpart in the two longest anagrams. The twelve-letter anagram also expresses the negative point in this dramatization, the “trapassare-morire” (passing time – dying), illustrating the thematic hierarchy on which this flow is based.

Figure 18:

(a) The most important anagrams produce the dramatization of the rhyme. (b) The poem’s beginning guides the spatial metaphor of the sparrow’s solitude.

The first line has, in addition to the important end-of-line generating section, an equally significant generative nucleus at the beginning. It is made up of short anagrams that have the merit of highlighting equally important structural aspects. The meaning of the nucleus is clear from the first two anagrams that reach, respectively, to “in disparte” and “dintorno” (Figure 18(b)). The two convergences outline a topographical system “d’in sù, dintorno, in disparte,” that uses the spatial metaphor to express the solitude of the rock thrush: the bird is “d’in su,” high up in the tower, while “dintorno,” around him, the joyful springtime life unfolds with respect to which he is “in disparte,” apart.

9.2 The generative and anaphoric function of “disparte”

This clear differentiation stands out above all if we consider in Figure 18(b) the first anagram that reaches “disparte,” a word which by itself generates with four anagrams the important rhyme “spassi-trapassi.” If we superimpose Figure 18(a) and 18(b), we can easily see the structural co-operation maintained by the initial anagram and by the last two anagrams in generating jointly the rhyme “spassi-trapassi” (Figure 19(a)), through the tension of both the shorter and the longer combinatory flows.

The structural function of “disparte” is, in fact, even more complex. Let us observe the ascending flow of the two internal anagrams of “di-spart-e” that arrive at “pas-sero soli-tar-io” (Figure 19(b)). In the utterance, “disparte” is a locative adverb with an adjectival function (tu pensoso in disparte), and this adjectival function refers anaphorically with art-tar to its semantic equivalent “soli-tar-io.” However, the anagram pas-spa performs in “pas-sero” a function that is not directly anaphoric, since it qualifies the bird’s narrative and psychological character (the rock thrush as a bird apart).

Figure 19:

(a) The structural cooperation between the first line’s beginning and end. (b) The locative, adjectival, anaphoric and psychological function of “disparte”.

All these processes indicate why the contribution to the formal structure of the word “disparte” is a determining one, leading us to the conclusion that it has an evident hierarchical priority in the generative process. Consequently, this example makes us particularly aware of how to interpret the generation of the text. The order of the utterance does not correspond to the actual priorities that have guided its composition, since this is re-assembled from several nuclei whose combinatory fabric conveys the structural hierarchy that has generated the text.