Accessibility / Report Error

Meaning and significance in poetic translation


When translating a pragmatic text, special attention must given to the concepts, and to the relationship between them, so as to safeguard their communicative function and their tendency to univocality. In poetic translation, there is a predominance of the material elements of the sign and a shift to the higher level of oblique signification, i.e., to the semiotic level of significance, allowing multiple readings by breaking the external referentiality.

Translation; Translator; Poems; Jacques Prévert; Carlos Drummond de Andrade

Na tradução do texto pragmático, a atenção deve voltar-se particularmente para os conceitos e para a relação entre eles, de modo a salvaguardar a sua função comunicativa e a sua tendência à univocidade. Na tradução poética, predominam os elementos materiais do signo e a passagem para o nível superior da significação oblíqua, para o nível semiótico da significância, o que permite a leitura múltipla pela quebra da referencialidade externa.

Tradução; Tradutor; Poemas; Jacques Prévert; Carlos Drummond de Andrade


Meaning and significance in poetic translation1 1 The article was published in French in the Canadian magazine Meta (v.2, p.217-222, 1996),

Mário Laranjeira

Mario Laranjeira is a retired professor from the Faculty of Philosophy, Letters and Humanities of the University of São Paulo and a translator of works on French philosophy, literature and essays. In 1997 he won the Jabuti award for the translation of Poetas de França hoje. He was visiting professor in Toulouse, Rennes and Bordeaux, in France, and lives in São Paulo. He has over forty publications, including authorial books, translations and articles. @ -


When translating a pragmatic text, special attention must be given to the concepts, and to the relationship between them, so as to safeguard their communicative function and their tendency to univocality. In poetic translation, there is a predominance of the material elements of the sign and a shift to the higher level of oblique signification, i.e., to the semiotic level of significance, allowing multiple readings by breaking the external referentiality.

Keywords: Translation, Translator, Poems, Jacques Prévert, Carlos Drummond de Andrade.

It is not the aim of this article to discuss whether or not poetry is translatable. Let us start from the fact that poetry has always been and always will be translated, and examine, preferably, the specifics of translational operation whose object is poetry in its most characteristic textual expression: the poem.

Let us observe, first, that poetry can be translated without necessarily involving poetic translation, as it has unfortunately often been done by many translators, especially in France. Let us precisely try to find in the text what constitutes the expression of poetry and see which operations in the act of translation enable "transferring" this poetry to the translated text; in other words, how to translate a poetic text so that it is expressed as poetry in the target language-culture.

In the typological distribution of texts there are no impervious, walled in compartments, but rather two poles towards which the texts are drawn, occupying spaces with more or less undefined boundaries.

On the one side is the pole of the concept, the meaning, and on the other is the pole of the signifier, the material element of the sign. Some texts are marked by the primacy of the concept. What is important to preserve are ideas, facts, relationships, processes. There is commitment to a reality outside the text, with a rationality considered as objective, with a logic that is governed by the criterion of truth. In this case, the relationship between the signified and the signifier appears as completely arbitrary, and the linear character of the signifier is imposed. This is reduced to the condition of the vehicle of the concept, which leads me refer to this type of texts as vehicular texts. This category includes, inter alia, the so-called scientific, technical, informative, pragmatic, and argumentative texts. Thus, in the report of a physical experience or of an operation, in the demonstration of a geometry theorem, in the instructions for the use of a device, in the description of a car accident, the primacy of the concept over the materiality of the sign is imposed, and the relationship with an external referent requires a linear and univocal reading for the function of the text to be totally fulfilled.

Therefore, the translator who rewrites a text of this kind should first and foremost ensure that the meaning of the text is respected as strictly as possible; and to that end the choice of vocabulary and sentence structure should be based on the concepts conveyed, in order to allow this clear and univocal reading that will avoid "misunderstandings" and ensure the equivalence of the target text to the source text with respect to its performance from the standpoint of the communicative function.

In turn, there are texts that tend towards the pole of the signifier, in which the materiality of the sign prevails over the concept: the "phoneme" retrieves its "sound" value, the written text often takes on iconic aspects, the arbitrariness of the sign is weakened in favor of the motivation of the relationship between signified and signifier, and this ceases to be a mere vehicle of the first in order to determine it, to produce it. Linear reading is replaced by retroactive and tabular reading that redefines the relationship between the elements of the discursive chain. Considered from both the linguistic and semanalytic standpoint, syntactic linearity is shattered, allowing the semiotic to penetrate the symbolic (cf. Kristeva 1974, p.40, 67-9 passim) under the action of the subject's bodily drivers (pulsing). It is this new way of producing meaning that happens within the text through the interplay of forces that subtend the meaning which is called significance, as opposed to referential meaning.

Significance is responsible for opening the meaning to multiple readings, all of them plausible, and this is one of the trademarks of the poetic text, as opposed to the univocity of the vehicle text. As it can be seen, we are facing a very different way of signifying, which is the trademark of the poetic text, of the poem. According to Michael Riffaterre (1983, p.11), "the poem tells us one thing and means another, and this is explained entirely by how the poetic text generates its meaning." Riffaterre ( 1983 p.13ss.) reserves the term meaning for the information provided by the mimetic text2 2 "Mimesis is characterized by a semantic sequence with continuous variation, because representation is based on the referential character of the language" (Riffaterre, 1983, p.13). , and uses the term significance to designate this formal and semantic unit that contains the levels of obliquity. These levels indicate that the text causes the limits of mimesis to explode and should be read and interpreted in another dimension, that of semiosis, which exceeds the strictly linguistic level. Significance is a manifestation of semiosis.

The translator of a poem should thus have before his text an attitude quite different from that of the translator of a vehicle text. While the latter translates especially the meaning, the former must, in his rewriting operation, pass on to his text the specific significance of the original poem, which is its identification card.

But as the translator's activity always begins with reading, this reader-writer must take into account all these facts that are observable and identifiable as responsible for the semantic obliquity of the poem, i.e., the textual markers of significance, and work to recover them in the text he produces. These markers are numerous. I will mention only a few among the most important ones.

To exceed the limits of mimesis and attain significance, the reader-translator must overcome certain obstacles, the first of which consists of agrammaticalities. If we assume that grammar is the basis that supports the pillar of referential significance, all violence against grammatical normality can be seen as a nullification of mimesis, i.e., agrammaticalities are indications that the text should be read at another level. The meaning, which is threatened at the mimetic level, is recomposed at the higher level of semiosis. Thus, agrammaticalities, which by themselves are not sufficient to constitute the semiotic level of the text should, however, be considered as keys of significance.

I should point out that contrary to how it is used by generative grammarians, the term agrammaticality is used here in a broad rather than a restrictive sense. It may designate from the minimal cases of disturbance of syntactic linearity to those extreme cases that lead to hermetics or stumble on nonsense. The verse itself, whether regular or free, can be seen as a type of agrammaticality to the extent that it is defined as a disturbance of grammatical linearity:

The syntactic order of the elements in the sentence and of the sentence in the paragraph is, in poetry, crafted and challenged by the verse. Unlike prose as a "discourse that moves forward," the verse shifts the elements and superposes principles of meter and parallelism to grammatical linearity. (Adam, 1985, p.221)

Similarly, the lexical, semantic, syntactic and phonic recurrences that draw the reader's attention for their unusual character can be considered as agrammaticalities in a broad sense.

The translator, who should "translate agrammaticalities" rather than "eliminate them" at the risk of losing one of the keys of significance, will many often feel uncomfortable by the fact that different languages have different grammars and, therefore, agrammaticalities are not the same or are of a different nature. Thus, the act of placing the adjective before the noun may in some cases be an agrammaticality in French, Portuguese and Romance languages in general. This agrammaticality cannot be translated by an agrammaticality of the same nature in English, by the simple fact that in this language the adjective precedes the noun as a rule. The translator should therefore violate another point of grammar to recover an equivalent level of agrammaticality.

Another textual index of significance that should be pointed out is the dual sign. Michael Riffaterre (1983, p.113) defines the dual sign or nodal point as "an equivocal word situated at the point where two sequences of semantic or formal associations intersect." When the dual sign happens to be the title, its strength as an obliquity factor is remarkably increased because the title is often the matrix of which the poem is the expansion. As the dual sign is actually a single signifier that in a given language conveys two different meanings, its use in poetry is a special case of "wordplay" or "pun". Well, it is known that wordplay is generally untranslatable by the fact that in another language usually there will be a different signifier for each of the meanings of the dual sign. The translator of poetry should use his imagination and mastery of the resources of his own language to recover the poetry effect which, in the original, stems from the use of a dual sign; otherwise, an element of the significance will be irretrievably lost. This is the case of the translation of Carlos Drummond de Andrade's (1976, p.120-1) Poema-orelha by Jean-Michel Massa, of which I transcribe the first four verses below:


Voici l´oreille du livre

par où le poète écoute

si on parle mal de lui

ou si on l´aime.

It is a word-for-word translation of the Portuguese original. But Drummond had "played" with the word ear which, in Portuguese, is a dual sign: it means the outer ear and the cover flap of a book in which the editor often includes comments on the contents of the book or the opinion of critics and readers. This ambiguity does not exist in French; the obliquity of signification is lost. Voici l' oreille du livre is perceived, in French, as a figure of invention, as a new metaphor, whereas in Portuguese the "ear" of the book is an expression commonly used and attested by dictionaries. The interest and originality of the Brazilian text lie in that the poem is the expansion of the matrix title; this means that the grammar of the significance of the text rests precisely in exploring the relationship between the two meanings expressed by one and the same signifier.

The textual interpretants constitute a third indication of the shift from the mimetic level to the semiotic or significance level that poetic translation should take into account. "The shift from meaning to significance necessitates the concept of interpretant, that is, a sign that translates the text's surface signs and explains what else the text suggests," adds Riffaterre3 3 As seen, the concept of interpretant in Riffaterre has a wider meaning than in Peirce (1977, p.43). (1983, p.107).

Let us examine, for example, this short poem by Jacques Prévert (1985, p.38-9):


C´est ma faute

C'est ma faute

C´est ma très grande faute d'orthographe

Voilà comment j'écris


The title of the poem is a textual interpretant in that it refers to the Confiteor, a prayer which is part of the Christian tradition and that in each language-culture has a canonical, official formula, a fixed form. Prévert reproduces verbatim part of the canonical text in French: "C'est ma faute, c'est ma faute, c'est ma très grande faute." We are immersed, through this interpretant, in the isotopy of the Christian ritual, of sin and forgiveness ... But to the reader's surprise, right in the middle of the poem there is a rupture caused by the dual sign faute4 4 In French the word " faute" can mean both an "error" of any kind" and "sin or guilt." , which although belonging to the isotopy of religion casts us, by its determinant "d'orthographe", in the isotopy of the "school" in which the entire sequence of the text is situated. The mode of expansion of the matrix or the grammar of significance is therefore focused on a textual interpretant (Confiteor) and on a dual sign (faute).

The translation of this poem by the Brazilian poet and translator Silviano Santiago is transcribed below (Prévert, 1985):




Que enorme erro de ortografia

Eis como escrevi


(My mistake

My mistake

What a huge spelling mistake

Here is how I wrote


The translated text entirely lost the poeticity of the original for the simple reason that the translator was unable to maintain the grammar of significance, i.e., in the target text the textual interpretant, which would be the canonical text of the Confiteor in English ("My fault, my fault, my most grievous fault ... "), and the dual sign disappear.

I would therefore propose the following translation, which takes into account the aforementioned remarks:


Minha culpa

Minha culpa

Minha máxima culpa em ortografia

Vejam como escrevi


Thus, in addition to maintaining the interpretant and the shift from the isotopy of "religion" to the isotopy of "school", some other elements of significance such as the rhythm of the original and the primacy of the material element of the sign over the concept can also be maintained. Undoubtedly, by translating "giraffe" ("girrafe" Santiago's translation) for "bassia" (literally "cuvette" or "bassin" in French; basin in English), I did something totally inconceivable in pragmatic translation but perfectly plausible in poetic translation. Indeed, when Prévert chose the word "giraffe" to end his text, he did not do it because of its semantic components (animal, mammal, quadruped, long neck, etc.), but for two specific reasons: first for the phonic recurrence (orthographe/giraffe) and then because the double consonant "f" in that word is at the level of spelling only, with no phonic or phonemic consequence. It is the same with the word "bassia" (whose correct spelling in Portuguese is "bacia"), which rhymes with "ortografia" and contains a strictly spelling error. Only the material elements of the word were considered. Well, Santiago's option, by attempting to maintain the conceptual elements of "girafe" (girafa in Portuguese) lost the phonic recurrence and added a phonetics error to the spelling error: in Portuguese the intervocalic "rr" (double "r") is pronounced differently from the intervocalic "r" (single "r").

Other textual markers of signification deserved to be examined here because of the poetic translation of the poem, but the limits of this article require some restrictions. I will limit myself, therefore, to mentioning one more before moving on to my conclusion. I am talking about what J.-M. Adam (1985, p.29) referring to an article by Jacques Aris calls visual legibility of the poetic text. We have seen that one has access to the signification of the poem by performing a two-step reading: the first is linear, mimetic; and the second is retroactive, tabular and semiotic. There is, however, a pre-reading that is strictly visual, based on the spatial distribution of the textual mass on the page. It does not participate in the linguistic discursivity, but presents itself as a global and achronic, no-sequential perception, as in fine arts.

"Seeing in poetry an art of the word is situating the message in time and thus favoring the sequential decoding that establishes a linear hierarchy of the linguistic materials to be taken into consideration in a given order. Taking, on the contrary, the page as a place of poetic expression means immediately giving primacy to a globalizing decoding [...]." The poem will first show itself as a spatialized macro-sign ... (Delas and Filliolet, 1973, p.176).

The main function of visual legibility is to generate the effect of the poem or the "poem-effect" (cf. Adam, 1985, p.29). By casting a glance at the page containing the text, one sees that it is a poem and not a newspaper article, a letter or a story, and this creates in the reader a predisposition to a poetic, non-referential reading that will seek oblique signification, poetic significance.

For the translator of poetry, the translation starts with transposing visual legibility. A sonnet should be translated for a sonnet, a poem in free verses for a poem in free verses, and so on. Doing otherwise would mean straying from the translation towards a free recreation or, referring to the term used by Joachim du Bellay back in the sixteenth century, towards simple innutrition. Translation has always been committed to the visual legibility of the original. This commitment allows some flexibility, no doubt, but the translator cannot ignore it and should try to recover it. Respect for visual legibility plays an important role in poetic translation at all times, in that it is the guardian of certain traits that place it in its time and in its cultural space; but this importance grows when it comes to the texts of some modern poets like E. E. Cummings,5 5 On the role of spatiality in the translation of E. E. Cummings, see article by Guy Leclerc (1987). or of neoconcretist poets, for example.

In short, if one can speak of faithfulness in translation, faithfulness in poetic translation will consist in recovering in the target text the textual markers of significance, so that the target text can be not only a poem in the receiving language-culture, but a poem that is homogeneous with the original poem in what constitutes its poetic identity.



  • ADAM, J.-M. Pour lire le poème Bruxelles: De Boeck, 1985.
  • DELAS, D.; FILLIOLET, J. Linguistique et poétique Paris: Larousse, 1973.
  • DRUMMOND DE ANDRADE, C. Réunion/Reunião, poèmes choisis Traduits et préfacés par Jean-Michel Massa. Paris: Aubier Montaigne, 1976.
  • KRISTEVA, J. La révolution du langage poétique Paris: Seuil, 1974.
  • LECLERC, G. Traduire la poésie, c'est faire de la poésie. Quelques jalons dans l´approche d´un poème de e. e. cummings. Revue d'Esthétique, Toulouse, n.12, p.117-119, 1987.
  • PEIRCE, C. S. Semiótica São Paulo: Perspectiva, 1977.
  • PRÉVERT, J. Poemas Seleção e tradução de Silviano Santiago. Rio de Janeiro: Nova Fronteira, 1985.
  • RIFFATERRE, M. Sémiotique de la poésie Paris: Seuil, 1983.
  • 1
    The article was published in French in the Canadian magazine
    Meta (v.2, p.217-222, 1996),
  • 2
    "Mimesis is characterized by a semantic sequence with continuous variation, because representation is based on the referential character of the language" (Riffaterre, 1983, p.13).
  • 3
    As seen, the concept of interpretant in Riffaterre has a wider meaning than in Peirce (1977, p.43).
  • 4
    In French the word "
    faute" can mean both an "error" of any kind" and "sin or guilt."
  • 5
    On the role of spatiality in the translation of E. E. Cummings, see article by Guy Leclerc (1987).
  • Publication Dates

    • Publication in this collection
      22 Nov 2012
    • Date of issue
      Dec 2012


    • Received
      13 July 2012
    • Accepted
      27 July 2012
    Instituto de Estudos Avançados da Universidade de São Paulo Rua da Reitoria,109 - Cidade Universitária, 05508-900 São Paulo SP - Brasil, Tel: (55 11) 3091-1675/3091-1676, Fax: (55 11) 3091-4306 - São Paulo - SP - Brazil