tu stai menandomi per la gamba would be literally correct but misleading. In Italian one should say mi stai prendendo per il naso, thus substituting an English leg with an Italian nose. If literally translated, the English expression, absolutely unusual in Italian, would make the reader suppose that the character (as well as the author) was inventing a provocative rhetorical figure – which is completely misleading as in English the expression is simply an idiom. By choosing nose instead of leg a translator puts the Italian reader in the same situation as the original English one. Thus only by being literally unfaithful can a translator succeed in being truly faithful to the source text. Which is (to redeem the triviality of my example) like echoing Saint Jerome, patron saint of translators, that in translating one should not translate verbum e verbo sed sensum exprimere de sensu (even though the notion of the right sense of a text, as we shall see, can imply some ambiguities).

Besides, in the course of my experiences as a translated author I have always been torn between the need to have a translation that respected what I believed to be my intentions, and the exciting discovery that my text, independently of my early intentions, could elicit unexpected interpretations and be in some way improved when it was re-embodied in another language.
I do not need to develop these points in depth at this stage because the following pages will reveal them. What I want to emphasise is that many concepts circulating in translation studies (such as adequacy, equivalence, faithfulness) will be considered in the course of my lectures from the point of view of negotiation.

Negotiation is a process by virtue of which, in order to get something, each party renounces something else, and at the end everybody feels satisfied since one cannot have everything.
In this kind of negotiation there may be many parties: on one side, there is the original text, with its own rights, sometimes an author who claims right over the whole process, along with the cultural framework in which the original text is born; on the other side, there is the destination text, the cultural milieu in which it is expected to be read, and even the publishing industry, which can recommend different translation criteria, according to whether the translated text is to be put in an academic context or in a popular one. An English publisher of detective novels may even ask a Russian translator not to transliterate the names of the characters by using diacritic marks, in order to make them more recognisable to the supposed readers.

A translator is the negotiator between those parties, whose explicit assent is not mandatory. There is an implicit assent even in the reading of a novel or of a newspaper article, as in the former case the reader implicitly subscribes a suspension of disbelief, and in the latter relies on the silent convention that what is said is guaranteed to be true.
A last remark. The chapters of this book were born as lectures and when lecturing one does not include bibliographical references, except for canonical authors. Moreover a reasonable bibliography on translation studies today would contain hundreds of titles. I have only put footnotes where I felt in debt to an author for a particular suggestion. But I have refrained from tracing the origin of many other ideas because they are now current.7

NOTES

1. Now published as Experiences in Translation (Toronto: Toronto U. P., 2001).
2. A more consistent version of both my Toronto and Oxford lectures, including many other reflections and examples, is now published in Italian under the title of Dire quasi la stessa cosa (Milano: Bompiani, 2003).
3. Raymond Queneau, Esercizi di stile (Torino: Einaudi, 1983).
4. Gérard de Nerval, Sylvie (Torino: Einaudi, 1999).
5. Since I shall repeatedly quote some of these translations, let me provide the references here: Le nom de la rose, tr. Jean-Noel Schifano (Paris: Grasset, 1982); Der Name der Rose, tr. Burckhart Kroeber (München: Hanser, 1982); El nombre de la rosa, tr. Ricardo Pochtar (Barcelona: Lumen, 1982); The Name of the Rose, tr. William Weaver (New York: Harcourt/London: Secker and Warburg, 1983). Le pendule de Foucault, tr. Jean-Noel Schifano (Paris: Grasset, 1990); Das Foucaultsche Pendel, tr. Burckhart Kroeber (München: Hanser, 1989); El pendolo de Foucault, tr. Ricardo Pochtar and Helena Lozano Miralles (Barcelona: Lumen-Bompiani, 1989); El pendel de Foucault, tr. Antoni Vicens (Barcelona: Destino, 1989); Foucault’s Pendulum, tr. William Weaver (New York: Harcourt/London: Secker and Warburg, 1989). L’île du jour d’avant, tr. Jean-Noel Schifano (Paris: Grasset, 1996); Die Insel des vorigen Tages, tr. Burckhart Kroeber (München: Hanser, 1995); La isla del dia de antes, tr. Helena Lozano Miralles (Barcelona: Lumen, 1995); L’illa del dia abans, tr. Antoni Vicens (Barcelona: Destino, 1995); The Island of the Day Before, tr. William Weaver (New York: Harcourt/London: Secker and Warburg, 1995). As for Baudolino, always with the same title, see the translations of Jean-Noel Schifano (Paris: Grasset, 2002), Burckhart Kroeber (München: Hanser, 2001), Helena Lozano Miralles (Barcelona: Lumen, 2001), Carmen Arenas Noguera (Barcelona: Destino, 2001), Marco Lucchesi (Rio de Janeiro: Record, 2001), William Weaver (New York: Harcourt/London: Secker and Warburg, 2002).
6. Cf. my The Role of the Reader (Bloomington: Indiana a U.P., 1979 and London: Hutchinson, 1981) and The Limits of Interpretation (Bloomington: Indiana U.P., 1990).
7. See also for a consistent bibliography Mona Baker, ed., Routledge Encyclopedia of Translation Studies (London: Routledge, 1998).

I. The plants of Shakespeare

When I look in Webster’s Dictionary, among the definitions of translate I find ‘to transfer or turn from one set of symbols into another’. That is more or less what happens when we use Morse code and we translate from alphabetic letters to Morse signals or vice versa.

However Morse code involves a process of transliteration, that is, of substituting letters of a given alphabet with letters of another one, and letters are meaningless. Using Morse code, operators could correctly translate messages written in a language they do not understand into a series of dots and dashes. But Webster’s, under the entry translation, also says ‘a rendering from one language into another’. Since the signs of any language (be it verbal or other) signify something and are thus endowed with meaning, we can figure out that, given a set of symbols a, b, c, . . . z and another set of symbols α, β, γ, δ, . . . ω, we are entitled to substitute an item from the first set with an item from the second set if and only if, according to some rule of synonymity, a is equivalent in meaning to α, b to β, and so on.

What does ‘equivalence in meaning’ mean? There is a widespread opinion that meaning is exactly that which remains unchanged (or equivalent) in the process of translation, but such an assumption runs the risk of entering a vicious circle. Another suggestion1 is that the equivalence implied by synonymity and translation is a referential equivalence: a given word A used in a language Alpha is synonymous with the word B used in a language Beta if both are seen to refer to the same thing or event in the real world. However, referential equivalence is no better than equivalence in meaning.

We generally believe that it is correct to translate husband as mari or marito, and undoubtedly Mary’s husband refers to the same person as le mari de Mary or il marito di Mary. But one only need open a dictionary in order to see that husband, in English, can also mean a navy manager or a steward, while mari or marito cannot.

Let us assume, however, that synonymity exists, that equivalence in meaning is a value rigidly established by a linguistic convention, and that a machine can be provided with rules that allow it to operate according to that convention, so that it can switch from one symbol to another even though it does not understand the meaning of these symbols.

With this in mind, I accessed Babelfish, the automatic translation system provided by AltaVista. I gave it a series of English expressions and asked it to translate them into Italian. Then I asked AltaVista to retranslate the Italian expressions back into English. Only in the last case did I follow a more complicated path, that is: English–Italian–German–English. Here are the results:

1. The works of Shakespeare → Gli impianti di Shakespeare → The plants of Shakespeare
2. Speaker of the chamber of deputies → Altoparlante dell’alloggiamento dei delegati → Loudspeaker of the lodging of the delegates
3. Studies in the logic of Charles Sanders Peirce → Studi nella logica delle sabbiatrici Peirce del Charles → Studien in der Logik der Charlessandpapierschleifmaschinen Peirce → Studies in the logic of the Charles of sanders paper grinding machines Peirce
   Let us stick to case number 1.

AltaVista undoubtedly has definitions and dictionary translations in its ‘mind’ (if AltaVista has a mind of any description), because it is true that the English word work can be translated into Italian as impianti and the Italian impianti can be translated into English as plants. Plainly we must give up the idea that to translate means only ‘to transfer or turn from one set of symbols into another’ because, except for cases of transliteration like Morse code, a given word in a natural language Alpha frequently has more than one corresponding term in the natural language Beta.

Besides, the problem does not only concern translation but also the very comprehension of a language Alpha on the part of its native speakers. What does work mean in English? According to Webster’s, work can be an activity, a task, a duty, the result of such activity (such as a work of art or a literary masterpiece), a