Vittoria Prencipe, Università Cattolica del Sacro Cuore
Translation Universals: Assets and Limits of research methodology
Abstract: The paper aims to provide a new interpretation of Translation Universals as a basis for identifying a text as a translation, sustaining that these are primarily influenced by the cultural context and the addressees of an interpreted text rather than determined by the act of translating.
1. Introduction
Scholars of Translation Studies (TS) have long been concerned about the relationship between Source Text (T1) and Target Text (T2). This bond has been commonly defined as equivalence. The concept has however triggered ongoing debate among translation scholars and particular criticism among literary scholars. Equivalence, as a mathematical concept, is easily defined.1 Yet, it is difficult to observe a mathematical equivalence in texts translated from natural languages. Thus, translation scholars have tried to determine equivalence at various levels in texts. Along with lexical equivalence – only seen in isolated cases – a structural equivalence has been identified, which mainly entails the correspondence of various elements constituting a text, and therefore occurring only among languages similar in structure. Moreover, two more kinds of equivalence have been hypothesized: textual equivalence, which considers the text as a unique sign whose sense is transferred to an equivalent text, and dynamic equivalence, obtaining the same communicative function of the translated text2.
The concept of equivalence has enabled scholars to overcome strict dichotomies, often too strict for translation, allowing general consideration of the facets←35 | 36→ constituting a text and the use of the text within a specific culture. The latter perspective has been particularly neglected by scholars for years.
Thirty years ago the entire branch of TS, that is Descriptive Translation Studies (DTS), started focusing exclusively on the analysis of translated texts, highlighting characteristics long considered secondary. Firstly, the translated text was seen as an independent text, with a specific role within a target culture. Through analysis referring to the T1, to other translated texts and to the original literature belonging to the target culture, the identification of the defining traits of a translation was made possible distinguishing translated texts from texts which have not been translated, resulting in a specific, independent group of texts3. These features are commonly defined as Translation Universals (TU)4 and became a tool, often conflicting with equivalence, to investigate the relationships between T1 and T2.
The purpose of this paper is to show how, although with positive effects on TS, the concept of TU can be forced, too. The paper will proceed as so: a definition of TU from the historical context in which it originated; a contrastive analysis of two Latin translations of Nicomachean Ethics, focusing on some of the previously identified TS.
2. Origins and definition of Translation Universals
The shift in perspectives within TS, has led to the development of more accurate tools for research. The ever-increasing importance of the role of corpora within TS has enabled scholars to analyse a great number of texts in a short time. Thanks to the use of corpora of texts in L1 and L2, Gideon Toury (1978) isolated preferential translation strategies and provided the concept of norm. These norms are options regularly taken up by translators at a given time and in a given socio-cultural situation.←36 | 37→
A specific context entails regularity and preferential option. The underlying shift of both of these as universals is subsequent. Mona Baker (1993, 243) suggests using corpora to reach the ‘elucidation of the nature of translated text as a mediated communicative event’. To do so, scholars must create tools to permit the highlighting of common features of translated texts independent of interference of a determined linguistic system, and which are not in line with a determined socio-cultural situation.
There are two main methods used to identify TU: the first formulates a hypothesis on the basis of contrastive analysis between T1 and T2, consequently this method focuses on the translation process and on the work of the translator. The second consists of contrastive analysis between T2 and texts which can be compared in the target language, highlighting the features of the so-called language of the translation, shedding light on the influence of the latter in the target language and culture5. Unfortunately, the distinction between TU as features of the process and features of the product has been often neglected.
It is Andrew Chestermann (2004a) who systematises the terminology. He distinguishes (i) S-Universals (source-related universals), i.e. universal differences (or differences that are subject to become universal) emerging from comparing T1 and T2 and (ii) T-Universals (target-related universals), i.e. universal differences between translated text and comparable original texts in the target language. The latter are the differences which determine a translation on the basis of its sole linguistic structure.6
Definitions of universals do not take this dichotomy into account. The starting point of Chestermann's critical analysis consists of a definition which refers only to what he calls T-Universals:
[…] we can define a translation universal as a feature that is found (or at least claimed) to characterize all translations: i.e. a feature that distinguishes them from texts that are not translations (Chestermann 2004b, 3).
Whilst considering terminological nuances, the following features are considered universals: explicitness, disambiguation and simplification, strong preference for conventional grammaticality, tendency to avoid repetition, and general tendency←37 | 38→ to exaggerate features of the target language. Still, Chestermann and Baker do not specify whether, to be considered as such, a translation needs the presence of one or all the universals listed above. Applicative studies generally indicate the simultaneous presence of one, or at least two universals.
3. Comparing translations
This paper will consider three of the features listed above. In particular, the general tendency to exaggerate features of the target language, namely normalization, simplification, and explicitation (a.k.a. explicitness). Simplification is active at several levels: at a lexical level it can be characterized as the process / product of reducing word count; at a syntactic level, it is found in substituting complex structures, e.g. by changing non-finite clauses with finite ones and by suppressing suspended periods; finally, there are various kinds of stylistic simplification, such as the tendency to use paratactic constructions in place of long sequences of sentences or hypotactic constructions. Explicitness can be characterized as the process of expressing the progression of thoughts more clearly, by interpreting obscure, contracted, or compressed passages. A clearer text is thus achieved by adding sentences and logical connective, such as conjunctions, and by introducing new information in the target text, developing short descriptions or rendering of implicit data. Finally, normalization is the process of constructing the target text using a language that is customary for the addressee. The Universals listed above are S-Universals, and are the outcome of the comparison between T1 and T2.
The starting point of our analysis is given by the hypothesis that all the features introduced are specific to the process of translations, in virtue of their being general features of every process of interpretation oriented at explaining a text to some specific end user. The source text used to exemplify our hypothesis is the Latin medieval and Humanistic translation of Ethica Nicomachea of Aristotle. Note that, since we deal with a philosophical text, in which every word has a precise meaning and little space is dedicated to descriptive or narrative structures, phenomena like TU rarely occur, so their presence in translations will have a greater theoretical impact. The translations considered here are quite relevant, both for their historical context and for their addressee: the first, the so called vetusta translatio, by R. Grosseteste (1229–1235), revised by Willem of Moerbeke as requested by St. Thomas Aquinas, who will use it as the a base for his comment; the second, by L. Bruni (1496), was destined to become a long-standing, fundamental text for academic teaching.←38 | 39→
Bruni openly criticizes the medieval vetus interpres on several occasions, which is why a comparison between the two versions is so interesting (Marchesi 1904, 78 ff.; Franceschini 1955). The vetusta translatio is certainly based on a very precise philological work and the correspondence with the source text is almost complete, both from a lexical and a structural point of view: the length of Latin sentences coincides with the Greek text and the syntactical structures are rendered with the nearest equivalent. This behaviour leads Bruni to accuse the vetus interpres of not knowing the language from which he translates, nor the subject of the text.
Thus, the text does not respect the Universals of simplification and explicitation, as one would expect; we assist a phenomenon which is opposite to normalization, at a lexical and syntactic level, consisting in being wholly oriented to the source text and giving rise to a non-standard text, i.e. a text which is difficult to recognize as a standard text of the target language. An example taken from the first chapter of book II follows:
(1) Διττῆς δὴ τῆς ἀρετῆς οὒσης, τῆς μὲν διανοητικῆς τῆς δὲ ἠθικῆς, ἡ μὲν διανοητικὴ τὸ πλεῖον ἐκ μὲν διανίας ἔχει καὶ τὴν γένεσιν καὶ τὴν αὔξησιν, διόπερ ἐμπειριας δεῖται καὶ χρόνου, ἡ δ’ ἠθικὴ ἐξ ἔθους περιγίνεταιταὃθεν καὶ τοὔνομαοἔσχηκε μικρὸν παρεκκλῖνον ἀπὸ τοῦ ἔθους. ἐξ ὃυ καὶ δῆλον ὅτι οὐδεμία τῶν ἠθικῶν ἀρετῶν φύσει ὴμῖν ἐγγίνεται
(2) Duplici autem existente virtute, hac quidem intellectuali, hac autem morali: ea quidem intellectualis, plurimum ex doctrina habet et generationem et augumentum. Ideo experimento indiget et tempore. Moralis vero ex more fit. Unde et nomen habuit parum declinans a more. Ex quo manifestum, quoniam neque una moralium virtutum natura nobis insit (Vetusta translatio).
(3) Cum vero duplex sit virtus: alia intellective, alia moralis: intellectiva quidem plurimum ex doctrina generationem habet atque argumentum. Quapropter experientia indigent tempore. Moralis autem ex assuetudine sit. Unde et nomen habet parumper a more declinans. Ex quo manifestum est nullam moralium virtutum natura nobis inesse (Aristotelis Stagyritae Ethicorum libri ad Nicomachum Interprete Leonardo Aretino).
I) Analysis of text 1: Apart from the obvious lack of articles in the Latin text and the inversion in the rendering of ἀρετῆς οὒσης due to the syntactical difference between the genitivus absolutus in Greek and the ablativus absolutus in Latin, we find a one-to-one correspondence at lexical level, with respect to both the form and the content. The syntactical structure is also the same: there is perfect correspondence in syntactical constructions, e.g. the rendering both of the genitivus absolutus and of the conjunction διόπερ; the word order is also the same. More←39 | 40→over, it is interesting to observe how the focus on the etymology of words in the source text is excellently rendered in the translation. Whereby, in the original text we find ἠθικὴ which derives from ἔθος ‘habit’, in the Latin translation of Grosseteste we find a correspondence between moralis and mos, a habit that entails a typical attitude, a custom.
II) Analysis of text 2: As for morphological constructs, Bruni’s translation adheres more to the source text. The adjective διανοητικός is not rendered with the more generic intellectuali, but with intellective, this term being composed of the suffix -ivus, which, like the Greek -ικός, entails causality. As for syntactic constructs, the ablativus absolutus, which corresponds to the Greek genitivus absolutus, is seen as an archaism (from the II century A.D). We cannot however state that the use of the most common construct in the XV century, cum plus conjunctive, occurring in the text by Bruni, is explicitation. We can rather define the use of the ablativus absolutus by vetus interpres as an archaism, justified by the necessity to show greater adherence to the source text, and the use of cum plus conjunctive is then to be intended as normalization. As for stylistic construes, it is worth noting Bruni’s choice of assuetudo, instead of mos, to translate ἔ to. There are two explanations for Bruni’s choice. (i) It is possible that mos had become more difficult to understand for the audience of the time or that its semantism had subsequently specialised to indicate only “institution” or “custom”. This would explain why it is used in a different context in the sentence which immediately follows. Thomas Aquinas, in the commentary to the passage, explains the meaning of mos adding sive consuetudo. In this case, Bruni, who presumably was aware of Aquinas’s commentary, would have carried out a process of normalization. Still, Bruni’s choice would make sense if the target audience were uncultivated, and did not know classical Latin, whereby the term mos, within that specific context, would have been incomprehensible. (ii) On the contrary, if the target audience were a specialised one, the strategy adopted by the translator could be defined not as normalization but as explicitness. A specialised audience would have understood the most ancient meaning of mos and the semantism it shares with assuetudo; this audience would nonetheless appreciate the translator’s ability to highlight the different facets in meaning of the two terms. Finally, it is interesting to observe Bruni’s use of the abstract term experientia to correspond to ἐμπειρία, whereby vetus interpres uses the concrete term experimentum. In fact, even if both terms are used interchangeably in classical Latin, the abstract one is linguistically closer to the original.←40 | 41→
4. Conclusions
On commenting these results we tried to connect TU with translation strategies, dictated by the demand for the best text for the intended audience. It is clear how the process of normalization is in tension with the need to render the content of the source text in the best way, as codified by its author. Hence, translations based on T1-oriented interpretations, typically addressed to a specialized audience, do not manifest this universal, whereas translations addressed to a general or cultivated audience, namely texts in which the clarity of the language is more important than the perfect correspondence with the source, are commonly constructed as normalized texts. Furthermore, the cultural level of the audience is the decisive element in the choice of a process of simplification vs explicitation. If we aim to provide a text for a wide or non-specialized audience, we will choose simplification. Whereas, if we aim to provide a text for a specialized audience, we will choose explicitation. Note that these TU are two aspects of a unique strategy: explanation. In the first case, where simplification is at work, the explanation is directed towards people of an average cultural level, while in the second case, where explicitation is at work, explanation is directed towards people of a higher cultural level.
On interpreting the genesis of TU in this way, I try to shed light on the limits of TU’s method of analysis, limits which depend on the unclear theoretical and methodological systematization of TU. My conclusion is that the universals of translation, intended as T-universals – on the basis of the terminology systematised by Chestermann – are the result of the universals of translation intended as S-universals, so that the features of the process determine the feature of the product. Furthermore, the S-universals are the result of strategies of interpretation adopted by the translator, who certainly is an interpret, to better achieve his principal aim: communicate to an intended audience. Finally, two other points can be stressed: 1) it is untrue that the concept of TU and the concept of equivalence are incompatible. From our point of view, the translator, who constantly makes choices at a strategic level, tends to select the best strategy to communicate a message to his audience with an equivalent meaning of what is transmitted by the source text. Hence, the very requirement of equivalence justifies the occurrence of S-Universals in the process and of T-Universals in the target text; 2) the influence of the structure of the target language should not be neglected. Even if we are convinced that anything can be communicated in any given language, we are also sure that the target language can drastically determine the way in which a message is communicated. As Jakobson (1959, 233) emphasized, the structure←41 | 42→ of the target language can make explicit / neglect some of the semantic traits transmitted by the words of the source language. In such cases, the pursuit of equivalence necessarily leads to a process of explication / simplification.
While point 2) is of great importance for defining a frame in which to compare texts and translations written in specific languages, point 1) is essential to understand the very process of translation: T-Universals are not only the consequence of the process of interpretation of texts, but are such that they work for the sake of equivalence. Indeed, if equivalence was not pursued, there would be no space for such systematic alterations of the target text in the process of translating. Our concluding proposal is the following definition of T-Universals: given a source text, a T-Universal is a trait deriving from systematic alterations of the target text depending on an interpretative strategy aimed to provide the best equivalent text for a given audience.
Bibliography
Baker, M. (1992): In other words. A Coursebook on Translation. London/New York.
Catford, J. (1969): A linguistic theory of translation: an essay in applied linguistics. London.
Chestermann, A. (2004a): Beyond the particular. In: Mauranen, A. / Kujamäki, P. (eds.): Translation Universals. Do they exist?, Amsterdam, 33–50.
—.(2004b): Hypotheses about translation universals. In: Hansen, G. / Malmkjaer, K. / Gile, D. (eds.): Claims, Changes and Challenges in Translation Studies, Amsterdam/Philadelphia, 1–13.
Even-Zohar, I. (1978): The Position of Translated Literature within the Literary Polysystem. In: Neergard, S. (1995) (ed). Teorie contemporanee della traduzione Milan.
—.(1990): “The Literary System” in Polysystem Studies. In: Poetics Today 11/1, 27–44.
Franceschini E. (1955): Leonardo Bruni e il “vetus interpres” dell'Etica a Nicomaco. In: Medioevo e Rinascimento. Studi in onore di Bruno Nardi, 299–319.
Jakobson J. (1959): On Linguistics Aspects of Translation. In: Brower, A. (ed.): On Translation. Cambridge, 232–239.
Marchesi C. (1904) L’Etica Nicomachea nella tradizione latina medievale. Messina.
Nida, E. (1964): Towards a science of translating with special reference to principles and procedures involved in Bible translating. Leiden.←42 | 43→
Nida, E. / Taber, C. (1969) The theory and practice of translation. Leiden.
Prencipe, V. (2006): Traduzione come doppia comunicazione. Un modello Senso – Testo per una teoria linguistica della traduzione. Milano.
Salsnik, E. (2008): From Translation Universals to the Italian of translations. In: Montella, C. / Marchesini, G. (eds.): I saperi del tradurre: analogie, affinità, confronti, Milano, 101–132.
Toury, G. (1995b). Descriptive Translation Studies and Beyond. Amsterdam: John Benjamins.←43 | 44→ ←44 | 45→
1 In mathematics, a relation is called equivalence if, and only if, it satisfies three properties: reflexivity (xRx), symmetry (xRy => yRx: if a weighs as b, b weighs as a), transitivity (xRy e yRz => xRz: if Luke is the brother of James, and James is the brother of George, then Luke is also the brother of George).
2 The question of equivalence, not being the subject of this paper, is hugely simplified here. The problem, however, is a very complex one and often appears in the reflections of translation theorists and translators who, like me, believe in the necessity of an exisiting relationship between T1 and T2. For details, see: Prencipe (2006), cap. 5.
3 Cf. Even-Zohar Polysystem Theory (1978, 1990), which sees culture as a polysystem, within which literature and translated literature constitute mutually independent, interacting each other and with the other subsystems.
4 This term is not always accepted. After initial enthusiasm for this hypothesis, there are today increasingly frequent doubts about its operation. In some cases we can find less binding appellations, such as shift, regularities (Chestermann 2004a). Toury (1995), as we will see, prefers to speak of norms, typical translator behaviour which is reflected clearly in the translated text, unconnected to the prescriptive norms that for years trained the translator. The label TU is used here, although I am well aware of the problems that it contains and which will be highlighted in the next paragraph.
5 Cf. Salasnik (2008, 103).
6 This definition is not broadly accepted. Baker (1993) uses, for instance, the term ‘Universal’ to refer to T-Universals, even though many of her examples derive from a comparison with T1 and the proper term should therefore be S-Universal. The fact that several scholars use these notions in such different ways make it difficult to compare results and so, further analytical description of the methodology is required.