Interpreting the Meaning of Translation

Andrew Chesterman

                            [2006. Interpreting the meaning of translation. Mickael Suominen et al. (eds), A Man of Measure. Festschrift in Honour of Fred Karlsson on his 60th Birthday. Turku: Linguistic Association of Finland, 3-11.]



Is “translation” a universal category? This question is examined via some contrastive etymological analysis of words meaning “translation” in a number of languages, mostly non-Indo-European. Different languages seem to give different emphases to three central semantic/semiotic features: difference, similarity, and mediation. Perhaps “translation” is a cluster concept.


1. Translation universals?

One recent trend in Translation Studies has been the search for what several scholars have called translation universals (see e.g. Mauranen and Kujamäki 2004). Other scholars have preferred to use labels such as regularities, patterns, general tendencies or translation laws. All these terms refer to the underlying intuition that translations seem to share certain linguistic features regardless of the language pairs or text types concerned. Some of these features can be formulated as differences with respect to source texts (such as: a tendency for translations to reduce repetition, or to be more standardized in style, or to be marked by interference). Other potentially distinguishing features are defined with reference to non-translated native texts in the target language (such as: a tendency to use a more restricted lexis, more simplified syntax, fewer target-language-specific items). The impulse to look for such universals stems partly from similar movements in general linguistics since Chomsky, and partly from computer programs enabling the quantitative analysis of large electronic corpora of various kinds. Insofar as evidence for translation universals is found, we can speculate that the causes for such widespread features may ultimately be cognitive ones, relating to the ways translators process and store language material.


This research trend has not gone without criticism. One of the most cogent points has been made by Maria Tymoczko (among others). She argues (1998, 2006[1]) that such research is based on the flawed assumption that there is a priori some universal concept of translation in the first place, which could serve as a basis for collecting a corpus. The lack of such a universal concept thus also makes it impossible to test hypotheses about universals. Different times and cultures may well conceptualize the notion of “translation” in very different ways. To what extent might such conceptualizations overlap? Could there be a universal prototype notion of translation? (Cf. Halverson 1998.)


This question, and hence the theoretical validity of the underlying basis of claims about translation universals, can be approached in two ways. Working top-down, we might seek to establish necessary minimal characteristics of all translations at a purely conceptual level. Working bottom-up, we could investigate empirically the different ways in which the kinds of texts we call “translations” in English have been conceptualized in different languages and cultures, and look for shared or differentiating features. Two recent collected volumes edited by Theo Hermans (2006) contain several discussions of the surprising variety of conceptualizations of the general idea of “translation” in non-Western cultures, including more detailed examinations of some of the examples to be presented below (e.g. Chinese, Indian languages), and other languages which I do not mention. The approach taken in this paper is partly top-down and partly bottom-up.


Moving top-down, we could start with the abstract level of semiotic analysis. This is precisely what is done by Ubaldo Stecconi in some recent work (2004). He develops a theory of the semiotic “Foundation” of translation, based largely on his interpretation of some parts of Peirce. Stecconi proposes three key semiotic features for a universal category of translation. They are: similarity, difference and mediation. “Similarity”, in that there must be some relation of similarity between source and target texts. “Difference”, because the languages are different and total identity of meaning etc. is usually impossible. And “mediation”, because the translator stands in some sense between two sides, mediating between them. I will not enter here into a detailed discussion of Stecconi’s proposal; nor will I discuss the possibility of formulating the features in Langackerian terms, which may occur to some readers. But I shall make use of these three features in analysing some bottom-up data. On the assumption that the features are good candidates for specifying a universal category for “translation”, I am interested in seeing whether different languages might nevertheless give different weights to different features, highlighting one rather than another.


The data come from contrastive etymology. The basic idea is to compare the etymologies, and hence the underlying conceptualizations (and implicitly too the reflected meanings – Leech 1974: 19), of words with a similar meaning. In my case, I start with the English words translation, translate and interpret, and examine the etymologies of the standard translations of these words in other languages. The other languages have not been selected systematically at this stage. Indeed, the present report is no more than notes on a few languages for which I have found helpful informants.[2] However, as I hope to show, even a small and non-representative sample of languages provides interesting food for thought.


2. Some data

Let us start with some familiar ground.  The words denoting translate, translation in Standard Average European (SAE) languages derive from roots in Latin and Classical Greek. The basic notion is that of carrying something across, from Latin transferre or Greek metapherein. A SAE translation is etymologically a metaphor, whereby something is, in some sense, something that it literally is not. That man is a pig; this article is (in Finnish) tämä artikkeli. The semantic elements that are highlighted in this construal of the notion are (a) something (say ‘X’) remains the same, the something that is carried across; and (b) there are two contexts involved, which we can call the source and target contexts. X is thus transferred from source to target, across a border. This border is traditionally conceived of as a linguistic border, but it may also be defined differently. Definitions of ‘X’ also vary, but traditionally this is usually held to be the meaning, roughly speaking. Our average European construal thus stresses the preservation of identity, some notion of sameness or similarity, across a border of difference. 


Within the Indo-European languages we can nevertheless already distinguish some different nuances in construal. Consider these three sets:


(1) Classical Greek metapherein, Latin transferre, English translate

(2) German übersetzen, Swedish översätta, Czech překladat

(3) French traduire, Italien tradurre, Spanish traducir, Russian perevesti


In set (1) the underlying cognitive schema is of carrying X across; here, the agent is conceived of as moving over together with X, like a messenger. In (2) the agent stands on the source side, putting or setting X across; X is transferred in a direction away from the agent. In (3) the agent etymologically leads X across; this suggests that the agent moves in advance of X, and the direction of movement is thus towards the agent. Rey (1992) suggests that the underlying metaphor here is of leading sheep. – It would be tempting to interpret these different nuances in terms of the different contexts of translating: into one’s native language (“towards” the agent) and into a second language (“away from” the agent). 


Other languages, however, construe the notion of translation very differently. In Finnish, the normal verb meaning ‘to translate’ is kääntää, whose basic meaning is ‘to turn’, e.g. in the sense of (transitively) turning a page or (intransitively) turning a corner, turning in a new direction. ‘A translation’ is correspondingly käännös, literally ‘a turn, a turning’. This construal is clearly different from the SAE one. It highlights difference, a new direction, entering a new context; what is not highlighted is any sense of preserving an identity, maintaining sameness. True, by implication there is a something that ‘turns’, and it is presumably still there in some form after the turning, but this is not foregrounded. The trope of ‘turning’ also occurs in equivalents for translation in other languages. Latin also used vertere, which had early cognates in several Germanic languages (Ārhammar 2004, Kilpiö 2005). Compare modern Lithuanian (verst ‘translate, turn around, turn upside down’). Dutch vertalen (‘translate’, etymologically ‘change the language’) ­also has an etymology which highlights difference; and see further below.


The Finnish verb kääntää, in the translation sense, goes back to Agricola’s time (16th century). However, from the early 19th century a second verb began to be used: suomentaa, literally ‘to finnish’ (Paloposki 2004). This verb obviously had, and has, a narrower meaning, restricted to a single target language. (Compare the archaic ‘to english’.) This verb too highlighs the target context, not the preservation of an identity. (Curiously enough, Finnish kääntää also has a slang meaning, ‘to steal’. So translating can perhaps also be seen as a kind of theft, a change of owner-identity, of belonging-ness. This reminds us that the classical god of translators was Hermes, who was also the god of thieves. )


Consider now the situation with words denoting oral translation. In English we have interpreter, interpreting, from Latin. The probable etymological root is ‘between prices’. The origin comes from the concept of trade, where goods are exchanged. The interpreter stands between the prices, or values, and ensures that there is adequate equivalence – equal value. This etymology thus stresses the mediating role of the interpreter. Finnish tulkata, and also the more general word tulkita, come from Swedish tolk ‘interpreter’, and its antecedents. The root meaning is ‘speak, make sense’ (cf. also Finnish tolkku ‘sense’). Both Finnish and modern English thus conceptualize written translation differently from oral translation. With respect to oral translating, however, both these languages highlight the notion of mediation, rather than difference or similarity. Interpreting is here construed etymologically as a rather different kind of activity from translating.


Let us now look briefly at some other languages. Do they highlight notions of difference, similarity or mediation? And do they have etymologically distinct terms for oral and written translating?[3]


3. Some more data

Indo-European. Modern Greek has two distinct terms. ‘To translate’ is metafrazo ‘to speak across’, which seems to  highlight difference. ‘To interpret’ is ermeneo, which implies explaining, i.e. mediating. Czech also has two sets of terms: for translating překladat, přeložit ‘to put, lay across’ (the feature of similarity); and for interpreting tlumočit. The latter comes from the Arabic (see below) and highlights mediation. Ukrainian makes a similar distinction: perekladáty ‘put, set across’ and tlumátšyty. Slovene uses prevajalec ‘lead over’ for both senses. Farsi has tarjuma-kardan ‘translate’ and ‘interpret’ (tarjuma ‘translation’, from the Arabic; and kardan ‘do’).


In Sanskrit, there are several words for the idea of translating. A translator is bhāsāntarakāri, which glosses as ‘ other language maker’; this highlights difference. On the other hand, some of the words meaning ‘translation’ seem to highlight other features. Chāyānuharanam means ‘loose translation’, and also ‘imitation, reflection (lit. ‘after-taking’), which suggests the feature of similarity. And anuvādah literally means ‘saying after, explaining’; this suggests the feature of mediation. An interpreter is dvibhāsāvādĒ ‘two-language speaker’ or bhāsāntaravaktā ‘other language speaker’. The second of these seems to indicate the difference feature.


Hindi also has two different sets of terms for written and spoken translation. Written translation is anuvād ‘saying after, explaining, repeating’ (anu ‘after’, vadah ‘speaking’); this suggests the mediation feature. Oral translating is done by a dubhāsiyā, a ‘two-language speaker’.


Uralic. The Hungarian word meaning ‘to translate’ is fordítani, whose literal meaning is ‘to turn something to the other side’. Like Finnish, this word seems thus to foreground the feature of difference, not similarity. The word for oral translation has different origins: tolmácsolni; an interpreter is a tolmács, from which German gets Dolmetscher. The etymology of these items may go back to the Hurrite language in Asia Minor, where talima  seems to have meant a mediator, someone standing between. Here again we see that interpretation is conceptualized primarily in terms of mediation. Cf. also the Czech and Ukrainian terms for interpreting, mentioned above.


Altaic. In Turkish the words meaning ‘translation’ highlight the feature of difference: the verb ćevirmek literally means ‘make turn’, i.e. change. But Turkish also uses another verb, tercüme etmek, and one word for a translator is tercüman, which derives from Arabic (below) and highlights mediation.


The Japanese for a translation is honyaku, where hon has the basic sense ‘turn, turn over, flutter’ and yaku means ‘substitute words’. The main semiotic feature here seems to be difference. Oral translation is denoted by the verb tsuuyaku suru, where tsuu has the basic sense ‘pass through, transmit, communicate’. Here the salient feature seems to be the preservation of similarity.


In Korean, words for translating and interpreting both seem to foreground the notion of mediation: ‘translate’ is tong yeok hada, where tong means ‘transmit, communicate’ and yeok means ‘explain’. The verb for ‘interpret’ is similar: dong si tong yeok hada.    


Afro-Asiatic. In Arabic, the central feature is closer to mediation, guiding towards understanding. Arabic targamah ‘translation’, turguman ‘translator, interpreter’, are loanwords from Aramaic. The cognate English dragoman still means a guide or interpeter. The verb targama ‘translate’ also means ‘write a biography’.


Sino-Tibetan. The Mandarin Chinese word for ‘translate’ is or fanyi. The verb fan has the basic meaning ‘flutter’, which suggests unstable movement, i.e. changes of state. For interpreting, the verb is kouyi, where kou means ‘mouth’. Lefevere (1998) suggests that the Chinese translation tradition differs from the Western one in that the Chinese have remained closer to the notion of interpreting, explaining, rather than the notion of fideltity or equivalence. We might see a reflection of this tradition in the very word itself in Mandarin, which highlights difference rather than similarity. (For more detailed discussion of the Chinese term and its history, see Cheung 2005.)


In Classical Tibetan, the same word skad-pa (skad ‘voice, speech, language’) is used for a translator and an interpreter, but the activities are conceptualized differently. To translate is sgyur-ba (‘change, turn’), and to interpret is grol-ba (grol ‘untie, release, remove obscurities, explicate). Translating thus seems to highlight difference, and interpreting highlights mediation.               


Austro-Asiatic. Vietnamese uses words for written translation that are cognate with the Chinese sense of ‘fluttering’, hence a highlight on difference. These are dich (‘change, move over’), phiźn-dich (phiźn ‘wave, flutter, turn upside down, inside out’), thông-dich (thông ‘penetrate, understand’). For oral translating different terms are used, which seem to highlight the feature of mediation: ląm thông-ngôn (ląm ‘do’, ngôn ‘word, speech’).


Malayo-Polynesian. Indonesian borrows menterjemahkan ‘translate’ from Arabic, but also uses manyalin, literally ‘copy, transfer’ (cf. Javanese salin ‘change dress’). Both mediation and similarity are involved here. Oral translating has a different term indicating a change of language: mengalihbasakan (alih ‘move, change’; bahasa ‘language’).


Dravidian. In Tamil, the same terms are used for the written and oral modes, and both stress the feature of difference, changing the language: molipeyarkka (moli ‘language’, peyar ‘transfer, change, turn over’).


4. Some conclusions

We have done no more than scrape the surface of a fascinating topic here. But even this mere scraping raises some questions. One interesting suggestion is that most of the modern Indo-European languages seem to give much more prominence to the similarity dimension than some other languages, even to a requirement of identity. This may partly explain the central role played by the notion of equivalence in western translation theory, and our need to develop other terms, such as adaptation, to describe freer types of translation (cf. Lefevere’s comment cited above). The fixation on the preservation of sameness has dominated most of the ideology of Bible translation since its early beginnings. Indeed, one of the reasons why these cross-language comparisons may be interesting is not that a particular etymology may be present in speakers’ consciousness, but the probability that it has influenced translation policies and practices in culturally highly significant ways. (See many of the papers in Hermans 2006.) On the other hand, it is true that we are only talking here of the relative highlighting of different features, not their inclusion vs. exclusion.


Most of the languages in this sample have different terms for oral and written translation, which suggests different ways of conceptualizing these activities. The oral mode is of course historically older. The etymologies of terms denoting interpreting seem to display the feature of mediation more frequently than those denoting written translation. It might also be interesting to see how many languages have, or develop, a hypernym covering both modes. In English, the term translation is often used by laymen to include interpreting. In German, the recent coinage of die Translation is an interesting case, covering both Übersetzung and Dolmetschen. Yet another question would be whether there are separate terms for the more general sense of interpretation. Finnish makes a distinction between tulkata (languages) and tulkita (in general), but English does not differentiate the two. German interpretieren, on the other hand, refers mainly to interpreting in the general sense.


So how can the meaning of the concept of translation best be interpreted? Do we have a universal concept? The answer may be yes, if we can allow such a concept to have a cluster shape, like the concept of a game, rather than a prototypical one (Tymoczko 2004, 2006). Cluster concepts are based on what Wittgenstein called family resemblances. A prototype concept, on the other hand, has some kind of prototypical instance  at the centre and less typical instances on the periphery. With respect to translation, however, I doubt whether we can posit a single prototype, even if we postulated a fictive one with exactly equal highlighting given to similarity, difference and mediation. What we seem to find, rather, is a cluster of closely related conceptualizations, some foregrounding one of the features proposed by Stecconi and some another. Stecconi’s three features seem to be explicitly or implicitly present in all the etymologies looked at here. An additional candidate feature might also be that of language itself, which occurs explicitly in many of the cases included.


A great many languages, of course, remain to be investigated from this point of view. A larger project could also look for correlations between particular constellations of relative feature dominance and empirical evidence of the ways in which “potential translations” of different kinds have been designated and classified in different cultures, i.e. correlations between semantic features and translation norms. It could also take into account other lexical items within the same semantic field, such as adaptation and version in English, and perhaps also transmission, transculturation, and representation (cf. Tymoczko 2006).


At the very least, the present preliminary study illustrates how the notion of translation has been interpreted in different ways in a number of different languages. It shows that not all these interpretations give the same priority to the preservation of sameness which characterizes the words denoting “translation” in many modern Indo-European languages.





Cheung, Martha P.Y. 2005. ‘To translate’ means ‘to exchange’? A new interpretation of the earliest Chinese to define translation (‘fanyi’). Target 17, 1, 27-48.

Halverson, Sandra 1998. Concepts and Categories in Translation Studies. Bergen: University of Bergen.

Hermans, Theo 2006. Translating Others. Vols 1 and 2. Manchester: St. Jerome Publishing.

Kilpiö, Matti 2005. Old English Vocabulary Dealing with Translation. Paper read at the symposium “Translation – Interpretation – Meaning”. Helsinki, 27-29.1.2005.

Leech, Geoffrey 1974. Semantics. Harmondsworth: Penguin.

Lefevere, André 1998. Chinese and Western thinking in translation. In Susan Bassnett and André Lefevere (eds.), Constructing Cultures. Clevedon: Multilingual Matters. 12-24.

Mauranen, Anna and Pekka Kujamäki (eds.) 2004. Translation Universals. Do they exist? Amsterdam: Benjamins.

Paloposki, Outi 2004. Macbeth ja Ruunulinna. Paljaasta suomennoksesta mukaelmaan – ja takaisin. Kielikello 2/04, 9-12.

Rey, Alain 1992. Traduire, interpréter : les mots pour le dire. Terminologie et traduction 2/2, 13-23.

Stecconi, Ubaldo 2004. Interpretive semiotics and translation theory: the semiotic conditions to translation. Semiotica 150, 471-489.

Tymoczko, Maria 1998. Computerized corpora and the future of Translation Studies. Meta 43, 4, 652-659.

Tymoczko, Maria 2004. Difference in similarity. In Stefano Arduini and Robert Hodgson (eds.), Similarity and Difference in Translation. Rimini: Guaraldi. 27-43.

Tymoczko, Maria 2006. Reconceptualizing translation theory. Integrating non-Western thought about translation. In Theo Hermans (ed.), Translating Others, Volume 1. Manchester: St. Jerome Publishing. 13-32.

Ārhammar, Nils 2004. Von lateinisch vertere zu lappisch jorgalit. Zur Onomasiologie der nordeuropäischen Bezeichnungen für den sprach- und kulturmittelnden Begriff ‘übersetzen/translate’. In Irma Hyvärinen, Petri Kallio und Jarmo Korhonen (Hg.), Etymologie, Entlehnungen und Entwicklungen. Festschrift für Jorma Koivulehto zum 70. Geburtstag. Helsinki: Société Néophilologique. 43-71.


[Back to list of Selected Publications]

[1] The paper includes some recent references that have appeared since the conference. They confirm the growing interest in non-Western conceptualizations of translation.

[2] Special thanks to all informants, especially Faruk Abu-Chacra, Bertil Tikkanen and Pertti Laakso.

[3] The transcriptions exclude some diacritics.