Andrei Fedorov and the Origins and Fate of Linguistic Translation Theories

This is a revised draft of an unpublished article of mine from 2013. References and some paragraphs are incomplete. There is no definitive version of the article: I update it from time to time.

In this paper I offer a sketch of what I take to be three founding moments in the history of linguistic approaches to translation theory. The sketch takes the form of a six-month westward journey starting in Moscow in July? [or the autumn of] 1953, then proceeding to Paris in December and finally to New York City in January 1954. It should be understood that what follows is not a contribution to the historiography of translation studies. It is speculative, far from thorough, and not based on archival research or interviews with students of the chief figures[1]. The purpose is to draw on an interpretation of the past in order to shed some light on the role of linguistic translation theories in present-day Translation Studies.

I should also immediately make clear that by ‘linguistic translation theory’, I do not mean specifically the application of some theory taken from the discipline of linguistics (in the way that John Catford’s 1965 book A Linguistic Theory of Translation was explicitly an application of UK linguist Michael Halliday’s theory of language structure). I simply mean an approach which sees translating as first and foremost a linguistic rather than a socio-cultural activity. That would certainly include applications of comparative grammar, sociolinguistics and text linguistics, but there is no restriction to branches of linguistics. Any theory that sees translation as first and foremost a special kind of reading and writing (or speaking/signing) would also be included. ‘Linguistic’ here is simply the adjective corresponding to the noun ‘language’. I am one of those who thinks that language has a high degree of autonomy from society and that therefore the linguistic aspect of translation work can be considered separately from its social aspect for theoretical purposes (though no such separation is possible for those who practice translation).

1. Origins in Moscow: language, not field

The relative importance of language in the study of translation figured in the very earliest debate in post-1950 European translation theory. In ?July 1953, Leningrad scholar and translator Andrei Fedorov (pronounced fyaw-duh-ruff) (1906-1997) published in Moscow his pioneering Vvedenie v teoriju perevoda (introduction to the theory of translation). Fedorov’s book seems to have been the first sustained argument (330 pages!) for a language-based rather than a literary theory of translation[2], and it immediately gave rise to a vigorous debate in the Soviet Union. The book was translated into Chinese in 1954; it "created a great impact" (Jin 1987:141), "attracted a large number of supporters" (but also opponents) (Luo 1984: 233) and was cited in an influential translation textbook by Lu Dianyang that was translated into English in 1958 (Pym 2016: 73-4; 81-2). It was also reported on in France by the Russian/French translator Edmond Cary (Cary 1957, 1958), and in that way came to the attention of French translation theorist Georges Mounin (1959, 1963:13-16)[3].

The debate about Fedorov’s book unfortunately tended to be about the practice of translating rather than about theory. Those who supported the linguistic approach said that the practice of translating was always and firstly a linguistic technique and as such should be informed by the findings of the scientific study of language. Those who opposed the linguistic approach said that translating is not mainly about language but about the fields of social activity in which they occur and the genres related to these fields. Translating for the theatre, for example, is more about creating a performable play than about words on the page; translating an article from the foreign press is more a matter of journalism than of language, and so on. Now as far as the practical activity of translating is concerned, the opposition is a rather sterile one; surely both views are true. What got lost in the debate was Fedorov’s real achievement—an advance not in how to see practice but in how to go about theorizing.

In his criticism of Fedorov, Cary wrote:

Whether one is translating poems or patents, one does need some knowledge of at least two languages. However that is only a starting point, one of the initial givens; it cannot form the objective foundation of any deep-going scientific study…Each ?genre/field…is sui generis, so distinctive that it needs to be considered separately, focusing on distinguishing features rather than common denominators. (1958:29, my translation)[4]

Now what Fedorov saw, I think, is that viewing translation in terms of fields and genres is highly problematic for any enterprise of ‘scientific study’. For if drama translation is part of theatre work and the translation of news stories is part of journalism, then there can be no general theory of translation. For that purpose, a common denominator had to be found, and the most obvious common denominator was language[5]. Fedorov mentions:

…a rather astonishing statement by Prof. A.A. Reformatskij, who in his article “Linguistic issues in translation” answered the question as to whether there could be a science of translation: “such a science cannot exist. The practice of translation may draw on the services of many sciences, but there cannot be a science of translation as such. This follows from the diversity of translational types and genres.” This argument is completely groundless. It may be a very complicated and difficult matter to systematize and generalize the various forms we see when we compare the regularities of two languages during the translation of different genres, different types of material, but that does not mean that the task cannot be accomplished. (1953:15, my translation)

Fedorov came under heavy fire for his views. He did not back down, though for the second edition, which appeared in 1958, he added the subtitle lingvisticheskie problemy (linguistic problems). This, as he pointed out in the preface to the 4th edition (1983), was a response to the criticisms: he was admitting that he was restricting himself to linguistic matters even though this was not sufficient for a complete understanding of translation. He thought that a complete understanding would require contributions from sociologists, psychologists, literary scholars and others (add page ref). On the other hand, his linguistic approach had two great advantages in the debate with the literary approach. First, focusing on different kinds of linguistic correspondence between source text and target was much more useful for translator training and for machine translation. Second, the linguistic approach seemed more objective and therefore more scientific--an advantage in an ideological environment where science was highly prized.

It should be noted that Fedorov’s focus on language in no way prevented him from losing sight of the different fields of translation. He understood ‘language’ in a broad sense to include stylistics and language specific to a field. Indeed a list of fields within which translation occurs is given in the very first paragraph of chapter 1, and Fedorov devotes lengthy sections of the book to specific discussions of journalistic and scientific-technical translation (25 pages), political translation (25 pages) and literary translation (60 pages). However, contrary to most literary scholars who took an interest in translation, not to mention the literary translators themselves, Fedorov did not see language merely as a means to a larger end. With Fedorov, perhaps for the first time, translation as such—conceived in his case as a linguistic activity—was explicitly described as an object for theoretical study, independent of any particular sphere of social activity, such as theatre or journalism, within which it may be practised. Before him, many had written about ‘translation’ but by this they meant literary translation, not translation in general. Nida had published a 360-page book about Bible translation, but as he wrote later “An earlier book (Nida 1947)… is essentially only a practical handbook… It was necessary to provide something which…would relate the specific area of Bible translating to the wider activity of translation in general. The present volume is an attempt to fill this need” (1964:ix); the first statement of this broader view was Nida (1959). Fedorov was as far as I can tell the first to formulate and elaborate the idea that one should theorize about translation in general by regarding it as language.

An interesting question that merits further research is the origins of Fedorov’s idea of a linguistic approach to translation. From a West European point of view, his own biography is a bit odd in this regard. While his book became the central text of linguistic translation theory in the Soviet Union and was familiar in parts of the world influenced by that country, Fedorov's own background was in literary studies and literary translation[6]. He was a published literary translator and he had been writing about literary translation since the mid 1920s. It seems that some of his teachers were linguists and literary formalists (Pym 2016:50), but he was certainly not a comparative grammarian. This made him quite different from the proponents of a linguistic approach in Europe, Canada and the United States in the 1950s and 60s: Vinay & Darbelnet (1958), Nida (1959, 1964), Mounin (1963), Catford (1965) and Kade (1968) were all trained in linguistics; none of them were literary translators or literary scholars (contrary to popular belief, Nida was not a "Bible translator"; he provided advice to Bible translators).

Apart from Fedorov's biography, one would need to look not only at the sources of his ideas in Soviet translation theory and practice, but also at the broader ideological, political and institutional context of thinking about language in the post-war Soviet Union. A good place to start, because it may help explain both the timing and some of the content of Fedorov’s book, was the astonishing publication in the summer of 1950, in the Communist Party newspaper Pravda, of a series of articles about linguistics signed by none other than Iosef Stalin. According to Medvedev (2003), the Georgian linguist Arnold Chikobava was personally asked by Stalin to write an article on linguistics, which was published in Pravda on 9 May 1950. Debate ensued, ending with a series of articles under Stalin’s own name beginning on 20 June (probably written with advice from Chikobava and the noted Russian linguist Viktor Vinogradov). These articles had the effect of bringing Soviet linguistics back into the international mainstream, for between 1930 and 1950, linguistics in the Soviet Union had been dominated by the bizarre but officially approved views of another Georgian, Nikolai Iakovlevich Marr (1865-1934). Amongst other odd claims arising from his peculiar view of the relationships among the ancient and modern languages of the Mediterranean area and the Caucasus, Marr saw different languages (or different historical "layers" he discerned within a language) as being related to different social classes rather than to different nations/ethnicities[7]. This is very different from the mainstream view that social class is reflected in superficial differences of pronunciation and grammar within a single language[8]. The burden of the Pravda articles was to denounce the views of Marr and his followers.[9]

Stalin’s intervention gave linguistics a very high profile when Fedorov was writing his book in the early 1950s. Perhaps Fedorov realized, in the aftermath of the Pravda articles, that certain ideas he already had could now be published if he discussed translation in terms of linguistics. In the very first paragraph of the introduction to the 1953 edition, Fedorov writes: "In the years when the "new theory of language" [i.e. Marrism] dominated, the theory of translation was undervalued or rather, completely sidelined. Even the possibility of a scientific linguistic approach to the study of translation was denied by the Marrists, despite the self-evident practical importance and theoretical interest of this task" (1953:3)[10]. Fedorov's linguistic approach was first introduced in a 20-page article that appeared in Vinogradov's new linguistics journal Voprosy Jazykoznanija (issues in linguistics)(Fedorov 1952), though if the title of his 1940 doctoral dissertation is anything to go by ("the linguistic foundations of translation theory")[verify this], Fedorov had been thinking about a general theory covering all fields for some time, perhaps inspired by his venture into German-Russian technical translation (Fedorov 1937-1941) as the USSR prepared for war with Hitler.

The first edition of his book appeared just a few months after Stalin’s death. It repeatedly cites Stalin’s statements denouncing the Marrist theory of language and asserting instead that a language is common to all classes within a society. Indeed the fifth chapter is entitled "Current issues in the theory of translation in the light of I.V.Stalin's writings in linguistics" and makes clear the relevance of the Pravda articles to translation (reference to Marr on p102). The later editions of the book do not refer to Stalin’s articles, no doubt because Stalin was no longer in good odour. However, I do not think Fedorov’s discussion in the original 1953 edition can be put down to mere genuflection to the Soviet leader. The Pravda articles did not simply ‘enable’ Fedorov’s book. Nor do I think it was merely 'convenient' for Fedorov to take a linguistic approach in the aftermath of Stalin's articles. The timing of his book may well have given him a big readership, but the discussion of the nature of language in the Pravda articles was related to what Fedorov wanted to say anyway. Indeed, it is perhaps not too much of a stretch to see a certain thematic resemblance between the notion that language is common to all social classes in a society and Fedorov’s idea that language is the common denominator for all the social frameworks within which translation takes place (journalism, theatre etc).

The Pravda articles pointed to and endorsed Marx and Engels’ view of language as “the immediate actuality of thought” and as being “as old as consciousness…practical consciousness that exists also for other men (The German Ideology, ch. 3, 1846:xx and ch 1:xx; cited at Fedorov 1953:63-4). This can be understood to imply that language does not simply convey our thoughts to others; it is the instrument of thought: no language, no thought. Language is thus to be seen first and foremost in psychological rather than sociological terms, and semantics/lexicology is not to be over-emphasized at the expense of syntax and phonology (Marr had denounced studies of syntax and phonology as "bourgeois formalism")[11]. Marx’ psychological view of language is strongly endorsed by Fedorov, whereas Fedorov's opponents (like Cary) emphasized the social side of language, as manifest in the various fields of social activity within which translating takes place. Debate about the nature of language is of course ongoing today; Chomsky, to take the best known example, endorses the notion that language is first and foremost an instrument of thought and only secondarily a means of communication. Any useful discussion of the merits of a linguistic approach to translation must be based on a conception of language. My own view, no doubt influenced by 40 years working as a full-time professional translator, is that translating should be seen in the first place as a mental activity that goes on in an individual’s brain, and only secondarily as reading and writing in a social context.

A final word on Fedorov: Translation had traditionally been regarded in the European tradition as a useful practical activity but hardly ever as something of intellectual interest. Translations of literature were often seen as inferior to their sources and therefore worthy of only occasional mention, certainly not systematic study. Translations of non-literary material, though commonplace, were never mentioned at all as objects of study. This changed with Fedorov, at least among those who could read him or learned about him from others. There was now a book-length argument conveying a sense that translation, in all fields, is worth studying for itself. Pym (2010, 2016 and elsewhere) has emphasized the ongoing importance of the various notions of equivalence developed in the 1950 and 60s. However before such discussions could start, it was necessary to establish that it was worthwhile investigating translation as a linguistic phenomenon in the first place. In other words, it was necessary to move beyond the dismissive view that ‘of course’ translation involves language (a point on which one can hardly disagree).

2. A little later in Paris: the translator steps forward

In December 1953, just months after Fedorov’s book was published in Moscow, the Fédération internationale des traducteurs was founded in Paris, to “organize the profession around the world, encourage the formation of groups of translators in countries where no such groups existed, defend translators’ interests and set out their duties” (Caillé 1963:170, my translation). As a result of these organizing efforts, a figure who had previously not been very visible began to come into view--the professional translator, the person who translates for a living as opposed to the bilingual who translates ‘on the side’: the child of immigrants translating for parents, the Greek professor translating Plato, the lawyer translating foreign contracts, the churchman translating scriptures, and more recently the fansubber or other volunteer translator. Thus,

alongside translation as such, the translator as such began to emerge[12]. The professional translator may translate agriculture in the morning, finance in the afternoon and a minister's speech the next day, yet he or she is not a part of the farming, banking or political spheres, but simply 'a translator'.

Now Fedorov had taken the study of translation to be a study of texts: “The task of the theory of translation is to discover regularities in the relationship between original and translation, make general deductions …from observations of particular translational occurrences, …” (Fedorov 1953/1958:15, my translation). Meanwhile the translators who assembled in Paris in 1953 were quite naturally interested not in texts but in themselves and in the services they provided: their economic and professional status in society, and how they could ensure that translators produced high-quality translations. Thus while the practitioners were interested in translators, Fedorov and other theorists were interested in translations. The translator as a figure of interest continued to appear, as in the past, only in the work of historians of translation or biographers of famous translators.

The focus on translations rather than translators was to continue among theorists, with the result that while the professional translator began to come into view in society, the translation producer did not come into view as an entity in the world of theory until the end of the century (Simeoni 1988-2008; Chesterman 2009). Starting in the late 1980s, theorists finally began to observe translators at work, using various kinds of recording equipment, in an attempt to understand what was going on their minds. But then in the late 1990s, the idea of a sociology of translation appeared, a trend that saw translators not as cognitive agents (using their brains to construct text) but as social agents entering into relationships with others.

It's true that Fedorov’s above-cited statement of the task of translation theory does refer indirectly to translators. The sentence continues: “[the task of the theory of translation is to….] assist translation practice, which might be guided by theory in searching for the requisite means of expression, and derive from it arguments and evidence in favour of certain solutions to specific problems …”. However there is a world of difference between theory assisting practitioners and theory studying them, whether in terms of their socio-economic activity or in terms of their language production work.

The interest in translations rather than translators can no doubt be explained by the twin disciplinary origins of translation theory. The discipline that had traditionally provided concepts for thinking about translation, and continued to do so, was literary studies—an obviously text-based discipline[13]. Then in the 50s and 60s, linguistics came to the fore with the writings of Vinay & Darbelnet, Nida, Mounin, Catford and Kade. However this disciplinary shift did not alter the focus on text, for during this period, Chomsky’s view of language, with its psychological emphasis on the mind of the speaking subject, was only just beginning to have influence. Linguistics was still by and large concerned with the description of observed language forms rather than with the language knowledge of the speaker, knowledge which is mobilized when people produce or receive speech. Thus the language-producing subject, in our case the translator, remained invisible to theory, which concerned itself with completed texts. Translation was seen in linguistics either as an applied field or as a matter of ‘languages in contact’, but the contact in question was geographical; nothing was said about the only place where contact between languages can actually occur—in the mind/brain of a human being[14].

When linguistics-trained theorists considered translation, their observations of linguistic forms were comparative in nature. Those taking the new linguistic approach did of course find that they had to go beyond the two linguistic systems, which are by nature ‘untranslatable’, and talk also (or instead) about texts: one does not really translate Russian into French; one translates texts which are written in Russian into texts written in French. It was not simply a matter of noting the degree of similarity of a certain Russian verb form (within the Russian system) and a certain French verb form (within the French system), for if two forms had a certain similarity (e.g. they both had to do with the past), that merely created a certain potential that the latter might be used to translate the former. It was necessary instead to compare real textual examples, noting perhaps that a Russian verb may in fact be translated, for reasons of stylistic preference, not by a French verb at all but by a French noun (<when he arrived> becoming ‘at the time of his arrival’, with a clause becoming a phrase). Still, even an approach based on textual rather than system comparisons inevitably placed in the background the subject who was producing the translation. By the time the two texts exist to be compared, the producer’s work lies in the past and is forgotten. As Barbara Folkart later expressed it: “The distinguishing characteristic of translation studies, I feel, should be that it is at least as concerned with writing as with reading. … So much of the discourse on translation is readerly—backward-looking, fixated on the already-said…Little of the canonical discourse on translation has anything to do with the actual business of making text” (2007, pp. xiv and 30).

The temptation to ignore the producer is also much greater when studying the written rather than the oral forms of translating. Written texts appear to exist independently of their writers and readers, in a physical object that can be seen and touched. When the theoretical study of oral translating got under way in the 1960s, eventually constituting itself in the early 1990s as Interpreting Studies, its focus was on the translator from the outset. In oral translating, the speaker and what is spoken are intimately linked. True, they are not completely inseparable, since an analyst can record and transcribe an oral translation, but such separation does not exist for the actual speakers and listeners in the translating situation: the language producer is right there. To paraphrase Yeats: “how can we tell the speaker from the speech?”

The backgrounding of the translator in the linguistic approach of the 1950s and 60s can be seen in the published form of Eugene Nida’s address to the Linguistic Society of America (the academic society of linguists in the United States), delivered when he was president of that organization (Nida 1969). Nida used two box diagrams of the translation process. In one, the translator is absent: it simply shows the sequence ST-Analysis-Transfer-Restructuring-Translation. In the other, the translator does appear but as ‘R2-S2’ (receiver 2 – sender 2)—little more than a waystation through which texts pass. Chapter 7 of his 1964 book Toward a Science of Translating begins with the statement that “No discussion of the principles and procedures of translation can afford to treat translating as something apart from the translator himself” (1964:145). Yet the chapter then plunges into these same box diagrams. Of course it would be absurd to say that Nida of all people was uninterested in translators. However what concerns us here is not Nida’s practical interest in translators but rather the place of the translator in his theoretical formulations. Nida innovated by focusing on the receptors of translational communication; he can be seen as a precursor of the target-oriented theorizing that emerged in Germany, Belgium and Israel a decade later. In other respects, his formulations were text-focused, not translator-focused, and this, I think, was typical of the period.

It would be some time before the two strands we have been considering came together in theoretical work: the linguistic aspect of translation and the person of the translator. The crucial step was to stop attending so much to completed texts (the source/translation text-pair) and to instead attend to that moment when a more or less conscious decision (to write wording X) is made in the translator’s mind just before (or perhaps just as) his or her fingers begin moving on the keyboard. There were anticipations of this in the 1960s. Rudolf Jumpelt[15], a theory-minded scientific and technical translator, and officer of the FIT, wrote that:

“translation is a process of choosing between complex variables; the theory of translation does not, therefore, set out to describe the objective phenomena of the languages in question as such, but rather the factors that lead to the choice between these phenomena…” (1961:xx, my translation).

In a similar vein, Jiří Levý (1967) famously suggested a game theoretical approach to translation as a decision process. That said, in my own view, the decisions and choices in question are not choices of wordings but choices of strategies (elaborate). It is certainly not the case that translators typically generate several possible target-language wordings and then choose one (I myself do that only when a passage is problematic).

In the 1970s, Toury (1980:51-78) brought the translator’s decisions into theory, but only indirectly, when he invoked the social norm governing the production of translations as a key factor in translation. The norm[16] is a society’s or subculture’s expectation of what a translation should be, as evidenced by the way texts are in fact translated. Norms are seen as dictating, in a broad way, the choices a translator makes. The focus at the time was on the norm’s social nature, but of course in order to function in a society, norms must be internalized in the mind/brains of individuals. They are not mere habits but forms of knowledge, which must be embedded in the translator’s mind if they are to govern the production of texts (i.e. those texts from the description of which theorists deduce the norm). Toury points to this fact, but indirectly, in a passage which is noteworthy for the way it manages to avoid using the noun ‘translator’: “…translational norms are intersubjective factors, influencing, and to a large extent even determining, the choice of translational solutions.….they act as a model, in accordance with which translations are actually formulated…every model supplying performance instructions may be said to act as a restricting factor on the selection of TL material…it opens up certain options while at the same time it closes others.” (1980:64). In retrospect, the line of inquiry initiated by Toury seems to have had the effect of removing attention from the translator, because the main point was to spend less time comparing source texts with translations and more time comparing translations into the target language with existing texts in the target language, in order to see how translations fitted into and affected the target textual polysystem.

It was not until the mid 1980s that a few people began to conduct empirical studies that observed translators (first students, then professionals) actually translating and tried to test hypotheses about what was going on in their minds. The language-producing translator had finally appeared as a focus of attention. Today, his or her mind is the centrepiece of linguistic translation theory, even what that mind is operating on linguistic strings generated by machines.

To sum up so far: Translators as such (as opposed to poets, journalists, doctors or lawyers who translate) had begun to achieve recognition with the founding of the International Federation of Translators in 1953. Meanwhile, also in 1953, Andrei Fedorov argued that a general theory of translation as such, rather than separate understandings of literary, journalistic, medical or legal translation, could be devised by taking language as the common denominator. However this isolation of language from the particular situations in which translating journalists or lawyers worked meant that translation came to be seen, under the influence of the linguistics of that time, as a matter of correspondences between textual wordings in two languages, rather than someone’s language-composing activity (activity determined externally of course by a social purpose, but nonetheless fundamentally a cognitive activity: the social determinants must be represented in the brain to have any effect on the translator).

3. Next year in New York: the translator vanishes

The idea of looking at texts separately from their producers received a boost within a few weeks of the founding of the FIT when, on 7 January 1954, in New York City, the first public demonstration of machine translation took place. Over the next few decades, huge public and private investments were made, in many countries, in research projects aimed at fully automatic machine translation. Though this goal has not yet been achieved (machine outputs continue to require editing by humans), MT can be seen as an incarnation of the concept of translation without translators. However it was not until the 21^st century that this notion came to be widespread: early MT systems were available only to large institutions whereas current systems have been easily available to the online public for the past 20 years. While only specialists understand how they work, these systems do convey every day to the general public the idea that translation is possible without translators.

Unlike the writings of Fedorov, Vinay & Darbelnet, Mounin and the other language-oriented theorists of the time, work in MT occasionally drew international public attention to translation[17], and specifically its purely linguistic aspect. An MT system is after all nothing but a recipe for replacing one sequence of signs with another.

There have been three main kinds of MT system, but until around 1990, there was only one. All systems were rule-based, that is, to simplify hugely, the program applied comparative lexical and grammatical rules to the source text. If the source text contained the word ‘poor’, the rule might say that if the following word has been analysed as a noun, and its lexical entry in the system’s dictionary marks it as a kind of human being, then translate it by word x (which means ‘having little money’); otherwise translate it by word y (which means ‘not very good’). Thus the ‘poor’ of ‘poor peasants’ would be translated by x, but the ‘poor’ of ‘poor mark on a test’ might be translated by word y; as for ‘poor worker’ the system might have trouble with this because it is ambiguous as between the two senses of ‘poor’.

The rule-based MT research that began in the 1950s required a vast amount of work comparing the lexicons and grammars of the source and target languages, and it’s possible that this lent tacit support to an approach to human translation which focused on comparing the visible wordings of the source text and the completed translation, rather than on the not so visible process in the translator’s mind at the moment of translation production. However MT never really became an overt model for the study of human translation. Aside from its obvious output failures, and its restriction to the levels of morpheme, word, phrase and sentence, MT researchers had failed to find a machine analogue for the subject-matter knowledge to which human translators appeal when deciding which of many possible meanings of a word or syntactic structure is the one meant by the source.[18]

By the 1980s, it had become clear—despite many breathless announcements of the imminent arrival of translating machines by futurologists, journalists, software makers and (with more circumspection) computational linguists—that no commercially viable rule-based system capable of replacing even beginner human translators would in fact be making an appearance in the near future. The systems with useful output were either extremely field-restricted (weather reports) or only good for conveying a general idea of the content of the source text (so that the reader could decide whether to send the text for human translation). The cost of human revision of the output (‘post-editing’) was, for all but a few purposes, too great for commercial viability.

In the 1990s, a statistical approach to MT emerged and it proved somewhat more successful than rule-based MT when dealing with unrestricted text input. However the algorithms involved bear no relationship whatsoever to the way humans process language. To give a hugely simplified idea of how this approach works, if a statistical MT program were processing the previous sentence, it would look up, in a database of source-target text pairs (e.g. human translations from English to the desired target language), the expressions ‘unrestricted’, ‘text’ ‘input’, ‘unrestricted text’ ‘text input’ ‘unrestricted text input’, ‘with unrestricted text input’ and so on. Statistical models would then be used to select, among the corresponding target-language expressions found in the database, the one which is the most probable equivalent of the English and also the most probable to occur (the most natural) as a target-language wording. Clearly this cannot provide a model for the study of human translation.

The same applies to the most recent ‘neural’ machine translation model, so called because it uses computer networks that were at one time metaphorically envisaged as being similar to networks of neurons in the brain[19]. Like the statistical approach, neural MT relies on ‘big data’—the knowledge embedded in huge sets of existing human translations. The machine learning techniques of Artificial Intelligence systems are used to learn from the data.

The failure of rule-based MT in the period 1950-1970, as the most visible application of a linguistic approach to translation, was perhaps one factor that helped usher in an era in which the focus shifted from translation as language-based activity to translation as intercultural communication. The new common denominator for the study of translation became “all translation is communication”. While the communication meme appeared to socialize translation, by seeing it as a set of relationships among people rather than between wordings, the emphasis turned out to be on the receiving end of communication (reception in the target culture) rather than on the sending by the translator, so that the translator’s language-producing activity was still not a focus of attention.

4. The fate of linguistic translation theory

After a brief period of dominance in Europe, from about 1955 to 1975, the linguistic approach to theorizing moved into the background, where it has stayed ever since. The various 'turns' have mostly been toward the social aspect of translating. The situation has of course been different within those translation schools that are focused on preparing professional translators rather than preparing researchers. These schools require many practical courses, which are inevitably language-based, so that the students can learn to compose text in their target language.

The linguistic approach developed during the 1950s and 60s lost favour in Western Europe partly because of the failure of rule-based MT and partly because of its emphasis on 'equivalents’. ...(expand)...

The linguistic approach of the early period did not have the effect of constituting the study of translation as an independent discipline, with its own apparatus of journals, handbooks, degree programs and conferences.[20] That had to await the widespread opening of university-based translation schools to meet the growing demand for non-literary translators. These schools potentially provided a space for organized thought about translation on the part of instructors as well as an audience for those few literary scholars who were interested in translation (notably those living in small cultures and speaking ‘small’ languages such as Czech, Slovak, Dutch and Hebrew). However, without the unified approach to translation made possible first by Fedorov and then by others, the translation schools might never have given birth to Translation Studies.

Whatever the shortcomings of the language-based approach, it did have the great merit of finally bringing to an end, after several centuries, the era in which European thinking and writing about translation consisted of nothing but interminable and tediously repetitive discussions of whether one should translate literally or freely. (Apparently no one thought of publishing a free translation and a literal one simultaneously on facing pages, in order to please everyone!) That debate, which has continued to the present in a variety of guises (semantic/communicative; formal/dynamic; foreignizing/domesticating), has always assumed, wrongly, that there are only two rather than several general ways to approach a translation, and it has often invoked lofty ethical or political positions in an effort to show that only one of the two ways is acceptable in general for all (literary) texts in all times and places[21].

The new linguistic approach did not answer the literal-or-free question but rather shifted attention away from it, by investigating how translators did in fact translate (the actual linguistic correspondences between particular wordings of source text and translation) rather than how they ought to translate in general or in principle[22]. An important benefit of the linguistic approach was that for the first time non-literary translations were seen as worthy of study. Generally speaking, non-literary translators (the overwhelming majority in the world today), whatever pronouncements they may make about the literal-or-free question on the rare occasions when they are asked to do so, in practice choose solutions from a smorgasbord of approaches as they work through a text[23]. If they are professionals with degrees in translation, their choices may be informed to some degree by what they were taught at translation school—typically a set of varying practical solutions to individual problems (for example, four ways to handle metaphors) combined with some general principles such as keeping in mind the field and readership of a text.

And what of today? Within linguistic translation theory, the focus has changed dramatically since the 1960s. The focus is now on translators rather than translations. Comparative lexico-grammar and comparative text linguistics have given way to observational studies of translators producing text, leading to hypotheses about what is going on in their minds as they work. In this minor branch of translation theory at least, things have slowly moved, since the mid-1980s, in the direction of overcoming Barbara Folkart's above-mentioned complaint about Translation Studies.[24]

Looking back almost 70 years later on our three-stop journey from Moscow to Paris to New York, the order of the three stops does seem to have been reflected in history: first the interest was in translations, then in translators, and now in machines. ….(expand)...

References

Ballard, Michel 1992 De Cicéron à Benjamin: traducteurs, traductions, réflexions.

Balliu, Christian 2005 ‘Clefs pour une histoire de la traductologie soviétique’ Meta 50(3), 934-948 [keys for a history of Soviet translation studies]

Bolaños Cuéllar, Sergio 1997 ‘Vigencia de la teoría de la traducción de Andrei Fedorov’ Forma y Función 10, 51-72 [validity of Andrei Fedorov’s theory of translation]

Bedford, Ian 1985 'Stalin on linguistics', Canberra Anthropology 8(1-2) 58-86.

Brang, I 1955 ‘Das Problem der Übersetzung in sowjetischer Sicht’, Sprachforum I, 124-34. [reprinted in Störig 1963 Das Problem des Übersetzens] [the problem of translation from a Soviet point of view]

Caillé, Pierre-François 1963 ‘La FIT a dix ans’, Babel 9(2) xx-xx [FIT is 10 years old]

Cary, Edmond 1957 ‘Théories soviétiques de la traduction’, Babel 3(4) 179-190. [soviet theories of translation]

Cary, Edmond 1959 ‘Andrei Fedorov: Introduction à la théorie de la traduction’, Babel 5(1), 19-20. [Andrei Fedorov's introduction to the theory of translation]

Cary, Edmond 1962 'Prolegomena for the establishment of a general theory of translation' trans. from French by Sidney Alexander, Diogenes 10(40) 96-121

Catford, John 1965 A linguistic theory of translation: An Essay in Applied Linguistics.

Chesterman, Andrew 2009 “The name and nature of translator studies.” Hermes 42, 13-22.

Chyzhevshkyj, D. 1956. Review of Fedorov 1953. Zeitschrift für Slawische Philologie 24(2) 426-29.

Coșeriu, Eugenio 1952 ‘Sistema, norma y habla’, in Teoria del lenguaje y lingüística general 1962. [system, norm and speech]

Dmitrienko, Gleb 2015 'Vers une science de la traduction? Contextes idéologiques, politiques et institutionnels du développement de la Théorie Linguistique de la Traduction en Russie soviétique (1922-1991)' [master's thesis, Université de Montréal] https://papyrus.bib.umontreal.ca/xmlui/handle/1866/13642

Fawcett, Peter 1997. Translation and Language

Fedorov, Andrei 1937-41 Теория и практика перевода немецкой научно-технической литературы на русский язык [theory and practice of translation of German scientific and technical literature into Russian]

Fedorov, Andrei 1952 ‘Oсновные вопросы теории перевода’ [basic issues of translation theory],Voprosy Jazykoznanija 1(5), 3-22.

Fedorov, Andrei 1978 ‘K истории становления теории перевода в CCCP’ [contribution to the history of the emergence of translation theory in the USSR], Babel 24(3-4), 144-49

Fedorov, Andrei 1983 'Как развивалась в нашей стране теория перевода' [how translation theory developed in our country] in Искусство перевода и жизнь литературы 155-170

Folkart, Barbara 2007 Second Finding: A Poetics of Translation

Garbovskij, Nikolai and Olga Kostikova 2012 ‘Science of Translation Today: Change of Scientific Paradigm’ Meta 57(1), 48-66

House, Juliane 2016 Translation as Communication across Languages and Cultures

Jin, Di 1987 ‘The debate of art vs science’ in Twentieth Century Chinese Translation Theory

Jumpelt, Rudolf 1961 Die Übersetzung Naturwissenschaftlicher und Technischer Literatur [the translation of scientific and technical literature]

Kade, Otto 1968 Zufall und Gesetzmäßigkeit in der Übersetzung [chance and regularity in translation]

Koptjevskaja-Tamm, Maria 1989 Linguistic translation theory in the Soviet Union (1950-1980s): a review, Reports from the Institute for Interpretation and Translation Studies, University of Stockholm (IITS Report No. 2)

Levý, Jiří 1967 ‘Translation as a decision process’ in To Honour Roman Jakobson vol II, 1171-1182

L'Hermitte, René 1987 Science et perversion idéologique: Marr, Marrisme, Marristes, Une page de l'histoire de la linguistique soviétique

Ludskanov, Aleksandar 1967/1969 Traduction humaine et traduction mécanique, self-translated from Bulgarian [human and machine translation]

Luo, Xinzhang (1984) ‘Chinese translation theory, a system of its own’ in Twentieth Century Chinese Translation Theory 2004

Medvedev, Roy 2003 ‘Stalin and Linguistics: an episode from the history of Soviet science’, ch. 10 of The Unknown Stalin, 200-209

Mounin, Georges 1964 Les problèmes théoriques de la traduction [theoretical problems of translation]

Mounin, Georges 1959 ‘La qualité en matière de traduction’ Babel 5(2), 84-88

Newmark, Peter 1981 Approaches to Translation

Newmark, Peter 2009 'The linguistic and communicative stages in translation theory', Routledge Companion to Translation Studies 20-35

Nida, Eugene 1947 Bible translating: an analysis of principles and procedures, with special reference to aboriginal languages

Nida, Eugene 1959 ‘Principles of translation as exemplified by Bible translating’ in RA Brower On Translation 11-31

Nida, Eugene 1964 Toward a Science of Translating

Nida, Eugene 1969 ‘The Science of Translating’ Language 45(3) 483-98

Pym, Anthony 2010 Exploring Translation Theories

Pym, Anthony 2016 Translation Solutions for Many Languages – Histories of a Flawed Dream

Reformatskij, A.A. 1952 ‘Lingvisticheskie voprosy perevoda’ [linguistic issues in translation], Inostrannye jazyki v shkole 6, 12-22

Revzin, I.I. & V.Ju.Rozenzweig 1964 Osnovy obshschego i mashinnogo perevoda [foundations of general and machine translation]

Simeoni, Daniel 1988-2008, various articles in The View from the Agent: Daniel Simeoni's "traductologies".

Stalin, I.V., Marksizm i voprosy jazykoznanija [Marxism and problems of linguistics], first published in the June 20, July 4, and August 2, 1950 issues of Pravda; reprinted by Foreign Languages Publishing House, Moscow. online version in English

Steiner, George 1975 After Babel: aspects of language and translation

Toury, Gideon 1976 ‘The Nature and Role of Norms in Literary Translation’, in Literature and Translation: New Perspectives in Literary Studies 1978 and in Toury 1980, 51-62

Toury, Gideon 1978 ‘Translated Literature – System, Norm, Performance: toward a TT-oriented approach to literary translation’ in Translation Theory and Intercultural Relations, and in Toury 1980, 35-50

Toury, Gideon 1980 In Search of a Theory of Translation

Vinay, JP and J Darbelnet 1958 Stylistique comparée du français et de l’anglais. [comparative stylistics of French and English]

Weinreich, Uriel 1953 Languages in Contact

Wilss, Wolfram 1982 The Science of Translation, an "updated and to some extent re-written" self-translation of Übersetzungswissenschaft: Probleme und Methoden, 1977

[1] With the deaths of Catford in 2009, Nida and Newmark in 2011, and ?Wilss in 2012, all the founding figures of West European/Canadian/American linguistic translation theory mentioned in this chapter are now gone.

[2] I am not claiming that Fedorov was the first person to write about translation in linguistic terms; I mean that the idea of a language-based theory of translation became well known, at least in the Soviet intellectual sphere, through his book. Its popularity is shown by the fact that there were four later editions, with varying titles, in 1958, 1968, 1983 and 2002. The 5th, posthumous edition is a repeat of the 4th but has the virtue of being easily available (in Russian) online at http://goo.gl/wnWPvI. A Centre for Translation Studies at Saint Petersburg University is named after him, and for a number of years after his death, there was an annual "Fedorov lectures" forum (http://translation.spbu.ru/fedorov.htm). Oddly, the article on Fedorov in the Russian Wikipedia does not mention any of the editions of his 1953 book.

[3] There is still no published translation in any West European language, though a translation into English by Brian Baer is to be published by Benjamins. In 1968/69, two students (R. Deresteau and S. Sergeant) at the École Supérieure de Traducteurs et Interprètes de l’État in Brussels prepared what appears to have been a complete French translation of the 2^nd (1958) edition, as a research paper for their Licentiate (Harris in Babel 57(1); Ballard p.15). There is also apparently an unpublished and almost complete German translation of the 2nd edition (Wilss 1982 p 52n). A 10-page summary appeared in German in 1955 (reprinted in Brang 1963). A review appeared in Italian in the July-August 1956 (or 1958(4)?) issue of Rassegna sovietica, and a Slavic philologist at Harvard wrote a very negative review in German (Chyzhevshkyj 1956). Perhaps because of Cold War barriers, and then the loss of interest in Russia in the 1990s after the Cold War ended, Fedorov’s book had little influence in English-speaking countries in particular as well as in other countries where a West European language predominates and few translation scholars are able to read Russian. East Europeans at the time would have read him in Russian (German theorist Otto Kade cites Fedorov in Russian in his 1968 book Zufall.... I first read parts of the 1953 edition in 2007. I had seen discussions of it much earlier, but only passing mentions in English (Steiner 1975: 237; Newmark 1981:9; Koptjevskaja-Tamm 1989:1-2; Fawcett 1997:34ff). For some recent discussion of Fedorov, see Bolaños Cuéllar 1997, Balliu 2005, Garbovskij & Kostikova 2012, Dmitrienko 2015, Pym 2016 (37-64) and Schippel 2017 (247-268).

[4] Cary later wrote (1962) that he was in favour of a general theory of translation that encompassed all fields and media, but he did not suggest any common denominator.

[5] For the third edition in 1968, Fedorov changed the title completely to "foundations of the general theory of translation: a linguistic essay", which nicely gets his point across by including both "general" and "linguistic". Cary’s discussion of Fedorov refers to language as a dénominateur commun, as does Mounin's (1963: 15); Jumpelt (1961), in discussing Fedorov, uses the German equivalent gemeinsamer Nenner. However Fedorov himself does not use this mathematical metaphor (Russian obshchij znamenatelj).

[6] It should be understood that in Russia, and elsewhere in the Slavic world, the study of language has never been as rigorously separated from the study of literature as has been common in western Europe since traditional philology gave way to linguistics on the one hand and literary/cultural studies on the other during the 20th century. (In Canada and the U.S., the linguistic aspect of philology gave way to anthropological linguistics--the study of indigenous languages--and finally to plain linguistics.)

[7] Marr had been developing his views of language since the late 19th century with no connection whatever to Marxist class struggle concepts, but in the mid 1920s, he realized that those views could, conveniently, be stated using Marxist terms such as "mode of production" and "superstructure". For a detailed view of Marr and the Marrist "school", see L'Hermitte 1987. Marr was never strictly speaking a Marxist. He was closer to the "Eurasianists" in his cultural-political outlook; in this view, the Russian Empire/Soviet Union was not part of the Romano-German world but a sui generis blend of the European and the Asian. In linguistics, Marr saw the Caucasian languages as remnants of an ancient "Japhetic" language which had been mostly displaced by Indo-European. In the decades that followed the Pravda articles, Marr's disciples succeeded in 'rehabilitating' him to some extent, and eventually he was hailed, by some linguists in the USSR and in France, as a precursor of socio-linguistics (!) because he had pointed to the relationship between language and class and between language and social change. The Republics of Georgia and Armenia issued stamps in his honour in 1996 and 2014 respectively.

[8] The obvious exceptions follow successful invasions, when for a brief period the ruling invaders speak one language, while the labouring masses among the conquered people speak another, as happened with French and English after the Norman conquest of England in the 11th century. Communication between rulers and ruled is then through bilingual intermediaries.

[9] According to Ian Bedford (1985), Stalin was not particularly interested in linguistics as such. What he did not like was Marr’s rejection of the relationship between language and nation, which Stalin criticized as non-Marxist. This was a subject of great interest to Stalin, in part because he (like Marr and Chikobava) was a member of one of the Soviet national minorities and a native speaker of a Kartvelian language, but mainly because he had often been preoccupied with the nationalities question (he wrote a pamphlet “Marxism and the National Question” in 1913 and he was the People’s Commissar of Nationalities from 1917 to 1923). The importance of the national question for the internal politics of the Soviet Union probably explains in large measure the appearance of these articles in the Communist Party’s newspaper.

[10] Later in the book, the charge against the Marrists was phrased somewhat differently. Questions about the nature of language that are important for translation theory "were either not posed or were answered wrongly" (1953:97).

[11] The following interesting passage occurs in one of the Pravda articles: “[Question to Stalin]: ‘Ideas,’ Marx says, ‘do not exist divorced from language.’ In what measure, in your opinion, should linguistics occupy itself with the semantic aspect of language? [Answer by Stalin]: Semantics must be assured its due place in linguistics. However, in working on problems of semantics and in utilizing its data, its significance must in no way be overestimated, and still less must it be abused. I have in mind certain philologists who, having an excessive passion for semantics, disregard language as ‘the immediate reality of thought’. It is said that thoughts arise in the mind of man prior to their being expressed in speech, that they arise without linguistic material, without linguistic integument, in, so to say, a naked form. But that is absolutely wrong. Whatever thoughts arise in the human mind and at whatever moment, they can arise and exist only on the basis of the linguistic material, on the basis of language terms and phrases. Bare thoughts, free of the linguistic material, free of the "natural matter" of language, do not exist....”

[12] The FIT is an umbrella group for national and regional organizations of professional translators, some of which had already existed for quite some time in 1953. The one of which I myself am a member (the Association of Translators and Interpreters of Ontario) was founded in 1920. However the appearance of an international organization probably gave professional translators a stronger profile and, being international, its presence asserted the general nature of translation as an activity, independent of any language or country.

[13] The Library of Congress classification of works on translation reflected the prevailing idea that translation was intimately connected with literature. In the 1950s, books dealing with translation (even machine translation!) were assigned to the class PN 241 ("translating as a literary pursuit") rather than P ("philology; linguistics") and books from that period are still to be found in PN; nowadays books on translation go into P306-310 ("language - translating and interpreting"). Among the general public, the idea is still widespread that translation has something to do with literature, even though the overwhelming majority of translators do not work in that field.

[14] Mounin, following Weinreich (Languages in Contact (1953:1), made this precise point about the individual speaker as the point of contact of languages (1963:3), but this focus on the speaker was not (yet) at all typical in mainstream linguistics.

[15] Jumpelt had read Brang’s summary of Fedorov: “He wants to develop translation theory into an independent scientific discipline because he sees the common denominator of the various translation types in linguistics [my translation]” (1961:6).

[16] Toury (1980:23, 50) introduced the concept by analogy with the notion of norm found in the work of the Romanian linguist Eugenio Coșeriu who, in his 1952 article ‘Sistema, norma y habla’, said that in studying language, one needs to attend not only to system potentials (Saussurean langue; Spanish sistema) and to actual wordings produced (Saussurean parole; Spanish habla) but also to the linguistic habits of a language community, what is typically said (or not said, even though the system allows it). This three-way distinction between what can be said, what people do say, and what is typically said has a parallel in translation: translators at a certain time and place can translate in various ways; they do translate in a certain way on a certain occasion; they typically translate in certain ways, this latter being the norm.

[17] International public interest in oral translation had been awakened a few years earlier during the Nuremburg trials. I distinguish general public interest in translation as an activity from the personal, practical interest of particular groups of people (traders, diplomats, army generals, tourists, immigrants).

[18] According to Revzin & Rozenzweig 1964 (but first suggested by them in 1958, according to Ludskanov 1967/69), appeals to knowledge of the world outside language were only necessary in literary translation. This proved to be a fundamental error. All types of language require such appeals.

[19] Google Translate used the statistical approach starting in 2006 and began implementing the neural approach for a few language pairs (all involving English as either source or target language) in 2016.

[20] The discipline now known as Translation Studies in English, with its apparatus of conferences, journals and doctoral degrees, was not constituted until around 1980. It first appeared locally in connection with translator training (as at ESIT in Paris) or regionally, but by the end of the century, it had been constituted worldwide.

[21] The theoretical answer to the question of how one ‘ought’ to translate had to await the shift in attention, in the late 1970s, to the reception of translations in the receiving culture. It then became clear that the question has no general answer. How people translate does depend in part on the effect desired by the commissioner/ publisher and the translator, but it also depends in great measure on existing linguistic conditions and expectations in the receiving society. As a result, under one set of conditions, people tend to translate more freely; under another, more literally. If conditions change, a new translation of a previously translated text may appear.

[22] The linguistic approach of the 1950s and 60s was not purely descriptive since the descriptions arrived at were used to prescribe ways of translating to students taking foreign language degrees and, increasingly, degrees in translation.

[23] For example, one might write in the style of a medical practitioner when translating a medical text for the general public. This would be a source-oriented approach. But there are three other possibilities, not just one: write in the style of the non-medical community to which the translation is directed, write in the translator’s own voice, or write in some fourth voice. ...(expand)...

[24] Observational studies do not constitute the whole of present-day linguistic translation theory. For example, House (2016) presents translation is being part of applied linguistics, not in the sense of 'linguistics applied' to translation (as with Catford) but in the modern sense of an effort to solve language-related social problems using insights from a variety of disciplines, not just linguistics. And in 2017, Routledge published a 450-page Handbook of Translation Studies and Linguistics.