Keywords

1 Translation Now

Translating is coming back to additional-language teaching. This is a general trend that has been going on for a few decades; it should be news to no one. Indeed, when I asked attendants at the LCNAU conference in Adelaide to raise their hand if they used translation when teaching languages, the forest of positive responses was so thick that I suspect that, in the particular field of language teaching in Australian universities, translation never really went away.Footnote 1 The prime question here could be not whether translation should be used, but what kind of translation activities are best suited to additional-language teaching (Königs asked the same question in the German context in 2000).

In some less enlightened parts of the world, however, translation has been effectively buried beneath the ideals of various immersive and communicative methods. It is still considered a very bad thing. A few decades ago the situation reached the point where, in France, proponents of the méthode directe had the use of L1 officially banned from the L2 classroom, and with it all possibility of using translation (Harvey 1996, p. 46). The prime reason for this was that translation activities stop students from thinking in L2, and thus hinder them from attaining spoken fluency—which is actually a good argument, if and when spoken fluency is the only aim we have when learning a language. On the other side of business, there are educationalists who actually want a lot of L1 and L2 mixes in the classroom but see translation as being too restricted for the kinds of multilingual skills required. For Zarate et al. (2004, p. 230), translation would be no more than “a reformulating activity that obscures all the challenges to intercultural communication”; what students really need, apparently, is “cultural mediation” (in the United States one would say “translanguaging”). A third mode of attack then comes from language educationists who see translation as a stratospherically difficult language competence, to be attained only by professional translators and thereby condemned to be of no profit or interest for lesser mortals. This is how a 2018 Companion Volume to the Common European Framework of Reference for Languages, which in 2001 included translation and interpreting as full parts of “mediation”, scandalously does not include them among the descriptors and scales for “mediation skills”—translation has apparently become so hard that it is off the scales for language learners (Council of Europe 2018, pp. 35, 107, 113). So translation is somehow either too banal or too difficult for the language class.

What interests me here is that translation can have a positive image in some university settings and such a negative image in what is still mainstream thought on language teaching. It is a problem not of what translation really is, but what people think it is, and how much they are prepared to explore it. There is a serious issue of branding at stake. And if you are on the side of translation, as I am, there is a serious need for rebranding in some quarters. But how can that be done?

Before responding to that question, let me seek a little inspiration in history, since it is in the history of language teaching that many of the current misconceptions and fears are based. For centuries prior to immersion ideologies, language teachers were using translation in their classes. Were they all really so wrong?

2 A Typology of Translation Activities

We know Roger Ascham used translation in his classes in the sixteenth century because he tells us quite a lot about it in The Scholemaster (1579), self-described as the plaine and perfite way of teachying, to understand, write and speake, the Latin tong. A perfect method seems worth following in some detail. After the instructor explains basic syntax and presents a lexicon, the learner is shown a relatively easy Latin passage and text-based teaching begins:

First, let him teach the childe, cherefullie and plainlie, the cause, and matter of the letter: then, let him construe it into Englishe, so oft, as the childe may easilie carie awaie the vnderstanding of it. Lastlie, parse it ouer perfitlie. (Ascham 1579, p. 8)

So to teach children Latin in a “plain and perfect way”, you first write out a translation for them? No, not quite. The learner’s initial understanding of the foreign here seems to be through spoken exchange (“cherefullie” is not a written mode), an oral discussion of words and meanings. Only then is that first understanding of the foreign concretized by a “construal” in the target language, presumably spoken by the instructor—by which we might understand a form of translation that includes explanations and probably fragments of the start text. Following that, the teacher runs through the spoken rendition, again speaking. Then the whole thing is repeated by the learner, speaking but not yet writing:

This done thus, let the childe, by and by, both construe and parse it ouer againe: so, that it may appeare, that the childe douteth in nothing, that his master taught him before.

This privileging of spoken exchange is perhaps motivated, since the text in question is Cicero, the archetypical orator. Yet the perfect method is supposed to be general in application, to both the spoken and written modes. It is only after three or so spoken treatments of the text that a written rendition is sought, along with first mention of the verb to translate and the presence of paper:

After this, the childe must take a paper booke, and sitting in some place, where no man shall prompe him, by him self, let him translate into Englishe his former lesson.

There is only one learner, of course, given the exclusive nature of sixteenth-century education. And the learner is not necessarily a man, despite the pronouns: Ascham’s most famous student, from 1548 to 1550, was Elizabeth Tudor, future queen. This learner is left to invent her own English translation; no model translation is in sight. And she is working on a substantial piece of text, rather than decontextualized words or phrases. The method proceeds:

Then shewing it to his master, let the master take from him his latin booke, and pausing an houre, at the least, then let the childe translate his owne Englishe into latin againe, in an other paper booke.

So the Latin text is taken away and then, at least an hour later, the learner has to back-translate from English into Latin, again in writing. In this case there is indeed a model translation, Cicero’s Latin, and the exercise would seem to have more to do with memorizing that text rather than inventing a rendition. That memorization then becomes a product to be evaluated:

When the childe bringeth it, turned into latin, the master must compare it with Tullies booke, and laie them both togither: and where the childe doth well, either in chosing, or true placing of Tullies wordes, let the master praise him, and saie here ye do well.

Affirmative pedagogy, of course: no scolding of errors, since “there is no such whetstone, to sharpen a good witte, and encourage a will to learnyng, as is praise” (Ascham 1579, p. 8).

Translation works both ways here, into and from English, and is informed by several purposes: first, understanding in spoken exchange and a spoken rendition in English, then written expression in English, and finally, yes, text-based memorization of the start text. Of these three, the translation into English is specifically designed to improve the learner’s capacity for expression, and this purpose is recommended for any vernacular: “what tonge soever he doth use” (p. 8).

These various types and uses of translation might be schematized as follows:

  1. 1.

    Initial construal, in spoken exchange, most probably mixing the two languages concerned. Note that the learner here is already able to write; they might be aged more than 12 or so; there is no suggestion of immersion in a monolingual exchange where the additional language is used exclusively. It is implicitly recognized that internalized or “mental translation” is one of the ways the learner is going to understand the L2 text anyway.

  2. 2.

    L1 translation (classically known as “version” and sometimes as “direct translation”), in written mode, going into the learner’s strongest language in order not just to show that the foreign text has been understood, but also to develop expressive resources in that language. As such, there is not necessarily any one model translation to be attained in this mode, in keeping with what is elsewhere recognized as the indeterminacy of translation (from Quine 1960).

  3. 3.

    L2 translation (classical “prose”, also misleadingly called “inverse translation”) is here used as a written learning exercise based on memory, but this need not be so. Prose also involved original composition in the L2, and translation into the additional language is, in our century, more than a pedagogical necessity for those languages where there are not enough competent foreign translators to handle the demand (as is the case for translations into English from Chinese, Japanese and Korean, for example, along with European languages with smaller numbers of speakers—in all such languages, translators regularly work into their L2).

    Beyond Ascham’s particular circumstances, one might add two further modes of translation that are common enough in the literature:

  4. 4.

    Checking translation, in either direction, is when translation exercises are used as a check on prior acquisition. Ascham’s two translation directionalities would indeed appear to be working in this way, and one can extend the category to all the tests and exams where translation is used to check that non-translational skills (such as grammar and usage) have been acquired.

  5. 5.

    Translation criticism should necessarily arise in an activity with more than one learner in the group, as different versions are compared and commented on. An ideal moment could be where two or more learners disagree about the best rendition, enter into debate between themselves, and discover at some point that their arguments need technical terms to describe translation, if not some useful theoretical concepts.

  6. 6.

    Full translation, which really deserves a better name (non-pedagogical? extramural translation?), would then be what learners do once translation skills have been fully acquired. Elizabeth Tudor went on to translate major texts from Latin, French and Italian (see Elizabeth I 2009), not as learning exercises but as translations to be read and used by those around her (including, for example, the first chapter of Calvin’s Institution de la Religion Chrestienne, rendered just 5 years after she had acquired translation skills from Ascham). This would be an application of Ascham’s method, although it is certainly not the only application: Elizabeth spoke in French, Italian and Latin to foreign emissaries, and gave extemporaneous speeches in Latin on visits to Oxford and Cambridge. Translation skills serve more than translation.

There are at least those six things that can be done with translation. And there are many more: just get any repertoire of activities for communicative language teaching and add another language to each scene. Inventing translation activities is not hard. A more difficult task is to say why the effort might be worth the candle.

3 A Potted History of Debates

A common story is that translation was used all the time in the bad old days when classical languages were being learned: students would have learned the additional language by doing no more than writing boring translations from and to it, without communicative context and with all kinds of interference of one language in the other. That translation-based methodology would then have been unthinkingly dragged across and applied to the teaching of modern languages under the name of “grammar translation”. Somewhere near the end of the nineteenth century, epiphany then occurred: teachers started discovering that languages have to be learned through immersion in spoken communication, as opposed to the bilingual scene of written translation. And variants on such immersion and communicative methods have just been getting better and better ever since.

There are several things wrong with that narrative, and indeed with the counter-narrative that would have a mindless century-respecting pendulum swinging back and forth, for and then against translation in additional-language learning (Kelly 1969/1976). I have taken some time to look at the main nineteenth-century textbooks used for teaching French, German and English as additional languages in Europe and the United States (the survey is in Pym 2016a), analysing them in much the same way as I have tackled Ascham above: who was being taught, at what level, for what purpose, and in what way. My general finding, echoing that of Siefert (2013), is that there was basically no such thing as “grammar translation” as a singular orthodoxy—the term was coined après coup, by detractors of translation near the end of the nineteenth century. Several other points come to light:

  • All the commercial textbooks claim to teach the additional language in the shortest time and in the best possible way, in keeping with the marketing principles used at least since Ascham, with not much change over the years. The early major bestseller, Meidinger, claimed that “learning from rules is the shortest and safest way to learn French” (1783/1799, p. 2, my translation here and throughout), so translation was used to illustrate grammatical rules. A little later, though, Seidenstücker, another bestseller, claimed his method was “imitating, as closely as possible, the natural way in which children come to gain knowledge and use their mother tongue” (1811/1833, p. iii), although he also made extensive use of translation as a learning activity. So commerce was served by contradictory arguments, and translation activities were used right across that spectrum.

  • All the textbooks that use translation as some kind of central learning activity are designed for learners above the age of 12 or so, where mental translation was likely to be involved in initial construal. Even the methods that purported to base themselves on the way young children learn were not designed to teach young children.

  • There are actually very few textbooks that do not allow space for spoken exchange, in addition to or alongside the written exercises. There are indeed some, but they seem to have been catering for the growing market in self-learning, and translation was indeed one thing that buyers could do at home by themselves.

  • The major innovation in the German textbooks from the beginning of the nineteenth century was careful attention to inductive learning, in keeping with the principles of Prussian New Humanism. This meant the learner was invited to discover grammar through contact with the language, with translation exercises being used to introduce points of comparative grammar—thus teaching about differences between languages rather than inviting interference.

  • The pedagogical progression in translation activities meant moving from simple to hard, with various checks on acquisition along the way. This involved using sequences of quite different translation activities. It also meant using decontextualized phrases and sentences at the initial levels, sometimes with quite comic results. Yet it would be unfair to characterize a whole method on the basis of such examples alone.

Of particular interest here is the way in which the inductive approach meant integrating translation as a kind of written code-switching. This produces variants rather like the spoken exchanges that Butzkamm (1980) calls the “sandwich technique”: L2 and L1 are alternated, as in “You’ve skipped a line. Du hast eine Zeile übersprungen. You’ve skipped a line.” Seidenstücker (1811/1833) uses techniques like the following, where the word-for-word vocabulary is given prior to the translation exercise:

Vous, ihr, avez, habt, livre, Buch, acheté, gekauft

Vous avez un bon père et une bonne mère. Avez-vous un livre? Le livre est bon. Nous avons acheté un bon livre. Le livre que vous avez acheté, est bon. […] (p. 2)

Yes, the sentences to be translated are decontextualized and rigorously non-authentic, but they only occur at the initial stages of the method.

Ploetz (1848/1877) uses italics to indicate the grammar point that the student should focus on when translating, in a book that makes heavy use of translation and appears to be designed for self-teaching:

Nos soldats ont combattu contre vos ennemis. (Lesson 13)

He also gives clues pointing to the grammatical differences between the languages, either by inserting notes or by adding numbers to indicate the desired target-language order:

Heute habe ich (Frz. ich habe) zwei Briefe dem Briefträger gegeben. (Lesson 30)

Hat man nicht eine2 herrliche1Aussicht…? (Lesson 41)

Another technique used to much the same effect is to make translation exercises easier by bending the start text to fit the target-language grammar. Here is an example from Ollendorff, where the English speaker is learning French:

Have you the bread?—Yes, Sir, I have the bread.—Have you your bread?—Yes, I have my bread. (1846, p. 10)

Spoken English would perhaps be happier with “Do you have the bread?” or “Have you got the bread?”, but here “Have you the bread?” is preferred because it is more likely to suggest the French “Avez-vous le pain?” (which could also be formulated in several different ways). This bending of language is quite common in the translation-based textbooks; it is one of the main features picked up in the later criticisms of translation activities. Once again, though, the bending and decontextualizing were mainly at the initial stages of the courses. More advanced lessons gave longer texts to be translated, allowing the learner to bring together numerous acquired skills.

As you go through the literature, you become aware not just of the many different ways translation activities were used, but also of the variety of arguments that were mobilized for and against translation. As noted above, diametrically opposed educational philosophies could nonetheless justify translation. One that stands out from the ruck is Claude Marcel’s insistence that, for learners who have already acquired written competence in L1, translation is actually the most natural approach, since it fulfils the role that immersion plays in the child’s learning of that L1:

The native expressions addressed to [the child learning L1] are always accompanied by tones, looks, and gestures, which explain them at once. The translation attached to the text [by the advanced learner of L2] interprets the foreign words at once, as the language of action interprets the native [language]. (1853, p. 93)

With translation, the L2 “ought, therefore, be understood by the learner in less time than the native language by the child” (1853, p. 93). That is, translation can save time, which is precisely what good marketing required.

Marcel later assumes that L1 acquisition progresses inductively through understanding speech, speaking, understanding writing, then writing (1869, pp. 11, 14). In L2 acquisition, on the other hand, the proposed order is reading, hearing, speaking and writing (1869, p. 22). In this way, Marcel attaches the advantages of his initial translations explicitly to reading (the first phase), then includes two-way translation exercises of the checking kind in his section on writing. That is, translation plays two quite different roles, at two quite different stages of the learner’s progression.

This is one of the insights that seem to have been forgotten. The experimental use of translation was interrupted, in some countries for many generations, as another kind of experience intervened.

4 Immersion Against Translation

In the second half of the nineteenth century, criticisms of translation came from two quarters, involving two quite different kinds of arguments.

The first was from German immigrants in the United States who had learned English as adults, basically through immersion in American society. If they could do it, then so could everyone else. In 1866 Gottlieb Heness opened a German-language school where young boys spoke German for four hours a day, five days a week, and they did indeed learn German very quickly. Heness claimed his students could speak fluently in 36 weeks, “a shorter time than any other method of study” (1867/1884, p. 7).

Heness’s basic claim was then picked up, schematized, and successfully marketed by Maximilian Berlitz, albeit with one major difference: whereas Heness had restricted his method to young learners, since “it is very difficult for the adult to understand and speak without translating” (Heness 1867/1884, p. 10), Berlitz offered the method to everyone. His marketing strategy recognizes that the best way to learn a language is actually to travel to the country where it is spoken, but attending Berlitz classes can do the same thing more cheaply and in a more organized way:

Instruction by the Berlitz method, is to the student what the sojourn in a foreign land is to a traveller. He hears and speaks in the language he wishes to learn, as if he were in a foreign country. He has however the advantage that the language has been methodologically and systematically arranged for him. (Berlitz 1888/1916, p. 4)

This “as if he were in a foreign country” means that translation is radically excluded from the Berlitz method, among other reasons because “knowledge of a foreign tongue, acquired by means of translation, is necessarily defective and incomplete; for there is by no means for every word of one language, the exact equivalent in the other” (1888/1916, p. 4). For the purposes of justifying his argument, Berlitz reduces translation to a word-for-word replacement exercise. This strategy was to become quite common.

The second main criticism of translation came from European academics who insisted that the classroom was a space where students should be taught to speak, rather than just write. The founders of what became known as the “Reform Movement” were not gratuitously phoneticians: Jespersen, Passy and Sweet. That said, only Jespersen radically excluded translation, and he did so in two contradictory ways. First, like Berlitz, he adopted a particularly myopic definition of translation, seen as a one-to-one pairing that hides the specificity of L2 and which embraces no explanatory elements. Taking the example of the many ways of rendering the English pronoun “it” in German (according to gender and case), Jespersen regrets that the learner who relies on any one translation will apparently fail to see those differences (1901/1904, p. 135). Second, he argues that there is little social need for translation: “while there is […] a constantly increasing number of people who need to express their thoughts in a foreign language, there are really very few who will ever have any occasion to exercise skill in translation” (Jespersen 1901/1904, p. 53). If you are not going to become a professional translator, forget about translation.

So translation was either too simple to convey useful linguistic knowledge, or too complex to be used in everyday life. Many language teachers would seem to have agreed.

5 Who Needs Translation?

Now, who might need to learn to translate? An example is actually given by Heness when he recounts how one of his students, having spoken German all day at school, replied when his mother asked him what he had been learning. The boy “began to stammer, unable to speak intelligibly”, then confessed: “‘Mother, if you will let me talk in German, I will tell you all;’ and he gave a full account of the day’s adventures in German” (Heness 1867/1884, p. 6). Heness presents the story in order to illustrate the success of his method, apparently for a land where everyday life would require no mediating bilinguals, no movement of content between languages, and perhaps no translators. Others, no doubt including the boy’s perplexed mother, might wonder whether extreme immersion will really save the world.

For almost all the nineteenth-century textbooks, languages are attached to national cultures, which are located in separate countries, and so an additional language is routinely called a “foreign language”. That is why immersion through travel was presented as an ideal way of learning a language: one crossed a border and went from one immersive space to another, as was the experience of the immigrant (who might become the instructor in the United States) and the “grand tour” (for the rich, presenting the ideal that poorer learners were condemned to imitate by paying Berlitz). That is also the way the European phoneticians justified their “direct” method: the learner imitates the crossing of a national border.

That said, when the German immigrants began to draw on their own experience, they occasionally recognized a rather different geometry. Heness’s justification for his pedagogy actually goes back further, to his experience in Germany when going to school meant progressing from a regional Bavarian variety to Hochdeutsch—not from L1 to L2, but from something like L1 to L1.5. Similarly, his American students could be immersed in German because it was not entirely foreign to them: they were going from one language to a cognate language, progressing through the overlaps. The application of nationalist immersion ideology did all it could to eclipse those overlaps, as we see in the case of the stuttering schoolboy, but it was nonetheless formulated from within a social space where the different languages were simultaneously present. Had the boy learned through translation, he should have been able to communicate proficiently within that multilingual space. Elizabeth Tudor, who was taught through translation, was able to move quite successfully through the multilingual spaces that constituted her court.

Who needs translations? One kind of answer concerns the traveller or the young child, and quite another pertains to language learners who can write and who become part of multilingual, multicultural societies, not as professional mediators but as citizens negotiating the linguistic engagements of everyday life. If such societies are to be inclusive, mediation of many kinds is necessary, and immersion need not be a priority.

In short, multilingual societies need translation. It is for this precise reason that the original 2001 version of the Common European Framework Reference for Languages recognizes four kinds of language activities through which learners acquire skills: reception, production, interaction and mediation, with mediation largely being the translation of oral or written texts (Council of Europe 2001, p. 15).

6 So What Is Translation?

As might be clear from this rapid overview of the last few centuries, one of the main ways to argue against translation is to see it as just one thing, mostly either as a simple language-replacement exercise (of little learning value) or as a complex professional activity (needed by very few people), with nothing in between. As noted above, both those reductive definitions are actually present in the 2018 Companion Volume to the Common European Framework Reference, where everything possible is done to keep mediation from being “reduced to translation”, along with repeated insistence that professional translators and interpreters use skills that are far beyond those accessible to normal language learners. The battle is far from over.

The history of teaching methods can nevertheless show that translation involves much more than word-for-word at one end, then specialized magic at the other. So, too, can minimal attention to what the trainers of professional translators have been doing for the past 70 years or so. This concerns not just the various complex models of translator competence (which unfortunately tend to reinforce the idea that translation is only for a few highly gifted specialists) but can more specifically draw on the various typologies of translation solutions, which would result from the discursive strategies that translation can entail.

One such typology, drawn from many previous attempts, is presented in Table 1, which is designed to encourage learners to consider the many ways that translation problems can be solved. (If there is no particular problem, then the translator is in “cruise mode”, at the top of the table.) Solving problems can be deceptively easy at the most abstract level: you copy some kind of expression, you change some kind of expression, or you put something else (those options are in the left-hand column). Each of those basic strategies then includes some general solution types (the middle column), which in turn can be broken down into open-ended lists of tricks that translators can and do use, depending on appropriate communicative circumstances. Any learner who can do all those things appropriately will be able not just to translate, but also to mediate effectively between languages in a very wide number of situations.

Table 1 A typology of translation solution types (cf. Pym 2016b, p. 220)
Full size table

This is not the place to elaborate further on the solution types. Let the table suffice to illustrate that translation is not just one thing.

7 Principles for a Pedagogy of Communicative Translation

If there is to be a general rebranding, the message for language educators has to be that translation is not simple; it involves a wide range of transformational skills that can be used to do more than translate in a word-for-word sense. It could also quite happily be attached to parts of “mediation” or “translanguaging”, if those terms make the educationalists happier and as long as the activity is based on texts of some kind—translation should retain the advantage of making learners pay close attention to written or spoken utterance, actual language use, and to construe meanings carefully from there. (This might be a Protestant virtue, sticking to prior text, as opposed to a sense of mediation that basically involves telling stories with whatever means available, but I am not here to argue theology.)

If translation is to be brought back into additional language learning in any general way, the arguments have to be rather more astute than simply pointing out that the bad old days were not really so bad. One argument is to insist on the range of translation solutions and the way different translation activities can suit pedagogical progression, to counter the characterization of translation as “too simple or too complex”. Another argument is to point out that translation can be more than a checking exercise: it is always a communicative activity, with the adjective in lights, since “communicative” is the word that language educators all over the world rate most highly in their descriptions of teaching methodologies (see Pym et al. 2013). So let’s take their favourite word and apply it to translation.

What would a manifesto for communicative translation look like? Here are some possible principles. Many of them draw on the lessons of history, while others address the problems of our present:

7.1 All Translation Is Communication

Translation is communication when it reaches a receiver (in written or spoken mode, since both are concerned here); it is communication when performed in a role play setting or similar exercise; and it is also communication when used to check on acquisition, since it communicates the acquisition level to the instructor. In short, translation is always communicative, in many different ways.

7.2 One Pedagogy Serves Both the Spoken and Written Modes

The professional distinction that is made between spoken translation (“interpreting”) and written translation (“translation”) is not useful in the learning of additional languages and should be dispensed with. Learners should be speaking and writing all the way through, as they have been doing for at least a few centuries.

7.3 The Spoken Is the Primary Situation

As seen in Ascham, the pedagogical use of written translation can and should begin from spoken translation, along with whatever other kinds of spoken checking operations might be desirable. This principle also applies to the training of professional translators: if they start from spoken interaction, they instinctively break with the literalism that most beginners assume is required in translation. When you translate out loud, there is less linguistic interference, particularly when you are communicating with a person who is right there in front you.

7.4 Translation Is for More Than Professionals

There has been a long, unspoken pact between translation scholars and translation professionals, where the former use the latter as a model to be reproduced, and the latter occasionally allow themselves to be studied in that light. Any such agreement should be broken. Translation skills can be used by anyone, since linguistic inclusiveness and interpersonal development start at grassroots level, in everyday life. There will always be high-risk events where highly paid professionals are required, but translation should be seen as a social activity open to anyone who is interested and motivated. There will always be opera singers (I hope), but everyone can still learn to sing (better).

7.5 Online Technologies Are Always Integrated

These days translation is open to all because anyone can use free online machine translation, with rapidly improving results for many language pairs. Since anyone can access the technology, everyone should be taught about the limits of machine translation and the ways it can be integrated into post-editing processes, with or without translation memory suites. Like mental translation, which has always been present but has rarely been recognized as part of the learning process, machine translation has to be explained and integrated into all additional-language learning, from the beginning.

7.6 The Full Range of Solutions Is Taught

Once they start to explore the limits of machine translation and the possible ways to improve it, learners tend to find that what the machine offers is rather limited in terms of translation solutions. It rarely ventures beyond the odd change of perspective and density, with the rest remaining as literal as possible (although there are curious omissions in neural outputs). Going beyond the machine thus means incorporating solutions at the top and bottom ends of Table 1, importing words and structures when desired, risking complex transformations, and editing content where necessary. Students should progressively be coaxed to a stage where they can play with the full orchestra of solutions.

7.7 Success Is Judged in Terms of Communication, Not Just Equivalence

So what is a good translation? Many different criteria come into play in different kinds of activities, so there can be no one-size-fits-all response. Each activity has its learning objective, and that is where success is to be judged. More important, simple backward-looking equivalence to the start text will often not be enough: we are no longer in the business of memorizing Cicero. Success more generally means fulfilling the function required in the specific communication situation, as has been accepted in translation studies for several decades now.

Together, those seven principles might invite educators to rethink translation activities in communicative terms. And they should do so across the board, both in additional-language teaching and in the training of professional translators.

8 What Is to Be Discovered?

The role of translation in additional-language learning is something that also requires serious research. Over the years I have had the honour of co-supervising three doctoral theses on the general topic, as well as leading a project for the European Commission. These few forays might serve as indications of the kind of research that can be done, in addition to the studies reviewed in Pym et al. (2013).

Peverati (2014) and our European Commission project (Pym et al. 2013) both focus on the opinions and attitudes of teachers concerning the use of translation in the additional-language classroom, both using the same questionnaire. Peverati finds that the transferability of translation-related learning (i.e., the degree to which skills acquired in translation activities can be used in other language tasks) is to some extent thwarted in the Italian context by teachers’ conceptions of translation as being no more than a tool for formalistic, contrastive language work. Pym et al. (2013, pp. 121–122) find that the resistance to translation differs in each country but is particularly pronounced in France and Spain, which happen to be among the countries with the lowest language-learning scores in Europe.

Ayvazyan (2017) presents the results of an experiment with 61 university-level learners of English in Catalonia, with translation activities and non-translation activities being alternated week by week. It is found that the use of translation activities correlates with a higher degree of student-initiated in-class interaction and thus a more pedagogically inclusive learning environment.

Artar (2017) reports on an experiment carried out in Turkey where 30 advanced learners of English were divided into two groups, one of which did translation activities while the other did non-translation activities. After 8 weeks, the scores of the translation group were significantly higher for writing activities, although there was no significant difference in the scores for spoken activities.

There are also ad hoc observations that merit some future research. For example, in my experience as a trainer of professional translators, I have regularly found that heritage speakers with relatively symmetrical bilingualism fare worse in translation tests than do late learners with more asymmetrical bilingualism. Other translator trainers have made similar observations. It could be hypothesized that late learners of L2 grasp the language through mapping operations that are similar to those activated in the translation process, and that heritage speakers tend to have had fewer occasions to develop mapping operations to the same degree. Further, late learners of L2 have relied on written images of the language, mapping written words, whereas early learners have a spoken experience of L2, less given to the written kind of mapping. This could mean that written translation skills are eminently transferrable, but that they might not be of the same nature as spoken production skills. As mentioned, Artar (2017) found that translation activities aided written performances but did not significantly affect spoken language skills.

Some of these findings obviously support a rebranding as communicative translation, as does the historical overview offered above. I am not drawing principles out of thin air. The more important point, though, is that ideas about translation should be open to empirical testing, without pretence to any absolute truth, neither for nor against. There are empirical studies that identify some beneficial effects of translation activities, and there are none, to my knowledge, that support the unthinking exclusion of translation from additional-language teaching.