Navigation – Plan du site

AccueilNuméros38Perspectives“Quite puzzling when I first read...

Perspectives

“Quite puzzling when I first read it”: Is reading for literary translation different from reading for post-editing?

Waltraud Kolb

Résumés

Alors que la recherche sur la traduction automatique littéraire a pris un essor considérable au cours des dernières années, on s’accorde encore largement à dire que, pour l’instant, les traductions générées par des machines doivent être suivies d’une post-édition humaine afin d’obtenir des résultats publiables. Cependant, le processus de post-édition diffère à plusieurs niveaux de celui de la traduction à partir du texte original. On s’intéressera ici à la manière dont les traducteurs, d’une part, et les post-éditeurs, d’autre part, lisent et abordent de manière créative le texte source, et comment ces approches différentes se reflètent dans les textes cibles finaux. En prenant pour cadres théoriques la stylistique cognitive et la narratologie, on montrera comment les traducteurs et les post-éditeurs développent leur propre compréhension de l’histoire, des personnages et des mondes évoqués par le texte source ; comment ils déduisent le sens du texte source et de son style et, par exemple, comblent les lacunes au cours de leur lecture. La discussion est fondée sur les résultats d’une étude empirique en deux parties dans laquelle cinq traducteurs littéraires professionnels traduisent une nouvelle d’Ernest Hemingway en allemand et cinq autres traducteurs littéraires professionnels ont post-édité une version traduite par DeepL de la même nouvelle fournie. Les participants ont été invités à réfléchir à voix haute pendant qu’ils travaillaient, et ces protocoles de réflexion à voix haute ont été utilisés pour explorer leurs démarches de lecture.

Haut de page

Texte intégral

Introduction

  • 1 https://www.nobelprize.org/prizes/literature/1954/summary [consulted April 8, 2023]
  • 2 https://hemingwayapp.com/ [consulted April 8, 2023]

1In 1954, Ernest Hemingway was awarded the Nobel Prize in Literature “for his mastery of the art of narrative […] and for the influence that he has exerted on contemporary style”.1 Distinguishing features of his writing include a preference for nouns and verbs (rather than adjectives or adverbs), low lexical variation, short and paratactic sentences, and frequent repetitions of words or phrases (Stewart, 2001). His style is still widely considered the epitome of easy-to-read straightforward prose, so much so that an app that promises to make the user’s writing “bold and clear” is named after him.2 At the same time, his texts are full of gaps and ambiguities so that readers need to take on a very active role, filling those gaps and interpreting those ambiguities. The same applies to translators, who are also—and eminently so—readers.

2Cognitive poetics, which applies cognitive science “to illuminate the study of literary reading” (Stockwell, 2007: 229), ascribes crucial significance to contexts of reading. Reading is seen as a purposeful activity in that it differs depending on why we read. Reading a work of literature for pleasure is different from reading it with the aim of writing an academic essay about it or translating it into another language. What about reading for post-editing purposes? Does it noticeably differ from reading for translation, and if that is the case, does this have a noticeable impact on the target text?

3The present article investigates these questions based on findings from a two-part empirical study in which five professional literary translators translated a short story by Hemingway into German and five different professional literary translators post-edited a German machine-translated version (DeepL) of the same story. The participants were asked to think aloud while they worked on their task (translating or post-editing), and their think-aloud protocols (TAPs) will be used to explore their reading processes.

  • 3 Translators on the Cover. Multilingualism & Translation. Report of the Open Method of Coordinatio (...)

4Hemingway’s language makes his texts interesting candidates to investigate the potentialities and challenges of machine translation (MT) in the literary domain. While research on literary MT has gained considerable momentum over the past few years (for an overview see, e.g., Rothwell et al., 2023), there is still wide agreement that, for the time being, MT needs to be followed by human post-editing (PE) to obtain publishable results.3 However, we do not know yet very much about what actually happens during literary PE, for instance, how post-editors read and how their reading processes differ from those of translators. In this article, I examine these processes against the background of cognitive poetics, looking at how translators and post-editors construct meaning, how they fill gaps, and how they arrive at interpretations of the story, the characters, and the world evoked by the source text.

The role of the reader

5Between the 1960s and 1980s, there was a marked shift of attention in literary theory away from authors and texts to the role of readers as active participants in literary communication. The post-structuralist view that the author is not the sole source of a text’s meaning crystallized in Roland Barthes’ famous announcement of the “birth of the reader”, which “must be ransomed by the death of the Author” (1967; quoted from the first English translation by Richard Howard). Influential theories and approaches which put the reader and the process of reading center-stage also came from the United States and Germany and have been loosely grouped under the collective terms of reader-response criticism and reception theory (Rezeptionsästhetik), with well-known exponents like Stanley Fish, Wolfgang Iser, or Hans Robert Jauss. While Jauss (1970) was particularly interested in how historical changes affect the reading audience, Fish’s chief contribution was the concept of interpretive communities (1980). Iser was mainly concerned with the ways meaning is being produced during the reading process, which he envisioned as a dialogue between text and reader. A literary work, he said, is “more than the text, for the text only takes on life when it is realized” (1972: 279). This realization, through the act of reading, is seen as an act of aesthetic re-creation, in which readers fill gaps and interpret indeterminacies or ambiguities, which abound in literary texts (and play a role in non-literary texts, as well), with readers striving, “even if unconsciously, to fit everything together in a consistent pattern” (Iser, 1974: 283). Readers fill such gaps in different ways, depending on their background, experiences, and perspective, among other things. In a much-quoted passage, Iser compares this individuality of the reading process with how “two people gazing at the night sky may both be looking at the same collection of stars, but one will see the image of a plough, and the other will make out a dipper. The ‘stars’ in a literary text are fixed; the lines that join them are variable” (1972: 287). In their interpretations, readers are never completely free as they are guided by the linguistic features chosen by the author (the “stars” in the sky). Umberto Eco, speaking from the author’s perspective, accordingly described his task when writing a novel as “orchestrating indeterminacies” (Eco, 1989: 39; my translation).

6Interest in the role of the reader and the act of reading as a co-creative act has been further strengthened with the growing influence of cognitive approaches in linguistics and literary studies, and the rise of new fields of research, especially cognitive poetics and the two subfields of cognitive stylistics and cognitive narratology. They are both concerned with “the experience of reading a text”, as literature “literally does not exist until it is read” (Stockwell, 2019: 12, 2), but they focus on different levels, thus bearing a complementary relation to each other (for more on the relationship between the two fields see, e.g., Shen, 2014). Exploring the mind-narrative nexus, cognitive narratology investigates narrative understanding and how cues contained in the text are used by readers (or viewers, listeners, translators) to construct the storyworld, i.e., their own mental model of places, objects, characters, actions, and events (Herman, 2013a). This, however, does not happen independently of situation and context, and as we will see below, readers “navigate storyworlds to the extent required for their purposes in engaging with a given narrative” (Herman, 2013b; emphasis added). Similarly, cognitive stylistics is concerned with the relationship between stylistic features of texts on the one hand and readers’ responses and interpretations on the other and the ways in which literary and linguistic information is processed by the human mind (Semino and Culpeper, 2002: ix). One of the first to draw on cognitive stylistics to explore how literary texts are read and processed by translators was Jean Boase-Beier (e.g., 2006, 2019; see also Tabakowska, 2000; Wright, 2016). Style, she says, “has always to some extent been viewed cognitively—as a reflection and an expression of mind. This is because style is obviously, even in the most formalist approaches, a reflection of choice in a way other aspects of language are not” (Boase-Beier, 2019: 88). The author’s choice, and the translator’s choice. What about the post-editor?

Related research

  • 4 An early study, which did not yet make use of these tools, was Gregory M. Shreve, Christina Schäf (...)

7Over the past decades, cognitive approaches have also gained significance in translation studies, and as a result, reading processes have also become a focus of interest. Such process studies typically use research tools like keylogging, think-aloud, or eye-tracking to investigate different modes of translatorial reading, also taking source text difficulty and different levels of expertise into account.4 Jakobsen and Jensen (2008), for instance, used eye-tracking data to compare levels of cognitive effort and processing associated with four types of reading, i.e., for comprehension, for subsequent translation, while translating orally and while translating in writing. In another eye-tracking study, Schaeffer et al., (2015) investigated target language activation during source text reading and the relationship between translation entropy and eye movements on the source text. Of particular interest in our context is Krings’ study, in which he compared traditional translation and PE processes based on TAPs and also investigated reading processes as part of translatorial decision-making. One of his findings is that in both modes, i.e., translation from scratch and PE, most effort goes into target text-related processes, followed by MT-related processes in the case of PE, with the least attention in both modes given to source text-related processes (2001: 529). Interestingly, however, he also found that the cognitive effort involved in source text processing, especially, reading of the source text, is higher for PE than for translation, which he explains with the fact that during PE the source text has to be compared with the MT output (ibid.: 327). Mesa-Lao (2014) reported the opposite; using eye-tracking data, he compared reading of a source text for translation with reading the same text for PE, analyzing task time, number of fixations, total gaze time duration, and number of transitions between source and target text, and found that PE involves considerably fewer fixations on the source text than translating from scratch as well as shorter fixation durations, which indicates a lower level of engagement with the source text in PE than in translation. In Vieira’s (2016) study of PE, which also included reading processes, eye-tracking was combined with think-aloud and subjective ratings, but no comparison was made between PE and translation from scratch.

Study design

8The material I will use to discuss reading processes during translation and PE was collected in a two-part study: In the first part (2009/2010), five professional literary translators (henceforth referred to as T-1 to T-5) translated a short story by Ernest Hemingway (“A very short story”, 1925/1930) into German; in the second part (2020/2021), five different professional literary translators (PE-1 to PE-5) post-edited a machine translation of the same story, generated by DeepL. DeepL was chosen as an MT system for this study as it is widely used in German-speaking countries and would be a likely choice for literary translators. At the time of the respective study, the participants had been working as literary translators between eight and 29 years; the post-editors had no previous PE experience, which is still typical in the field of literary translation but brings a slight bias to the study. The participants all worked in their usual work environment, in most cases at home; they were asked to think aloud while working on their task and audio-record their verbalizations. They also used keylogging software to produce their target texts; the first group used Translog (Jakobsen and Schou, 1999), the second group Inputlog (Leijten and Van Waes, 2013), as the first group had experienced some technical problems with running Translog (both tools log the same types of data, such as insertions, deletions, and pauses). Process and product data obtained in the course of the study thus include TAPs, keylogs, interim and final versions, printouts with handwritten annotations, other handwritten notes, and questionnaires for background information.

9In this contribution, I will explore how the participants of the two groups derived meaning from the words on the page, how they filled gaps and interpreted ambiguities, both on a narrative and on a stylistic level. To do so, I will use excerpts from their TAPs. Think-aloud protocols (Ericsson and Simon, 1984) have been used in translation process research from early on, despite a number of criticisms that have been voiced (see, e.g., Krings, 2001; Sun, 2011; Vieira, 2016; Jääskeläinen, 2017). One such criticism, i.e., potential inconsistencies between verbal reports and cognitive processes due to the lapse of time, mainly applies to retrospective verbalizations, while concurrent verbalizations (as used here) rely on information in the working memory and are therefore not affected by the passage of time. Another criticism is the potential impact of the think-aloud mode on the translatorial process as such. For example, Krings (2001) and Jakobsen (2003) have reported slow-down effects (see also Sun et al.,, 2020) or a tendency to process texts in smaller segments. As participants in both parts of this study worked under think-aloud conditions, comparative findings are not expected to be unduly distorted. Also, no evidence has come forth so far that thinking aloud significantly impacts the course or structure of the process (Sun, 2011; Sun et al.,, 2020). Needless to say, TAPs will never yield a complete account of a translation process (which goes for any other elicitation method, as well), but what they do yield is rich verbal data of different kinds, including reading aloud of snippets of source or target text or raw MT, more or less disjointed sequences of thinking aloud, incidental remarks, murmured running commentaries, emotional reactions (sighs, laughs), with no time lapse or filter between activity and verbalization—information that would be hard or impossible to obtain by other methods. Social communications such as explanations and rationalizations, usually addressed to the researcher, provide information on the participants’ reasoning leading to their translatorial decisions and, more generally, on their poetics and role perception, thus offering explanations for phenomena and patterns identified in the products. TAPs also allow us to observe different kinds of problem recognition behavior and uncertainty management during translatorial tasks, which, following Angelone (2010), can be either directly articulated (T-5: “this is not clear”, “I don’t understand this at all”) or indirectly articulated (T-1: “hm ... to open a hospital ... hospital ... hm”), or not articulated at all (pauses). As we will see below, the participants’ TAPs contain all of the above-mentioned types of verbal data.

Reading for translation, reading for post-editing

10In what follows, excerpts from the participants’ TAPs will be used to show how translators and post-editors read the source text and interpreted Hemingway’s choices, how they constructed meaning and developed a narrative understanding of the story, the characters, and the world evoked by the text. In “A very short story”, Hemingway tells us about a man and a woman who meet in a hospital in Padua in northern Italy at the end of WW I. The unnamed male protagonist is an American soldier in need of surgery; the female protagonist, called Luz, works at the hospital where he is treated. This is how the story opens (words/phrases relevant for the subsequent discussion are in italics):

One hot evening in Padua they carried him up onto the roof and he could look out over the top of the town. There were chimney swifts in the sky. After a while it got dark and the searchlights came out. The others went down and took the bottles with them. He and Luz could hear them below on the balcony. Luz sat on the bed. She was cool and fresh in the hot night.
Luz stayed on
night duty for three months. They were glad to let her. When they operated on him she prepared him for the operating table; and they had a joke about friend or enema. He went under the anaesthetic holding tight on to himself so he would not blab about anything during the silly, talky time. After he got on crutches he used to take the temperatures so Luz would not have to get up from the bed. [Hemingway, 1930 [1925): 83)

They would like to get married but do not have the requisite documents. After the armistice, he has to return to the United States and wants Luz to come with him. Luz, however, decides to not join him until he has found a good job:

  • 5 In the first US edition of 1925 and the subsequent 1930 edition, which was the source text the pa (...)

On the train from Padua to Milan they quarreled about her not being willing to come home at once. When they had to say good-bye, in the station at Milan, they kissed good-bye, but were not finished with the quarrel. He felt sick about saying good-bye like that.
He went to America on a boat from Genoa.
Luz went back to Pordenone5 to open a hospital. (ibid.: 84)

In Pordenone, Luz falls in love with an Italian major, and she and the male protagonist never see each other again. Hemingway’s story is often read as autobiographical. He wrote it just a few years after he had been wounded himself as a volunteer driver for the American Red Cross in northern Italy in 1918 and had fallen in love with an American nurse (Agnes von Kurowsky) in a hospital in Milan where he spent six months recuperating. In the first version of the story, the female character was even called Ag (see also Scholes, 1990: 46). Like the fictional couple’s relationship, theirs also did not last. What we are concerned with, however, is not whether Luz should be read as a fictionalized Agnes but rather how the study’s participants developed a narrative understanding of who she is—as Hemingway left much to his readers’ imagination.

11Before zooming in on the question of Luz’ identity and a number of other gaps and ambiguities and how the participants dealt with them, it will be of some interest to look at how much reading aloud of source text segments the participants did—after all, the source text serves as the “prompt which triggers narrative comprehension” (Dancygier, 2008: 54). Though the participants were not instructed to read aloud they frequently did so. This was also observed by Krings (2001: 321-322), who found that if participants in his study needed to read segments of the source text, they generally read them out aloud. He also found that source text-related reading processes occurred about twice as often during PE than during translation (ibid.: 327), as the post-editors needed to check the MT version against the source text. Table 1 shows how much of each participant’s TAP was devoted to reading aloud chunks of source text (calculated by subtracting segments of source text reading from the think-aloud transcripts). As participants might well have (and in all probability will have) also read segments of the source text silently, the numbers in Table 1 cannot be taken as accurate measures but can still serve to indicate general tendencies.

Table 1. Source text read aloud by participants, as percentage of each participant’s TAP.

Participant

Percentage

Participant

Percentage

T-1

17.46

PE-1

12.86

T-2

15.48

PE-2

14.21

T-3

10.00

PE-3

20.58

T-4

8.85

PE-4

19.82

T-5

14.75

PE-5

16.55

Average (mean)

13.31

Average (mean)

16.80

Median

14.75

Median

16.55

As Table 1 shows, post-editors did—on average—more reading aloud of source text material than translators. This is in line with Krings’ findings, though percentages are much lower in his study (2.69 % for translation from scratch, and 5.75 % for PE), which might also have to do with the fact that his study did not deal with a literary text. What is more, there is considerable inter-subject variation—something that has also been found to apply to the participants’ overall working style and working speed (see Kolb, 2023). For instance, the percentage for T-1 is almost twice T-4’s and well above the average (mean) and median for the group of post-editors. But as we will see below, reading and re-reading chunks of the source text, taken by itself, is not necessarily a reliable measure of the cognitive effort that goes into source text processing.

12One sentence (henceforth referred to as “the Pordenone sentence”) emerged as particularly interesting in this context, and I will use it as a starting point to explore the participant’s reading processes. It is the last sentence in the passage quoted above: Luz went back to Pordenone to open a hospital. DeepL translated it quite literally: Luz ging zurück nach Pordenone, um ein Krankenhaus zu eröffnen.

  • 6 Two studies of Hemingway’s story drawing on mental space theory are Elena Semino, “Possible world (...)

13For readers, engaging with literary narratives “entails mapping textual cues onto the WHEN, WHAT, WHERE, WHO, HOW, and WHY dimensions of mentally configured worlds”6 (Herman, 2013b; emphasis in original). Regarding the Pordenone sentence, the following questions arise: Who is Luz? What exactly does she intend to do, and what is it that she wants to open? And why does she even want to do so? The TAPs are in German, but the excerpts below relating to the Pordenone sentence (and a few others) have been translated into English for this contribution; italics indicate that something was originally read or said in English; three dots indicate a pause in speaking, if put in square brackets, words omitted by me; German terms that are relevant to the discussion have not been translated and are printed in small caps, with English translations or short explanations provided in footnotes.

  • 7 Lazarett: hospital for wounded or sick soldiers; military hospital.
  • 8 errichten: to build, construct, erect.
  • 9 eröffnen: to open.
  • 10 einrichten: to set up, furnish, equip, fit out.

T-1: Luz went back to Pordonone ... to open a hospital, Luz ... went back to Pordenone, this will ... maybe not be a hospital but a Lazarett,7 she will, yes, well … errichten8 a Lazarett, this may be it, to open a hospital [...] this will probably be the Lazarett but checking ... hm ... to open a hospital ... hospital ... hm ... Lazarett, base hospital is a Lazarett... ha! ... it will be a Lazarett ... we are in the war [...] Lazarett is also a military hospital, which means it is reasonable to translate it as Lazarett ... even if it was a [regular, W.K.] hospital, after all [...] not eröffnen9 but einrichten,10 errichten.

  • 11 aufbauen: to build (implying a prolonged process).

T-2: [translating earlier sentence:] Luz would not come home until he had a good job, this was already quite puzzling when I first read it, that he says coming home, of course, she could be American, the original one was also American, but because of the name Luz I somehow assumed that she is not from there [...] [translating Pordenone sentence:] Luz went back to Pordenone to eröffnen a hospital there, sounds like she baptized it and threw a champagne bottle against it, maybe to aufbauen11 a hospital there ... maybe ... she seems to be a sort of head nurse, which she was in real life, too.

T-3: Luz ... went back ... Pordenone [...] Luz went back to, where now is Pordenone? ... ah, in Friuli, Pordenone, to eröffnen a hospital.

  • 12 aufmachen: to open (slightly more colloquial than eröffnen).

T-4: [after first reading of story:] I am not yet sure about the story, Luz, whatever kind of name that is ... it sounds Spanish, most likely, hm, Luz seems to be a nurse or perhaps a young doctor ... but more likely a nurse, because it says she prepared him for the operating table and this is what the nurse does, surgery nurse ... perhaps ... and then it is strange that she ... aufmacht12 ... a hospital [...] [translating Pordenone sentence:] to open a hospital ... a nurse ... pff ... or even if she was a trainee doctor ... who aufmacht a hospital ... but to open a hospital [...] also, why does she aufmachen a hospital if she wants to go to America as soon as possible? ... this is not logical .... eröffnen ... no, this sounds as if she was the mayor who eröffnet the hospital [...] well, it is during the war, and probably it is not even a hospital but a Lazarett.

  • 13 Hospital (obsolete): (often a smaller) hospital.

T-5: [after first reading of story:] so first, is Luz a doctor or is she a nurse, I assume she is a nurse [...] I think she is not a doctor but a nurse, but this also is not clear, and he is an enlisted man and a nurse would get involved with him, if she was a doctor ... she would not [...] Luz was on night duty ... for three months, night duty .... so she will have been a nurse if she is on permanent nights [...] is she Italian or what, this is not clear, Luz, Luz ... Luzia, has she as an Italian nurse? ... this is simply not clear [...] [translating Pordenone sentence:] to open a hospital, what does she open a hospital for? ... that´s ridiculous [...] Luz went back to Pordenone ... and eröffnete a hospital, now I don’t understand this at all, but never mind, maybe a new hospital has been aufgemacht and she was there or something ... to open a hospital [...] careful now ... to eröffnen a Hospital,13 I love this word Hospital [...] I like it, it is so nicely old-fashioned [...] could he have been in a Lazarett? [...] tss, no idea whether he was in a civil hospital or a Lazarett.

PE-1: [post-editing earlier sentence:] Luz would not come home […] hm, this is strange because she, Luz is not, I think, Luz is not American, or is she? maybe she is? [...] [post-editing Pordenone sentence:] Luz went back to Pordenone [reading German MT] ok, I leave it like that [...] [post-editing later passage:] ok, so Luz is American, after all, because she had never known Italians before, now this fits in with the home-coming [...] there is also the question whether to write something like Lazarett because is takes place during the war, but I will leave it for now.

  • 14 zurückgehen: to go back.
  • 15 zurückkehren: to return.

PE-2: Luz went back to Pordenone to open a hospital [reading three more sentences in English, then reading German MT:] Luz ging zurück14 to Pordenone, here I would rather say kehrte zurück.15

PE-3: Luz went back to Pordonone to open a hospital, ok [reading German MT].

PE-4: Luz went back to Pordenone to open a hospital [reading German MT].

PE-5: [post-editing earlier surgery scene:] what is it you do as an anaesthesiologist, you prepare him for it [...] [post-editing Pordenone sentence:] I think it is strange that she is the one who eröffnet it ... this sounds as if she somehow in a sort of ceremony ... a gift ribbon or something, to eröffnen a hospital, probably a new one was aufgemacht there and she is part of it [...] it can hardly be her hospital, or can it? [...] to aufmachen a hospital I don’t really like either, but I don’t like eröffnen.

As the TAP excerpts indicate, the extent to which participants engaged with textual cues to build a coherent story varies between subjects, not all participants identified the same gaps and reflected on the same issues; what clearly emerges, though, is that the purpose of their reading (for either translation or PE) had an impact. One of the mappings done by the participants concerns Luz’ identity. Hemingway does not tell us anywhere who exactly she is, whether she is a nurse or a doctor, whether she is American or something else. The TAPs show that the process of “on-line inferencing to keep track of characters in the text and to up-date our representations of them” (Culpeper, 2002: 273) is a complex affair. Of the five translators, two (T-1 and T-3) did not question Luz’ identity. The others used various textual cues to come up with a coherent representation: her (possibly Spanish or Italian) name (T-2, T-4, T-5), her amorous involvement with an American soldier (T-5), her role in preparing him for surgery (T-4), her being on night duty for three months (T-5), and her role in the opening of a hospital (T-2, T-4, T-5). From T-2’s TAP we know that he was aware of the story’s autobiographical angle. He was especially puzzled by the phrase she would not come home because, for him, it contradicted other textual cues, such as her name. From the group of post-editors, only two briefly touched upon Luz’ identity when working on other sentences, but not when working on the Pordenone passage: For PE-1, the sentence about Luz’ reluctance to come home also triggered uncertainty regarding her nationality (as it did for T-2), which she considered resolved at a later point when she read that Luz had never known Italians before she got involved with an Italian major, which she took to mean that Luz was not Italian herself. PE-5 made a passing reference to Luz’ identity when reading the earlier surgery scene, calling her an anaesthesiologist, however, with no doubt or question mark attached. Some of the above excerpts (e.g., T-5’s reasoning that a doctor would not get involved with a low-ranking soldier but a nurse would) also illustrate how readers “rely on their knowledge of culturally and historically variable trait-codes to map textual cues onto individuals-in-storyworlds”, with these “repertoires of trait-names” deriving from a variety of sources, not least “folk psychology” (Herman (2013b), partly drawing on Chatman (1978: 123-126); on cognitive stylistic approaches to literary characterization see also Culpeper, 2002).

14But there are further ambiguities in the Pordenone sentence, apart from Luz’ identity. T-4 and T-5 wondered how the whole event fitted in with the overall storyline and struggled with constructing a coherent narrative (T-4: “why does she aufmachen a hospital if she wants to go to America as soon as possible?”); T-5 openly voiced her dissatisfaction with the source text and/or its author (“this is ridiculous”). Another concern for some participants was what exactly it was that Luz intended to do and what kind of hospital it was. T-1 and T-2 explicitly considered (and eventually chose) verbs with more specific meanings for to open, while others (T-4, T-5, PE-5) hesitated between two near synonyms, with one of them (aufmachen) being more colloquial than the other (eröffnen). For T-2, T-4, and PE-5, the verb even evoked a kind of official ceremony (“champagne bottle”, “mayor”, “gift ribbon”).

15Regarding the type of hospital, T-1, T-4, and T-5 spent quite some time debating whether the hospital was a civil or a military one and eventually chose the military option (Lazarett). PE-1 was the only post-editor who briefly considered Lazarett as an alternative but then decided to not change the MT version which does not specify the type of hospital (Krankenhaus). For German readers, the term Lazarett serves as an indicator that events take place during/at the end of a war. It appears in three of the translations, but none of the post-edited versions. Hemingway at no point in the story uses the term war but evokes the war setting via other cues: searchlights, front, armistice, battalion of arditi, major. Of particular relevance in this context is the term searchlights as it appears in the very first paragraph and contributes to setting the scene, so to say. All five translators rendered the term with the German Suchscheinwerfer, which is also a compound that contains the element to search (suchen) and will in all probability be read as a subtle evocation of a war context. Of the post-editors, three retained the MT suggestion of Scheinwerfer (a general term for a lighting unit that projects a strong beam of light, e.g., floodlight, spotlight, or headlight), with no indication in their TAPs that they were in any way uncertain about the choice of word. The other two post-editors, also without providing an explanation, replaced it with Suchscheinwerfer, thus offering their readers the same early cue of a war context as Hemingway and the translators did.

16As was to be expected, differences in the reading processes had an impact on the target texts. In the case of the Pordenone sentence, the five translations contain four different German verbs for to open, with slightly different meanings (eröffnen, aufbauen, errichten, einrichten), and two variants for hospital (three times Lazarett, twice Krankenhaus). The post-edited target versions are less varied; four of them are identical, retaining the terms supplied by the MT engine (eröffnen, Krankenhaus). PE-5’s version is the only one that uses different terms, i.e., a more colloquial verb for to open (aufmachen) and a regional, but otherwise unspecified variant for hospital (Spital). PE-1’s comments on two occasions that she will “leave it like that” signal some degree of uncertainty on her part, on which, however, she did not act, accepting the MT version of the sentence without any edits.

17What the above examples also show is that stylistic choices (such as a specific German term for to open or for hospital) also carry meaning. Another example is how some of the participants dealt with the repetition of good-bye in the scene preceding the Pordenone sentence. The translators (with one exception) devoted much effort to the question of whether it should also appear three times in their target texts. For example, T-1, in her 5th revision, eventually decided on reproducing the repetition, reading it as a “clue to [the text’s] meaning” (Boase-Beier, 2006: 4), as the following excerpt from her TAP shows: Hemingway repeats it three times, which means the protagonist does not want to leave, there is this hesitation ... and this keeps him there, and this is why I have to repeat it”. T-2 was more concerned with his own stylistic priorities, which contradicted Hemingway’s; the way he read the source text it seemed to him “as if he [Hemingway, W.K.] somehow doesn’t have a repertoire … well, this was his first short story, and, true, it is very distinctive and definitely written in this Hemingway style, but this can’t have been a deliberate stylistic choice”; he therefore decided to “jazz up the language a bit here” and do away with the repetition. T-3 also articulated her misgivings, complaining about Hemingway’s “monotonous” tone, but eventually went along with his choice—as did T-5, after going back and forth between repetition and variation. By contrast, T-4’s TAP (and her keylog) does not show any indication of uncertainty or hesitation when reading the three good-byes and then fluently typing in her translation (for a more detailed discussion of this passage see Kolb, 2011, 2021).

18DeepL reproduced the repetition, using the same German verb (verabschieden) three times, but omitting the kiss: Als sie sich im Bahnhof von Mailand verabschieden mussten, verabschiedeten sie sich, waren aber mit dem Streit noch nicht fertig. Er fühlte sich krank, weil er sich so verabschiedet hatte. (English gloss: “When they in the station of Milan had to [verb] good-bye, they [verb] good-bye-d, but were not yet done with the quarrel. He felt sick because he had [verb] good-bye-d like that.”) When the post-editors read the MT version, they all either explicitly or implicitly reacted to the repetition, which is even more conspicuous in German (PE-1: “of course this is not beautiful, verabschieden muss, then verabschiedeten sie sich, but it is also repeated in English, to say goodbye and kiss goodbye”). But their engagement with Hemingway’s choice remained superficial, they did not follow it up with any further deliberations as to the reasons or implications of the repetition. They were more concerned with the omitted kiss and correcting what they perceived as an MT error (PE-3: “there is now once, one, two, three times verabschieden, one of them completely wrong, that is, kissed good-bye”; PE-4: “the kiss is missing”). Three of the five translations and two of the post-edited versions reproduce Hemingway’s choice by using the verb verabschieden and/or the corresponding noun Abschied three times.

(Dis)Similarity between target texts

  • 16 Post-editese has been defined as “the expected unique characteristics of a PE text that set it ap (...)

19Given the findings above, the question arises how varied or similar target versions of the complete text are, and whether the post-edited target texts are indeed more similar to each other than the translations, as the above examples suggest. In the context of research on “post-editese”,16 Castilho and Resende (2022), for instance, computed variance scores for selected textual features, such as lexical density, lexical richness or sentence length, of Brazilian Portuguese translations and post-edited versions of two English novels, with mixed results for individual features (see also, e.g., Toral, 2019). No such detailed quantitative analysis is attempted here. To broadly assess how similar/different from each other the five translations on the one hand and the five post-edited versions on the other are, differences between target texts within each group were calculated with CharCut (Lardilleux and Lepage, 2017). CharCut is a character-based metric used to quantify how similar/different two strings of text are. It can be used to evaluate MT output by comparing it to a human reference translation, or to determine the edit distance between an MT output and a post-edited version of it by displaying insertions, deletions, and shifts. Here, it is used to determine whether the two groups of target texts (translations vs. post-edited versions) differ from each other in terms of in-group homogeneity/heterogeneity. Table 2 shows the CharCut distance scores for each target text in comparison with each of the other four target texts within the same group, with average (mean), median, and variance scores. Table 3 shows the average or mean in-group distance score for each target text and also includes CharCut scores for the distances between the five post-edited versions and the MT output (see also Kolb, 2023).

Table 2. Distances between target texts calculated with CharCut (arrows indicating direction of comparison).

Translation

CharCut

Post-edited version

CharCut

T-1

T-2

0.373

0.371

PE-1

PE-2

0.361

0.328

T-1

T-3

0.315

0.316

PE-1

PE-3

0.335

0.334

T-1

T-4

0.409

0.380

PE-1

PE-4

0.305

0.300

T-1

T-5

0.344

0.356

PE-1

PE-5

0.381

0.388

T-2

T-3

0.360

0.360

PE-2

PE-3

0.369

0.397

T-2

T-4

0.431

0.398

PE-2

PE-4

0.316

0.340

T-2

T-5

0.376

0.392

PE-2

PE-5

0.395

0.444

T-3

T-4

0.402

0.373

PE-3

PE-4

0.342

0.340

T-3

T-5

0.303

0.317

PE-3

PE-5

0.338

0.397

T-4

T-5

0.391

0.443

PE-4

PE-5

0.346

0.363

Average (mean)

0.371

Average (mean)

0.356

Median

0.373

Median

0.344

Variance

0.0015

Variance

0.0013

Table 3. Average (mean) distance of each target text from others in the group and distance of each post-edited version from MT output (CharCut).

Translation

Average in-group distance

Post-edited version

Average

in-group distance

Post-edited version

Distance from MT

T-1

0.358

PE-1

0.342

PE-1

0.282

T-2

0.383

PE-2

0.369

PE-2

0.314

T-3

0.343

PE-3

0.357

PE-3

0.320

T-4

0.403

PE-4

0.332

PE-4

0.258

T-5

0.365

PE-5

0.382

PE-5

0.366

The results in Tables 2 and 3 indeed suggest that texts within the PE group tend to be more homogeneous or similar to each other than texts in the translation group (higher CharCut scores indicate more difference/less similarity). Table 2 shows that both the average (mean) and the median of the distance scores are higher for the translations than the PE versions (0.371 vs. 0.356 and 0.373 vs. 0.344). Results of average in-group distances (Table 3) point in the same direction: two (out of five) translation scores are higher than the highest PE score (0.382), while two (out of five) PE scores are lower than the lowest translation score (0.343). Another measure to assess the (dis)similarity of the texts under consideration is variance, which shows the dispersion of data points (in this case individual distance scores) from the mean. Variance is higher for the 20 translation scores than for the 20 PE scores (0.0015 vs. 0.0013), meaning that the PE scores are less widely dispersed from the mean compared to the translation scores, which again points to greater homogeneity of the PE versions and greater heterogeneity of the translations, respectively. However, results do not reach statistical significance (p < 0.05), and given the small number of participants and the source text’s shortness (633 words), they cannot be more than rough indicators. As with regard to the amount of source text that was read aloud by the participants, inter-subject variation is high here, too. For instance, T-4 clearly stands out with consistently high CharCut scores: a distance score above 0.4 in relation to each of the other four translations (Table 2) and an average in-group score of 0.403 (Table 3), the only one above 0.4. One explanation for this result could be the tendency of this translator to be more explicit than the source text author, a personal translatorial preference she is well aware of and repeatedly commented on in her TAPs (for more on this see Kolb, 2013). In the PE group, PE-5 is the most different from the other PE versions, which is in line with what we have observed in the examples above—where this post-editor was the only one who changed the MT suggestions for to open and hospital, while the other four retained the MT output. Not surprisingly, PE-5 is also the most different from the MT output, meaning that it contains the most edits (Table 3). The highest number of edits in the final target text, however, does not necessarily mean that this post-editor also put in the most PE effort in terms of time or keyboard activity. In this case, PE-5’s technical effort was the second highest in terms of keystrokes, but she spent a shorter time on the task than any of the other four post-editors (for more details see Kolb, 2023) – which again demonstrates the significance of personal working styles.

Conclusion

20In his exploration of What We See When We Read, Peter Mendelsund, a visual artist and author, describes the act of reading in similar terms as Iser did, pointing out that “the images we ‘see’ when we read are personal [...] Every narrative is meant to be transposed; imaginatively translated. Associatively translated. It is ours” (2014.: 207). “Good books incite us to imagine” (ibid.: 203; emphasis in original), he says, and provocatively calls it a form of “robbery” if we as readers are told too much. Hemingway certainly does not tell us too much in “A very short story”, and as readers we need to make inferences, fill in gaps, and interpret ambiguities. The examples discussed above against a background of cognitive stylistics and narratology have shown that the way we read also depends on why we read, and that reading for translation is not the same as reading for post-editing. The translators who participated in this study tended to engage on a deeper and more personal level with the source text than the post-editors, even if they, on average, read aloud less of the source text than the post-editors. Especially when producing their first draft, the translators engaged more actively in mapping textual cues, constructing meaning, and considering the implications of choices made by the author on both the narrative and the stylistic level. In the case of PE, the MT system supplied such a first draft, thus priming to some extent the reading and interpreting processes of the post-editors—who saw little need to question the algorithmic integration of textual prompts beyond detecting and correcting MT errors and obvious inconsistencies (after all, the primary purpose of PE). Even though the study has also shown that there are considerable differences between individuals, we can expect that such priming effects, together with a possible proneness of post-editors to accept an MT version as good enough even though this version might otherwise not have been their first choice, will in many cases give rise to target texts that are less varied than translations from scratch. Given the limited scope of this study, findings primarily serve to indicate general tendencies and encourage future research. The use of eye-tracking, for instance, might yield interesting insights into translatorial reading processes for various types of literary texts (including narrative prose that is less straightforward than Hemingway’s) and in which ways attention on the source text and reading/interpreting processes are impacted by the PE task.

Haut de page

Bibliographie

Corpus Text

Hemingway, Ernest, “A Very Short Story”, In Our Time, New York, Charles Scribner’s Sons, 1930, p. 83-85 [first published in the United States by Boni & Liveright, 1925].

Works cited

Angelone, Erik, “Uncertainty, Uncertainty Management and Metacognitive Problem Solving in the Translation Task”, Translation and Cognition, Gregory M. Shreve and Erik Angelone (eds.), Amsterdam, John Benjamins, 2010, p. 17-40.

Barthes, Roland, “The Death of the Author”, trans. Richard Howard, Aspen, Volume 5/6, Item 3, 1967, https://www.ubu.com/aspen/aspen5and6/threeEssays.html#barthes [consulted April 8, 2023].

Boase-Beier, Jean, Stylistic Approaches to Translation. Translation Theories Explored, Manchester, St Jerome Publishing, 2006.

–––––, Translation and Style, London, Routledge, 2019.

Castilho, Sheila and Resende, Natália, “Post-editese in Literary Translations”, Information, Volume 13, Issue 2, 2022, p. 66, https://doi.org/10.3390/info13020066 [consulted April 8, 2023].

Chatman, Seymour, Story and Discourse: Narrative Structure in Fiction and Film, Ithaca, Cornell University Press, 1978.

Culpeper, Jonathan, “A Cognitive Stylistic Approach to Characterisation”, Cognitive Stylistics, Elena Semino and Jonathan Culpeper (eds.), Amsterdam, John Benjamins, 2002, p. 251-277.

Dancygier, Barbara, “The Text and the Story. Levels of Blending in Fictional Narratives”, Mental Spaces in Discourse and Interaction, Anders Hougaard and Todd Oakley (eds.), Amsterdam, John Benjamins, 2008, p. 51-78.

Eco, Umberto, “Vom offenen Kunstwerk zum Pendel Foucaults. Umberto Eco im Gespräch mit Jean-Jacques Brochier und Mario Fusco”, Lettre International, Volume 5, 1989, p. 38-42.

Ericsson, K. Anders and Simon, Herbert A., Protocol Analysis: Verbal Reports as Data, Cambridge/London, The Massachusetts Institute of Technology Press, 1984.

Fish, Stanley, Is There a Text in this Class? The Authority of Interpretive Communities, Cambridge, Mass./London, Harvard University Press, 1980.

Herman, David, Storytelling and the Sciences of Mind, Cambridge, MIT Press, 2013a.

–––––, “Cognitive Narratology”, the Living Handbook of Narratology, 2013b, https://www-archiv.fdm.uni-hamburg.de/lhn/node/38.html [consulted April 8, 2023].

Iser, Wolfgang, “The Reading Process: A Phenomenological Approach”, New Literary History, Volume 3, Issue 2, Winter 1972, p. 279-299.

–––––, The Implied Reader: Patterns of Communication in Prose from Bunyan to Beckett, Baltimore/London, Johns Hopkins University Press, 1974.

Jääskeläinen, Riitta, “Verbal Reports”, The Handbook of Translation and Cognition, John W. Schwieter and Aline Ferreira (eds.), Hoboken, NJ, Wiley-Blackwell, 2017, p. 213-231.

Jakobsen, Arnt Lykke and Schou, Lasse, “Translog Documentation”, Probing the Process in Translation: Methods and Results, Gyde Hansen (ed.), Copenhagen, Samfundslitteratur, 1999, p. 151-186.

Jakobsen, Arnt Lykke and Jensen, Kristian T. Hvelplund, “Eye Movement Behaviour Across Four Different Types of Reading Tasks,” Looking at Eyes. Eye-Tracking Studies of Reading and Translation Processing, Susanne Göpferich, Arnt Lykke Jakobsen and Inger M. Mees (eds.), Copenhagen, Samfundslitteratur, 2008, p. 103-124.

Jauss, Hans Robert, Literaturgeschichte als Provokation, Frankfurt am Main, Suhrkamp, 1970.

Kolb, Waltraud, “The Making of Literary Translations: Repetition and Ambiguity in a Short Story by Ernest Hemingway”, Across Languages and Cultures, Volume 12, Issue 2, 2011, p. 259-274.

–––––, “‘Who Are They?’: Decision-Making in Literary Translation”, Tracks and Treks in Translation Studies. Selected Papers from the EST Congress, Leuven 2010, Catherine Way, Sonia Vandepitte, Reine Meylaerts and Magdalena Bartlomiejczyk (eds.), Amsterdam, John Benjamins, 2013, .207-221.

–––––, “‘Hemingway’s Priorities Were Just Different’. Self-Concepts of Literary Translators”, Literary Ttranslator Studies, Klaus Kaindl, Waltraud Kolb and Daniela Schlager (eds.), Amsterdam, John Benjamins, 2021, p. 107-121.

–––––, “‘I Am a Bit Surprised’: Literary Translation and Post-editing Processes Compared”, Computer-assisted Literary Translation: The State of the Art, Andrew Rothwell, Andy Way and Roy Youdale (eds.), London, Routledge, 2023, p. 53-68.

Krings, Hans P., Repairing Texts. Empirical Investigations of Machine Translation Post-Editing Processes, trans. by Geoffrey S. Koby et al.,, Kent, Ohio, Kent State University Press, 2001.

Lardilleux, Adrien and Lepage, Yves, “CHARCUT: Human-targeted Character-Based MT Evaluation with Loose Differences”, Proceedings of the 14th International Workshop on Spoken Language Translation, 2017, p. 146-153, https://aclanthology.org/2017.iwslt-1.20 [consulted April 8, 2023].

Leijten, Mariëlle and Van Waes, Luuk, “Keystroke Logging in Writing Research: Using Inputlog to Analyze Writing Processes”, Written Communication, Volume 30, Issue 3, 2013, p. 358-392, DOI: 10.1177/0741088313491692 [consulted April 8, 2023].

Mendelsund, Peter, What We See When We Read, New York, Vintage, 2014.

Mesa-Lao, Bartolomé, “Gaze Behaviour on Source Texts: An Exploratory Study Comparing Translation and Post-editing”, Post-editing of Machine Translation: Processes and Applications, Sharon O’Brien, Laura Winther Balling, Michael Carl, Michel Simard and Lucia Specia (eds.), Newcastle upon Tyne, Cambridge Scholars Publishing, 2014, p. 219-245.

Rothwell, Andrew, Way, Andy and Youdale, Roy (eds.), Computer-assisted Literary Translation: The State of the Art, London, Routledge, 2023.

Schaeffer, Moritz, Dragsted, Barbara, Hvelplund, Kristian T., Balling, Laura W. and Carl, Michael, “Word Translation Entropy in Translation: Evidence of Early Target Language Activation During Reading for Translation”, New Directions in Empirical Translation Process Research: Exploring the CRITT TPR- DB, Michael Carl, Srinivas Bangalore and Moritz Schaeffer (eds.), Berlin, Springer, 2015, p. 183-210.

Scholes, Robert, “Decoding Papa: ‘A Very Short Story’ as Work and Text”, New Critical Approaches to the Short Stories of Ernest Hemingway, Jackson J. Benson (ed.), Durham, NC, Duke University Press, 1990, p. 33-47.

Semino, Elena and Culpeper, Jonathan (eds.), Cognitive Stylistics: Language and Cognition in Text Analysis, Amsterdam, John Benjamins, 2002.

Shen, Dan, “Stylistics and Narratology”, The Routledge Handbook of Stylistics, Michael Burke (ed.), London, Routledge, 2014, p. 191-205.

Stewart, Matthew, Modernism and Tradition in Ernest Hemingway’s, In Our Time, Rochester, NY, Camden House, 2001.

Stockwell, Peter, “Cognitive Poetics and Literary Theory”, Journal of Literary Theory, Volume 1, Issue 1, 2007, p. 135-152, 229-230.

Stockwell, Peter, Cognitive Poetics: An Introduction, London, Routledge, 2019.

Sun, Sanjun, “Think-Aloud Translation Process Research: Some Methodological Considerations”, Meta, Volume 56, Issue 4, 2011, p. 928-951.

Sun, Sanjun, Li, Tian and Zhou, Xiaoyan, “Effects of Thinking Aloud on Cognitive Effort in Translation”, Linguistica Antverpiensia, Volume 19, 2020, p. 132-151.

Tabakowska, Elżbieta, “Is (Cognitive) Linguistics of any Use for (Literary) Translation?”, Tapping and Mapping the Processes of Translation and Interpreting, Sonja Tirkkonen-Condit and Riitta Jääskeläinen (eds.), Amsterdam, John Benjamins, 2002, p. 83-95.

Toral, Antonio, “Post-editese: An Exacerbated Translationese”, Proceedings of MT Summit XVII, Volume 1, 2019, p. 273-281, https://aclanthology.org/W19-6627 [consulted April 8, 2023].

Vieira, Lucas Nunes, Cognitive Effort in Post-Editing of Machine Translation: Evidence from Eye Movements, Subjective Ratings, and Think-Aloud Protocols, PhD dissertation, Newcastle University, Newcastle upon Tyne, 2016.

Wright, Chantal, Literary Translation, London, Routledge, 2016.

Haut de page

Notes

1 https://www.nobelprize.org/prizes/literature/1954/summary [consulted April 8, 2023]

2 https://hemingwayapp.com/ [consulted April 8, 2023]

3 Translators on the Cover. Multilingualism & Translation. Report of the Open Method of Coordination (OMC) Group of EU Member States’ Experts, 2022, p. 58-59, https://op.europa.eu/en/publication-detail/-/publication/a4059b86-8317-11ec-8c40-01aa75ed71a1/language-en [consulted April 8, 2023].

4 An early study, which did not yet make use of these tools, was Gregory M. Shreve, Christina Schäffner, Joseph H. Danks and Jennifer Griffin, “Is there a special kind of ‘reading’ for translation: an empirical investigation of reading in the translation process,” Target, Volume 5, Issue 1, 1993, p. 21-41.

5 In the first US edition of 1925 and the subsequent 1930 edition, which was the source text the participants received, Pordenone was misspelt as Pordonone (it was corrected in later editions). How the participants dealt with this typo is beyond the scope of this contribution, which is why all spellings appear in their correct form here.

6 Two studies of Hemingway’s story drawing on mental space theory are Elena Semino, “Possible worlds and mental spaces in Hemingway’s ‘A very short story’”, Cognitive Poetics in Practice, Joanna Gavins and Gerard Steen (eds.), London, Routledge, 2003, p. 83-98; Gabriela Tucan, A Cognitive Approach to Ernest Hemingway’s Short Fiction, Newcastle upon Tyne, Cambridge Scholars Publishing, 2021, p. 251-254.

7 Lazarett: hospital for wounded or sick soldiers; military hospital.

8 errichten: to build, construct, erect.

9 eröffnen: to open.

10 einrichten: to set up, furnish, equip, fit out.

11 aufbauen: to build (implying a prolonged process).

12 aufmachen: to open (slightly more colloquial than eröffnen).

13 Hospital (obsolete): (often a smaller) hospital.

14 zurückgehen: to go back.

15 zurückkehren: to return.

16 Post-editese has been defined as “the expected unique characteristics of a PE text that set it apart from a translated text” by Joke Daems, Orphée De Clercq and Lieve Macken, “Translationese and post-editese: How comparable is comparable quality?”, Linguistica Antverpiensia, Volume 16, 2017, p. 89-103, here p. 90.

Haut de page

Pour citer cet article

Référence électronique

Waltraud Kolb, « “Quite puzzling when I first read it”: Is reading for literary translation different from reading for post-editing? »Palimpsestes [En ligne], 38 | 2024, mis en ligne le 30 novembre 2024, consulté le 23 mars 2025. URL : http://journals.openedition.org/palimpsestes/9109 ; DOI : https://doi.org/10.4000/12spc

Haut de page

Auteur

Waltraud Kolb

Waltraud Kolb est Assistant Professor de traduction littéraire au Centre d’études sur la traduction de l’université de Vienne. Elle a étudié la traduction (anglais, français, portugais/allemand) et est titulaire d’un doctorat en littérature comparée. Ses recherches portent notamment sur l’utilisation de la traduction automatique et de la post-édition dans le domaine littéraire. Elle est également traductrice professionnelle et membre du conseil d’administration de l’Association autrichienne des traducteurs littéraires. Ses dernières publications comprennent : “’I am a bit surprised’: Literary Translation and Post-editing Processes Compared,” in Computer-assisted Literary Translation: The State of the Art, Andrew Rothwell, Andy Way, et Roy Youdale (eds), London: Routledge, 2024, pp. 53-68 ; “Human and Machine Translation of Nonce Words in Literary Texts: Johann Nestroy’s ‘Talisman’ and its English Translations,” avec Wolfgang U. Dressler et Elisa Mattiello, Target 35 (4), 2023, pp. 540-572 ; “Brazilian short prose in German: A study of literary post-editing,” Revista Tradumàtica 2, 2023, pp. 3-86. https://doi.org/10.5565/rev/tradumatica.347

Haut de page

Droits d’auteur

Le texte et les autres éléments (illustrations, fichiers annexes importés), sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search