Stieg Larsson's Millennium series, beginning with the global bestseller Män som hatar kvinnor (2005; The Girl with the Dragon Tattoo [2008]), is part of a global hypercanon of popular fiction by writers like Dan Brown, Stephenie Meyer, and E. L. James. Unlike the work of those other writers, however, Larsson's novels were written in Swedish and anglophone readers encounter them in translation, as Nordic noir. English publishers of these novels have not only translated the books but repackaged them for a new market. Our computational comparison of the Swedish and English editions reveals that this publication process also included thoroughgoing edits. It turns out that the creative remaking of a transnational book series is less like a subtitled Swedish film than a Hollywood shot-for-shot remake. If a subtitled film emphasizes the interplay of an original and its translation, a remake absorbs that original into a new commercial enterprise. Our analysis considers the passages left on the cutting-room floor to illustrate this process of transcultural editing in a literary context.
The most obvious transformation of the first novel is its title: the original Swedish title, a literal English translation of which would be Men Who Hate Women, became The Girl with the Dragon Tattoo. This suggestive metamorphosis has been revealing in studies of social and literary elements of the novel (see, e.g., Åström et al.; Bergman; King and Smith). From our perspective here, the choice of the title is clearly part of the publishing machinery described by Karl Berglund elsewhere (see esp. “With a Global Market”; “Genres”). Berglund has offered a theoretical perspective on crime fiction that considers the genre as a book trade category rather than as a specific kind of literary content. This approach focuses on the market-oriented transformation of paratext, including international branding, marketing, and distribution. The latter perspective, however, does not allow us to see how the text itself is transformed.
If literary studies of translation tend to focus on especially relevant moments in a text, and studies of transnational editing and publishing processes have little to do with the text inside the covers, a new approach is necessary to reckon with how international marketing shapes the text itself. No one has analyzed the alterations to Larsson's novels in translation, which comprise changes to the text far beyond the kinds of emendations that fall under the rubric of house style. We took a systematic approach in our study of the translations by designing a computational method for identifying major alterations across all three novels. Previous models for comparing texts at this scale cannot capture the stream of significant and unmarked changes that are integral to the translation, which are visible in aggregate only through a line-by-line comparison. Our results show that the translation differs substantially from the source text, suggesting that the process of translation in the production of global commercial fiction is entangled with the process of editing.
We singled out major alterations to the text by looking for changes to paragraph meaning and length across the three novels. Our semiautomated workflow combines computational alignment with manual curation to create a basis for qualitative analyses of these alterations in the translation process. The digital method allowed us to isolate the heaviest cuts as well as the rare times that material had been introduced. This method, which works by aligning paragraphs, deliberately ignores subtle changes to meaning in favor of more obvious changes to length in order to identify passages that suggest the pen of an editor rather than a translator. Translation studies has long benefited from corpus linguistics methods.Footnote 1 This approach, like ours, pairs text-oriented results in the tradition of descriptive translation studies (DTS) with accounts of the importance of social and historical context—for instance, through polysystem theory.Footnote 2 Our perspective is different, however, because we are not interested in the translation itself. We examine the passages that have been most clearly transformed by substantial cuts or additions of semantic content. What we aim to achieve here is a study not of the translation per se but of the traces of heavy intervention into Larsson's text by the creators of its English version.
Our computational perspective on the Millennium trilogy in English reveals something that is oddly difficult to capture at the level of the sentence, particularly in novels that have struck many readers as underedited: the novels are significantly more streamlined in English. We found that, across the trilogy, about 6.3% of the paragraphs of each novel have been excised or shortened by at least 30%. The extent of this editing suggests that the publishing house treated the English translations of the Swedish novels as if they were original manuscripts. The persistent cuts, often to descriptions of characters speaking or internal thought processes, exhibit a principle of style in line with something like Stephen King's often-quoted advice to “cut it to the bone” and “[g]et rid of every ounce of fat” (12).Footnote 3 The edits thus reflect what might be characterized as commonsense revisions in the US context. We suggest that the push toward target-culture norms for this kind of fiction is stronger because the trajectory goes from Swedish, a language with a relatively small number of speakers, toward English and the US market.
The story of the alterations to these texts has many invisible heroes besides the author and the translator. Because their author, Larsson, died suddenly while the manuscripts were being prepared for publication in Swedish, the books are in fact known for his absence from the publication process and the subsequent struggle for control of his literary estate. In many ways, though, the books are typical of the process of transnational publication. Because of the high profile of the series, we have accounts of the process by key players who note (and sometimes lament) the impact that global publishing practices had on the final version of the texts in English. The substantive changes to the texts are the work of the broader system of actors always involved in a major commercial enterprise like the Millennium trilogy, produced by (among others) the source language publisher, scouts, literary agents, target language publishers, editors, and proofreaders. Once the translator submits the translation, the manuscript undergoes the full editorial process at the English publishing house, which has every power to make changes. A publishing studies perspective is crucial to understanding how this works. What we propose is thus a merging of cultural analytics methods and a sociology of translation perspective. In registering textual differences between the Swedish and English e-books, our model sheds light on the pattern of changes to the text of a bestseller adapted to the anglophone context within the global market.Footnote 4
From Deckare to Nordic Noir
The name for a blood-drenched beach read like The Girl with the Dragon Tattoo in Swedish is a deckare (an umbrella term covering all kinds of crime fiction, including thrillers and suspense), but in translation these novels are marked by Scandinavianness as well—they become Nordic noir. The recent wave of Scandinavian crime fiction in the global anglophone market raises a question: How do the constraints of a specialized genre shape the transformation of a text from one language to another? In this section, we contextualize the Millennium series within the US market for Scandinavian crime that it helped to create.
To understand why the Millennium series was so thoroughly reshaped for an anglophone readership, it is important to understand the linguistic power dynamic. In line with Lawrence Venuti, among others, we understand translation as never transparent or nonideological, but on the contrary always dependent on the cultural situation and norms of the country or countries to which the translation is targeted. Johan Heilbron has created a translation language hierarchy that maps languages as either central, semiperipheral, or peripheral, with English in a league of its own as hypercentral. Translating Larsson into English is a movement from the semiperiphery of global publishing (Sweden) to its center (the United Kingdom and the United States), and the direction of this movement matters. According to Heilbron and Gisèle Sapiro, “[t]he more the cultural production of a country is central, the more it serves as a reference in other countries, but the less material is translated into this language” (96–97). Since the anglophone book trade dominates global publishing, there is more pressure for books from the periphery to adjust to its cultural norms.
Swedish crime fiction gained some minor attention in the United States during the twentieth century, but with the publication of Larsson's Millennium trilogy (2005–07)—translated as The Girl with the Dragon Tattoo (2008), The Girl Who Played with Fire (2009), and The Girl Who Kicked the Hornet's Nest (2010)—that interest exploded.Footnote 5 Larsson climbed the US bestseller charts, and in 2010 he was dubbed “author of the year” by the book trade journal Publisher's Weekly: “With a combined weekly total of 202 weeks on the 2010 bestseller charts and an impressive 59 of those in the #1 spot, his Millennium trilogy . . . outpaced all other bestsellers” (Maryles). This success story led to an increasing demand for novels in the same vein, which, in turn, provided unprecedented commercial opportunities for Scandinavian authors, publishers, and literary agents.Footnote 6 Larsson is not the sole cause of the Nordic noir boom, but he is no doubt its most important author. When he entered the global hypercanon of popular fiction to become a megaseller on par with Dan Brown, Stephenie Meyer, and E. L. James, he broke every expectation of how big a Swedish author of popular fiction could become in English (Berglund, “With a Global Market”).Footnote 7
The impact of Nordic noir on the Swedish and US markets reflects the difference of scale between the two. In less than two decades, the Swedish book trade has shifted from a culture of literary import to one of export (Svedjedal; Berglund, “Turn”; Hedberg). The impact on the US market is less apparent, since only about three percent of all books published in the United States are translations (i.e., of non-English origin). But within this small share, translations from Swedish have almost doubled in the last decade, in part as a result of an increase in translated crime fiction, especially from 2012 onward.Footnote 8 Translations in general are rarely bestsellers, and the only Swedish author who has managed to break into the top ten charts since Larsson is David Lagercrantz—the author who took over the Millennium series following Larsson's death.Footnote 9 The English translations of the original Millennium trilogy thus affected perceptions of Scandinavian crime fiction in general among readers of the English edition. The English edition also affects the global, nonanglophone reception of the genre because English often serves as a transit language between semiperipheral languages like Swedish and other target languages (see Heilbron; Walkowitz 10–11).
How the Sausage Gets Made
Because of its runaway success, the Millennium series is a useful illustration of the rushed process of producing bestsellers in a publishing climate built on hype and the next big thing. Unverified accounts from people centrally involved in the preparation and publication of the novels suggest a heavy editing process with little authorial input. The Swedish publisher, Norstedts, ordered an English translation of the Millennium trilogy for the purposes of selling the rights in 2005—before the last two books were even published.Footnote 10 The translation was done rapidly by the American translator Steven T. Murray (1943–2018), credited under the pseudonym Reg Keeland.Footnote 11 Several years later, that translation was edited by Christopher MacLehose and his team at Querkus Press, and the US rights to this revised version were later sold to Knopf, which made further changes.Footnote 12 Millennium 2 and Millennium 3 were thus translated into English before their publication in Sweden, and Millennium 3 (as we discuss below) was translated from a version that had yet to go through the full editing process at Norstedts (Keeland, “America”). What followed the three books under discussion here was a global launch of the David Lagercrantz sequel to the Millennium trilogy, The Girl in the Spider's Web, which occurred simultaneously in twenty-five countries worldwide in August 2015.Footnote 13 Such an undertaking depends on rapid cotranslations into multiple languages and involves numerous people coordinated by Norstedts, the literary agents, and the rights holders. Under such conditions, a focus on the familiar translator-author dyad, and even on the broader cultural forces that shape translations, leaves much out of the story of how publishers turn a bestseller from a semiperipheral language into a translated book to sell in the major English book markets. For Larsson to get edited when he gets “Englished” is simply how it works, end of story.Footnote 14
We build here on the idea that translation and editing blend together in the phased process of publishing. Karin Stetting coined the phrase transediting to argue that a “certain amount of editing has always been included in the translation task” and that minor cultural and situational adaptations are often necessary (371). Eva Hemmungs Wirtén connects this argument to practices within contemporary publishing to describe the “Swedification” of Harlequin novels: “The category of ‘translation’ and that of ‘editing’ close in on each other, and the line between translator and editor is unquestionably a fluid one. In short, translators edit and editors translate—and this is what the process of transediting involves” (125–26). To study the process of transnational publication from English to Swedish, Hemmungs Wirtén rebuilds the trail from one to the other, from the published book in English to the translator's manuscript “as it came to the publisher” to the editor's changes (read against her own notes on the translation and the translation manuscript) and, finally, to the translated book itself (129). This method tends to highlight the subtle acculturation of the romances to a Swedish context. By contrast, we used an automated process that ignores subtle changes in favor of unsubtle ones, one that cannot recognize internal phases or steps in the process of translation but is sensitive to paragraph-level changes in length in the final text. This outcome-oriented examination of alterations to the text disregards the choices of individual actors. It reveals that thoroughgoing changes to the text are a powerful but normal part of producing commercially oriented translations.
When bestsellers travel globally, they pass not only through the prism of language but through the editing and—importantly, though it is not our central focus here—the marketing departments in the target language. Several scholars of Nordic noir have shown that the branding changes for the market segment are particularly clear in paratextual elements like titles, covers, and supplementary maps (Broomé; Stougaard-Nielsen, “Nordic Noir”; Berglund, “With a Global Market”). We show that a similar claim can also be made for the literary texts themselves. We call this melting down and reshaping of the novel's material literary recasting.
Methods
Our approach to these novels is informed by the sociology of literature, which aims for a systematic analysis of the interconnections between literature and the surrounding society, and by the trend toward sociological approaches in large-scale textual analysis more generally. As James English and Ted Underwood have noted, the two approaches share a focus on quantitative findings, systematic ambitions, and a much closer affinity with the social sciences than the discipline of literary studies holds in general.Footnote 15 In paying attention to the publishing context of the texts we analyze, our project responds to Katherine Bode's sustained argument for the importance of source knowledge and, particularly, of data curation in the computational analysis of digital material (Reading 12–25; World 17–35). By using results derived from an algorithmic-based tool to identify the traces of a corporate production process, we use computational reading to answer a sociologically oriented research question: What can the final shape of a translation tell us about how genre-specific expectations might inflect the editorial process?
We used textual alignment to aggregate alterations to the Millennium trilogy in translation. To study the alterations made to the Millennium series between Swedish and English, we worked from a corpus of the e-book versions in the original Swedish and in translation into American English. We developed a semiautomatic workflow based on computational textual alignment to detect changes in the translations. We used LF Aligner, which relies on the Hunalign algorithm, for the alignment and created a Python script to detect changes to paragraph length.Footnote 16
Given parallel texts in digital form, LF Aligner can automatically pair sentences and paragraphs together across several languages, including English and Swedish. When we tried to align the Millennium trilogy by sentence, however, the program broke down because the original text and the translation differed too much. This was the first computational evidence that the alterations in the English translation were extensive. We then aligned the novels by paragraph, which the tool could manage.
After the initial alignment, we wrote a Python script that singled out all paragraphs that are either missing or differ in character length by 30% or more in the translation.Footnote 17 We checked these manually to correct alignment errors and then ran the aligner again. At this point, we did a second and final manual check to remove paragraphs that differ primarily because of linguistic differences between Swedish and English or as a result of the kinds of expected contractions or expansions in translation that we would not count as editing. Of the 1,347 paragraphs identified as shortened by the script, 163 (12%) were removed manually because we identified the changes as not due to editorial decisions beyond the normal scope of translation. We have treated (for example) the discrepancies in the three passages in table 1 as attributable to linguistic differences or simple translation choices and not as changes in content. This is up for debate: Is cutting a person's first name not a change of content? Is “said” the same as “undrade” (“wondered” in Swedish)? Is “emailed” interchangeable with “växlade mail” (“exchanged email” in Swedish), which emphasizes a reciprocity missing in the English? Though such choices are suggestive, they are relatively close to the original. Here, we are interested in more dramatic cuts and editorial additions.Footnote 18 In table 1 and in subsequent tables, we provide as literal a translation as possible, alongside the text of the US English edition.
Table 1. Three Examples from the Millennium Trilogy Treated as Nonediting: Paragraphs 3254 and 3724 of Millennium 1 and Paragraph 5437 of Millennium 3
These examples of passages that we manually removed from the computer-generated corpus demonstrate the difference between the perspective of a reader familiar with both languages and the perspective of a script that made a rough guess at which sentences said the same thing and then counted characters to match them up. The script does not consider translation choices at the level of individual words that might invite a more nuanced semantic analysis of the translations across the three novels. Our approach is in line with what Sarah Allison has called reductive reading. Allison argues that the reductive process of using computational methods forces literary scholars to be transparent about the elements they are leaving out. To single out a subset of passages based on the presence or absence of a discrepancy in length defamiliarizes the literary material and, in turn, enables us to take a fresh perspective on the text (1–35). There is something clearly reductive about identifying alterations only as changes in the length of paragraphs, yet it makes visible a pattern of editing-in-translation that would be hard to find without computer-assisted alignment and character counts. To Bode's point that it is essential to clarify the source of the texts under analysis: the body of texts in the following sections are not the novels in Swedish and English but a set of passages exhibiting the most extensive editorial alterations made during the process of translation.
We found that a large majority of the alterations in the English translations are cuts. In total, 1,184 paragraphs in the Millennium trilogy have been edited out or shortened by at least 30%. By contrast, only 109 paragraphs have been expanded, and only 4 include additions of completely new material. The cuts are quite evenly distributed across the three titles in the trilogy. In absolute numbers, Millennium 3, the longest of the books, has the most material edited out (452 paragraphs), but by relative frequency Millennium 1 saw the greatest number of cuts—6.7% of its paragraphs have been omitted or significantly shortened. The mean value of cuts for the whole trilogy is 6.3%. Our more fine-grained analysis suggests that these numbers are fairly consistent throughout the series (see figs. 1 and 2). Taken altogether, all three titles appear to have been quite heavily edited in the translation process, with 1 in 100 paragraphs removed completely and 1 in 20 paragraphs shortened substantially. The editing patterns, quantitatively speaking, seem very similar for the first two parts of the trilogy, but the final book appears to have been edited in a somewhat different fashion—it has, for example, a significantly larger number of expansions than the other two novels.
Fig. 1. Absolute numbers of paragraphs edited out or shortened in the English translation of the Millennium trilogy: count per book.
Fig. 2. Relative frequency of paragraphs edited out or shortened in the English translation of the Millennium trilogy: percentage per book.
Editing Out: Streamlining Larsson
One of the things that surfaces most clearly is the systematic streamlining of dialogue and related inner monologue. Consider the passage in table 2, taken from Mikael Blomkvist's first appearance in the trilogy, where a journalist is questioning Blomkvist about the outcome of his trial for libeling the crooked financier Hans-Erik Wennerström.
Table 2. Paragraphs 52–56 of Millennium 1
Both the replies and the material around them have been cut. Blomqvist's reply is terser in the English: “I think not,”Footnote 19 and his internal response to the reporter's body language has been cut entirely.
This passage is typical for the English translation, where lengthier replies are cut down and descriptive phrases and the inner thoughts of characters entirely omitted (see further examples in table 3).
Table 3. Paragraph 1585 of Millennium 1, Paragraph 4486 of Millennium 2, and Paragraphs 5255–57 of Millennium 2
Since dialogue is important in contemporary plot-driven popular fiction, it is no surprise that cuts tend to cluster around dialogue-heavy passages, and in particular description of what characters do and think when they are not directly speaking. It is not only such passages that have been pruned, however. To illustrate the pervasiveness of the editing-in-translation, we reproduce these paragraphs from the beginning of Millennium 1, which describe the first of a series of flowers sent anonymously to an aging tycoon. The passage has been cut by about a fifth (8,509 characters in the Swedish original have been reduced to 6,782 in the English). This scene is hardly spare in translation, unless you compare it with the Swedish original (table 4).
Table 4. Paragraphs 16–18 of Millennium 1
By comparing the final products, it is possible to see that, in Swedish, you learn more about the flower, more about the botanist, more about the challenges of figuring out who might have sent the flower. If a few choices about what to cut might raise questions about the role of interpretation in translation or in editing—like “Indigenous Australians” glossed as “aboriginal,” the curiously emphatic exoticization of the flower, or the insistence that any number of the millions of Swedes might have tried to grow this “difficult-to-grow flower”—the larger pattern seems to be, simply, trimming. In English, there is a little less of everything. Joan Acocella quotes this passage in her critique of Larsson's style in 2011 and quips, “I am basing these judgments on the English edition, but, if this text was the product of extensive editing, what must the unedited version have looked like?” In a discussion of translation in the Swedish book industry, Nils Pennlert quotes a piece of the same passage as evidence that, in the move from Swedish to English, there is “hårda krav på översättarna att ge texterna en pregnans och en förtätning som skapar framåtrörelse” (“a strong demand for the translators to give texts a conciseness and tightening that creates forward movement” [our trans.]). See table 5 for an example of editing that boils Larsson down to a very different substance.
Table 5. Paragraph 2043 of Millennium 1
Our corpus shows analogous streamlining throughout these novels—there are hundreds of examples similar to the ones above. Larsson in the original Swedish is far from spare. In translation, Larsson's tendency toward verbosity is muted but still present. The prose, before it was edited for US readers, was even less efficient.
John-Henri Holmberg—an old friend of Larsson's—has characterized the translation of Larsson as “unfaithful to the original” (34). This familiar formulation drives home a key point here: that The Girl with the Dragon Tattoo was intended not as a faithful rendering of Män som hatar kvinnor but rather as an original product in its own right that took the Swedish novel as raw material. Moreover, we might well consider it as an example of how contemporary popular fiction in a peripheral or semiperipheral language might be more deeply reshaped to suit the norms of the target language when the target language is a central one. If terse prose is the gold standard for crime fiction in the United States and the United Kingdom, which constitute the center of global publishing, then the deep cuts across these books should come as no surprise.
Although the vast majority of the alterations are editorial cuts, there are just over one hundred paragraphs that have been expanded. The majority of the additions are small, clarifying expansions in the dialogue. For example, “När kan du göra det?” (“When can you do this?”) is translated “Can you say when you will be able to provide some concrete information?” (Millennium 2, par. 2507). This emendation does not reflect a pressure to cut to the bone, but these rarer instances of explanatory expansion also indicate the thoroughgoing editing of the novels in translation. In some cases, those passages that have been edited in gloss the Swedish context or emphasize an exoticized Nordicness, revealing what critics have taught us to expect: that the texts have been edited in translation to meet the expectations of Nordic noir. Sometimes, the added text explains elements of the Swedish context to the reader. For example, “kokkaffe” (literally, “boiled coffee”) is expanded to “plainly boiled in a pan in true Norrland style” (Millennium 1, par. 536), and a four-sentence biographical footnote is supplied in the translation to explain a reference to Anna Lindh, a popular Swedish politician who was assassinated in 2003 (Millennium 2, par. 2975). Such additions are closer to the kind of culturally specific transediting described by Hemmungs Wirtén, and the power dynamic of transnational publishing means that more contextual explanations may be necessary in an English translation from Swedish (as a semiperipheral language) than if Swedish were the target language of a book written in English. To make a work of fiction fit in a new cultural context, things like “kokkaffe” and references to former Swedish foreign ministers need to be explained. Moreover, such glosses contribute to the light exoticization of Scandinavia and its politics that puts the Nordic in Nordic noir. The maps of Sweden and Stockholm included at the beginning of the English-language editions—not part of the Swedish originals—work in a similar fashion. If cultural glosses are relatively common in translation, here they are part of a larger editorial process that includes marketing and paratext in the making of crime fiction from Scandinavia into Nordic noir.
We also found a third, more dramatic type of expansion, in which wholly new passages appear to have been introduced in the English. One of the more interesting cases concerns Anita Kaspersson, a victim in Millennium 3. In the Swedish, she is introduced to the reader only after her death. In English, the reader meets her alive, during the moments before she is killed, in a six-paragraph passage focalized from her perspective. It is not unusual for the reader to get a glimpse of the victim's perspective before a murder takes place, even if the character is not central to the novel—this is a way to create suspense in contemporary crime fiction. What is strange is that this perspective is given only in the translation. Such passages are only found in Millennium 3 and reveal a final and critical element of the process of translating the global megaseller: its incredible speed. The presence of passages in English that do not appear in the Swedish edition suggests that, in the rushed and convoluted publishing process, the translation of this book into English was produced before the book had been edited in Swedish, and so passages cut by the Swedish editors are retained in the translation (Keeland, “America”). Such discrepancies illustrate the free-for-all of transnational global publishing at speed and challenge any lingering assumption that the original of this text is the published Swedish edition.
In the commercially important segment of the book trade, it is not uncommon for rights to have been sold to numerous countries before the original publication (Berglund, “With a Global Market” 80–82). Simultaneous worldwide original publication—as in the David Lagercrantz case—and translation processes that involve several translators working in tandem to meet a firm deadline are becoming more common.Footnote 20 The amount of work that happens between the acquisition of translation rights—often based solely on draft translations of a couple of chapters along with a synopsis—and the final publication is substantial. All this taken together fosters a translation climate where producers prioritize speed over text quality, and where fidelity to an original can be a barrier to the more important pursuit of smoothness in the target language (“smoothness” here meaning easy accordance with readerly expectations). Somewhat ironically, then, this might lead to a situation where the most widely read translated novels—the bestsellers—are the most rushed, and thus the ones where the differences from the published originals are the largest.
With the results of our computational study as empirical foundation, we delineate a process of literary recasting in the context of contemporary big publishing, its power dynamics in terms of source and target languages, and the often rushed processes with many actors involved that characterize the most commercial segments of such translational events. The changes we identified are mostly cuts, and range from pruning dialogue to lopping lengthier paragraphs to deleting minor characters and entire scenes. We also detected a few additions that explain or emphasize Nordic references, and one remainder paragraph edited out of the Swedish original, evidence of the hasty translation process in which English translators and Swedish editors worked simultaneously from an unpublished manuscript to create different products.
As a result of all this, the Swedish original version and the English translation of the Millennium trilogy differ substantially. Some commentators on the translation have argued that, in the presentation of a different version of Larsson's novels to the anglophone audience worldwide, the meaning of the originals was lost. By focusing on the publication context, we show how the extensive changes to the novels can be understood as typical for popular fiction that travels from a (semi)peripheral language into English. Rather than being treated as finished works of fiction (to be translated as accurately as possible), such books are treated as manuscripts, as works-in-progress that can and should be extensively reworked before publication in order to improve the novels with target readers in mind. It is thus not a process of translation that also includes editing, but a process of textual recasting that includes translation.
To understand the textual changes we point to here, it is clear that we must look beyond the texts themselves. Translations of commercially important works of fiction always include many actors, they are almost always rushed, and they are often characterized by a multistep translation where a first raw translation (accomplished by one or several translators) is subsequently revised by editors not familiar with the original text. This holds true for many contemporary translations of popular fiction, but the pattern is likely to be stronger when the United States or the United Kingdom are the target countries. National book trades where translations are rare (again, three percent in the United States) may treat them as works-in-progress to a greater extent.
The cuts and additions to the English versions of the Millennium novels that exceed the scope of standard translation choices may well have contributed to the novels’ breakout success. The novels apparently did not even need to be written by Larsson, as the continued success of the series has shown. As commercial translations are tied less and less to an original text produced by an author, they may be increasingly regarded as immaterial rights, content that can be adjusted to fit a new target audience as closely as possible—material to be melted down and reshaped.
To address this phenomenon in a broader international context is work for the future, but our work is suggestive, particularly for other small languages. We know that this brazen degree of intervention does not work the same way in the other direction. When Dan Brown enters the Swedish market no similar significant transformation takes place. The Millennium series reflects the intersection of the hypercentrality of English, the dominance of the United States and the United Kingdom in global publishing, and the market incentives at play when work in a so-called minor language hits the global market. The script we wrote in order to collect all the things cut from and added to these books ultimately shows not the reasoning behind the translator's rendering of a passage, nor a consistent process of cultural adaptation, but the remaking of a novel for the international reader. This process cares little about authorial intent or fidelity to an original. It is the market's expression of the death of the author: Larsson as a globally best-selling author is not the person with authority over the text, but the trademark used to sell a text-based product—in all its different versions across the globe.
