by Jim Hubbert
This article briefly discusses quality assurance (QA) for translated entertainment content, including dubbed and subtitled film and television dialog. I will argue that such translation is often inadequately vetted for quality, and that similar QA approaches should be used for entertainment as well as other types of content.
Translation involves paid creation of work product. To ensure accountability as well as accuracy, translators should not be the final judge of their work, so to gain assurance that the work product accurately represents the source content, the client typically has a third party review the translation for accuracy.
“Accuracy” can be hard to define but is recognizable enough for real-world purposes. If we give the same patent to five experienced translators, we expect to see very similar results due to the highly specific nature of the material. We also expect to mostly agree on what constitutes accuracy for this type of material. However, for entertainment content, we would have to settle for diverse results from our panel of translators, and opinions would vary concerning which version best reflected the intent of the original.
In a recent article in this publication, Dr. David Petersen drew some instructive contrasts between translation that emphasizes “mapping” of source language (SL) word choice and phrasing onto the target language (TL), vs. translation that allows paraphrasing based on underlying SL meaning. The goal of the latter approach is audience-friendly work product that reads as if it were originally written in the target language.
Dr. Petersen suggested that the former approach, which he calls source language-based translation, has at least two advantages over the latter approach, called target language-based translation. First, SL-based translation tends to better meet client expectations―a requirement in the real world―because the work product can be quality-assured in a transparent, accountable manner; second, SL-based translation requires the translator to stay firmly grounded in the source material, limiting the temptation (especially under deadline pressure) to “write around” content that is difficult to map onto the target language. TL-based translation is less likely to read (or sound) like a translation, but the more discretion the translator has in rendering the material, the more difficult it becomes to quality-assure the work in terms of some objective standard.
Nevertheless, entertainment translation must be strongly TL-oriented. If the localized content does not emotionally engage its audience, it fails as work product―there is not much demand for “boring but accurate” entertainment translation, except in academic contexts. Dubbing and subtitling also involve numerous constraints (principally on-screen time constraints) that do not apply to, say, literary translation. This drives entertainment translation further toward emphasizing TL style and cultural norms, and gives translators far more non-transparent discretion than would be the case with an SL-based approach.
Yet, given the dictum that translators should not quality-assure their own work, how should QA for entertainment translation be carried out? Should the “subjective” nature of the material be reason enough to leave difficult decisions largely or entirely up to the veteran translator?
In practice, entertainment translation is far less subject to rigorous QA than are other types of translated material. Yet dispensing with QA entails risks. Nearly always, when translated screen dialog fails to achieve the intent of the original, viewer comprehension and enjoyment is compromised. Any careful sampling of circulating dubbed or subtitled film and television programming quickly uncovers numerous problems that could easily have been prevented with a QA process not unlike that used for “noncreative” translations.
An effective, collaborative translation approach should leverage distinct skill sets among the individuals involved. For example, it makes little sense to have a highly qualified reviewer checking work product from someone with basically similar but less developed skills, unless the goal is translator training. Instead, the most productive QA pass for any translated material will be performed by a native speaker of the source language, with strong source language writing skills and advanced TL skills. The qualified SL native can, so to speak, rapidly and accurately “load” the source material into neural RAM―usually far more rapidly and accurately than the translator, whose primary skill focus should be writing in the target language. The reviewer then scans for material discrepancies between source and translation. For the SL native, such discrepancies are not merely easy to spot; they almost rise up off the page.
Note that polishing the translator’s phrasing is an editorial rather than a QA function. If time and resources permit, review by a TL native with screen writing skills may be useful, and this individual does not necessarily need foreign language skills. However, it will typically be most efficient to have a single-stage QA pass performed by a qualified SL native.
For entertainment content, discrepancies between source and target occur primarily at the level of story and characterization, rather than specific words and phrases. Space does not permit listing examples of mistakes by dubbing and subtitling translators; in any case, given the inherent difficulty of the work, it is amazing how well most translators perform. Still, it is rare for the TL native to have spent thousands of hours, from childhood on, being exposed to SL entertainment content, building up a stock of associations between a wide variety of dramatic contexts and the expressions and communication strategies characters use in those contexts. As such, SL native-speaker review is just as critical for entertainment content as for other types of material, regardless of the translator’s level of expertise.
Note that translation QA by the project producer or director is unlikely to result in reliable error detection. These participants in the production chain are usually most concerned with how the translation “plays” for audiences, rather than whether it faithfully represents the original.
While entertainment content varies widely in terms of difficulty, the major pitfalls for translators tend to lie in segments of dialog that obliquely signal character motivation, indirectly reference character relationships, hint at later plot developments, and so on. These key nuancing lines occur throughout the dialog and are often difficult for TL natives to reliably identify. Dropping or distorting such nuances can significantly impact the viewer’s understanding of the content, blunting the intended effect of the dialog at best and leading to misunderstanding or incomprehension at worst.
Following are just a few of the problems a reviewer may encounter. These examples indicate the general QA orientation for entertainment translation: faithfulness to the intent of the original. Given differences across cultures, this macro perspective is the best way to ensure that the target audience experiences the content as closely as possible to the way the original audience did.
- Misinterpretation. For example, if “get out of here” means “don’t pull my leg”, translating the phrase literally will skew the scene, even if “leave now” appears to fit.
- Failure to include important units of meaning, or adding meaning not found in the original. (Minor alterations are, however, sometimes justified to clarify an unfamiliar context for the target audience.)
- Distorting characterization by overemphasizing some aspect of the character through skewed word choice, resulting in an altered and/or less nuanced impression. An example would be use of overly feminine speech conventions for a female character in a position of responsibility, such as law enforcement officer, CEO, politician etc.
- Following the SL wording too closely and thus compromising its intent. This sort of distortion may be quite subtle. For example, rendering “what’s with you?” as (in effect) “what’s the matter with you?” could skew audience understanding of the relationship between characters; or a character may say something to divert attention from her real feelings, but the translation, while “literally” correct, does not convey this hidden agenda.
As noted, TL error detection requires solid communication skills (verbal and written) in the source language; the reviewer needs to be adept at identifying nuances and pivot points in the dialog that, over the course of the story, deliver a deeper experience for the audience when effectively adapted. Finally, it is the reviewer’s job to identify and explicate variances from the source, not to rewrite the translation. Final resolution of any problems identified is primarily the responsibility of the translator.