Dealing with source text ambiguities

Many translation projects typically involve just one or two language pairs. Even when several language pairs are involved, we all too often manage projects as a collection of single language pair projects. Consequently, very plausible but misleading translations of ambiguous source text segments are not always flagged as problematic, much less edited or corrected. Worse yet, our clients may be unwittingly delivering inconsistent — or even different — messages in different countries.

This failure to flush out ambiguities is especially prevalent in projects with longer documents and tight deadlines imposed by the client. We all have clients who pressure us to either deliver parts of longer translations as they are finished, or in the case of multiple language projects, to deliver each language as soon as each is completed.

The following three examples illustrate the challenges of uncovering hidden ambiguities in source text when we manage multiple language projects by using a traditional “silo” approach (managing the project as simply the sum of a collection of single language pair projects).

Example 1: “All meetings and telephone conversations of any substance must be documented.”

Recently I served as both Italian and Portuguese editor of a 10,000-word project consisting of text — including this sentence — involving an online training program for a large insurance company. Two other language pairs that I’m familiar with were involved — Spanish and German — for which we had received edited translations. Following my review of the Portuguese text, which required minimal editing, I plunged into the Italian text. Something did not seem quite right about the Italian segment containing the phrase “conversations of any substance,” an issue that did not arise during my review of the Portuguese translation. In fact, when I placed all four translations side by side, at least three very different interpretations emerged, which later called into question what the English writer was really trying to say:

Italian: Tutte le riunioni e le conversazioni telefoniche su qualsiasi argomento devono essere documentate [unedited draft].

Spanish: Todas las reuniones y conversaciones telefónicas de cualquier naturaleza deben ser documentadas [edited translation].

German: Alle wichtigen Meetings und Telefonate müssen dokumentiert werden [edited translation].

Portuguese: Todas as reuniões e conversas telefônicas, de qualquer nível de importância, devem ser documentadas [unedited draft].

Both the Spanish and Italian translators treated substance as a particular topic, while the German and Portuguese translators considered substance to mean something like “the quality of being important.” Given the overall context of the document, the Italian and Spanish translators got it wrong, while the German and Portuguese translators were closer to the mark… or were they also a bit off?

The German translation simply says that “all important meetings and telephone calls” must be documented, while the Portuguese draft translation says that telephone calls and meetings of “any level of importance” must be documented. The Portuguese text is much more inclusive and very close to a literal reading of the source text. The Portuguese translation (“any level of importance”) ranges from the very lowest level of importance — sounds almost like “unimportant” — to the highest level of importance. This seems to be a baby step away from saying that “All meetings and telephone conversations must be documented.” After all, they all have some substance, or they would not exist.

In contrast, the German translation is a bit more limiting. Only important calls and meetings should be documented. At this point, a looming client deadline quickly put an end to this mental agonizing. As one of our ATA colleagues said several years ago, “Translations are never finished, they are simply abandoned!” Consequently, this is what we delivered to the client after the following in-house edits:

Italian: Tutte le riunioni e le conversazioni telefoniche importanti devono essere documentate [with my edit to draft translation].

Spanish: Todas las reuniones y conversaciones telefónicas de importancia deben ser documentadas [with my edit of the edited translation].

German: Alle wichtigen Meetings und Telefonate müssen dokumentiert werden. [no change to original edited translation].

Portuguese: Todas as reuniões e conversas telefônicas, de qualquer nível de importância, devem ser documentadas [no change to draft translation].

As you can see, I superimposed the German rendition that “all important meetings and telephone calls must be documented” on the Italian and Spanish translations. However, I left the Portuguese translation alone, so it still says that meetings and telephone calls of any level of importance must be documented, while the other three languages say that only important ones must be documented.

When I expressed my post-translation-delivery doubts to our German team (indicating that the Portuguese translation was still in my opinion closer to the source English text), they indicated that a literal German translation — Alle substantiellen Meetings und Telefonate müssen dokumentiert warden — would have let them off the hook, but was not what the author meant to say. They added, “we still think our translation is correct!”

So here we have it in a nutshell. Should we translate everything we are handed literally, or do we translate the meaning that we think the source text author was trying to convey? In fact, after reflecting on the German team’s last comments, it struck me that the source text was the source of the problem. It is quite likely that the author meant to say, “Any meetings and telephone calls of substance must be documented.” If that were the case, then the Italian, Spanish and German translations we delivered to the client were on the mark, while the Portuguese translation was a bit off.

Of course, one might say, “why didn’t you just ask the client what the author meant?” Given that this project actually involved 10,000 words going into nine languages (nine translators and nine editors), it was virtually impossible to raise more than a handful of significant questions with the client. It was also a bit late to seek client clarification in the editing process. Because this example did not involve any matters of life and death (such as a medical instrument or medication with ambiguous instructions), we simply had to move on.

Example 2: “Maintaining job documentation/descriptions, and salary survey information; pricing jobs”

Shortly after unraveling the mysteries of documenting telephone calls “of any substance,” our project manager asked me to edit a Brazilian Portuguese translation of a list of 277 human resource personnel activities (3,150 words). The English source text of each activity appeared in a separate Excel cell. I soon stumbled on the phrase “pricing jobs,” which appeared just after a semicolon.

I was a bit uncomfortable with the Portuguese rendition: precificação de cargos, and more so with the Spanish translation: calcular precios de trabajos. With assistance, I determined the gist of the Russian and Korean translations, which led me to believe that “pricing jobs” had something to do with salary determination.

The original translations of “pricing jobs” appear in Figure 1. At this point, I asked our project manager to seek client clarification. The client said “pricing jobs” was a shorthand way of saying “defining salaries for specific positions.” Armed with this information, we edited the Portuguese and Spanish translations and left the Russian and Korean translations as submitted (Figure 2).

Example 3: “Response: Get medical attention if you feel unwell”

Another Portuguese editing job, which was part of a five-language project, came to my desk as I finished agonizing over “pricing jobs.” The source text consisted of 1,500 words for product labels. The text seemed straightforward enough, until I came to the word response followed by a colon and a detailed description of first aid/first responder activities. The Portuguese translator translated response as Resposta de emergência. This 300% word expansion did not seem directly supported by a literal reading of the source text. Then I decided to view the translations of response in all five languages (Figure 3).

While one might argue that the Chinese, German and Spanish were perfectly (and literally) correct, it became apparent that the French and Portuguese translators used the total context (the verbiage that followed the colon) and took some liberties to come up with their translations. These two translations reflect the underlying meaning of the word response and convey what the source text author was trying to say. In fact, one could conclude that the source text was the source of the problem due to an ambiguity that was initially hidden.



In an ideal world, we should strive to manage multilanguage projects in a “multilingual” manner during all project phases. A multilingual approach can uncover errors and ambiguities in source text that are often unrecognized when we manage in a traditional silo manner, meaning overall project manager, language team 1 (translator, editor), language team 2 (translator, editor), language team 3 (translator, editor) and so forth.

I propose four actions to break down the traditional silo approach to multilingual project management and improve the quality of translations by uncovering and tackling the hidden ambiguities that lurk in clients’ source text:

First: Translators should be encouraged to ask project managers questions during the initial translation phase, especially when they find themselves translating words, phrases and segments that they do not understand after their own research. Project managers need to find answers to questions about source text meaning and share answers with all members of all translation teams as soon as possible.

Second: Translators and editors should be encouraged to submit “translator notes” or “editor notes” covering any remaining uncertainties with all project submissions, including final submissions. When no such notes are present, the project manager is left with the impression that the translation team is equally confident that all words, phrases and sentences have been correctly translated. In most projects, there is always some part of the translated and edited text that is at least slightly tenuous.

Third: We should encourage more cross comparisons of target texts among language pairs during the editing process. While a “super-editor” of multiple languages would be ideal, there are not that many retired popes we can call on. Nevertheless, we should take advantage of the third and fourth working languages of editors. For example, we could use a single editor with command of multiple languages to edit two (or more) languages of a multilanguage project. Alternatively, we could provide editors with both the target text for editing plus one or two other target texts in the editors’ other working languages (or languages they have studied in the past) to be used as reference documents.

Project managers will then need to deal with any issues (ambiguities) raised by editors working with two or more target texts, and decide if they need to seek client clarification of any source text segments. Client answers, in turn, may dictate the need to reexamine one or more target segments of the other language pairs.

Fourth: Project managers need to be increasingly aware of specific, major and indisputable translation errors (and subsequent edits) due to a misreading of the source text by the translator and editor of any of the language pairs they are managing. Then, project managers should reexamine selected parts of the remaining edited translations. If one translator made a major error due to a misreading of the source text, it is highly likely that another translator in a different language pair could have made the same error, which the editor may have missed.

Admittedly, implementation of these recommendations will require additional time. However, even a partial, ad hoc implementation should help mitigate the negative effects of a “silo approach” to project management, yield significant improvements in translation quality by uncovering and resolving source text ambiguities, and help our clients deliver information in multiple languages that is more consistent.