Recently, DeepL introduced a new tool called Clarify. After translating a text using DeepL Translate, users can click on a magnifying glass icon and see questions about their text. Was term X meant this way or that? Should this expression be translated with the original language in parentheses? When I first tested out this new feature, my initial reaction as an academic translator was that the tool was going to make it even harder to justify the merits of a human translation over machine-generated translation. After some consideration, however, I’ve reached just the opposite conclusion. In fact, I think Clarify can actually show the researchers for whom I work exactly where my value lies. Clarify is revealing at least part of the work I do with every sentence of every translation.
Example of DeepL’s Clarify
Let’s look at an example from Weber’s Die protestantische Ethik: [1]
„Der Pietismus ist auf dem Boden des Calvinismus in England und besonders in Holland zuerst erwachsen, blieb durch ganz unmerkliche Übergänge mit der Orthodoxie verknüpft, und vollzog dann gegen Ende des 17. Jahrhunderts in der Wirksamkeit Speners seinen Eintritt in das Luthertum, teilweise dogmatisch umfundamentiert.“
DeepL offers a translation[2] and Clarify asks four questions:
- Pietismus: Der Pietismus wird oft als „evangelische Mystik“ bezeichnet. Soll diese Information in die Übersetzung aufgenommen werden?
- Ja, füge diese Information in Klammern hinzu.
- Nein, diese Information ist nicht erforderlich.
- Orthodoxie: Es gibt für diesen Begriff mehrere Übersetzungsmöglichkeiten. Geht es hier um die orthodoxe Kirche, oder um Orthodoxie im Sinne von „traditionell“ oder „konform“?
- Es geht hier um Orthodoxie im Sinne von „traditionell“ oder „konform“.
- Es geht hier um die orthodoxe Kirche
- Speners: Es ist vielleicht nicht jedem Leser klar, wer Spener war. Soll die Übersetzung kurze Details liefern?
- Ja, es soll in einer Fußnote darauf hingewiesen werden, dass Spener ein deutscher Theologe war.
- Nein, es soll nur der Name verwendet werden.
- Luthertum: Der Begriff „Luthertum“ hat keine genaue Entsprechung im Englischen. Wie sollte er übersetzt werden?
- Ersetze ihn durch eine Beschreibung dieses Begriffs (z.B. „the teachings of Martin Luther“)
- Ersetze ihn durch einen ähnlichen Begriff (z.B. „Lutheranism“).
These are legitimate questions, and if I were working on translating these two segments, I would ask myself the same things. In this example, it has found all of the issues that I myself would need time to consider while crafting my translation – in other examples I have found that it has missed important questions. The tool has even offered some suggestions on how to solve the issues, although these are by no means the only possible solutions. What Clarify cannot do, however, is answer the questions. As an academic translator, that is my job. To know which translation to choose, further research will likely need to be done, for example looking into how “Lutheranism” is used in similar literature or whether an explanation of Pietism is necessary (and if so, how much explanation). This is what good academic translators have always done – it has just been work that was usually invisible.
Making invisible work visible
This new feature asks some of the right questions when it comes to translations. It reveals at least part of the work that academic translators do with every sentence they translate. It is possible that researchers will use Clarify to translate their own articles sentence by sentence, answering each of Clarify’s questions, deciding whether Clarify has missed any issues, and doing any necessary research to decide which option is best for each particular case. However, I do want to note that for most academic sentences I have tested with this feature, Clarify has asked between two to four questions – when multiplying that by the number of sentences in the average article, it will amount to a rather staggering number of questions to answer. Partly for that reason, it may actually turn out to be a tool that reveals the painstaking work academic translators have been doing all these years and the value of finding a translator with the expertise to both recognize which questions need to be asked and know how to go about finding the answers.
[1] Weber, M. (1984). Die protestantische Ethik: Eine Aufsatzsammlung (J. Winckelmann, Ed., 7th ed., p. 115). Gütersloher Verlagshaus. (Original work published 1920)
[2] DeepL’s initial translation: “Pietism first developed on the basis of Calvinism in England and especially in Holland, remained linked to orthodoxy through very subtle transitions, and then, toward the end of the 17th century, under the influence of Spener, made its entry into Lutheranism, partly with a new dogmatic foundation.”