Courting Disaster: The Perils of Imprecision

Transcription to translation & beyond

Imagine you are an attorney, let’s say practicing criminal law. It doesn’t matter here whether you are counsel for the defense or the prosecution. You are working on a criminal case where the defendant faces serious consequences in the event of a conviction. Let’s also assume that the statements made by a witness during police investigation are crucial for the outcome of the case. Last but not least, to complete our trial scenario: unlike everyone else present in the courtroom, the witness does not speak English very well, so an accredited court interpreter will assist with the testimony.

That key witness is now on the stand during trial and it is your turn to question earlier statements, referring to a copy of the transcript. You ask the witness under oath a question about an essential issue that was also addressed during the earlier police interview to probe whether witness statements are consistent.

According to popular belief, trial lawyers don’t ask questions to which they don’t already know the answers. In this instance, the witness gives a fundamentally different answer, in apparent contradiction with the earlier statement. In such a situation, the text of the earlier statement might be presented to the witness to inquire about that contradiction, with a question such as: “Is it not true that in your statement [made on such-and-such date, in such-and-such place and circumstance] you said, and I quote: […] ?” The witness emphatically denies having made the earlier statement: “I have never said anything like that!”

The witness was truthful in his statement to police, a well as on the stand.

How can that be? Perhaps a more pertinent question here is: why would a court interpreter, i.e. the author of this article, address witness credibility? Let’s rewind to an earlier frame of our fictitious trial scene to unravel the mystery…

[…] the witness does not speak English very well, so an accredited court interpreter will assist with the testimony.

A first and obvious candidate for closer inspection here would be the accuracy of the original transcript and its translation. All too often, translated transcripts are accepted as accurately reflecting original statements without further scrutiny. Of course, when transcribed statements from a recorded interview are in the same language that is used in court, any doubt concerning the quality of the transcript can be addressed simply by comparison with the original recording. In that case, anyone hearing the recording may spot a significant difference between what was said and what is reflected in its transcript.

Moreover, court reporters have a profession with standardized quality criteria and, equally important, a widely recognized system regulating their credentials. This affords a great deal of certainty that a given transcript made by an appropriately credentialed court reporter is both accurate and complete.

The fundamental trouble with translated transcripts

What about transcripts of statements made in a foreign language? Unfortunately, no corresponding quality assurances exist for those. There is no equivalent uniform assessment available to measure the competency of transcribers in other languages. This by itself significantly enlarges the room for potentially costly errors. But that is not the worst contribution to uncertainty.

Once the transcript is made, the next step is translating it into the language of the proceeding, i.e. English. More often than not, the same person providing the transcript also offers its translation. Not that there is something inherently wrong or suspect about the same person doing both tasks. The problem is that all too often this individual is the only one capable of reviewing the material for accuracy. In other words: where an English speaking witness may have a potential back-up resource available in every other English speaker present in the courtroom who is able to compare a transcript with its originating recording, in the case of a non-English speaker that recourse is strongly diminished, if not absent altogether.

That is why a transcriber in a foreign language has an obligation to multiply vigilance and diligence while transcribing. If at all, few are in a position to check for possible problems at later stages of a trial process. This is precisely the reason that I always work with a reviewer when I provide a translated transcript; someone who is equally committed to accuracy in comparing the recording and its transcript and can therefore certify the final translation as a true reflection in English of the original statements. In my view, this second step of certification is absolutely necessary to ensure that translated transcripts are comparable in quality to transcripts made in English by a court reporter. It is a matter of equity and procedural integrity to do so.

The issue of translated transcripts and their perils is worthy of more in-depth treatment, which I will address in a future article. At this point, I must emphasize that when a single individual transcribes and translates without independent review to confirm accuracy, assuming the quality of a translated transcript, particularly when doubt or dispute arise, is extremely dangerous. While time and resource constraints may be a consideration, it is important recognize this highly cost-effective means of establishing the reliability of a translated transcript. Of course, the originating recorded material must still be available as a reference for verification.

When a transcript and its translation are done procedurally correct by committed professionals performing in accordance with acceptable standards of quality and ethics, such situations are not only avoidable, but unnecessary.

Double trouble: back-translating translated transcripts

Aside from the reliability of a given translated transcript, our scenario contains a second source for concern, somewhat buried in a comment on the apparent contradiction by the non-English proficient witness:

In a situation like this, the text of the earlier statement might be presented to the witness to inquire about that contradiction […]

Ordinarily, i.e. in the case of an English speaking witness, this is a simple and effective way to confront different statements: which of these is true, if any? With a non-English proficient witness it’s not that simple, because the transcript offered is in English! And that opens the door to back-translation issues. When you have a statement translated from a given language into English, and then translate it back to the original language, a subtle variation in the first translation can have a disproportionate, exponential effect on the understanding of the statement in the final (back-translated) version.

The exponential nature of variance issues inherent to chain translations probably isn’t too obvious, so I’ll illustrate that with a a few examples.

One analogy serves to clarify the essentially unreliable nature of relays: the fairly well-known telephone game during which a whispered “secret” message is passed from one person to the next until the last person in the group repeats it aloud. The final version is then compared with the original message. Of course, the object of the game is the hilarity caused by an accumulation of nuances that dramatically transformed the message as passed along the chain. In the setting of a court proceeding such differences ar far less amusing when they occur.

A second example illustrates the predicament of chain translations more directly: by feeding a short text in the English language into an automated translation service, translating it into Spanish, and then back into English. The example fragment is taken from the short story “A Clean, Well-Lighted Place” by Ernest Hemingway. Note how the description (action) is “suddenly” attributed to another person, and two seemingly subtle variations in verb use give a very different meaning to the described interpretation of events:

The original fragment:

The two waiters inside the cafe knew that the old man was a little drunk, and while he was a good client they knew that if he became too drunk he would leave without paying, so they kept watch on him.

The same, back-translated to English from an automated translation into Spanish of the above, with my emphasis added:

The two waiters inside the café knew the old man was a little drunk, and while he was a good customer I knew that if he became too drunk he would leave without pay, so I watched.

To be sure, the point of this simple exercise is not at all to criticize the quality of the translation service in question; rather, it is to illustrate the exponential nature of progressive errors in chain translations. In this brief example, three significant differences appear: changes in the grammatical person (“they” > “I”), and changes in verbs (“leave without paying” > “leave without pay” and “they kept watch” > “I watched”) which, taken together, tell a rather different story.

This is why the otherwise not uncommon practice of presenting a witness a transcript of earlier statements should be treated with utmost care in the case of a non-English proficient witness, i.e. when subsequent translations are involved.

I hope that this description clarifies the vital importance of ensuring the accuracy of translated transcripts. Although the example scenario offered at the beginning is fictional, the underlying problematic situation is very much real. During my years of working in courtroom situations, I have encountered several instances where seemingly minimal variations in language use had truly dramatic effects on outcome.

As is often the case, details matter…

Join in and share your thoughts!