Author: patenttranslations

Righting a Wrong

A monolingual attorney’s guide to challenging an erroneous translation


Most patent attorneys have seen their share of sketchy translations. Machine translations routinely end up in prosecution histories and, on occasion, are even submitted to court. Likewise, translations produced with more concern for price and speed than expertise, which may have been suitable for getting the general gist of a document, are commonly used later in proceedings as if they were authoritative submissions. The first two posts in this series outline how to assess the likelihood of errors being present and how to then determine if a specific part of a translation includes a mistake. Because translation errors can falsely make it appear that something was (or was not) disclosed at a particular date, they can have a major impact on your case. Fortunately, it is almost always possible to set the record straight in a way that leaves little room for doubt.

What’s Right Is Right

Attorneys are sometimes hesitant to challenge an existing translation, even if they know it to be wrong, because they worry that the other side will claim that whether a translation is right or wrong is a matter of opinion. It is not. As I have described elsewhere, while more than one correct wording of the same gist may be valid, it is essentially always possible to determine and demonstrate when a translation is incorrect. In fact, given time for reflection on a specific sentence, it is vanishingly rare for translators to disagree among each other in this regard. Both in litigation and as part of editing and review processes, I have seen thousands of errors corrected and have never encountered a situation in which the correct translation could not be clearly established.

The Whole Enchilada

When commissioning a correct translation that will be submitted in order to challenge the existing translation, you will need to decide between a full and a partial translation. In the last installment, we saw that it is unlikely for a successful challenge to be made concerning the overall meaning of an entire document. Individual words or phrases (or sometimes a group of phrases considered together) are most likely to contain demonstrable errors. Thus, from a time and cost point of view, it would seem sensible to translate only the actual sentence in dispute. Bearing in mind that faulty translations are generally faulty in more than one place, commissioning a new translation will probably result in numerous changes in the description throughout the document. Thus, another advantage of a partial translation is that it does not burden attorneys with extra material to review, and does not risk unintended consequences from changes in other parts of the translation.

There are, however, numerous advantages to having the entire text retranslated. The first, and most obvious, is that a full translation heads off any complaints of cherry-picking. But it should also be remembered that consideration of the full context of the sentence does in fact impact the way in which it should be translated. Consider how differently “approximately one hour was required for impregnation” would be translated in a resin strengthening patent and in an article on cattle breeding. And although it is possible to ask the translator to consider the whole document when ordering a translation of just one sentence, the production of a full translation makes this consideration apparent to all. A complete translation may also be useful in demonstrating the inaccuracy of an existing translation, not only in the portion at issue, but throughout, which makes a stronger case for rejecting it. Lastly, while not impossible, it is difficult to provide suitable certification for a translation of just a portion of a document, and even if you do arrive at a suitable wording for the statement, some translators may be reluctant to sign such a certification.

Getting It Right

Considerable care should be taken in commissioning the replacement translation. As the other side will have every reason to want to discredit it, the translation must be error-free in its entirety. In dealing with translation agencies, you should be involved in choosing the lead translator on the project. At this juncture, even if you don’t expect to have to call the translator as an expert witness, it makes sense to choose them as if you do. It is also a good idea to discuss the wording of the certification or declaration that will accompany the translation at the outset, including any summary of the translator’s qualifications that may appear above their signature.

web image2


It is important to leave enough time for a good translation. Don’t make the mistake of waiting to the last minute and then settling for the only available translator. The ones with time on their hands are unlikely to be the best. Remember also that, while most translators can produce about three thousand words a day of polished work, quality improves across the board when they can take one or two days between completing their first draft and reviewing their own work. For an important translation of this sort, word-by-word checking by at least one other linguist is also essential and will add a few days to the turnaround time.

While you probably do not want to bias the translator by exposing them to the existing translation, there is no benefit to having them work entirely blind. There is nothing untoward in explaining the context in which the translation will be used and indicating those portions of the document that are of particular importance. If the translator is experienced in legal translation, this knowledge will allow them to apply the most suitable translation techniques, resulting in a translation that is better able to withstand scrutiny.

If you are not sure that submission of the correct translation will be enough to overcome the existing translation, it may be a good idea to have the translator write a report on why the existing translation is wrong. In most cases, these matters are clear enough that no other translator will disagree, allowing the matter to be settled decisively.

Translation Credibility

A Monolingual Attorney’s Guide to Assessing the Reliability of a Translation

Patent attorneys are increasingly aware that translations can be challenged in litigation, IPR proceedings, and even USPTO prosecution. Even if you are careful to source expert translations for foreign language documents that land on your desk, some cases come with their own translations. Attorneys working on litigation often deal with translations that have been commissioned either by their own clients or the other side’s clients early in the prosecution of a patent, for example, to satisfy IDS requirements. Your clients may also provide their own translations of documents that they consider to be key in the dispute at hand. In other cases, translations may be asserted by an opposing party, with no way for you to determine how credible they are. Fortunately, you do not have to be bilingual to spot potentially unreliable translations. Testing for problems can be fast and painless.

In this three-part series, I’ll talk about ways in which monolingual attorneys can spot dodgy translations of prior art and what needs to be done when you find them. This first post covers some basic “smell tests.” Going forward, I’ll suggest some strategies for translations that fail them.

It Doesn’t Rockets Science
The first indication of a dodgy translation is the presence of major grammatical errors. A few odd word choices or unwieldy turns of phrase do not necessarily spell trouble, especially if you have reason to believe that the writer is not a native speaker of English. On the other hand, if rules that you associate with basic literacy are being broken and the grammar is so bad that the document is hard to understand, further investigation is needed.

Although it is common to see translations that were not prepared by native speakers of English, relying on those that are obviously poorly executed carries significant risk. The problem is, if the translator’s command of English is limited, they literally do not know what they are saying. Imagine yourself translating a patent into another language armed with nothing but a dictionary and your high-school (or for that matter, university level) Spanish or French. How confident would you be about submitting your effort to a patent office or a court? In practice, people who do not have a fluent and confident command of the grammar, vocabulary, and colloquialisms of a language not only produce confusing sentences, they almost always simplify or even omit things that are difficult to express. (Yes, that’s right, unskilled translators will actually leave out important parts of a sentence when they don’t know how to say it in the target language.) What is more, they are incapable of evaluating the true meaning of what they write, such that their best effort will never be more than a best guess.

None of this means that an ungrammatical translation is necessarily wrong. The portion of the description that you are interested in may well accurately reflect the source text, but it would be unwise to assume this without checking, especially if a major part of the case rests on the translation being right.

Another clear sign of trouble is a statement that does not make technical sense. If the description appears to violate a physical law, or the logic that connects the phrases in a sentence escapes you, ask yourself whether it is more likely that the original patent or scholarly article was written by a madman with no knowledge of science, or translated by someone who did not understand it. Likewise, if you are reading within your field of expertise and have never heard of the techniques or equipment being described, it is unlikely that this is due to ignorance on your part. In the big bad world of international translation agencies, bilingual people without technical expertise are routinely asked to translate things they do not understand. And when the going gets tough, these translators wing it. Think of all the bilingual people you know. Now think about how many of them would be able to read and understand the science behind the documents you work with.

In the bad old days, when specialized technical translators were rare, attorneys got used to picking around the bad parts and filling in the blanks (for instance, guessing that when the translator wrote “acetic acid ethyl” they probably meant “ethyl acetate,” and working out from context that “turning current” referred to three-phase current). But even if you can make sense of such products of poor understanding, there is a very high chance that the meaning you take away will not correspond to the meaning of the original document. As with the linguistically challenged, translators who are out of their technical depth routinely resort to simplification and paraphrasing, and because the words they are attempting to reproduce have no real meaning to them, the result is much the same as one would expect from a person who did not know the source language. Remember that for every misunderstanding you spot, there are likely to be several more that you don’t.

Too Good to Be True
Oddly enough, another potential warning sign is an overly pleasant reading experience. This is due to a phenomenon in translation known as “smoothing,” in which a translator inaccurately rewrites the text in a way they find to be more pleasing than the original. An extreme example of this is a single patent claim rewritten as multiple complete sentences because the translator thought that would be easier for the reader to understand. You might also worry if the translation of an academic paper is free of technical jargon and run-on sentences. Smoothing is a type of interpretive translation and is not suitable for translations for evidence. When the cards are down, courts and patent offices care only about what was actually disclosed in the original publication.

Who Done It?
If you have your doubts about a translation on your desk, one route to reassurance is to check the provenance. If the translation has been certified or bears the name of the translation company, you know that the provider at least had enough confidence in their work to attach their name to it. Although certification is not a guarantee, it is a clue. If the translation is made by a translation agency, you can consider the reputation of the agency and the extent to which they specialize in your field. Personal certifications made by an individual translator may indicate that the person has worked alone, without review by a second translator, which may lower your confidence. Obviously, a grammatically unpleasant translation prepared by a foreign patent attorney or scientist deserves more confidence than a similar document coming from the desk of a student. When the origin of the translation is entirely unknown, particular care should be taken, as it is possible that the translator had no idea that anything more than a rough draft was required and never imagined that their work would land on an attorney’s desk.

Although most attorneys know this already, it bears mentioning that machine translations should always be subject to maximum scrutiny. In particular, while services that use statistical translation engines, such as Google and Microsoft, often read quite well, it is routine to find words, phrases, and ideas in machine translations that were entirely absent from the source text or, conversely, for the translation to be missing large chunks of what was originally there. For more information, see my discussion of why this happens and what to do about it.

Houston, We May or May Not Have a Problem
None of the above, however, should be taken to mean that every poorly written translation of unclear provenance must be thrown out. On the contrary, in many situations there are definite advantages to staying with an existing translation. If red lights do start flashing when you read a translation, all that really tells you is roughly how high or low your confidence level should be. As you may have guessed from the above discussion of smoothing, even a great sounding translation can be inaccurate and, by the same token, the chunkiest, most tortured writing sometimes gets the job done.

For an example of how translation errors can be decisive, look at the court’s findings regarding translation (highlighted) in Mitsubishi v Barr, in which Patent Translations Inc. was able to prove that a key prior art document did not disclose what a flawed translation seemed to show.

In the next installment of this series (March 2016), I will describe how to estimate the potential impact of a mistranslation and some good practices for getting a more accurate understanding of the fidelity of an existing translation. (If you can’t wait for the next installment, give me a call and I’ll be happy to tell you what I know over the phone.)

Translating for Legal Evidence

The Northwest Translators and Interpreters Society (NOTIS) and Seattle court interpreters will be hosting a day of workshops and panels on October 4th (first Sunday after International Translation Day) and have been kind enough to ask me to speak on translating for legal evidence.

Evidentiary translation is an area in which translation (written) and interpretation (spoken) converge. As differs from localization, where the goal is often to start with a source text and make something new that fits the target culture, and literary translation, where creativity is almost as important as understanding, both court interpreting and translations for legal evidence require an exact and accurate reproduction of all the content, without embellishment or modification. A court interpreter cannot speak for the witness or offer their own interpretation of the facts. The Washington State courts say:

“A language interpreter shall interpret or translate the material thoroughly and precisely, adding or omitting nothing, and stating as nearly as possible what has been stated in the language of the speaker, giving consideration to variations in grammar and syntax for both languages involved. A language interpreter shall use the level of communication that best conveys the meaning of the source, and shall not interject the interpreters personal moods or attitudes.”

General Rule 11.2 Code of Conduct for Court Interpreters

Because attorneys with less experience in this field tend to assume that every translation they work with will satisfy the rules of evidence and non-specialized translators tend to assume that they should polish, clarify, and edit the text of every translation they work on, it is not uncommon for both to be unpleasantly surprised.

My talk will present best practices for working on evidentiary translations and cover a set of guidelines (recently outlined by Françoise Herrmann in the Translation Journal) and a method known as “conservation of lexemes,” which can be used to produce and evaluate translations for evidence.

The exact time of the workshop as well as the details of other presentations will be posted soon at:

Best Practices in Translations for Legal Evidence: How to Protect Yourself and Your Clients

Two years ago we started on a new edition of the ATA Patent Translator’s Handbook. While that project is still cooking, in the process of writing one of the chapters, I realized that much of the advice we were planning to give to patent translators could be used by anyone working on evidentiary translations. The ATA Chronicle was recently kind enough to publish it, and you can find a copy here.

I wrote the article for translators but legal professionals who work with translations should also know what is in play in this special type of translation, including guidelines for certification, communication policies and the role of existing translations.

If you are interested in a more technical description of the requirements and processes for patent translation, please see the book chapter that I wrote on the subject.

A Great Question: what attorneys can learn from an NYU patent translation course

Recently, I had the honor of being invited to talk about translation in the New York University SCPS Masters in Translation program, as a guest author in Francoise Herrmann’s Patent Translation course.

The students in this class came prepared with many questions that impressed me with the depth of thought that they demonstrated. One question, however, posed by Emily Whelan, went right to the heart of a common problem in patent translation: one with dramatic consequences for both prosecution and litigation attorneys.

The class had been assigned the translation of a Belgian patent filed in 1929, which set out to improve on what was at the time a new invention: crepe-rubber shoe soles. While the early versions of this newfangled footwear were waterproof and resistant to wear, they were also so slippery that, when walking on wet ground, it was almost impossible to move forward. The solution proposed in the patent was to provide “dessins à parties saillantes et rentrantes” (patterns having raised and recessed parts) on the bottom of the sole, as shown in the drawings.


Here, the student felt that the literal translation of, “having raised and recessed parts,” made the text somewhat cumbersome. This is where both patent attorneys (who probably didn’t find anything cumbersome about “raised and recessed parts”) and ordinary translators (who probably did) should pay attention. Like all good translators, the student had learned to avoid calques, which is to say, slavish word-by-word copying of the source text, which sounds unnatural in the target language. Examples of this sort of bad translation can be found on some Chinese restaurant menus (“cooked pig with five fragrant”) and signs for tourists (“annoying parking.”) In fact, interpreting and recasting the source text in the words that seem most suitable in the target language (“five-spice roast pork”/”no parking”) is at the center of the translator’s job description. In other words, translators are trained to believe (usually with good reason) that the more they avoid using clunky, awkward phrasing, the better they are doing their jobs. In this case, the student saw the unwieldy phrase, “patterns having raised and recessed parts,” and considered replacing this with the phrase “corrugated patterns.” There is no question that “corrugated patterns” reads more smoothly and, in light of the picture above, it seemed to be fitting. Her concern was that this phrasing was much shorter than the original and might leave out some of the meaning.

The student is to be applauded. Many, if not most, ordinary translators would not have hesitated to render this as “corrugated patterns” but while this rolls nicely off the tongue, it is very far indeed from the original disclosure. The problem is that it does not cover non-corrugated arrangements that nonetheless have raised and recessed parts such as this:

or even this:



In fact, even the linear pattern having raised and recessed parts that we first saw


need not be limited to corrugated configurations, as that would exclude jagged sawtooth patterns, or irregularly-stepped linear patterns, for example.

To select an accurate translation, the translator must be aware that the drafting attorney chose claim language that was deliberately broad. Without this awareness, the best practices of non-specialized translators, who want nothing more than a comfortable reading experience for their client, often result in the scope of the description being drastically narrowed. As a result, an attorney reading only the text of the translation could be led to believe that the patent does not disclose non-corrugated patterns.

A great deal of patent language sounds unnatural, verbose and stilted to the uninitiated, and this routinely leads to excessive editorial smoothing in lower-quality patent translations. 

Universities offering courses in patent translation are still relatively rare, but it is good to know that NYU is doing its part in making the word of translations a little safer for attorneys. 

What’s Right Is Right

“In the end, isn’t translation a matter of personal opinion?”

Construction in Progress (photo by Jurvetson [flickr])

The question was thrown at me by an attorney for the other side who hoped what she was saying was true. If it were, and translations could only be evaluated in the same way we make decisions about fashion or flavors of ice-cream, then the one she had in her hand would be just as valid as the one that my client had submitted.

Obviously, the answer is, “No.” But the question the attorney put to me is an interesting one, and deserves a longer answer. For one thing, this wasn’t the first time I had heard it. The notion that translation is a nebulous art, ill suited to clear rules or standards, is not uncommon, especially among those who have a little multilingual knowledge. That said, I have been working in technical and legal translation for more than a quarter of a century and, outside of the very special context of a court room, where some people can be particularly disinclined to change their minds, I have never seen two translators remain in disagreement over the proper translation of a phrase for more than a few minutes. Invariably, a short discussion is enough to satisfy one of the translators that the other is right, and an evidence-based consensus is quickly achieved.

So where do non-translators (and even some novice translators) get the notion that there is no such thing as a wrong answer?

Part of it comes from the general idea that language itself is mysterious. It is, after all, amazing that so many of us manage to generate complex and flawless grammatical structures without even knowing the rules. (Be honest, gentle reader, if I were to ask you to give an example of the future subjunctive mood, would you be able to do so?) And then there is our awareness of the ambiguity that pervades our communication. If you have ever spent time with a teenager who has just learned how to tell “That’s what she said” jokes, you will know that there are very few short utterances that cannot be taken in two ways.

Another potential source of confusion is bilingual dictionaries. The entry for the French word “adhésif,” for example, is likely to include such English translations as, not only “adhesive” and “glue,” but also “sticker” and “seal.” If they are all listed in the dictionary, who is to say which one is correct?

Going further, even amateur translators will have come across situations in which the same idea can be expressed in two different ways. There is not much difference between saying that, “the cargo is carried by the vehicle” and saying that “the vehicle carries the cargo.” Both are possible, so isn’t the translator’s preference the ultimate arbitrator?

There are two fallacies at work here. The first lies in assuming that, because human choice is involved, the choice is inherently arbitrary. The second is imagining that, because more than one possible correct translation can be conceived,  all translations must be correct.

In any form of complex communication, ambiguity increases as the sample length decreases. One bit could mean anything, and a handful of bytes in a data transmission is generally useless without knowing which packet it came from. So while it is true that an “elongate member,” mentioned by itself, might mean any number of things to a thirteen year old, when we hear that it is “eccentrically coupled to a rotary drive means,” the possible interpretations narrow significantly. Context, in short, determines which readings are right and which are clearly wrong.

It is also context that tells us which of the many terms in the bilingual dictionary will be appropriate. Even if “sticker” is one of the terms listed under the entry for “adhésif,” it is simply incorrect to translate “collés par un adhésif liquide” as “bonded by a liquid sticker.”

The question of how to select the most suitable phrasing is a little too complex to address in a blog post, but is covered in some detail in my chapter on literal translation in the ATA Patent Translator’s Handbook. Suffice it to say, there are rules and, while there may be more than one possible right answer, there are also unquestionably wrong answers.

Forming an option is indeed part of the translation process, but not all options, or translations, are equal.


It’s official. I will have the pleasure of presenting at this year’s International Japanese English Translation (IJET) Conference in our home town of Seattle.

I am particularly looking forward to this, as this will be the first time that I will do a presentation with someone else, as a team effort. My partner is Masa Kajiki, who runs MK Translation Firm in Osaka, and is an exceptionally talented translator in his own right.

The presentation is called: Beyond Translation for Information: How to Become Indispensable in J to E Patent Translation for Filing and Litigation, and the blurb reads as follows.

“This seminar will explain what US and Japanese attorneys are looking for in both translations for filing and translations for litigation support. We will present a methodology for meeting the requirements of literal translation, as well as strategies for dealing with particularly difficult constructions in translations for filing, which are rarely seen outside of the patent field.

The session will include a short workshop, in which attendees will be asked to try their hands at some classic J>E patent translation stumpers and evaluate the possible solutions. People are also invited to email us in advance ( with difficult phrases and patent-specific terms that they would like to see discussed.

This session is best suited for translators who already have some experience in the field of patents or who have attended seminars covering the basics of patent translation in the past.”