- Reproduce the meaning. Patents are usually very technical so you must understand the text well to reproduce the right meaning. There is often ambiguity in implication; therefore, you need to have a certain degree of knowledge to understand the text. This does not mean that you must have a master’s degree in the subject matter, though of course that helps; it means that you need to be able to research very well.
- Reproduce sentence breaks and carriage returns. A lot of translators like to break longer sentences to the short ones because they feel that it will be easier for the reader. Do not do this! A judge at patent proceedings may sometimes reject a translation without even looking at the content only because the sentence breaks are not in the same place as in the source document. The sentence is the fundamental unit in law, and it needs to be respected. The solution for this is: if you have a long sentence and you feel that it is difficult to track, use semicolons in the places where you would use periods and you will get the same effect yet respect the sentence. Keep the carriage returns where they are in the original, too. The carriage returns sometimes have a meaning in patents so it is best just leaving them alone.
- Be consistent with vocabulary and phrasing.This advice is generally true for any translation, but it is even more important in patent translation due to the specific ways that people read and analyze the patent. If you call an element “a shaft” in one instance, you should not call the same element “a rod” or “an axle” in another part of the translation. That rule applies to phrasing also. If you have a phrase in one place that is translated as “a shaft for driving a flywheel”, do not translate it in another place as “a shaft by which the flywheel is driven.” Terminology needs to be consistent throughout the whole document.
- Maintain one-to-one correspondence between the source and the target. This is where literal translation for patents may be different from literal translation for localization, for example. In practice, one-to-one translation means that a translator will be able to go to the courtroom and draw a diagram on a whiteboard, underline words in the source text, and draw an arrow to where it is translated in the target. This exercise shows that everything in the source text is shown in the target text, and that everything that is in the target source was in the source text. Imagine yourself in the courtroom and wanting to see if everything corresponds one-to-one. How does one do that?
- When the equivalent is very well established (usually but not always, when the equivalent is listed in the dictionary). When called upon in the court you can just pull out the dictionary. Additionally, if there is something clearly established by convention and you can show that.
- When using source lexemes would lead to undue confusion, or create a highly unnatural style.