If you translate patents you’ll know that they generally come to you in pdf format. Only few customers do the below job and send you an editable text. Moreover, they often are not so clear to read. It derives that the best way to translate them is printing the document and translate it “manually”, without any CAT.
Copyright © ProZ.com, 1999-2020. All rights reserved.
Scientific patents have generally a lot of formulas, numbers and repetitions (I think for example about laboratory procedures that are quite similar, often it is sufficient changing figures) and it’s a shame having to translate them without any help.
Here you can find some tips to translate patents (but other similar documents as well) in an easier and faster way.
First of all, invest in a professional OCR (optical character recognition) software. You can find many free OCR on the web but they do not work as good as professional ones. Remember that the pagination of this kind of documents quite often is not so good. Professional OCR software also have language dedicated character recognition, to improve the final outcome.
Use the software to “read” and convert the pdf document in a editable format (generally .doc or similar formats). If the original text is good enough, you have a ready-to-translate document. You’ll surely have to edit some parameters (font, margins, number of columns, etc.), but the better the original document, lesser “post-editing” job is required. Anyway, it is almost always worth while spending some more time on this kind of editing, since the result is that you can use a CAT to translate it.
If the original document is of bad quality, so that the OCR output is unusable, you have another possibility, at least for European patents (I don’t know if it is the same for USA and Japan patents too).
You can register (for free) on the EPO (European Patent Office) website: www.epoline.org/, where you can find all filed patents with their history and the document. The text is again in pdf format but often it has a better quality. Moreover, the text you can find on EPO is the master reference in case of doubts or handwritten notes in the text you have. You can search by title, EP number, date, applicants and some other parameters.
Now you have the editable text, you can use your favourite CAT to translate it.
Apart from terminological consistency (if the patent is very long it’s easy to loose consistency), you can benefit from repetitions and less than 100% match. Below you can find some examples:
- Claims: they often are a repetition or nearly a repetition of the previous text.
- Laboratory procedures (for chemical/scientific patents): they often are quite similar if not the same. It is sufficient changing some figures or names
- Written formulas: if the OCR has done its job, you can just not translate them or do “copy source into target”.
- Figures on tables: the same as above. Remember to verify the exactitude of figures and formulas when doing the final reading/proofreading before delivering the translated text.
- Long lists of chemical names: you can control in a better way not to forgot any of them. By simply copying the pdf document it’s quite easy to lose some term or to repeat it twice.
Moreover there are some typical sentences that repeat through the text and the CAT can help you in the first translation and in future ones. If you translate patents similar between them or related to the same argument, building a “patent translation memory” can help you saving time for future translations too, as you can find some similar or identical sentences. And the terminology is always consistent.
After translation you have to clean up the text and format it according to customer requirements. These tasks can seem to retard the translation time, but they can be a valid help to speed up the translation phase.
To further speed up your work you can use a speech recognition software, even if I can say, by my experience, that for chemical and scientific names it does not work so well, or at least it needs a great deal of training.
Patents repeated sentences and syntactical formulas can be boring while translating a patent, but with some extra work they can be a great support for translator.
I apologise for any mistakes, as I’m not an English mother tongue.
Silvia Barra is a technical translator, with a huge experience in patents. Her main fields of activity are chemistry, materials, energy, industrial processes, environment and health & safety.