Dr. Aline Azevedo Larroyed, one of my contacts in Linkedin was so kind as to send me one copy of her thesis on patent law whose title is “Translation accuracy and dissemination of disclosure of patent information: An Analysis of Translation and its Influence on Patent Law” and can be bought through “Amazon”.
The thesis was successfully defended at Maastricht University on the 25th of September 2019 and was the result of Dr. Aline Larroyed’s four years of scientific investigation, three-year PhD at Maastricht University and one year as a visiting researcher at the Max Planck Institute for Innovation and Competition.
It is difficult to summarize a 419 page book and with such a high density of information in a blog entry but I will try to draw your attention to the most relevant aspects I found.
The book-thesis begins by answering the main research question: “How does accuracy of translation influence the dissemination of the disclosure of patent information on patent law?”
- “The advent of quality machine translation and the changes on its interaction with human translation characterizes a true revolution to patent rights and represents the basis of the patent system in its current configuration”.
- “Without the improvement of machine translation, the patent system would not have been reshaped into a global structure”.
- “Machine translation has universalized the reach of patent information disclosure”.
Aline also lets us know the main challenges she faced during the period of research:
- Academic resistance towards the topic, mainly from IP academics.
- The issue has a high level of novelty. It was necessary to create a methodology to approach it.
The thesis is made up of 8 chapters:
It deals with the economic justification of the patent system. Disclosure is found to be the main linking between patent systems and innovation systems. When a patent application is filed abroad, the requirement of “sufficiency of disclosure” will only be met if a proper translation of the patent application is carried out.
This chapter deals with the methodology used. It begins by posing 3 subquestions to the main question:
- What is the current configuration of the patent system and what does translation represent to it?
- Does translation disclose patent information?
- How does translation influence patent law on each different legal stage of a patent?
The technology from where granted patents were chosen was that of “green patents”; “wind power” and “solar energy. 100 patents were selected among the applications filed between 2013 and 2015. The set of languages chosen for the analysis was: English, French, German, Italian, Spanish, Chinese and Portuguese. The latter was the main reference as a target language in the process of analysis.
The methodology chosen to assess the quality of the machine translation provided by “patent translate” was LISA (Localization Industry Standards Association) Quality Assessment (LISA-QA) default model, a system created for managing the quality translations certificated by ISO 9001.
This chapter analyses the patent system as a communication system. According to the author, translation can be seen both as a linking element, which defines the identity of the patent system, and a disruptive one which can compromise its identity.
This chapter portrays translation as a vehicle to patent rights.
The author states that there is an undeniable hegemony of the English language when it comes to patent translation and that even though the help of machines has increased immensely the speed of production, it cannot be said that machine translation attains publishable quality without human review.
This chapter makes reference to the regulations on the “European Patent with unitary effects”, and to the fact that the linguistic regime relies on the improvement of Machine Translation, since the Council Regulation 1260/2012 states that “the maximum period for the development of high-quality machine translations cannot be considered to exceed 12 years”. According to that regulation we are supposed to enjoy a high-quality machine translation in 2024. The author regards this as feasible, especially after the replies by some of the patent examiners interviewed saying that the last two years had made a lot of difference to translations from Asian languages, like Japanese, Chinese and Korean, with an emphasis on Japanese.
The author underlines as well that unlike what most people think, translating patents is not easy at all.
Patent translation and its influence on the disclosure of technological information
This chapter analyses the level of disclosure of patent information through machine translation. According to Jeanne Fromer “patent disclosure is so central to the patent system that the way it performs or underperforms can directly affect technological progress and its main premise, which is fostering innovation”.
The patent document is poorly structured and does not contain some of the most pertinent technical information. Those problems allow if not encourage, the writer to under divulge. On the other hand the translated texts will probably repeat or even worsen the deficiencies of the original text.
This chapter presents the results of the analysis using the LISA-QA method. A sample of 100 patents from solar and wind technology are analyzed from the point of view of Machine Translation from English, German, Italian, French and Spanish into Portuguese. Another study takes into account translations from Chinese into English.
When it comes to translations from European languages into Portuguese, French presents the least mistakes whereas it is German the language with the highest number of mistakes. There is a higher number of mistakes in translation from Chinese into English. The author concludes that the correct translation of technical terms (terminology) from Chinese into English still has a long way to go for a better result. Machine translation into English can be sufficient to disclose the contents of Chinese patent but not very accurately.
Patent translation and its influence on cross-border agreements.
Translation plays a key role in patent-related international agreements. The PCT is a good example of a wide range of language and translation features. The role of Lingua Franca is played by English. The European Patent Convention is cited as another example of a patent-related international treaty where the language regime and thus translation are of paramount importance. The European Patent Convention also gave rise to the “European Patent with Unitary Effect” project which, as explained previously, relies on Machine Translation to simplify the linguistic regime.
This chapter devotes a point to Spain’s action against the “European Patent with Unitary Effect” project before the Court of Justice of the European Union (CJEU). Whereas the Advocate General stated that “there is no doubt that European researchers can understand patents published in those languages”, the author replies that that statement sounds vague and unfounded, as not all the European researchers can understand or make the best use of a patent document written in German, for example. In this regard, I missed a reference to the “English only” position that Spain defended as a last ditch effort to try to stay in the “Unitary Patent” project.
In this chapter, several examples of inaccuracies in the translation of international treaties to Portuguese are cited: Articles 23b and 27.3 of the TRIPS and Article 4F of the Paris Convention, for instance.
Chapter VII The influence of translation on the different legal stages of a patent
This chapter studies the effects of translation on the different stages of the timeline of a patent, which is not as obvious as it could seem at first sight.
The author highlights that during prior art search; patent examiners mostly rely on machine translation (80%), and very rarely demand a human translated document to confirm a decision (20%). Nevertheless, the most relevant document will be human translated at oppositions and infringement cases (50%-50%).
From the interviews we know that Brazilian examiners never use human support in translation and that Japanese and European Office patent examiners consider human support as necessary in special cases. There is an agreement that an accurate examination implies, overall, the full access and understanding of foreign prior art.
Another opinion that can be inferred from the interviews is that European and Asian Offices are considered as very accurate in finding prior art. The USPTO can be less accurate, what makes it easier to have a patent granted in the US system.
Several examples of translation problems found in examination proceedings, office actions and opposition cases are provided:
EP695351; A conflicting interpretation of the term animal in English, French and German
T1332/12; A decision of lack of inventive step was overturned due to an inaccurate machine translation of a prior art document.
This chapter also deals with the main aspects in patent court cases involving translation and three relevant questions emerge:
- Does translation affect the decisions in infringement cases involving patents?
- Does it mislead these decisions?
- If they do (or at least have the potential to) mislead these decisions, what could be done to minimize these negative effects?
From interviews it came up that patent attorneys think that:
- It is not rare that an inaccurate translation can be used with the intention to bias a case.
- The use of speech and words by a human translator will evidently make a difference in influencing the case’s decision.
- The difference is often difficult to be identified and described.
The conclusion in this regard was:
“The analysis evidenced that inaccurate translations can bias decisions related to infringement. This can be related to subtle inaccuracies in translated prior art, to infelicities concerning the original texts or to inaccurate translations purposely presented to change the scope of protection and influence judicial decisions”.
Translation costs can be considered as part of the highest ones during patent prosecution. Different jurisdictions may offer different problems related to translation and communication (European Patent attorneys have more problems with translation and communications in Latin America than in Asia due to bureaucratic demands).
Chapter VIII .
Summary , conclusions and recommendations
Office, opposition and infringement cases; the influence of translation on patent law and its snowball effect.
The conclusion is that patent translation exerts prevalent, all-encompassing, pervasive influence on patent law.
Translation discloses patent information and enables the universal dissemination of innovative invention. The conclusion was that the level of disclosure of technological information through machine translation is almost 80%.
Translation and its influence on international agreements as a network effect.
The influence of translation on cross-border agreements and their internalization as national or regional regulation tend to follow the logic of a network effect. It is a constant effect which pervades all the structures of the whole system.
Finally, the author makes a series of recommendations:
- To increase the level of integration and interaction between patent offices and patent courts:
- Agreements for establishing regional patents, like the European Patent with Unitary Effects.
- Agreements on reducing translational costs.
SOME OTHER RECOMMENDATIONS:
- Offices should reduce as much as possible the bureaucracy related to translations.
- Patent Offices and patent courts should be seen as part of a global system.
- The opposition procedure should admit the possibility of the applicant to request a human translation of prior art documents which were crucial as the basis for an examination decision.
- Patent judges should be specialized in patent law.
RECOMMENDATIONS CONCERNING TRANSLATION AND DISCLOSURE IN THE LANGUAGE REGULATION OF THE EUROPEAN PATENT WITH UNITARY EFFECTS AND THE UNITARY COURT.
- The system should encourage the use of English as an intermediate language.
- An SME of a member state should always have the right to bring the litigation to a chosen brand of the unitary court, when being sued by a bigger company.
RECOMMENDATIONS TO IMPROVE SEARCH TOOLS AND EXPERT SYSTEMS IN TERMS OF OPTIMIZING THE USE OF MACHINE TRANSLATION TO DISCLOSE AND DISSEMINATE PATENT INFORMATION.
- New platforms and tools can be developed aiming to translate the content of patents into other types of discourses, such as legal reviews (and not only associated legal cases), videos or texts with educational purposes involving groups of patents.
- Some older Asian (or other national/regional) patent documents are not available for translation in search engines. Patent Offices should work on their files and search engines in order to give access to these patents.
SOME RECOMMENDATIONS FOR FUTURE DEVELOPMENT AND STUDIES
- How do national patent language regimes influence the dissemination of the disclosure of patent information?
- How can national patent language regimes and regulations improve in order to contribute to a more harmonized and efficient patent system at the global level?
- What are the effects of regional or international agreements to disclose and disseminate patent information?
- Is English as lingua franca the solution to reach an ideal level of universal disclosure of patent information?
- How can patent writing affect patent translation?
- How can national legal cultures affect the dissemination of disclosed patent information?
- What is the relationship between language, dissemination of patent information and linguistic rights?
- How has translation influenced the dissemination of innovative inventions from a historic perspective?
- Can patent information be used for language learning?
- How does language and translation affect the integration of a patent national system to the patent system at the global level?
- How can language be used to bias patent cases?
- What are the professional requirements for a patent translator? What kind of education can create qualified professionals to better perform this function?
- What is legal discourse and what is scientific discourse in patent documents? How can they best interact in order to disclose patent content?
- What is the role of political, ideological and cultural discourse patterns on national patent systems?
- What kind of new search engines could be developed in terms of maximizing the disclosure of patent information for different social groups?
FALLACIES AND GENERALIZATIONS TO BE AVOIDED IN DEALING WITH THE SUBJECT OF PATENT TRANSLATION AND DISCLOSURE
According to the author, the first fallacy over the relation between machine translation and disclosure is the idea that machine translation is not to be taken seriously and will never replace the work of a human translator.
The second fallacy goes to the opposite direction of the first one and is related to the belief that the quality of machine translation has improved so much that, with the support of drawings, patents are easily and totally disclosed.
Another common fallacy is presuming that studies on language or translation related to the patent system should be restricted to language departments.
I am afraid that what was intended to be a brief summary of this interesting thesis by Aline Larroyed has turned into a lengthy text. I apologize but it was really difficult to reduce the minimum number of pages needed to convey its main ideas.
Undoubtedly, Machine Translation is bound to be a game changer in the future of the patent system. This thesis can be regarded as a trailblazer , and a huge amount of studies on translation and patents are waiting to be written.
Thank you very much indeed Aline for your thorough work and for being a pioneer in this field.
Leopoldo Belda Soriano