A patent attorney at a mid-size US firm called me a while back about a problem that shouldn’t have existed. They had a PCT application that was entering national phase in China, and the Chinese patent attorney on the other side had rejected the translation three times. The first rejection was for terminology inconsistency — the same component was referred to by two different Chinese terms in different parts of the specification. The second rejection was for a formatting issue: the claims weren’t numbered according to CNIPA conventions. The third rejection was for something more subtle — a background section that the Chinese attorney felt introduced matter not present in the original specification, which in Chinese patent prosecution can trigger an objection for adding new matter and, in the worst case, invalidate the claim in that jurisdiction.
The translation had been done by a provider that the US firm had used for years for general patent translation. Good provider, reasonable rates, decent turnaround. The problem was that the provider didn’t specialize in CNIPA national phase entry. They translated the PCT text into Chinese, but they translated it as text. They didn’t translate it as a CNIPA filing document. The distinction matters.
PCT national phase entry is, in the abstract, a simple concept. You file an international application under the Patent Cooperation Treaty, it gets published, and then you enter the national phase in each country where you want patent protection. As part of that national phase entry, you need to provide a translation of the application into the official language of the target patent office. The translation has to meet the requirements of that patent office. And those requirements are not the same across offices. That’s the part that causes problems.
The three offices I see the most issues with are the USPTO, the EPO, and the JPO, and the issues are different at each one. I’m going to walk through them, but the broader point is this: if your translation provider is treating every national phase entry the same way, you’re going to get rejected at one of them, and the rejection will be for something that a specialized provider would have caught automatically.
USPTO is probably the most straightforward of the three in terms of translation requirements, which is not the same as saying it’s simple. The USPTO requires that the translation of the specification and claims be accurate and that it conform to US patent application format. In practice, this means the claims have to be in the correct claim numbering format, the specification has to follow the US practice for sections (background, summary, detailed description), and any drawings must have reference numerals that are consistent between the drawings and the specification text. There are specific formatting requirements for the claims section — the preamble, the body, the dependent claim format — that differ from, for example, EPO practice. A translation into English for USPTO national phase entry that follows EPO claim formatting conventions will be formally correct but non-compliant with US practice, and the examiner may issue a notice requiring correction.
I’ve also seen USPTO reject translations for what they call “informalities” — not substantive issues with the patentability of the claims, but procedural issues with the filing itself. Incorrect page numbering. Missing the US-specific information disclosure statement format. The drawing sheets not labeled in the US convention. These are small things individually. But each one generates office action correspondence, which extends the prosecution timeline and adds to the applicant’s legal costs. If you’re entering national phase in ten countries and five of them issue informalities on the translation, you’re paying your patent attorney to respond to five sets of correspondence that were avoidable.
The EPO is more complex in a different way. The EPO operates in three official languages — English, French, and German — and the language of the national phase entry depends on which EPO contracting state you’re entering. If you’re entering national phase in Germany, the translation has to be in German. If you’re entering in France, it has to be in French. And the EPO has specific requirements about how the translation of claims is presented relative to the original claims, particularly for claims that were amended during international phase examination. The translation has to reflect the amended claims, and the format has to clearly show which claims are amended and what the amendments are. If the translation doesn’t do this, the EPO will issue a communication requiring correction, and the timeline for national phase entry is tight enough that corrections are expensive.
The JPO has a particular issue that trips up non-specialized translation providers. Japanese patent practice has very specific conventions for claim structure and terminology that don’t map cleanly to English or European patent conventions. The Japanese term for “comprising”, for example, is broader than the English “comprising” as interpreted by US courts. A direct translation of a claim using “comprising” from English to Japanese can unintentionally narrow the claim scope because the Japanese equivalent carries different jurisprudential weight. Conversely, a translation from Japanese to English that uses “comprising” without understanding the Japanese source term’s broader scope can unintentionally broaden the claim in ways that create prior art vulnerability.
JPO also has strict requirements about the relationship between the translation and the original text. The Japanese Patent Act, specifically Article 36, requires that the description, claims, and drawings of a Japanese application be in Japanese and that the scope of the claims be supported by the description. When translating into Japanese for national phase entry, if the translation adds any content that isn’t supported by the original PCT application, it can be rejected for adding new matter. But “adding new matter” in Japanese practice is interpreted more broadly than in some other jurisdictions — even restating a concept in slightly different Japanese terms can trigger this objection if the Japanese examiner believes the restatement has subtly changed the technical scope. This is why JPO national phase translation requires not just a bilingual translator but a translator who understands Japanese patent claim construction.
The timeline pressure is the thing that makes all of this more acute than it needs to be. PCT national phase entry deadlines are fixed. You have either 30 or 31 months from the earliest priority date to enter the national phase in each designated country, depending on the country. There is no extension for most jurisdictions. If your translation isn’t ready, you either file incomplete and deal with the consequences, or you abandon the national phase entry and lose the patent in that country. For a PCT application that was filed with a significant number of designated countries — and large tech companies routinely designate 15 to 30 countries — the translation workload is substantial, and the deadline is the same for all of them.
This is where the translation provider’s project management becomes as important as the translation quality itself. The provider needs to be able to coordinate translations across multiple language pairs simultaneously, ensure that each translation meets the specific format requirements of the target patent office, and deliver everything within a window that leaves enough time for the patent attorney to review and file before the national phase deadline. The large patent translation providers have systems for this. Smaller providers and general translation providers often don’t, and the result is a cascading set of delays that puts the attorney in the position of either filing with translations they haven’t had time to review or missing deadlines.
I’ve been involved in cases where the translation provider quoted a two-week turnaround on a ten-country national phase entry, but the provider was a general translation service that didn’t understand that each country’s translation needed different formatting. They delivered all ten translations in two weeks, but six of them had issues that required correction. The corrections took another three weeks. By the time the corrected translations were ready, the attorney had less than a week to review all of them before the national phase deadline. They filed, but they filed without the level of review they would have wanted. Two of the ten countries issued office actions on the translations within the first three months of prosecution.
The cost structure of PCT patent translation is something that in-house IP counsel at large companies think about a lot, because the volumes are enormous. A typical PCT application has 15,000 to 40,000 words of specification text plus claims. At national phase rates for patent translation — which run from $0.20 to $0.50 per source word depending on the language pair and the provider’s specialization — a single-country translation costs $3,000 to $20,000. Multiply that by 20 designated countries and you’re looking at $60,000 to $400,000 per PCT application in translation alone. For a large tech company filing 50 to 200 PCT applications per year, annual translation costs for national phase entry can run into the tens of millions of dollars.
At those volumes, the per-word rate matters less than the rejection rate and the correction cycle time. A provider that charges $0.05 less per word but generates a 30% rejection rate on national phase filings is more expensive than a provider that charges $0.05 more per word and generates a 2% rejection rate. The rejection rate drives attorney time, extension fees, deadline risk, and in the worst case, abandoned national phase entries. Those costs dwarf the per-word savings.
There’s also a less obvious cost that I don’t see discussed much, which is the cost of terminology inconsistency across the national phase entries. When a PCT application enters national phase in 20 countries, each translation is done independently — at least, that’s the default with most translation providers. The English-to-Japanese translation is done by one team, the English-to-German by another, the English-to-Chinese by a third, and so on. Each team makes independent terminology choices. Two years later, when the patent is being litigated or licensed in multiple jurisdictions, the claim terms in the Japanese grant might not map cleanly to the claim terms in the German grant or the US grant. If the claim terms are materially different, the patent’s enforceability across jurisdictions becomes uncertain. The patent attorney has to go back and reconcile the translations, which is time-consuming and expensive and might not be possible if the grants have already issued.
The solution to this is a glossary — a controlled terminology list that the translation provider builds before starting work and that all translation teams use for the national phase entries. This glossary needs to cover not just the technical terms but the legal terms: “comprising,” “consisting of,” “selected from,” and all the other transition phrases that carry claim-scope implications in different patent systems. If the glossary is done well, every national phase translation uses the same terminology for the same concepts, and the resulting patents have claim terms that map consistently across jurisdictions. If the glossary isn’t done, or if the provider doesn’t use one, the terminology drifts and the patent’s cross-jurisdictional coherence degrades over time.
The other thing that large patent filers are starting to do, and this is relatively new, is using translation memories and terminology databases across their entire patent portfolio. Instead of treating each PCT application as a standalone translation project, they maintain a translation memory that captures all the terminology choices, claim constructions, and formatting decisions from previous filings. When a new PCT application enters national phase, the translation provider uses the translation memory as a reference, which speeds up the work, reduces inconsistency, and gives the filer a searchable record of how every term in their patent portfolio has been translated into every target language. This is valuable for portfolio management, for litigation preparation, and for the kind of cross-jurisdictional claim term reconciliation that I mentioned earlier.
Not every translation provider offers this level of portfolio-level consistency. It requires infrastructure: translation memory management software, glossary maintenance processes, and quality assurance workflows that check new translations against the portfolio’s existing terminology. But for filers with large portfolios, the investment pays for itself in reduced rejection rates and improved cross-jurisdictional enforceability.
Artlangs Translation provides PCT patent application translation for national phase entry across 230+ language pairs: jurisdiction-specific formatting for USPTO, EPO, and JPO requirements, portfolio-level translation memory and controlled terminology management, coordinated multi-country delivery within national phase deadlines, and legal review by patent attorneys in each target jurisdiction. Because a translation that’s accurate in one patent system isn’t necessarily compliant in another.
