Thousands of examined and unexamined patent applications, utility models and granted patents are translated every year from Japanese to English in United States, Japan, Great Britain and other countries. The quality of these translations is sometime excellent, and sometime not very good. It is not difficult to find samples of bad translations from Japanese. For example, the Japan Patent Office provides English summaries of all published unexamined (Kokai) patent applications in English (see my collection of links to various vital services provided by JPO at: http://www.ptranslation.com/find-japanese-patents.htm). These summaries are very useful as they provide important information to people in foreign countries who do not read Japanese. However, although most of the time, these English summaries are more or less understandable, they are clearly written by native Japanese speakers who are probably specialists in their technical fields, but definitely not professional translators into English, as their fluency in English leaves much to be desired. In fact, the English of these translations is often quite hilarious. In contrast, English summaries of patents that were originally written for instance in German or French, available for instance from the World Intellectual Property Organization (WIPO) at http://www.wipo.int/ipdl/en/ or from the European Patent Office at

http://ep.espacenet.com/.../home.hts are almost always written in clear and grammatically correct English, including the proper technical terms. Clearly, these summaries were written by professional technical translators whose native language was probably English, or whose fluency in English is almost indistinguishable from that of an educated native speaker of English. Sometimes I do find mistranslations in these English summaries, which is probably due to insufficient knowledge on the part of the native English speaker of the original language, but it is very rare, and I am always surprised when this happens. English summaries of Japanese patents on the Japanese Patent Office website, on the other hand, do not really contain mistranslations based on misunderstanding of the original language (because they are written by native Japanese speakers) and the technical terms are usually correct, but the English is sometime so bad that I have to read the Japanese text at least twice before I can figure out what the English text means.

The Japanese Writing System—The First Line of Defense in Japanese against Foreign Speakers or Translators

One has to learn at least two thousand kanji characters to be able to read simple texts, but many more, about twice as many, to be able to translate technical and medical texts.

There are very good reasons why it is more difficult to find qualified translators who can translate well patents from a language such as Japanese or Chinese than, for instance, from German or French. Japanese is a much more difficult language to learn than German or French. I am not saying that it is easy to learn German, French, Russian, etc.—but I am saying that it is much more difficult for a native speaker of a European language to learn a language such as Japanese or Chinese. One level of difficulty in Japanese (what I would call the first line of defense against foreign speakers or translators) is presented by the actual writing system. In order to learn any language well, a student of the language has to be able to read it and write it. In Japanese, this means that this student will have to learn two alphabets, katakana, which is used mostly for transcription of foreign words, hiragana, which is used for Japanese words or to indicate grammatical aspects such as tense or prepositions (which are really postpositions in Japanese), and several thousand Japanese characters called kanji, (which means "Chinese characters", as most of these characters were originally adopted from Chinese). This means that one has to learn at least two thousand kanji characters to be able to read simple texts, but many more, about twice as many, to be able to translate technical and medical texts. Unless you are Japanese and you live in Japan where you start learning kanji characters in kindergarten and then spend the rest of your life surrounded by quite complicated kanji characters everywhere you go, you will need to spend a lot of time, probably several years, memorizing characters. You don't have to do that to learn French or Russian. On the other hand, it is also true that there is a finite number of characters and character combinations and anyone can learn them, it is just likely to take a long time.

Another problem with translation of Japanese patents is poor legibility of characters in older copies of Japanese patents. An Unexamined (Kokai) Japanese Patent Application has four small pages which fit a large page (letter size in US or A4 size in Japan) only because the characters are quite small. When the size of the characters is reduced, some combination of strokes inside a complicated character, which may easily consist of more than 20 strokes, some of them very small, will appear as dark rectangles referred to by professional technical translators as "those freaking blobs". The problem is, when the meaningful portion inside a character is illegible, the whole character is illegible and the translator has to make an educated guess. This is unlikely to occur with a patent application in French or German, unless it is many decades old, and even then, only one or two letters in a word would be illegible, not most of the meaningful portion of the word. Translators of Japanese patents, on the other hand, have to systematically eliminate various potential characters when dozens of similar characters may match a given blob, depending on the context. The big problem therefore, is that when the translator does not really see the actual character, he or she is guessing, and thus does not know for certain which character it is. If I were to include a translator's note every time when I see an illegible character, I would not have any clients left. Therefore, I usually just take my best guess without letting the client know. After 20 years of translating Japanese patents on a daily basis, my guess is a pretty safe one, I would hope, but it is still a guess. Thanks to improved technology, more recent patent applications are published in a large font which is clearly legible. However, many patents that need to be translated as evidence of prior art are quite old, twenty, thirty, fifty or more years old (the oldest Japanese patent that I translated last year was from 1937, but thankfully, the text was very legible).

But even getting past the hurdle represented by a very complicated writing system, so complicated that the system went through several reforms since 1868, which were designed to simplify it, and did simplify it to some extent, but also made it even more complicated because a new layer of "simpler" kanji characters that must be memorized again in addition to old characters by the translator were created in this manner, is only a first step toward fluency in Japanese for a foreign speaker.

The Second Line of Defense—The Incredibly Infuriating "Nihongo no Bunpo" (Japanese Grammar)

The next hurdle, a formidable one, is the Japanese grammar. To a native speaker of English or another European language who is just starting to learn Japanese, Japanese grammar looks like a jungle full of unknown and dangerous animals. The parts of speech in English, German, French and many other European languages are virtually identical, or at least very similar—after all, grammatical rules and conventions in these languages were often developed and categorized for centuries to mirror Latin grammar. In English or French, it is fairly easy to determine what is the subject, a noun, or a verb, or where a word begins and ends. But Japanese grammar obviously does not reflect Latin grammar at all. It is up to the reader to determine where a word begins or ends in the text—there are no spaces between words in written text, and in fact, it is not clear what exactly represents one word in Japanese, or whether one word is in fact a legitimate structural unit in Japanese. There is usually no distinction between singular or plural. At least most of the time, there is none: you can put plural in, if you are so inclined, but most of the time, every noun would have to be translated into English as "part(s), hook(s), wheel(s), widget(s), knife/knives, etc., if you insist on a correct translation. A patent translator can do this perhaps once or twice in a patent claim to be as precise as possible, but in the rest of the patent, he or she will have to take an educated, but sometime risky, guess, while constantly looking at the figures, if there are figures, to see if these figures indicate one ore more elements. That is why, when a patent lawyer asks me whether the Japanese original uses singular or plural, my honest answer is usually "I don't know". Japanese is also not very consistent when it comes to distinguishing between tenses, for instance there is no real future tense in Japanese, although a distinction is made between present and past. And although there is no "real future" in Japanese, some adjectives (so called "keiyo doshi" adjectives) can be in the present or past tense. The preferred tense in patents, however, is the undetermined -ru form, which is really no tense at all, a form that is similar to the infinitive in European languages. This can also cause problems in Japanese patents, although not as much as the lack of distinction between singular and plural. Japanese also often uses past tense where English would normally use present tense and vice versa (but this problem also exist in German and other languages).

A much more important problem is the fact that a Japanese sentence does not necessarily need to have a subject or an object, although just like German, it will invariably end with a verb. The cavalier attitude of Japanese language to subject is due to the fact that context is much more important in Japanese than for instance in English—so important, in fact, that context plays a fixed grammatical role. The subject can be represented by several so called particles—wa, ga, wo (wo is mostly used for object in Japanese, but sometime needs to be translated as subject into English), indicating mostly the position of a noun in Japanese and its relationship to other parts of speech in the sentence. But the subject, represented by one of these particles, can be located several sentences before the present one, and by the time the translator has reached the present complicated sentence, full of somewhat weird and possibly subversive technical terms, he or she may have forgotten what the subject actually was, especially if it was on the previous page. However, if you can trace back correctly the connections in the sentence, you can always find the proper subject in the context of several sentences, as long as you understand these sentences (and as long as they make some sense in Japanese, which is not always the case).

Another annoying characteristic of Japanese grammar from the viewpoint of a native speaker of a European language, especially in patents and technical and medical articles, is the tendency of the language to string together several kanji characters to create new technical terms in Japanese, similar to the way in which German compound nouns can be strung together to create long German words. But because every character is sort of like a picture, it has a meaning, several meanings, in fact, and a good translator will be able to find the corresponding technical terms in English, if there is one, from dictionaries, or from the Internet, or supply his own term if these characters do not seem to represent a legitimate term. However, technology has made the job of technical translators much easier in the last decade or so, because the Japanese Patent Office website can be searched in Japanese to establish whether a certain string of characters is in fact a technical term and whether there are English summaries in which this Japanese term has been translated into English. This is particularly helpful with complicated medical terms, such as names of bones, muscles and tendons, or names of species of fish or algae, etc. (there are many Japanese patents in this field for obvious reason).

What Does Boiled Fish Paste Geometrical Shoe Heel Design Looks Like? Is It By Any Chance Similar to a Guêpièrre Bra?

Transcription of foreign words into katakana, on the other hand, can be much more difficult to figure out (as mentioned above, katakana is one of two Japanese alphabets which is mostly used for transcription of foreign words indicating the proper pronunciation of these words in Japanese). Technical translators are often asked to translate documents on diverse and sometime quite arcane subjects. Last year for example, I became something of an expert on shoe heel design, as I was translating a number of patents and utility models on this subject from Japanese, German and French. The oldest patent was in French, it was from 1920s, I think. Then there were quite a few German ones from the Third Reich era, with the German eagle holding the swastika in his crooked talons on the front page, and a few in Japanese from the 1960s. One of them was in handwritten Japanese. The inventor, who had a rather nasty handwriting, was clearly fond of seafood because the geometrical shapes and patterns of shoe heels were described in this handwritten patent document with frequent references to sushi and other traditional Japanese cuisine items. I suppose every Japanese reader can imagine what a "kamaboko" pattern will probably look like, but what image will the term "boiled fish paste pattern" evoke in an American reader? (Kamaboko means boiled fish paste—I eat it a few times a month, but my guess is that most of my customers probably have no idea what it is). A more recent and equally fascinating subject, which I had the pleasure of translating for the past couple of years, involves intricate lingerie design, in particular some daunting improvements of the bra design. This is in fact not an easy subject because it involves complicated and detailed concepts relating to mechanical engineering, textile materials, some chemistry and a lot of special terms used in the fashion industry by lingerie designers. Plus, not being a woman myself, I never really had much first hand knowledge with the subject at hand, except for some rudimentary introduction when I was a teenager, which was long time ago (and the bras' design has changed quite a bit since then). But, as always, the most difficult technical terms relating to any technical field, including lingerie design, from the viewpoint of a Japanese patent translator, are foreign terms transcribed into Japanese katakana. It is easy enough to decipher "outer bra", "corset", or "bodice" from transcription into katakana, ("outaah bura", "kohsetto", "bodisu"), but it is really time consuming to figure out "guêpièrre" from katakana, especially if you had no idea that such a word existed in the first place, or on what language was the transcription into katakana based. Another problem with katakana words is that these words are often misspelled in Japanese because Japanese speakers often don't know the correct spelling of the original term in the foreign language. For instance instead of "gepiyehru", which is how the word should be transcribed into Japanese, the word could end up as "kepiyehru", in which case it would be virtually impossible to track down the term. In fact, at first I thought that the original word might be an erroneous katakana transcription of "Gefühl" ("feeling" in German), and wasted some precious time researching "Gefühl lingerie" on the Internet—and since there is a lot of Gefühl lingerie, I was led on a wild goose chase for quite a while before I figured out that the original word must be in fact French.

Technology Does Make Life Easier These Days for Japanese Translators

After more than 30 years of trying to learn the Japanese language, and more than 20 years of translating Japanese patents into English almost on a daily basis, I must say that there are fewer "wild goose chases" such as the one mentioned above, mainly thanks to Internet resources that are available instantaneously to technical translators, including the capability to Google terms in Japanese and English and the capability to search the Japanese Patent Office website in Japanese and in English. An Internet search for a technical term may take a bit longer than opening a dictionary, but every dictionary is by definition limited and obsolete by the time it is available for sale, let alone many years later. On the other hand, the capability to store technical terms on the Internet is without limitations, and even very recent terms can be usually found online quite easily (although older terms are often much more difficult to find on the Internet).

A few years ago I visited the Japanese Language Department at the Charles University in Prague, my old alma mater where I studied Japanese in the seventies. Technology has changed the way the language is taught there as well. Instead of a big blackboard by the front wall on which our professor was writing Japanese and Chinese characters (he sure did look like a bearded sage from an old Chinese illustration of a Zen textbook), there were at least a dozen computer stations in the class, equipped with specialized character teaching software and connected to the Internet. In fact, one of the Japanese teachers told me that he was also at the moment translating a Japanese biotechnology patent into Czech, and he checked out my website while we were talking to see if there is any useful information there.

Thanks to modern technology, in particular Internet, we can perhaps hope that in spite of the formidable obstacles to a good translation from one language into another, especially between languages that are as different from each other as Japanese and English, the changes will be relatively minor, as the translator no longer works in isolation from the "real world" using mostly only the reference which is stored in his or her brain and in the dictionaries.