English-to-Japanese Translation Programs: Can They Do the Job?

We review four Windows-based E/J translation packages

In our April issue, we looked at the output of four of the top Windows-based Japanese-to-English translation programs and considered some of the problems involved in automating the process of translation. In this follow-up article, we cross the fence to examine the state of English-to-Japanese translation software.

by Steven Myers

After our article on Japanese-to-English translation programs appeared in April ("Can Computers Translate?," page 35), several readers wrote to express their surprise at the "low quality" of the English translations rendered by the four commercial programs we reviewed. There seems to have been a strong general impression that the technology behind machine translation (MT) had reached a far more advanced level than that exhibited by the products we reviewed.

Some readers pointed out that all of the developers of the programs reviewed were Japanese, and suggested that perhaps they would fare much better in a comparison of English-to-Japanese translation packages. In this follow-up article, we provide just such a comparison - but the results are not much better.

Human intelligence vs. the artificial variety

Before we analyze the output produced by each of the products, though, let's go over once again the main difficulties faced by MT software in general. As described in the April article, accurate translation between Japanese and English requires far more than just a thorough knowledge of both written languages. The skilled "literary" translator (of books, articles, letters, etc.) must also possess a solid understanding of cultural issues, have a strong sense of flow and structure in the target language, and be able to think in terms of ideas and concepts (as opposed to just mapping sentences from one language to another).

Technical and commercial translators, meanwhile, generally need expert knowledge in one or more specialty areas. They must be highly flexible writers as well, since they often will have to tailor their translations to the needs of individual clients. A law firm, for example, may insist on a precise, sentence-by-sentence translation of a Japanese document at the expense of natural-sounding English, whereas a software company would be more likely to demand that manuals and press releases be rewritten to completely eliminate any awkwardness in the target language.

It is these creative aspects of the translation process, with their requirements for flexibility and aesthetic sense, that present the greatest obstacles for MT programs. The fact is that, even though there has been a tremendous amount of hype surroun ding the progress of machine translation software in the past few years, professional translators don't feel the least bit threatened by today's MT systems.

So what, exactly, is MT in its present form good for? In the hands of an experienced translator, the current crop of machine translation systems and online dictionaries can serve as powerful tools for speeding up the translation process. Indeed, Tosh iba calls its ASTRANSAC product a "translation accelerator," which is a far more accurate description for all of the products reviewed here than "machine translation system."

For organizations that must choose relevant documents to translate from a large volume of text, a machine translation system can be extremely helpful for quickly determining the overall subject matter of the documents. Special dictionaries can be cre ated for each specific type of job and used in conjunction with technical dictionaries to greatly reduce the time spent looking up arcane terms. Most of the reviewed products also have an "interactive mode" that allows translators to make line-by-line mod ifications of the output.

Products such as those reviewed in this article are now commonplace in multinational offices throughout Japan, and all are readily available in almost any Japanese software shop. Be warned (again), however, that for most documents, it is unrealistic t o expect an even moderately accurate translation from any current commercial MT system.

The promotional literature that comes from some of these companies is quite misleading and seems aimed specifically at busy professionals who are not bilingual - those not about to sit down and check through the translated text themselves, and therefo re susceptible to outrageous marketing claims. ("Translate a nuclear physics article from a Japanese academic journal!" proclaims one.) Before getting taken in by effusive ads, however, you might want to have a bilingual colleague take a look at some samp le output.

Comparing the programs

For this review, we started with three sentences that we felt were fairly straightforward and within the capabilities of current MT technology. (See the "Sample translations of simple sentences" sidebar.) The sentences are all short and, with the exce ption of sentence #3, do not use ambiguous vocabulary.

There was not much significant variation among the four programs for these simple sentences, and while the Japanese is by no means polished (there are grammar mistakes, improper use of politeness levels, and such), the Japanese reader can, for the mos t part, understand the gist of each sentence. With sentence #3, though, none of the programs was able to recognize that the word "take" should be translated in this case as æ‚é rather than Žæ‚é. ASTRANSAC also had a hard time with the expression "get of f at" (`‚ō~‚è‚é) and pretty much made a mess of the whole sentence. In general, though, it is safe to say that at this level of sentence complexity, all of the products in this review can render roughly accurate translations.

As sentences become slightly longer and more complex, however, the general quality of the translated output changes dramatically, and idiosyncratic differences among the programs begin to emerge. Let us stress, though, that most of the product manuals for translation packages are pretty clear on the point that sentences (like the ones in this test) require a fair amount of pre-editing and should not be translated in their "raw" form. (See the April article for more on this requirement.) But just about any document will contain many sentences of even greater ambiguity and complexity, so it is instructive to take a look at how such sentences are handled by the programs without pre-editing simplification.

(1) The first sentence (see the "Sample translations of more difficult sentences" sidebar) was taken from a news report and contains examples of words with multiple meanings. ASTRANSAC did the best job with the term "First Lady" (•Ä''å"-̕vl); PC-T ranser made the mistake of performing a literal translation (Å‰‚̏-‚̐l-- "the first woman"). All of the programs choked on "Whitewater affair." jLondon and ASTRANSAC used the term Ž-•¿ (matter, circumstance) instead of Ž-Œ (incident, affair). LogoVist a failed to recognize Whitewater as a proper noun, rendering the phrase as ‹}-¬‚ÌŽ-Œ (the rapid stream incident). Also, ASTRANSAC and LogoVista both had trouble with "recall," using ŒÄ‚Ñ-߂· (which has the nuance of calling back to a person) and "P‰ñ‚·‚ é (withdraw, revoke).

(2) The second sentence, while certainly short enough, contains an irregular use of the phrase "make it to ~" that confused all of the products. LogoVista came closest to getting it right with ~‚·‚邱‚Æ‚ª‚Å‚«‚é (be able to do ~). LogoVista was also t he only one to recognize that "the 18th" refers to a date and not a numerical identifier or a number of times. This type of sentence is a prime example of casual language with minimum contextual information that most humans have no trouble understanding, but machines tend to stumble over.

(3) Sentence number 3 was taken from the Computing Japan newsbriefs. Here, some of the systems had trouble with this particular meaning of "explosive," using more literal terms like "š"­« and "š"­•¨ rather than "š"­"I. None of the products ma naged to correctly translate the term "hiring practices": LogoVista came up with the wrong type of practice (-ûK, as in practicing the piano), while ASTRANSAC and PC-Transer both opted for ŽÀs (execution, performance, action). Interestingly, PC-Transer rendered "recruiting" as V•º‚ð•åW‚·‚é, as in the recruiting of soldiers.

(4) With the fourth sentence, we were trying to see if any of the programs could recognize a common figurative expression ("forever kick myself"). Not surprisingly, all of the products performed a literal translation of this expression as some variati on of ‰i‹v‚ÉŽ„Ž(c)g‚ðR‚é, which does not convey the intended meaning at all. All of the programs also had trouble with "issue" as used in this sentence. ASTRANSAC converted this to -â'è (problem), while LogoVista used -Á•¿ (as in a stock issue).

(5) On sentence 5, all four of the programs used a different term for "sleep disorder." jLondon had the most amusing translation with ‡-°•s-@sˆ× (illegal sleeping behavior). ASTRANSAC came up with ‡-°•s"ˆê (disunified sleep), while PC-Transer cal led it ‡-°-ŽG (messy sleep). Also, despite the presence of the word "library," none of the systems translated the term "check out" correctly, opting for ƒ`ƒFƒbƒN‚·‚é and '²‚ׂé rather than ŽØ‚è‚é.

Isn't there anything better?

After seeing results such as those described in this article, a common question we hear from many bilingual persons is, "Do people really pay money for this software?" Yes, they certainly do, anywhere from ¥50,000 to ¥200,000 per package.

Again, the trick is in knowing how to properly use these products and avoiding unrealistic expectations. By developing their own dictionaries for each different type of job, translators can manage to avoid many of the ambiguity problems described. In addition, these packages are highly useful for many types of technical translations, especially those that involve maps, diagrams, lists, etc.

Do these products really represent the state-of-the-art in machine translation software? In terms of commercial, shrink-wrapped software, these are among the best that we've come across so far.

We'll continue to look for other machine translation software packages that rival or surpass the ones we've seen so far, so keep watching these pages for more reviews and comparisons. And if you can recommend a particular program that you think we sh ould look at, send us your own evaluation along with contact information.

Prior to joining the Computing Japan staff, Senior Editor Steven Myers worked as a professional Japanese-to-English translator.

The human translations used for comparison in this article were provided by Massachusetts-based Japanese Language Services.


Sample translations of simple sentences

English originals:

1. The man took off his jacket and sat in the chair.

2. How did it go last night?

3. Take the Hibiya line and get off at Ginza.

jLondon EJ:

1. l‚Í"ނ̃WƒƒƒPƒbƒg‚ð'E‚¬AˆÖŽq‚ɍÀ‚Á‚Ä‚¢‚1/2B

2. ‚»‚ê‚͍ð"ӂǂ̂悤‚ɍs‚Á‚1/2‚(c)H

3. Ginza‚É‚¨‚¢‚ÄHibiyaƒ‰ƒCƒ"‚ðŽæ‚èA~‚è‚È‚³‚¢B

ASTRANSAC:

1. ‚»‚̐l‚Í"ނ̃WƒƒƒPƒbƒg‚ðŽæ‚菜‚«AˆÖŽq‚ɍÀ‚Á‚1/2B

2. ‚»‚ê‚͍ð-é‚ǂ̂悤‚ɍs‚Á‚1/2‚(c)B

3. "ú"ä'Jƒ‰ƒCƒ"‚ð‚Æ‚è‚È‚³‚¢B‚»‚µ‚āA‹âÀ‚É-£‚ê‚ċ߂«‚È‚³‚¢B

LogoVista EJ:

1. 'j‚Í"ނ̃WƒƒƒPƒbƒg‚ð'E‚¢‚ŁA‚»‚µ‚Ĉ֎q‚ōÀ‚Á‚1/2B

2. ‚ǂ̂悤‚É‚»‚ê‚͍ð-és‚Á‚1/2‚(c)H

3. Hibiya ü‚ð‚Æ‚Á‚āA‚»‚µ‚Ä‹âÀ (Ginza) ‚É‚¨‚¢‚č~‚è‚È‚³‚¢B

PC Transer/ej:

1. ‚»‚Ì'j«‚́A"ނ̃WƒƒƒPƒbƒg‚ðŽæ‚è‹Ž‚Á‚āA‚»‚̈֎q‚ɍÀ‚Á‚1/2B

2. ‚ǂ̂悤‚ɁA‚»‚ê‚͍ð"ӁAs‚Á‚1/2‚(c)H

3. Hibiyaü‚ð‚Æ‚Á‚āAGinza‚ō~‚è‚È‚³‚¢B


Sample letter translation

Several ads and promotional brochures for MT software products depict an American businessman writing a letter in English, running it through the software, then dashing the translated letter off to his counterpart in Japan. Just how realistic is this scenario?

Here are four machine translations of a short English letter. In addition to general awkwardness, many of the same problems shown in the previous examples (such as finding the right meaning for a word based on the context in which it appears) can be s een here as well. Instead of Žx•¥‚¢ for payment, we find Šñ•t (donation) and •ñV (reward, compensation). Other notable errors include ˆÍ‚Þ (surround) instead of "¯••‚·‚é (enclose), and ÜŽ^ (praise, compliment) for "complimentary."

Original:

Dear Mr. Iijima,

It was a pleasure meeting you at last week's conference. I greatly enjoyed having dinner with you and learning about the new magazine you are publishing. In response to your request, I have enclosed a complimentary copy of our latest product for revie w. I have also included payment for a one-year subscription to your magazine.

Thank you very much for your time, and good luck on all of your endeavors!

jLondon EJ:

eˆ¤‚È‚éIijima‚³‚ñC

æT‚̉ï‹c‚É‚¨‚¢‚Ä‚ ‚È‚1/2‚Ɖ‚±‚Æ‚ªŠy‚µ‚Ý‚Å‚ ‚Á‚1/2BŽ„‚́A‚ ‚È‚1/2‚ƃfƒBƒi[‚ð‚Æ‚é‚Ì‚ð'傢Šy‚µ‚݁AV‚µ‚¢ŽGŽ‚ɂ‚¢‚Ä‚ ‚È‚1/2‚ðŠw‚Ô‚±‚Ƃ͏o"łµ‚Ä‚¢‚éB‚ ‚È‚1/2‚Ì-v‹‚Ɍĉž‚µ‚āAŽ„‚͍ČŸ"¢‚Ì‚1/2‚ß‚ÌŽ„'B‚̍ŐV‚̐»•i‚Ì'¡'æ-{‚ð"¯••‚µ‚1/2BŽ„‚Í1"NŠñ•t‹ à‚Ö‚Ì•ñV‚à‚ ‚È‚1/2‚ÌŽGŽ‚ÉŠÜ‚ñ‚Å‚¢‚1/2B

‚ ‚È‚1/2‚Ì"w-͂̂¤‚¿‚Ì‚·‚ׂĂ̂ ‚È‚1/2‚ÌŽžŠÔ‚ƍK‰^‚ð‚ ‚È‚1/2‚É‚Æ‚Ä‚àŠ´

ŽÓ‚·‚éI

ASTRANSAC:

"qŒ[A"Ñ"‡-lA

‚»‚ê‚͐æT‚̉ï‹c‚Å‚ ‚È‚1/2‚ɉŠy‚µ‚Ý‚3/4‚Á‚1/2BŽ„‚Í‚ ‚È‚1/2‚Æ‚Ì-[H‚ðH‚ׂ邱‚Æ‚¨‚æ‚ÑŒö•\‚µ‚¢‚éV‚µ‚¢ƒ}ƒKƒWƒ"‚ɂ‚¢‚Ä'm‚邱‚Æ‚ð"ñí‚É‹Žó‚µ‚1/2B‚ ‚È‚1/2‚̃ŠƒNƒGƒXƒg‚ɉž‚¶‚āAŽ„‚͍čl‚̂߂̉äX‚̍ŐV‚̐»•i‚̈¥ŽA‚̃Rƒs[‚ðˆÍ‚ñ‚3/4BŽ„‚Í‚ ‚È‚1/2‚̃}ƒ KƒWƒ"‚Ö‚Ì‚P"N‚Ì-\-ñ‚É'΂*‚é•ñV‚ð‚³‚ç‚ÉŠÜ‚ß‚1/2B

‚ ‚È‚1/2‚ÌŽžŠÔ‚¨‚æ‚Ñ‚ ‚È‚1/2‚Ì"w-͂̂·‚ׂĂ̏ã‚Ì‚æ‚¢‰^‚É"ñí‚ÉŠ´ŽÓ‚·‚éI

LogoVista EJ:

eˆ¤‚È‚é Iijima Ž (Mr. Iijima) A@

‚»‚ê‚͐æT‚̉ï‹c‚É‚¨‚¢‚Ä‚ ‚È‚1/2‚ɉï‚Á‚ÄŠì‚Ñ‚Å‚ ‚Á‚1/2BŽ„‚Í'傢‚É‚ ‚È‚1/2‚ƈꏏ‚É-[H‚ðH‚āA‚»‚µ‚Ä‚ ‚È‚1/2‚ªo"łµ‚Ä‚¢‚éV‚µ‚¢ŽGŽ‚ɂ‚¢‚ÄŠw‚Ô‚±‚Æ‚ðŠy‚µ‚ñ‚3/4B‚ ‚È‚1/2‚Ì-v¿‚ɉž‚¦‚āA‚̓Œƒrƒ...[‚Ì‚1/2‚߂ɉäX‚̍ŋ߂̐»•i‚̏µ'҂̃Rƒs[‚ð"¯••‚µ‚1/2 BŽ„‚Í‚ ‚È‚1/2‚ÌŽGŽ‚É‚P"N‚Ì'èŠúw"nj_-ñ‚Ì‚1/2‚ß‚É"¯‚¶‚­Žx•¥‚¢‚ðŠÜ‚ñ‚3/4B

‚ ‚È‚1/2‚ÌŽžŠÔ‚Æ‚ ‚È‚1/2‚Ì"w-͂̂·‚ׂĂ̏ã‚Ì-ǂ¢K‰^‚ð'å•Ï‚ ‚肪‚Æ‚¤I

PCTranser/ej:

eˆ¤‚È‚éIijimaŽA

æT‚̉ï‹c‚Å‚ ‚È‚1/2‚ɉ‚±‚Ƃ́AŠì‚Ñ‚Å‚ ‚Á‚1/2BŽ„‚́A'傢‚É‚ ‚È‚1/2‚ª"­•\‚µ‚Ä‚¢‚éV‚µ‚¢ŽG‚ɂ‚¢‚Ä‚ ‚È‚1/2‚ÆŠwK‚ŃfƒBƒi[‚ðs‚¤‚±‚Æ‚ðŠy‚µ‚ñ‚3/4B‚ ‚È‚1/2‚Ì-v¿‚É"š‚¦‚āAŽ„‚Í"á•]‚Ì‚1/2‚߂ɉäX‚̍ŐV‚̐»•i‚̏܎^‚̃Rƒs[‚ðˆÍ‚ñ‚3/4B‚Ü‚1/2AŽ„‚Í‚ ‚È‚1/ 2‚ÌŽGŽ‚É1"N‚̐\ž‚É'΂·‚é•ñ‚ðŠÜ‚ñ‚3/4B

‚Æ‚Ä‚à‚ ‚È‚1/2‚ÌŽžŠÔ‚Æ‚æ‚¢‰^‚Ì‚1/2‚ß‚É‚ ‚È‚1/2‚Ì"w-͂̂·‚ׂĂ̏ã‚Å‚ ‚肪‚Æ‚¤I

Human translation (provided by Japanese Language Services):

"Ñ"‡-l

æT‚̃Rƒ"ƒtƒ@ƒŒƒ"ƒX‚É‚Ä‚¨‰ï‚¢‚Å‚«‚Ü‚µ‚1/2‚±‚Æ‚ðŒõ‰h‚É'¶‚¶‚Ü‚·B-[H‚̐Ȃł́A‚²o"Å'†‚̐VŽGŽ‚ɂ‚¢‚Ä‚¨˜b‚ðŽf‚¢A'å•ÏŠy‚µ‚¢‚ЂƂƂ«‚Å‚µ‚1/2B‚²-v-]‚É‚¨‰ž‚¦‚µ‚āA‚²ŽQl‚Ü‚Å‚ÉŽ„‚Ç‚à‚̍ŐV»•ii‚Ì-³-¿ƒTƒ"ƒvƒ‹**j‚ð"¯••‚¢‚1/2‚µ‚Ü‚*B‚Ü‚1/2AŒäŽ‚Ì"NŠÔ'èŠ úw"Ç'ã‹à‚à"¯••‚¢‚1/2‚µ‚Ü‚*‚Ì‚Å‚Ç‚¤‚1/4‚¨Žû‚ß‚­‚3/4‚³‚¢B*

‹Md‚È‚¨ŽžŠÔ‚ð‚¢‚1/2‚3/4‚«A‚Ü‚±‚Æ‚É‚ ‚肪‚Æ‚¤‚²‚´‚¢‚Ü‚µ‚1/2BV‚1/2‚ÈŽ-‹Æ‚Ì‚²¬Œ÷‚ð‚¨‹F‚è‚¢‚1/2‚µ‚Ü‚·B


Sample translations of more difficult sentences

English originals:

1. First Lady Hillary Rodham Clinton grudgingly responded to a Senate committee investigating the Whitewater affair, saying she recalled almost nothing about her law firm billing records.

2. Can you make it to our next meeting on the 18th?

3. The explosive growth of job listings and recruiting services on the Internet is having a significant impact on traditional hiring practices in Japan.

4. I will forever kick myself for not buying your December 1995 issue before it sold out.

5. He checked out all the books in the library that had anything at all to do with sleep disorders.

jLondon/EJ:

1. 'å"-̕vlƒqƒ‰ƒŠ[RodhamƒNƒŠƒ"ƒgƒ"‚́AWhitewaterŽ-•¿AŒ3/4‚¤‚ð'²¸‚µ‚Ä‚¢‚éã‰@ˆÏˆõ‰ï‚É‚¢‚â‚¢‚â‚È‚ª‚ç"1/2‰ž‚µ‚1/2@"ޏ-@"ޏ-‚Ì-@-¥¤‰ïLƒŒƒR[ƒh‚ɂ‚¢‚Ä‚Ì‚Ù‚Æ‚ñ‚lj1/2‚àŽv‚¢o‚³‚È‚(c)‚Á‚1/2B

2. ‚ ‚È‚1/2‚Í18"Ô-ڂɂ¨‚¢‚ÄŽ„'B‚ÌŽŸ‚̉ï‹c‚É‚»‚ê‚ðì‚邱‚Æ‚ª‚Å‚«‚é‚(c)H

3. ŽdŽ-ƒŠƒXƒg‚Ì"š"­"I‚Ȑ¬'·‚¨‚æ‚уCƒ"ƒ^[ƒlƒbƒg‚̃T[ƒrƒX‚ð•åW‚·‚邱‚Ƃ́A"ú-{‚Ì"`""I‚È'ÀŽØKŠµ‚ւ̏d-v‚ȃCƒ"ƒpƒNƒg‚É‚ ‚Á‚Ä‚¢‚éB

4. Ž„‚́A‰i‹v‚ɁA‚»‚ꂪ"„‚èØ‚ê‚é'O‚É‚ ‚È‚1/2‚Ì1995"N12ŒŽ"­s•¨‚ð"ƒ‚í‚È‚¢‚1/2‚߂ɁAŽ„Ž(c)‚ðR‚éB

5. "ނ́A‡-°•s-@sˆ×‚ª•K-v‚Å‚ ‚é‚悤‚ɉ1/2‚Å‚à'S'RŽ‚Á‚Ä‚¢‚È‚(c)‚Á‚1/2}'ŠÙ‚Ì‚·‚ׂĂÌ-{‚ðƒ`ƒFƒbƒN‚µ‚1/2B

ASTRANSAC:

1. •Ä''å"-̕vlƒqƒ‰ƒŠ[‚q‚‚„‚ˆ‚‚ƒNƒŠƒ"ƒgƒ"‚́Au"ޏ-‚ª‚Ù‚Æ‚ñ‚ǃŒƒR[ƒh‚ð‹L"ü‚·‚é-@'¥‰ïŽÐ‚ÉŠÖ‚*‚é‰1/2‚àŒÄ‚Ñ-߂³‚È‚(c)‚Á‚1/2Bv‚ÆŒ3/4‚Á‚āA‚v‚ˆ‚‰‚"‚...‚-‚‚"‚...‚'Ž-•¿‚ð'²¸‚·‚éAã‰@ˆÏˆõ‰ï‚É•s³•s³‚ɉž"š‚µ‚1/2B

2. 'æ‚P‚W‚̏ã‚̉äX‚ÌŽŸ‚̉ï‚É‚»‚ê‚ðì‚邱‚Æ‚ª‚Å‚«‚é‚(c)B

3. ƒWƒ‡ƒu‚Œ‚‰‚"‚"‚‰‚Ž‚‡‚"‚Ì"š"­«‚̐¬'·A‚¨‚æ‚уCƒ"ƒ^[ƒlƒbƒgã‚Ì•â[‚·‚éƒT[ƒrƒX‚́A"ú-{‚Ì'†‚ɏ]‚Ì'À'݂·‚éŽÀsã‚ɏd-v‚ȉe‹¿‚ð‹y‚Ú‚µ‚Ä‚¢‚éB

4. Ž„‚́A‚»‚ꂪ"„‚èØ‚ê‚1/2'O‚ɁA‚ ‚È‚1/2‚Ì‚P‚QŒŽ‚É‚P‚X‚X‚T"N-â'è‚ð"ƒ‚Á‚Ä‚â‚ç‚È‚¢‚1/2‚߂ɁAŽ„Ž(c)g‚ð‰i‹v‚ɏR‚낤B

5. "ނ͐‡-°•s"ˆê‚ōs‚¤‚*‚ׂĂʼn1/2‚Å‚àŽ‚Á‚Ä‚¢‚1/2}'ŠÙ'†‚Ì-{‚ð‚*‚ׂă`ƒFƒbƒN‚µ‚1/2B

LogoVista EJ:

1. ƒqƒ‰ƒŠ[Eƒƒ_[ƒ€EƒNƒŠƒ"ƒgƒ" (Hillary Rodham Clinton)ƒtƒ@[ƒXƒgƒŒƒfƒB[‚͏ã‰@ (Senate) ˆÏˆõ‰ï‚ª‹}-¬‚ÌŽ-Œ‚ð'²¸‚µ‚āA"ޏ-‚ª"ޏ-‚Ì-@-¥Ž--±Š‚ªƒŒƒR[ƒh‚𐿋'‚ð'-‚邱‚Ƃɂ‚¢‚Ä‚Ù‚Æ‚ñ‚lj1/2‚àŽv‚¢o‚³‚È‚(c)‚Á‚1/2‚ÆŒ3/4‚¤‚±‚Æ‚É•s³•s³‚É•Ô"š‚µ‚1 /2B

2. ‚ ‚È‚1/2‚͉äX‚́A‚P‚W"ú‚ÉŽŸ‚̃~[ƒeƒBƒ"ƒO‚É‚»‚ê‚ð‚*‚邱‚Æ‚ª‚Å‚«‚é‚(c)H

3. ƒWƒ‡ƒuƒŠƒXƒg‚ƃCƒ"ƒ^[ƒlƒbƒg (Internet) ‚̏ã‚ɃT[ƒrƒX‚ð"ü‚ê‚邱‚Æ‚Ì"š"­•¨¬'·‚Í"ú-{ (Japan)‚É'΂µ‚Ä‚Ì"`""I‚ȌقÁ‚Ä‚¢‚é-ûK‚ɏd-v‚ȉe‹¿‚ðŽ‚Á‚Ä‚¢‚éB

4. Ž„‚́A‚»‚ꂪ"„‚èØ‚ê‚é'O‚ɁA‚ ‚È‚1/2‚Ì‚P‚X‚X‚T"N‚P‚QŒŽ‚Ì-Á•¿‚ð"ƒ‚í‚È‚¢‚±‚Æ‚É'΂µ‚ĉi‹v‚ÉŽ(c)g‚ðR‚é‚Å‚ ‚낤B

5. "ނ͐‡-°‚ÉŠÖ‚µ‚ďáŠQ‚ð‚*‚é‚1/2‚ß‚É‚Ü‚Á‚1/2‚­‰1/2‚Å‚àŽ‚Á‚Ä‚¢‚1/2}'ŠÙ‚Å‚·‚ׂĂÌ-{‚ðƒ`ƒFƒbƒN‚µ‚1/2B

PCTranser/ej:

1. Å‰‚̏-‚̐lƒqƒ‰ƒŠ[Eƒƒ_ƒ€EƒNƒŠƒ"ƒgƒ"‚́Agrudgingly‚ÉWhitewaterŽ-‚ð'²¸‚µ‚Ä‚¢‚éã‰@ˆÏ‰ï‚ɉž"š‚µ‚1/2 \"ޏ-‚ªŒ˜‚¢˜-ñ‚ª‹L˜^‚·‚é"ޏ-‚Ì-@-¥‚ɂ‚¢‚Ä‚Ì‚Ù‚Æ‚ñ‚lj1/2‚à"P‰ñ‚µ‚È‚(c)‚Á‚1/2Œ3/4‚Á‚āB

2. ‚ ‚È‚1/2‚ªA‚»‚ê‚ð‰äX‚ÌŽŸ‚̃~[ƒeƒBƒ"ƒO‚ɏ÷"n‚·‚邱‚Æ‚ª‚Å‚«‚éa'æ18H

3. ƒCƒ"ƒ^[ƒlƒbƒg‚̏ã‚ÌŽdŽ-ƒŠƒXƒg‚ƐV•º‚ð•åW‚µ‚Ä‚¢‚éƒT[ƒrƒX‚Ì"š"­"I‚Ȑ¬'·‚́A"ú-{‚Å"`"I‚ÈŒÙ-pŽÀs‚ւ̏d-v‚ȉe‹¿‚ðŽ‚Á‚Ä‚¢‚éB

4. Ž„‚́A‚»‚ê‚Ì'O‚Ì1995‚Ì"­s•¨‚ª"„‚ê‚1/2‚ ‚È‚1/2‚Ì12ŒŽ‚ð"ƒ‚¢Žæ‚ç‚È‚¢‚1/2‚߂ɉi‰"‚ÉŽ„Ž(c)g‚ð‚¯‚éB

5. ‡-°-ŽG‚Å‚*‚é‚*‚ׂĂ̐l‚ɉ1/2‚àŽ‚Á‚1/2}'ŠÙ‚ŁA"ނ͂·‚ׂĂÌ-{‚ð'²‚ׂ1/2B

Human translation:

1. ƒqƒ‰ƒŠ[EƒNƒŠƒ"ƒgƒ"'å"-̕vl*‚̓zƒƒCƒgEƒEƒH[ƒ^[‹^˜f‚ð'²¸'†‚̕ďã‰@ˆÏˆõ‰ï‚É'΂µA•vl‚Ì-@-¥Ž--±Š‚ÌŽx•¥‹L˜^‚ɂ‚¢‚ẮA‚Ù‚Æ‚ñ‚Ç‹L‰¯‚É‚È‚¢‚Əa-ʂÅ"š‚¦‚1/2B

2. ŽŸ‰ñ‚Ì18"ú‚̉ï‹c‚ɏoÈ‚Å‚«‚Ü‚*‚(c)B

3. ƒCƒ"ƒ^[ƒlƒbƒgã‚̏AEˆÄ"à‚¨‚æ‚ÑŒÙ-pƒT[ƒrƒX‚ª"š"­"I‚É'‰Á‚µA]-ˆ‚Ì"ú-{‚ÌŒÙ-pŠµs‚É'1/2'å‚ȉe‹¿‚ð‹y‚Ú‚µ‚Ä‚¢‚éB

4. ŒäŽ1995"N12ŒŽ†‚ð"„‚èØ‚ê'O‚É"ƒ‚í‚È‚(c)‚Á‚1/2‚±‚Æ‚ðAŽ„‚͈ꐶ‰÷‚â‚Þ‚Å‚µ‚傤B

5. "ނ͐}'ŠÙ‚ŁA‡-°áŠQ‚ɂ‚¢‚ĉ1/2‚ç‚(c)‚ª'‚(c)‚ê‚Ä‚¢‚é-{‚Í‚*‚ׂĎ؂è‚1/2B

Contact information


Translation software
for English Win95

The translation programs reviewed in this article run under Japanese Windows. Neocor Technologies (La Jolla, CA), however, has released both English-to-Japanese and Japanese-to-English translation packages that run on either of the 32-bit versions of English Windows (95 and NT) without requiring additional support software to handle the Japanese input/display. Both Typhoon MT (J/E translation) and Tsunami MT (E/J translation) come with their own front-end processor (FEP), two fonts, and complete Japan ese text-editing capabilities.

The software is entirely OLE 2.0 compatible, so that Japanese text can be integrated with other Win95 or NT OLE-compliant programs. The translation engine used in Typhoon MT and Tsunami MT was developed by AI Laboratories in Tokyo and features an augm ented dictionary and kanji search system tailored specifically for native English users.

Neocor Technologies

Phone +1-619-483-2524

Fax +1-619-483-2586

email: info@neocor.com

http://www.neocor.com


PC Transer/ej

Nova Corporation, phone 03-3351-3356, fax 03-3351-5766; ¥198,000.

JLondon/EJ

Kodensha Corporation, phone 06-628-8880, fax 06-628-3841; ¥98,000.

ASTRANSAC

Toshiba Corporation, Software Products Division, phone 0423-40-6244, fax 0423-40-6010; ¥63,000.

LogoVista E/J

Catena Corporation, phone 03-5690-8578, fax 03-5690-1290; ¥97,000.




About Steven Myers