Is Machine Translation Ready for You?

Article Featured Image

Anticipation is growing that translation software (known as machine or auto-translation) is ready for prime time. With it, you’ll be able to drop text into a field, press a button, and receive the results automatically. To see how far we’ve come, my staff and I ran a few tests using a variety of translation software, ranging from free to expensive. The results almost took our breath away.

When you input in English: For U.K. English, press hash (#).
The result is:
German: For Great Britain English, press a big mess.
Japanese: For the United Kingdom English, publications commutated meat dishes.

Phew! As a translation professional, I wiped the sweat from my brow in relief. I can keep my day job; software is not ready to replace me just yet. But the time is coming—soon—when translation software will be available at high quality and for low cost.

Indeed the future for auto-translation is bright. Although no software has conquered perfection, one day it will. Then you will be able to press one button to translate your interface, Web site, voice prompts, email, and telephone conversations into multiple languages and regenerate prompt files with text-to-speech.

Machine Translations That Almost Work
For general business documents, help pages, and thick manuals, machine translation is powerful and saves thousands of dollars. Software can be trained for repeated expressions, such as This agreement is by and between, or Click on the print icon. Most manuals are now translated by auto-translation or memory-aid software and then cleaned by human drudge translators, who tediously perform the same task every day.
Auto-translation results improve the closer the target language’s grammatical order is to the original. Intelligibility also improves when fewer adjectives, metaphors, imagery, and technology expressions are contained in the text.

When you input in English: Your ticket is scheduled for pickup on June 1st at 1:22 p.m.
The result is:
Mandarin: Your ticket in June 1 schedule hitchhiking at 1:22 p.m.
Russian: Your ticket is planned for a sensitive element on June 1 for 1:22 p.m.
German: Your ticket will be held for recording on June 1 around 01:22 pee emm.
Japanese: 1:22 PM your ticket is the June 1 will be scheduled for shipment.

Current Deficiency
Machine translation will not attract customers or sell your product. Not yet. It has no persona, creates no corporate image, and engenders no emotional impact. IBM makes good auto-translation software, but doesn’t use that  software to translate its main Web pages. Why? Because IBM understands that there is more to attracting sales than word matching. Sales require expressions with cultural impact and relevance, a need not met by software. But if you are a reporter searching for articles around the world on a certain subject, machine translation is brilliantly effective in narrowing the scope.

For speech professionals, auto-translation can have dire consequences. The most serious problems are introduced by complex sentences with multiple subphrases that involve from and to expressions, dates, and times. When translated, these phrases must often move to a different part of the sentence, or are spoken in flipped order, and degenerate into incomprehensibility. For example:

When you input in English: Please say the name of the city you are starting from, followed by the city name you are traveling to.
The result is:
Spanish: Please to say the name of the city you yes to be starting from, to follow by the city name you yes to be traveling to.
Mandarin: Your name says you are from, city name follow start your travel city.
Japanese: Depending on your journey you have to be followed by the name of the city began to say the name of a city.

Optimal results from machine translation are achieved with one subject, one verb, and one direct object. Keep sentences short. Do not abbreviate, shorten, or add metaphors. Remember that out of sight, out of mind becomes invisible insane, and keep text simple and direct if it goes to translation software. Considering the time it takes to train translation software to sometimes be correct, you could have hired a professional translator to always be correct. Now. At this time in history. But the future is coming.

Sue Ellen Reager is CEO and founder of @International Services, a global translation services company and developer of localization software. She can be reached at sueellen@internationalservices.com.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues