"Translation is an art," Cal Poly Spanish associate professor Fernando Sanchez said. Graphic by Melissa Wong

Smartphones are making language translators more accessible to everyone, but one Cal Poly professor says he doesn’t expect students to stop studying modern languages and literatures any time soon.

Despite the rise in popularity of translation applications, such as Ortsbo and Vocre, Cal Poly Spanish associate professor Fernando Sanchez said learning a second language is still important for Cal Poly students’ success.

Translators can be effective tools because everyone needs dictionaries occasionally, Sanchez said. Still, he warns those using translators to be careful, as certain words might be taken out of context or misunderstood by a computer. Human interpretation is more effective in translating not only words, but syntax and intent also, he said.

“Translation is an art,” Sanchez said. “It’s not just a matter of finding the matching word, but it’s also interpreting.”

Human translators are most capable of perfecting this art because a person can decipher a conversation better than a computer, Sanchez said. It is obvious to native speakers when text is translated poorly because meaning and context are often lost, he said. People who rely on translators will narrow the way they interact grammatically with the world.

“From a philosophical perspective, they will never see what is correct or real,” Sanchez said.  “They will be just living in an illusion.”

Despite this, mobile device translation applications are growing popular and have had high user growth in recent months.

On Oct. 11, Intertainment Media Inc. announced the expansion of its mobile translator application, Ortsbo, to Android devices. Ortsbo allows users to communicate with others through text translations in 53 different languages.

The company also announced plans for voice-to-text translation technology. With Ortsbo’s expansion comes the opportunity for even more user growth; according to Intertainment’s press release, Ortsbo already had a 21 percent increase in iPhone users in September.

Competing application Vocre, released by myLanguage in mid-September, has already capitalized on the concept of voice-to-text translation. The application won “Best mobile App” and “Audience Choice Award” at the TechCrunch Awards in San Francisco in September.

Applications such as Vocre and Ortsbo use a specific method of translation called statistical machine translation. It is the main translation process for most translators because it is less costly and saves more recourses, said Cal Poly computer engineering lecturer Foaad Khosmood.

In statistical machine translation, the translation of a word or phrase is looked up in a large compilation of information called a repository, Khosmood said. Statistical machine translation is more accurate than other types because it takes advantage of online resources already available, instead of trying to program specific grammatical rules into the translator, he said.

Even so, statistical machine translation still faces barriers that prevent it from being completely accurate. There are two major problems associated with it: disambiguation and alignment, Khosmood said.

Alignment focuses on the pattern of words and is difficult to perfect in more complex sentences, Khosmood said. Certain languages have different noun-verb orders than others, making translation confusing, he said.

Disambiguation decides the context in which a word is used, an important factor in the process of translation, Khosmood said.  He said he estimates 70 to 75 percent of words can be correctly interpreted, but that disambiguation distorts many translations.

“Am I talking about ‘dog’ like I’m dogging someone, or is it like a pet?” Khosmood said.

It is words with multiple meanings that create problems for translation applications. Statistical machine translation has made a lot of progress in recent years, but still has a long way to go, Khosmood said.

“It’s sort of the best game in town right now,” he said. “But you know, it’s far from perfect.”

No matter how advanced technologies become, human translations will always be more reliable, said computer engineering senior Aaron Burke. Burke speaks Chinese and Japanese and studies translation. Translators are becoming more and more accurate, but aren’t as capable as humans, he said.

“There’s always that bit of limitation,” Burke said. “Computers are still dumb machines and only know what they know.”

Accurate translation is not the only way students benefit from becoming bilingual, Sanchez said. Students who study a second language will have an enriched life, he said. He said his own perception of the world has been broadened because of his experiences with language.

“I live in a different world, different dimensions,” Sanchez said. “I would say that reality is richer because I have the ability to understand the world in two ways.”

Students who learn another language will also have advantages over their non-bilingual competitors as they enter the job market, Sanchez said.  Students majoring in areas such as communication, engineering, agriculture and business administration could all have heightened success if they choose to learn another language, he said.

“People who speak more than one language will be more marketable and valuable because they can interact with more than one culture,” Sanchez said.

Brooke Sperbeck wrote this article.

Join the Conversation

2 Comments

  1. Well, while human translators may be superior to computer translators right now, they probably won’t be for long. The technology for understanding/translating the nuances of languages, such as words that have multiple meanings, context, etc, etc, is just starting to emerge. Take for example IBM’s computer, Watson. While playing Jeopardy, Watson could understand the questions, many of which were fairly convoluted. I expect that programs will be as good as (and probably better than) than human translators within just a few years.

Leave a comment

Your email address will not be published. Required fields are marked *