Machine Translators Are No Match for the Human Touch

200 million translations. Every day. This is how many phrases are typed into the most commonly used online translation tool. Since its beginnings in the early 2000s, machine translation has come a long way, with more and more complex algorithms resulting in more accurate translations. Yet, machine translation cannot replace a human for many reasons. When it is imperative that the message gets across accurately, machines are often not the best choice.

How does automatic machine translation work?

All “automatic” machine translators operate using statistics instead of rules. They review patterns of millions of documents that have already been translated by humans. Then they make a choice based on the most frequent occurrence of that situation. In Spanish, for example, “darse cuenta” is translated into “realize” by humans most of the time. Based on those statistics, a machine will ignore the word-for-word translation “give account” and translate the phrase as “realize”.

Since more documents are available in English than any other world language, machine translators usually use English to translate between languages. A language will be translated into English first, then that is translated into the target language. Sometimes, more than two iterations of translation can occur, making the translated phrase less likely to perfectly resemble the original.

Why isn’t machine translation good enough?

Although machines can be great at translating basic phrases, especially into English, the very procedure that it uses prevents it from competing with a human translator. Statistics within the programs can’t understand the feel or intent of the original text. They may use words that are most probable, but destroy the meaning of the phrase. A human translator can pick up on the tone of the text and choose the correct words using that information.

Because language is based on rules, a machine will have problems translating more complicated grammatical concepts. For example, it will struggle with the difference between imperfect and preterite past tenses because it usually uses English, a language that does not distinguish between the two grammatically.

Lastly, a machine does not know who your audience will be, so it will not necessarily choose words that apply to the situation. If you are speaking in a causal setting, it will may use more formal speech that would sound strange to your casual audience.

Need an example?

The following text appeared on a website that sells Argentine Wine. Of the English versions, one was written by a human translator, and one was written by everyone’s favorite online translation tool. Chances are, distinguishing between the two won’t be too hard.

Original text:

Después de una excelente cosecha como la que le precedió, la cosecha 2009 muestra sus virtudes en este vino base Cabernet Sauvignon, mas el ensamble de tres variedades de gran personalidad que encontraron en San Rafael el terruño ideal para la expresión de sus mejores cualidades. Vino aun de color rojo violáceo intenso a pesar de los años en botella, ya en la copa se nos muestra intenso y seductor con aromas especiados que se entremezclan con nítidos y frescos aromas a frutas de ciruelas, cerezas negras y moras, mientras que se van desprendiendo lentamente los aromas tostados que recuerdan a granos de café molidos.

Translation 1:

After a bumper harvest as that which preceded the 2009 vintage shows its virtues in this cuvée Cabernet Sauvignon, but the assembly of three varieties of great personality that found in San Rafael ideal for the expression of his best qualities terroir. Wine intense purplish red even though the years in the bottle color, and in the cup shows intense and seductive with spicy aromas mingle with crisp, fresh fruit aromas of plums, black cherries and blackberries, while van slowly peeling roasted aromas reminiscent of ground coffee beans. (Source)

Translation 2:

After a great harvest like the one that preceded it, the 2009 harvest shows its virtues in this cuvée Cabernet Sauvignon. It’s an expressive mixture, articulated by three varieties of great personality that are found in San Rafael, the perfect region to bring out its best qualities. The wine still preserves a strong purplish-red color, in spite of the amount of years gone by since it was bottled. Once poured into the glass, it remains intense and seductive. Its spiced scents mix together with clear and fresh fruit aromas of plum, black cherry and blackberry, while its toasted scents release slowly, reminiscent of the delicious smell of ground coffee beans. (Source)

Results

Translation 1 was done by machine, and Translation 2 was created by a human translator. That was probably pretty obvious. The first translation was difficult to make sense of. It got the point across with a few misinterpreted words, but did not sell the product like the second translation.

Machine translation is a handy tool to have. Mainly in a pinch when you need a basic understanding of a phrase or to read a sign while traveling. However, machine translation is very limited and it loses its value when important documents need to be translated. In many cases, the human touch is necessary to translate messages correctly using an understanding of the subtleties and rules of language.

Source

About Global LT

Global LT, formerly Langua Tutor, was founded in 1979 and for over 39 years, our mission has been to empower corporate employees to live and work successfully anywhere in the world with our language and cultural training, translation and interpretation services as well as expatriate destination and global workforce talent development programs.

Headquartered in Troy, Michigan, with representatives in Boston, Los Angeles, Phoenix, Pittsburgh, San Jose, Hong Kong, Frankfurt, London and Shanghai, Global LT employs over 100 corporate employees worldwide, with a dedicated network of over 2000 tutors, translators, trainers and specialist who deliver our services.

Please fill out a contact form for more information or call + 1.248.786.0999