Four Typical Mistakes in Chinese Translations of Google Translate

Google Translate has always been a leader in the field of machine translation (MT), but what’s embarrassing is that in Chinese translation results of Google Translate, there are some typical low-level errors.

However, the reason why these errors are not solved may just reflect the advantages of Google Translate, that is, Google pays more attention to and is good at understanding and translating meanings of sentences, rather than their naturalness or fluency. This is beyond some other machine translation software and even many human Chinese translators.

In order to maintain the natural fluency of their translations, many Chinese translators would even omit important words and meanings in the original text. In this way, the target text may look beautiful, but in fact it is not as faithful and complete as Google Translate in expressing meaning of the original text.

I’m not here to evaluate pros and cons of machine translation and human translation, but just talk about some advantages and disadvantages of Google Translate and human translation.

And here are some of the most common problems in Google Translate, which may be helpful for Chinese MTPE projects (if using Google Translate) and other related projects (such as machine translation software development).

No “enumeration comma”

At present, there is no “enumeration comma” (、) in target texts of Google Translate, which is very common in Chinese. Google Translate would convert all commas in English texts (,) into a Chinese-format comma (，). This would be a problem for juxtaposed Chinese nouns.

“著名” (famous) is written as “着名” (wrong word)

This is a lower-level mistake. Google doesn’t seem to know how to write “famous” in Chinese correctly, but instead replaces it with “着名”. This error is not easy to see because “著” and “着” look similar, and Chinese readers of the target texts would still understand it.

Abuse of “被” (be/being)

Abuse of the Chinese character “被” (be/being) relates to the emphasis on semantic translation by Google mentioned above. Engineers of Google Translate should be aware that passive sentences in Chinese are far less than English. However, in Google Translate, the passive sentence in English is often literally translated into Chinese containing the character “被” (be/being).

Such translation certainly does not affect understanding, but for many sentences, reading with passive sentences could be awkward. But even for Chinese human translators, it’s not an easy job to transform many passive English sentences into active ones.

So, Google Translate simply gives up on such conversions. The advantage of this is that at least the original meaning would be retained. Instead, when some human translators find it hard to change “passive” into “active”, they would rather delete or even misinterpret some difficult sentences, in order to achieve seeming fluency, in order to avoid possible pickiness of employers or clients.

Word-for-word translation

Like all other machine translation applications, Google Translate also has a problem of word-for-word translation.

There doesn’t seem to be a solution to this for machine translation in near future. But even in word-for-word translation, Google Translate generally does a better job than other machine translation tools in “understanding” and translating meaning of the original English texts, which also reflects the advantages of Google Translate in semantic understanding.

Above are the four typical errors and problems that exist in Google Translate. Compared to other companies, Google is unparalleled in big data, but might find it not as convenient as translation agencies in collecting human translation data, because Google doesn’t provide human translation services.

After all, few human translators would submit errors on Google Translate pages. In comparison, translation agencies could collect human translations easily via translation memories (TM), to improve their own MT tools.