Fewer Errors from Humans
While machine translation is attractive for its speed and low cost, it is riddled with errors. Many words are spelled the same in two different languages, yet carry completely different meanings. Machine translators will not catch these differences to ensure the correct meaning is inserted into the translation because they are incapable of comprehending the subject matter.
Machines Don’t Recognize Cultural Differences
Many phrases that have a cultural significance in one location do not carry the same meanings in other locales. Machine translation will usually translate the words literally, losing the cultural meaning. I have found that only a human translator is capable of identifying the cultural meaning, then translating that meaning in order to produce the same results.
Not All Words Translate Literally
Many words do not have a literal translation from one language to another. In these situations, I have seen unpredictable results from machine translators (with most being senseless), whereas human translators have always been capable of identifying the meaning behind the word, then translating that meaning using the correct words for the language.
Machines Are Not Dialect Specific
There are oftentimes many distinct dialects within a single language. Low cost machine translators often use only the predominant dialect regardless of the specific audience we intend to address. Believe me, showing a Mandarin presentation to a Cantonese audience can be embarrassing. Human translators are quite capable of identifying and utilizing the correct dialect our audience requires.
Machines Do Not Incorporate Literary Devices
Machine translators cannot distinguish literary devices like alliteration, metaphor, and other wordplay. Again, I have witnessed some interesting results come from automated translating programs that were completely inaccurate. The human translator understands these devices and the meaning behind them, and is capable of using the proper vocabulary for the target language.