Clear English gives good machine translation
Companies want to supply information to international audiences. Translation and localisation by professional translators is expensive. We have economic problems, and companies want to decrease their costs. Sometimes, machine translation is a good method of decreasing the cost of translation.
Machine translation is the automatic translation of text using only software without the help of a human translator. Machine translation is also known as computer translation or automated language translation.
Machine translation has a bad reputation. Many people can show examples of bad machine translations. However, many organisations use machine translation successfully.
For machine translation to give good translations, the source text must be optimised for machine translation. Therefore, machine translation is not suitable for literature or for poetry.
Many free machine translation tools are available on the Internet. Unfortunately, free machine translation has two large problems:
- You cannot specify the translation of particular words and phrases.
- You cannot specify text that must not be translated. [Note]
Therefore, free machine translation is not as good as commercial machine translation. However, if text is optimised for machine translation, the quality of free machine translation is usually satisfactory.
For good machine translation, the source text must be very clear. Correct spelling and correct grammar are important. This section gives three other important guidelines about how to write for machine translation.
In English, a word sometimes has more than one meaning. For best results with machine translation, use a word with its primary meaning. For example, both of the following sentences are grammatically correct:
- Hit a key to continue.
- Press a key to continue.
Although many technical manuals contain instructions to 'hit' a key, the word 'hit' is not used with its primary meaning. For machine translation, the first sentence is not as good as the second sentence. Possibly, a machine translation will not be correct.
A phrasal verb is a type of verb that has two words. Frequently, the meaning of a phrasal verb is different from the combination of the two words. Frequently, the parts of a phrasal verb can be separated by a noun. For example, the following sentences are both grammatically correct.
- First, carry the most important project out.
- First, carry out the most important project.
Instead of a phrasal verb, use an alternative one-word verb. For example, instead of 'carry out', use 'complete', 'do', or 'perform'. If a phrasal verb is necessary, keep the parts of the verb together.
Frequently, long sentences have complex sentence structures. Complex sentence structures can cause problems for machine translation and for people.
A clause is a group of words that contains a subject and a verb. Ideally, do not use more than two clauses in a sentence. In the following example, the two options are equivalent. The first option has one sentence and three clauses. For machine translation, the second option is better:
- The software that is supplied with the XYZ machine that we bought yesterday is not satisfactory.
- We bought the XYZ machine yesterday. The software that is supplied with the XYZ machine is not satisfactory.
To avoid complex sentences, make sure that sentences have fewer than approximately 25 words.
The text in this article is optimised for machine translation. Therefore, we expect a free machine translation to have satisfactory quality.
We will use free machine translation to translate this article. Then, translators will evaluate the quality of the machine translation. We will publish the results.
Note. The source text was in a Word document. Therefore, with Google Translate (http://translate.google.com), we could not specify that some text must not be translated. (This note is not part of the text that was used in the evaluation.)