Machine translation syntax

Machine translation (MT) is a method for translating text automatically from one language to another. Machine translation is used to speed up the process of delivering products, services, and content to non-English speaking markets, and to increase the number of languages in which products, services, and content are available.

Machine translation takes sentences in a source language as input, and produces a translation of the source sentences in a specified target language. The output of machine translation is sometimes post-edited by human translators, and sometimes published without a review by a human.

The style of the source language has significant impact on the quality of the translation and how well the translated content can be understood.

If machine translation will be used to translate your content, use the following guidelines in addition to the global English syntax guidelines.

Guideline

More information

Write in Standard English.

Machine translation (MT) software is programmed to recognize the rules of Standard English grammar and punctuation. Therefore, content that follows these rules has the best chance of being translated correctly.

Good for MT

If you do not want to add tags to every picture, you can add tags to a batch of pictures.

Not good for MT

If adding tags to every picture sounds time-consuming, don’t worry—you can add tags to a whole batch of pictures at once.

Ensure that your content is free of grammar, punctuation, and spelling errors.

Unlike human readers, MT cannot compensate for most errors in grammar, punctuation, and spelling. For example, if “window” is misspelled as “widow,” MT will translate “widow” as “a person whose spouse has died,” which will look very strange in the machine-translated content.

Avoid writing sentences of more than 25 words, but do include determiners such as a, the, your, or this to avoid ambiguity.

The longer a sentence is, the more decisions the MT software has to make about how the clauses and phrases are related. The fewer the number of subjects, verbs, and phrases in a sentence, the better the chance that the sentence will be translated accurately.

Good for MT

When you publish your webpage to a website or post it to a server, be sure to publish the subfolder also. The subfolder contains all the supporting files that are necessary to properly display your drawing.

Not good for MT

When you publish your webpage to a website or post it to a server, be sure to publish the subfolder as well, because it contains all the supporting files necessary to properly display your drawing.

Avoid writing very short sentences and headings.

With very short sentences and headings, MT may not have enough information to figure out the syntax of the text. Ensure that a short sentence or heading has the syntactical information that MT needs to translate the text correctly.

Good for MT

Empty the container.

The empty container

Access is denied.

Not good for MT

Empty container

Access denied.

Do not use sentence fragments.

Because sentence fragments are, by definition, fragments of language instead of complete expression, they often lead to poor machine translations.

Reduce the ambiguity of verb-like words that end in “-ing.”

A verb-like word that ends in “-ing” can be highly ambiguous because it can function as an adjective, noun, or gerund. Whenever it isn’t clear from the syntax what the function of such a word is, disambiguate it. Note that the following sentence is ambiguous:

  • This is blocking code.

You can remove the ambiguity by rewriting this sentence as follows:

  • This is blocking the code.

  • This is the blocking code.

For more information, see Words Ending in -ing

Reduce the ambiguity of verb-like words that end in “-ed.”

A verb-like word that ends in “-ed” can be highly ambiguous. Whenever it isn’t clear from the syntax what the function of such a word is, disambiguate it. You can do so in three ways, depending on the context:

Add a determiner (a, an, the, this) either before or after the “-ed” form.

Good for MT

They have an added functionality.

Not good for MT

They have added functionality.

Turn the “-ed” phrase into a clause that uses a form of the verb be.

Good for MT

You should also configure limits for the backup size that are based on the amount of disk space available.

Not good for MT

You should also configure limits for the backup size based on the amount of disk space available.

Note

If a modifier such as “based on” can modify more than one noun, MT will join it to the closest noun. In this example, note that the based on can modify either limits or size. Because size is the closest noun, that’s the one that the MT system will join it to. But note that this is not correct. Based on modifies limits. The example under “Good for MT” makes this clear by adding “that are” (a plural verb that can only modify the plural noun “limits”). You can also disambiguate the sentence in this example by writing it as two sentences, as is done in the next example.

Write the idea as two sentences.

Good for MT

You can also configure limits for the backup size. These limits should be based on the amount of disk space available.

Avoid linking more than three phrases or clauses by using coordinate conjunctions such as and, or, or but.

For every and, or, and but in a sentence, MT has to determine which sentence elements are being coordinated, and that becomes more difficult as the number of these coordinate conjunctions increase in a sentence. Therefore, try to keep them to a minimum.

Good for MT

Other computers may be able to acquire usage rights and play the songs if they have a connection to the Internet. However, devices such as portable CD players and car stereos will not be able to acquire those rights. Therefore, they will not be able to play protected content.

Not good for MT

Other computers may be able to acquire usage rights and play the songs if they have a connection to the Internet, but devices such as portable CD players and car stereos will not be able to acquire those rights, and will not be able to play protected content.

Do not use abbreviations unless they are standard abbreviations, such as CD for compact disc.

If you use an abbreviation that is not in the MT dictionaries, the abbreviation will not be translated. For example, if you decide to abbreviate “not permitted” in a table as “NP” after first mention, “not permitted” will be translated, but NP will not be translated. It will remain NP in the target language, and may be confusing or unintelligible to the reader.

Good for MT

To open SQL Server Configuration Manager on the target computer, double-click the file.

Not good for MT

To open SQL Server CM on the target computer, double-click the file.

Use correct capitalization.

Correct capitalization helps improves MT because capitalization clearly marks beginnings of sentences and identifies proper nouns. Note that the sentences in the following examples may not be translated the same way because “Start” is not capitalized in the “Not good for MT” example.

Good for MT

Learn how to use the Windows Start menu to open programs and folders.

Not good for MT

Learn how to use the Windows start menu to open programs and folders.

Note

Because MT can use capitalization to disambiguate words, sentence-style capitalization is better for headings than title capitalization.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
13.58.44.229