Modern technical writing tools and methodologies have encouraged the use of automated tasks to reduce the amount of duplicate content by single sourcing content across documents and media. This benefits companies by reducing the total amount of content, and, therefore the time to create or update that content and the documents it goes into.
Reducing the amount of content also has a significant impact on the cost of translation. Translation is typically charged per word. And if you are translating into multiple languages, that is an additional saving for each language your content is translated into.
So the reducing the content you have to translate means more savings. Well, for automated documentation processes, it is not as simple as that.
Challenges with translating automated documentation
Automated processes split up sentences and recompile them when the document is produced. For a translator, this means translating words or partial sentences so the context is lost.
Not all languages work the same way, so translating content that is used in an automated process presents a number of challenges. It is important to take these into consideration when setting up your production environment to ensure that content is used in a consistent context.
Here are five common challenges with translating content used in single sourced or automated production environments.
Declension
English uses very little declension – the change to a noun or adjective based on its context, for example, male/female, singular/plural. Other languages, especially Slavic languages are heavily declined.
This presents a challenge when a word is automatically inserted in the text, because the form of the word may need to change based on the context it is used in.
A common scenario is cross references. Using cross references within the text means that the context can change.
For example:
“See Table 1”, is not the same as “Table 1”.
In the first instance “Table 1” is the direct object, and in the second it is the subject.
Plurals
In English there is either singular or plural – one or many. In other languages, both adjectives and nouns may change depending on the number of objects, for example, 1, 2-3, 4-10 or more than 10.
This presents a challenge when the total count changes due to the production process.
A common scenario is a list of features that change depending on the product variant.
For example:
“This product has the following features:
- Feature A
- Feature B
- Feature D”
The number of features in the list can change beyond the thresholds for one variant, and not the other.
Modal verbs
Modal, or auxiliary verbs modify another verb, for example, “I can type”. Not all languages use modal verbs for the same modifications.
This presents a challenge when splitting the content of sentences to allow it to be changed at publishing time.
A common scenario is hanging sentences used to introduce a list of features that change depending on the product variant.
For example:
“You may want to consider:
- Walking to work,
- Taking the bus, or
- Learning to drive.”
Non-equivalence
Not all words can be translated one-to-one between languages. This may be due to cultural reasons, but also because the concept does not exist, for example, industry jargon.
This presents a challenge when automatically inserting words, or short sentences into content during production.
A common scenario is when using variables to insert a specific word into a sentence.
For example:
“Please can you pass the marmalade.”
Marmalade is a specific type of fruit preserve. Not all languages have a word for marmalade.
Left-to-right vs Right-to-left
English is read from left to right where as Arabic is read from right to left.
This presents a challenge when content is automatically compiled from a sequence of text inserts.
A common scenario is error messages where a series of parameters are linked together.
For example:
“Error 101: Your [PC] is [unable to connect] to your [Router] because the [Cable] is [Not Attached]”.
The order of the content inserted into this sentence may be grammatically incorrect in the target language; therefore, simply translating the component parts is not enough.