Recent Posts View All

Blog - Alignment: Getting to know your content

Alignment: Getting to know your content

posted 2007-03-25 17:23:01
by Kevin Moss

It’s amazing how diverse the Globalization space is and how quickly it is growing. There really are so many perspectives on what Globalization is and where it benefits business. And along with this are the accompanying attributes of each category such as technology, business process improvement, and expansion into new global markets. Working in this arena, I find my daily activities vary with the numerous facets of the business.

Recently, I have been working with a practice known as ‘Alignment’. Alignment quite simply put is the task of taking previously translated content along with the source content it is associated with and generating Translation Memory (TM). The TM is typically generated from aligned segments, where a segment is a unit of content such as a sentence. This TM is in turn used for future translation projects using Computer Aided Translation (CAT) tools.

There are three techniques commonly used to align content:

• Manual alignment


• Automated alignment


• Computer aided alignment

There is a another approach -- do not align at all and revisit the translations. We’ll discuss this topic in a minute. So how do you choose which one to use? Here's a little more detail on each technique. Making some decisions early on will save you a great deal of time.

Manual Alignment
When the content format and structure is very different between languages. You want to look at things from a manual perspective when the content is formatted completely differently in each language. For example, paragraphs are in a different order, content is organized into tables in one document but simple paragraphs in another...and the list goes on.

Automated alignment
When you are certain the structure of the content matches perfectly between languages. The WorldServer Globalization Management Solution provides a lot of power here from a workflow and project management perspective. If the content originated from a common template or CMS system, you can expect a high level of perfectly aligned documents. Recently I ran a large Macedonian project and 95% of the files aligned perfectly because of the structure of the content.

Computer aided alignment
When there are subtle differences in the content being aligned or text can be extracted and placed in a intermediate format. When things don't line up perfectly, firing up the WorldServer Desktop Workbench is a great way to finish the process of alignment. Splitting and Merging coupled with all the keyboard shortcuts provide an expedient solution. Generally, you want to look for 60% or greater aligned segments when aligning using a tool.

When not to align?
The content is in two different formats or the format is not conducive to alignment. Also, when the 'translated' content is not really a translation. For example, the translation is not a literal translation but a re-write.

The last point is highly pertinent because in many cases what is assumed to be a translation is really a piece of original work based on a topic versus an actual translation. It’s easy to pick these out in some languages, but others such as CJK or Arabic languages are difficult if you are not familiar with the general linguistic flow of the language. These types of issues can crop up in seemingly structured content.

It pays to know your content. Selecting the right approach and not making any assumptions will save you a lot of time.

Share:

Back to the Blog

Back to Top