- May 5, 2017
- Posted by: Wordminds
- Category: Uncategorized
For those of us who work with translation memory (TM), we come across the terms CAT (computer-assisted translation) and TM regularly, and perhaps even the names of the most popular TM tools, such as SDLX, DejaVu, and Trados. If you have overlooked them previously, it’s time to reconsider. These are by no means the latest buzzwords in our industry, they do add value to what we do. The question is: how accurate is TM?
For those who aren’t wholly familiar with TM just yet, it is essentially a database that stores source phrases and sentences that have already been translated. Whenever you translate new pieces of text, you can use the localisation and translation programs to retrieve related translations, saving a great deal of time. TMs make localising changes easy. For example, if you need to make a few changes to a manual, you can use TM where the original version will already be stored. You just make the changes you need to and import the new source text. We are able to build TMs for each client to help save costs in the future and ensure consistent translations with previously localised materials.
Types of texts that work well with TM
There is a myriad of texts that TM works well with, especially content that has repeated areas of text. Examples include large websites, instruction manuals, and software manuals such as FAQ sections or help files.
Examples of Translation Memory
Let’s consider an example. Typically, Translation Memory tools will organise segments into three categories: new content, fuzzy matches, and then repeated content. For instance:
- New content: Jack ran up the hill
- Fuzzy match: The boy, Jack, sprinted up the hill
- Repeated content: Jack ran up the hill
As you can see, the fuzzy match is very similar to the new content, but it’s not exactly the same. The Translation Memory tools award percentage scores according to the degree of similarity. For instance, a 100% score means that the segment is the same, whereas an 80% score means it is almost the same.
The benefits of Translation Memory
There is a great many benefit to using TM in our translation and localisation projects, but the three core advantages are:
- Consistency – when we reuse approved translations, we save a great deal of time which means we can take on more projects.
- Faster turnaround times – translators spend less time working on repeated content.
- Lower costs – by leveraging content that has already been translated, you save on manpower and therefore the client saves on costs, so it is a win-win situation.
Does Translation Memory effect the quality of translations?
The main effect of TM tools is related to the “error propagation.” This means that if the translation for a certain segment is not accurate, it’s more likely to be the wrong translation which will be reused the next time that source text or a similar one is translated.
TM can differ from one tool to another and with regards to their structuring. While they do help to save time and money, we cannot disregard the accuracy of professional and experienced linguists.