Everything You Need to Know About Translation Memory

With the escalating demand on localization and translation nowadays to cope with internationalization, computer-assisted translation tools have been created to help translators accomplish their work effectively.

One of the core elements of the computer-assisted translation tools is the translation memory (known as TM).

According to the Common Sense Advisory (CSA), 41% of all human translated content is done using translation memories.

What Is Translation Memory?

A translation memory is a database that is filled up with previously translated phrases, sentences, paragraphs, or sentence-like units such as headings and titles that help translators and linguists in their translation processes.

Each source text and its translation that are stored in the translation memory are called “translation units.” The following is an example of such units:

Source Text (English): How are you?

Target Text (Danish):  Hvordan har du det?

Torjoman believes that machines can also help translators to produce the most accurate results. Therefore, translation memories, glossaries, and termbases are an important part of Torjoman’s translation process.

How Does Translation Memory Software Work?

Translation memory is an important element in any translation management system owing to its role in facilitating the whole process of translation.

The process starts when a translator uploads the source file through a translation management system (e.g., CAT tools), then the system automatically checks the source text and retrieves any matches from the translation memory.

The translator will be able to view and compare both the source text that is generated from the database and the suggested translated one.

In fact, a translation memory has a smart function named as “rating system” that determines the accuracy of pairing the source text with an equivalent translated one.

The accuracy is defined through the following three different matches:

Perfect Match

This typically happens when the source text perfectly matches the suggested translated text in the translation memory. In this case, the percentage of matching is 100% and translators accept the text as it is.

Fuzzy Match

The matching is called “fuzzy” when only parts of the text correlate to the data in the translation memory. The percentage is less than 100 and ranges from 75 to 99%. Usually, a translator, here, decides to make changes and adjustments to the text as the fuzzy match only helps translators or linguists to speed up the process of translation by saving their time.

No Match

If there is no match between the newly added content and the stored segments in the translation memory, a translator has to start the process of translation from the scratch. Then, the system is notified that a new translated content will be added.

Indeed, this will enrich the database of translation memory as the new content will be useful while accomplishing other future translation projects.

Repeated Segments

There may be a repeated segment in a source text, and it has to be translated only once. Translation memory helps translate repeated segments automatically during the translation process.

Translation memory can be created through one of the following two ways:

• Empty Translation Memory

This one is built up by a translator at the beginning of a translation project and, therefore, all the new content will be added to the database.

• Translation Memory from Existing Translations

Translators can create it by using an “alignment” tool that automatically transforms the source text and its translated one into translation units that can be used in future translations.

Using Translation Memory Is a Win-win Situation

• It saves time and effort either for individual translators or translation agencies. Since translation memory automatically matches segments from the original text with identical translated ones, it definitely saves time and makes the process of translation effortless to a larger extent.
• Translation memory also ensures the highest quality especially in translation projects where more than one translator are working together since it allows them to use only one database and, consequently, any discrepancies will be avoided. Torjoman employs the highest innovative technologies during the process of translation including computer assisted translation (CAT) tools, and powerful translation management systems (TMS).
• Clients do benefit from translation memory software as they receive their work as fast as it is possible with the highest quality that meets their consumers’ expectations.
•  Using translation memory is important especially in technical and legal translations where many segments are repeated.

Get most out of translation services by vising Torjoman.