All about translation memoriesDefinition, function and benefits

10 October 2023
Posted by: Nereida BIRDWELL
Categories: Confidentiality, Document translation, Terminology

At Birdwell, professional translators use translation memories for improved efficiency and accuracy when translating documents. Translation memories are not machine translation tools (like Google Translate) where the translation is generated entirely by a machine. Instead, they are designed to be used by human translators to facilitate their work and improve translation quality.

What’s more, for Birdwell, translation memories are invaluable assets guaranteeing the confidentiality and quality of translations. You can rest assured that each memory is carefully created, client by client, in specific textual domains, ensuring that there is no unauthorized distribution to third parties.

A point on terminology: a translation memory is often referred to by professionals as a “TM”, an abbreviation of “Translation Memory”. This should not be confused with “MT”, which stands for “Machine Translation”, and refers to a translation performed by an Artificial Intelligence engine.

What is a translation memory?

A translation memory (or TM) is a linguistic database used by translators and translation professionals to store and manage segments of translated text. These segments can be sentences, of varying length, or even entire paragraphs. The main purpose of a translation memory is to help improve the efficiency and consistency of translations, particularly for repetitive or similar translation projects.

In practical terms, a translation memory works by storing pairs of translated sentences (the original sentence and its equivalent translation) in a database. When a translator is working on a new project, the computer-aided translation (CAT) software examines the text to be translated and searches for matches in the translation memory. If similar or identical segments have been translated in the past, the system automatically suggests the previous translations, enabling the translator to reuse existing work. The CAT tool generates a report (see image 1), which categorizes the matches according to the level of similarity of the new segments to those already translated.

As such, the match categories include:

• exact matches (100%, Context Match, Perfect Match),

• partial matches (75-99%),

• approximate matches (50-74%),

• new words (or non-matches) (less than 50% similarity).

Image 1. Grid of TM match categories as displayed in a report.

How do professional translators use translation memories?

Professional translators use translation memories in a number of ways in order to translate documents more efficiently and accurately.

1- Reuse of translated segments:
One of the main uses of translation memories is the reuse of previously translated text segments. When a translator encounters a similar text segment in a new document, they can search the translation memory for a matching translation. This saves time, and ensures consistency of terminology throughout the project.

2- Terminology management:
Translation memories enable translators to manage domain- or client-specific terminology efficiently. Segments containing terms can be stored in the translation memory once translated, to ensure that this terminology is translated consistently throughout the project. Another way of managing terminology is by using a Glossary.

3- Customization of translations:
Professional translators can adjust the translations stored in the memory to suit the specific needs of the project or client. They can modify, delete or add elements to the stored translations, depending on the context or particular requirements. Depending on the level of similarity between a new segment to be translated and the match found in the TM (see the section on match categories above), the translator must therefore modify the suggested text to a greater or lesser extent, to ensure that the final translation matches the source text as closely as possible.

4- Quality improvement:
Translation memories can help translators avoid common errors and meet linguistic quality standards. By comparing current translations with those stored in the memory, translators can quickly detect and correct inconsistencies or errors.

5- Collaboration:
One of the major advantages of translation memories is their ability to facilitate collaboration between translators, enabling several professionals to work on a joint project while maintaining linguistic consistency. Each translator can access the shared translation memory to benefit from translations previously added by other team members.

6- Time saving:
By using translation memories, translators can speed up the translation process, which is particularly useful when it comes to meeting tight deadlines or handling large volumes of text.

7- Saving money:
The use of a translation memory is a smart way to reduce translation costs. Agencies do not charge the same rate for all the text match categories. It would be illogical, and dishonest, to charge clients for content that they have previously paid to have translated. So the longer you work with Birdwell, the greater your chances of making real savings.

In short, translation memories are translation aids that translators use to optimize their work. Using translation memories has a number of advantages: it saves time and money, facilitates consistency in both translation and terminology, and reduces errors.

Professional translators always exercise their linguistic and cultural expertise to produce high-quality translations, using translation memories as complementary resources.

Did you know? When CAT tools, and therefore translation memories, first came onto the market, most translators were opposed to them. Today, they are an indispensable work tool, making life much easier for linguists – and project managers.

What’s more, translation memories coupled with Artificial Intelligence tools, harnessed by professional translators, now make it possible to work even faster, while ensuring quality and cutting costs! However, that will be covered in a future article (coming soon).