What is translation memory?
Translation memory is a type of software that has the ability to store and reuse terms and phrases as content goes through revisions and updates. It removes some of the traditional obstacles to speedy, accurate translation—duplication, complexity, and time—and replaces them with a process that is streamlined and efficient.
How does it work?
Translation memory compares sentences in a document we are translating with their translation in a previously translated document. The software stores this pair in a database or "memory." When a previously translated sentence reappears in a new document, translation memory automatically gives us the translation. Translation memory also gives us "fuzzy" matches for sentences that are similar, but not identical, to sentences we have previously translated.
Translation memory means that we never have to translate, research, or look up the same phrase twice. The software remembers phrases and sentences as we translate them. If they are repeated, the software will offer the previous translation back to us, allowing us to leverage our work. The more documents we translate using translation memory, the larger our memory becomes. This means we can translate more words, faster and more accurately.
Translation memory also gives us the ability to replicate the design and formatting of any document we translate—flawlessly and seamlessly. Many translation agencies outsource the formatting to a typesetter or desktop publishing company. This involves the back-and-forth of multiple proofs among the client, the translator and the typesetter. We do it in-house with translation memory software. Our software separates the text from the formatting instructions. Then, after the translation is complete, the text can be placed back into the page layout template where it automatically acquires all the formatting associated with the original, including pagination, bullets, fonts, and graphics./
Translation memory has the added advantage of being technologically fluent. It supports everything from HTML/ASP/JSP/SGML/XML and XSL formats to FrameMaker, PageMaker, QuarkXPress, InDesign, Ventura and Interfleaf/Quicksilver files. It even supports the latest NET-driven Web technologies and related formats such as ASP.NET.
The effect is especially striking for updates. For example, in a service manual that is updated annually, we may find that more than 50% or more of the content can be leveraged from earlier translations.
What about new clients who have not established a translation memory database? We will create a translation memory database for them, using documents that the client has previously had translated. This increases our efficiency, and ensures consistency of terminology and style.
LWT Communications uses translation memory because we believe that translated documents are part of a company's intellectual property. We think that once you've made an investment in translation, you should get a good return on that investment. And you can do that by leveraging and reusing previously translated material as widely as possible. It's a way of safeguarding and protecting an important asset.