Glossary

It is an agreement between a data controller (such as a company) and a data processor (such as a third-party service provider). It regulates any personal data processing conducted for business purposes. Recently, it has become an important part of business relations in the translation industry due to the entry of the General Data Protection Regulation in the EU.

Each CAT tool divides a text into segments. Then, it begins analysing the text against a relevant translation memory. As a result, each segment is given a score, i.e. a match ranging from 1 to 99%. Matches are important because they speed up the translation process and provide discounts. As a result, you save a lot on a bigger translation memory and similar texts stored in it. There are different types of matches:

• 101% match (also called: ICE match or a context match) is an identical segment in an identical context, which means we the translator can be sure this match can be approved, because the segments before and after it are also identical.
• 100% match (also called: identical match) is an identical segment, but the segment before or after it is different, which means it still needs to be checked by the translator.
• 99% - 95% matches – almost exact matches; it is usually enough to check the tags or particular words in the segment
• 95% - 75% matches – they are called high and medium fuzzy matches which show similar texts in the translation memory where some changes are needed
• 75% - 50% matches – low fuzzy matches usually useless for the translation process and most CAT tools do not show them at all; as a result such segments are considered “no matches” for which you need to pay the whole price
• Fuzzy repeat matches (95% - 75% internal matches) – these are similar segments found not in the translation memory, but across one file or more files in one project. Not all vendors provide discounts on them, but you can always try to convince them.
• Repetitions – are types of internal matches, i.e. identical segments found across one file or more files in one project. However, their special status is due to the fact that most vendors charge less for them, while not willing to pay less for the others.
• Machine translation matches – if you use an MT engine for your translation project, each segment for which no fuzzy match was found in the TM is propagated with machine translation from this engine

It is a label that define the way a certain information should be processed. In CAT tools generally, tags are used to mark parts of a text that correspond to:

• different formatting i.e. different fonts, font sizes, line breaks or colors
• internal code (e.g. HTML, index entries or cross-references)
• other special characters

However, sometimes they are just minor (or even invisible) changes.

Localization is the process of adapting a product, service, advertising or any other content to a specific market. It should be considered different from translation because it actually goes way beyond. Localization requires you to convert text from one language to another, taking into consideration all the different factors such as: time zones, currencies, symbols, colours, legal requirements, website links, product beliefs and cultural references. The goal is to give your content the look and feel of having been created specifically for each target market you’re aiming at.

It is a broader term for a translation vendor, meaning an organization that offers language-related services (i.e. translation, interpreting, localization, transcreation, transcription, subtitling, voice-over). An LSP can include a translation agency, a company, or an individual.

It is defined in ISO 18587:2017 as automatic translation of text from one natural language to another using a computer application. Machine translation is a hot topic in the translation industry right now with everyone talking about it both in respect of the human factor and the tools used in the process. MT quality has dramatically improved over the past few years. Many enterprises and language service providers have turned to MT for faster and more cost-effective translation.

It is also known as a confidentiality agreement (CA) under which the parties agree not to disclose any information covered by it. It is especially important when translating sensitive or confidential information contained in documents. Many translation vendors fail to properly protect their customer’s information which can lead to many legal issues. You have a choice: to sign your NDA or your vendor's NDA. This agreement can be signed between the client and LSP, and between LSP and translator/reviewer.

Service defined in ISO 18587:2017 as editing and correcting machine translation output There are two main levels of post-editing, light and full; their use depends on the purpose of the translation output and the client’s requirements:

• full post-editing - process to obtain a product comparable to a product obtained by human translation
• light post-editing - process to obtain a merely comprehensible text without any attempt to produce a product comparable to a product obtained by human translation

This agreement defines the scope of expectations and needs of the client in relation to the service delivered by the software provider. It can outline the following aspects:

• minimum time of system availability
• time of providing a solution to reported issues
• procedure of reporting and monitoring issues
• consequences of failure to comply with the terms of agreement (e.g. compensation)

A termbase (TB), also known as terminology database, is a glossary of industry-specific and company-specific terms, which you can manage outside your translation memories. It is a glossary of terms in a CAT tool that you can create to be able to search for individual words and display them among translation results. It is useful if there are multiple translators working on the same project and it’s important to keep accuracy and consistency regarding the main terms. It can be single-language or multilingual. A term base shows the translators which terms are new, approved, forbidden, or preferred together with their additional description, definition or picture.

It is an electronic collection of source language and target language segment pairs. Its purpose is to facilitate the retrieval and use of previously translated content. It is a database of translated sentences (segments), which is created during the translation process [source: ISO 17100:2015 Translation services — Requirements for translation services]. A translation memory has been at the core of the industry since the late 1990s and its importance has not faded even with the arrival of new technologies.