Glossary of translation terms

By GeaSpeak Team | 2022-01-15

In this glossary you will find some of the most popular terms used in the translation industry.


  • 100% match: The TM recognizes identical segments and translates them automatically by suggesting a 100% match.


  • Alignment: It allows to create matching segment pairs out of two separate but identical documents, one in the source language and another one in the target. These segment pairs can then be imported into your translation memory.
  • Auto-Propagation: A feature that applies your translation of a segment to other segments in the same or different files. Such segments need to have identical or similar source content. If the segment is not identical, you will of course need to adjust the autopropagated content accordingly.


  • Batch: When a translation file is divided into multiple files, the group of files is called “batch”.
  • Bilingual file: A bilingual file mainly consists of two columns. One column containing the source text divided into segments, and another one containing the target text divided in the same way. Thus, the original segment is shown next to its translation. It could also contain the type of segment (100% match, fuzzy match) and comments, where applicable.
  • Brief: A document compiling all relevant instructions and guidelines for a translation project. Most frequently translation agencies provide the final customer a document to fill in relevant information such as target language and audience, preferences on how to handle currency, measure units, numbers, tone to be used, localization of brand or product names, glossaries, URLs and any other project specifics such as character restrictions, etc. The brief is then provided to the linguists taking part in the project for them to follow all client guidelines.


  • CAT tools: CAT stands for Computer-Aided Translation. This term refers to technologies used to streamline translation and localization processes. 
  • Closed caption (CC): Unlike subtitles, closed captions include not only the video dialogue but also non-speech elements such as sound effects, soundtrack and other background noises. They serve the purpose of aiding people with hearing difficulty by describing them the audio. Subtitles, on the other hand, only include the video narration or dialogue, and serve as an aid for people who can’t understand the video language.
  • Context match: To be a context match, the translation memory segment must be a 100% match and the previous and following segment must be a 100% match as well (i.e., the segment context has to be the same).


  • Do-Not-Translate (DNT): Do Not Translate terms, as their name suggests, are those words/phrases which should not be translated. They may be proper nouns, placeholders or words that the client prefers to keep untranslated for some other reason.
  • DTP: The use of computer software to design and create digital files for printing or online viewing, such as emails, PDFs, decks, flyers, brochures, newsletters, etc. which most frequently contain text and images. In the translation industry, it often refers to the process of creating the fully localized formatted file.
  • Dynamic PE: Unlike regular postediting tasks, where a fixed rate is paid, dynamic post-editing tasks are paid based on the effort needed to edit the MT output. The final rate is calculated based on the amount of editing that is needed for each segment, from no edits to a full edit. Also, this model ensures that all translators are paid a minimum percentage of their full word rate even if no changes are made to the segment.


  • Editing: Second step in the TEP process. Editors examine translations in order to improve their overall quality of writing and to make sure it conveys the same meaning as the original text.


  • Full Post-Editing (FPE): Full post-editing consists of a human post-editor making corrections to translations made by software (machine translation). Corrections required are not only basic issues as in LPE, but are also concerned with structure and style. The result will be a high human-quality translation.
  • Fuzzy match: When a similar segment is found in the TM, a fuzzy match (e.g. a match of 79%) is suggested. This segment usually needs to be modified to fit the current source segment.


  • Glossary: A glossary is an alphabetical list of terms relating to the same subject with their own definitions. In the translation industry, a glossary also includes the translation for each term to be used for a specific project or client.


  • In-Country Review (ICR): An In-Country Review is a process where the translated files are delivered to the final client. The latter then reviews it with the help of a native speaker. This native speaker assesses the translation to ensure that the terminology used is appropriate for the target region or locale. This process normally happens after the translation and editing steps.


  • Language pair: A language pair consists of the language of the original file (the source language) and the language it needs to be translated to (the target language). For example: English (United States) -> Portuguese (Brazil).
  • Language Quality Inspection (LQI): A Language Quality Inspection (LQI) is a process that measures the quality of translations received. It involves inspecting a sample of translated files, usually at an early stage of the localization process. LQI experts look for residual errors and classify them according to different categories. Feedback from LQI experts is provided to translators for them to make any  adjustments in translation. Just as for LQEs, during the LQI process the translation is given a score.
  • Language Service Provider (LSP): An agency or company which provides language services such as translation, localization or interpretation, among others. 
  • Language Sign Off (LSO): LSO is the final language check on files that have come back from DTP processes and are in the final format (mostly PDF). It is performed to enhance the quality by eliminating errors that are easy to miss while working on the segmented files or that appeared during file conversion, like layout issues.
  • Light Post-Editing (LPE): Light post-editing consists of a human post-editor making minimal corrections to translations made by software (machine translation). These corrections will be made only when necessary to make the translated text understandable. They are mostly major grammar and spelling mistakes.
  • Linguistic Quality Evaluation (LQE): LQE is the process where external reviewers linguistically assess translations by completing a scorecard, usually in a dedicated platform, where they include any changes along with comments when required. Translation is given a score (and qualified as Pass or Fail) and translators are given the opportunity to rebut any corrections they deem inaccurate or unfair.
  • Localization: The process of translating a product, service or website in one language into another language while adapting it for the target culture.
  • Localization Kit (Lockit): A Localization Kit (Lockit) is a set of tools and files necessary for the correct completion of a localization task. It is sent by the client, and typically contains instructions, source and reference files, glossaries, style guides and translation packages that are specific for the task.
  • Logs: A document containing the detailed word count of a project.


  • Machine Translation (MT): A process performed by a software whereby a written text in one language is rendered into another language without humans being involved.
  • Machine Translation Post-Editing (MTPE): The process of editing machine translation output to achieve the required quality for translation. Postediting may be light, when quality requirements are low, or full, when a great human-quality translation is expected.
  • Migration: This task consists of making adjustments to a translation originally intended for a paper format so that it fits an electronic format. Migrations usually involve small linguistic changes and tag addition. E.g., migrating a paper survey translation into an electronic format may involve adding tags required for HTML format and also replacing strings such as “Circle the right option” with appropriate text like “Select the right option”.
  • Multi-Language Vendor (MLV): A vendor providing language services into multiple languages.


  • Non-breaking space (nbsp): A special type of space character which prevents a line break in the position where it is placed. This helps avoid that a number is split from the symbol following it, like the percentage symbol. It is also used with proper nouns consisting of more than one word, to prevent them from being split into different lines.


  • Package: A file containing the required materials to work on a translation project, namely the files to translate, the translation memory and the termbase. It may also include source files and detailed word count. They are specific to each CAT tool and can only be opened with the relevant software.
  • Perfect Match: In most CAT Tools, a Perfect Match is identical to a 100% Match. However, in SDL Trados, a Perfect Match is a type of context match that, instead of comparing a segment to a Translation Memory, compares it to a set of bilingual files: if the segment and its surrounding entries are checked and confirmed to be identical to those of other bilingual files, that is a Perfect Match, also known as segment match.
  • Project Manager (PM): A Project Manager is a person working for a translation agency who is mainly responsible for organizing translation projects and ensuring that these projects are completed and delivered in a timely manner. Sometimes they are also in charge of keeping in touch with translators, and help them if they have any issues.
  • Pronunciation Guide (P-Guide): A pronunciation guide is a file containing words with possibly unclear pronunciation or words that have more than one possible pronunciation which are gathered as an aid for voice talents. A P-guide job involves extracting “difficult” words and strings of text from the translated document and annotating the entries with information on how to pronounce these texts. For annotation, the use of International Phonetic Alphabet (IPA) characters may be required. Guidelines are provided on how to identify “difficult” words and word combinations, and on how to annotate pronunciation.
  • Proofchecking: Not many companies use this task, but for those who do, proofchecking is not the same as editing. An editor focuses on style, whereas a proof-checker should focus on accuracy and eliminating types of mistakes like omissions, mistranslations, incorrect numbers, incorrect proper nouns, spelling, grammar errors, hyphenation mistakes, correct use of punctuation, consistent and correct measurements, consistency with client guidelines and glossary entries. Although it sounds pretty much like an editing, style matters are totally out of scope, and that’s what makes the difference with a proofchecking. Proof-checking is the final check carried out by a native speaker before file delivery to the final client.
  • Proofreading: This step does not involve the same for every Service Language Provider. Some translation agencies call “proofreading” what is regularly considered as “editing”, but for the vast majority, the proofreading step is the final reading of a translation, without comparing the source and target text, in search of residual typos, punctuation mistakes or any other final adjustments required. Unlike an LSO step, the proofreading takes place during the production -and not the post-production- step, meaning that the linguist performs it while working on the CAT tool, not on the final formatted localized file.
  • Purchase Order (PO): A document issued by translation agencies detailing the word count of the project completed and the amount of money the linguist will be paid for it. Once a translator receives a PO, they should issue their invoice according to the translation agency’s financial procedures.


  • Quality Assurance (QA): The quality assurance step may involve different things depending on each company, but most frequently it has to do with going through quality assurance reports flagging spelling, terminology and inconsistency issues, fixing any problems, providing feedback to relevant linguists and also having a final read of the translation. Other companies also call “QA” to a final reading of the formatted localized files searching for typos, spelling and format issues, what is most frequently called an “LSO”.


  • Rebuttal: A process by which linguists can accept or reject the corrections introduced to their work during linguistic quality evaluations. Linguists can reject corrections or penalizations to their translation by providing reasons why they think their version was better than the corrected text or why they think the change should be considered preferential and not penalized.
  • Rush: A rush project or job is a request which needs to be completed either as soon as possible or by a deadline which is considerably tight and quite shorter than expected in relation to the project scope or volume. Usually special rates are offered for completing this kind of jobs, sometimes paid as a percentage over the regular rate.


  • Screenshot review (SSR): When a project comprises the translation of content for a mobile application, the equivalent to the regular “LSO” step is the so called Screenshot Review. Once the app has been localized, screenshots are taken and provided to a linguist to perform the Screenshot Review, or SSR. This task consists of going through screenshots provided to confirm they have no issues, the text is showing correctly, there are no layout problems and no text is missing or misplaced. For this task, source screenshots are provided along with a report specifying what text is out of scope, what legacy translations should have been followed, etc. There is usually a Word file report containing all relevant screenshots where the linguist is required to provide a comment stating if the screenshot looks okay or what adjustments are needed. Then there are further rounds as appropriate (SSR1, SSR2, etc.) to confirm any changes have been correctly implemented.
  • Search Engine Optimization (SEO): The process of optimizing a website’s configuration, content and link popularity so that it can be easily found by search engines such as Google or Yahoo. Using keywords in the right places, i.e. titles, headlines, website’s link, among others, is one of the most common ways of optimizing a website’s configuration. 
  • Segment: Usually a segment is a sentence, but it could also be just a few words, a number or a link.
  • Source: The original text to be translated.
  • Style Guide: A document issued by a particular client that sets its basic style, grammar, punctuation and terminological guidelines and conventions around a specific language and locale, such as es-419 (Latin America Spanish). By using this document and adhering to these guidelines, it is ensured that all texts from such client will be localized consistently. This also contributes to making the localization process more efficient. 
  • SubRip Subtitle (SRT): An SRT file is a plain text of the subtitles of a video that contains not only the script of the video but also the timing of each line so that the subtitles match the audio. SRT software is used to create and edit this type of files. 
  • Subtitling: The process of translating spoken dialogues into written texts on a screen. Subtitling is a type of audiovisual translation which has its own set of rules.


  • Tag: Tags are basically placeholders. They show and reflect specific formatting information like subscript or superscript or for elements like footnote or index markers.
  • Target: The translation of the source text.
  • TEP: TEP is the process by which most language service providers operate to ensure a high-quality and accurate translation. TEP stands for Translation, Editing and Proofreading. Once a translation is completed, it is edited and proofread by another linguist who has the necessary industry expertise as well as the linguistic skills to accomplish this task.
  • Termbase (TB): Terminology databases (or termbases or TBs for short) are more similar to electronic dictionaries than to translation memories. They store single words or expressions with one translation only.
  • Testing: A service provided by LSP to make sure that products and apps work effectively and smoothly in all cultures, languages and devices. Testing can be functional testing (ensures that software and hardware products meet all functional requirement specifications), localization QA testing (verifies localization quality to make sure your product will satisfy the customers in the target cultures), regression and maintenance testing (makes sure software updates and enhancements work properly) or UX testing (ensures your product can be easily used by customers and will have positive interactions). 
  • Track changes: A feature in Microsoft Word and in some CAT tools allowing to see the changes introduced to a translation by another person, e.g. a reviewer. Changes can then be accepted or rejected. Trados Studio, for instance, offers this feature and several views so that you can see the original translation, the reviewed version with visible changes or the final corrected version with changes implemented. This is very useful for sharing feedback and assessing translation quality based on the number and nature of review corrections.
  • Transcreation: Transcreation combines translation and creation. Transcreators have more artistic license. They are true to the original, but can incorporate copywriting as well. They have to preserve not only the original tone and intended meaning, but also the original feelings and emotions that brands look to evoke in the audience. The main goal is to give the target culture the same emotional experience as the source culture.
  • Transcription: The process of converting oral texts into written texts.
  • Translation: The process or result of rendering a written text in one language into another one, conveying not only the meaning of the words but also the intention.
  • Translation Memory (TM): Translation memories (TMs for short) are databases that store source and target sentences as a matching segment pair. They are used to ease the translator’s workload when working in similar documents.
  • Turn Around Time (TAT): It refers to how long it will take a linguist or provider to complete a project. Sometimes projects are required to be completed by a certain deadline. Sometimes the client does not know when the project will be assigned, but they do know that from the moment the job is assigned, there will be a certain number of days to complete it, e.g., 3 days. That is the TAT for the project. Also, if you are a freelancer or service language provider, you may be asked what is your best TAT to complete a request.


  • Voice over (VO): Both voiceover (VO) and dubbing involve replacing the original audio of a video, movie, etc. with a localized version of it. The difference resides in that voiceover does not intend to reflect the tone of the original audio, but only to convey its content. On the contrary, dubbing is intended to convey the same emotions as the source audio and be as less noticeable as possible (i.e., it aims at giving the audience the impression that actors speak the audience’s language). For this reason, VO is most frequently used in documentaries and training content, while movies, television shows, etc. use dubbing instead.


  • Word Count (WC)/Weighted Word Count (WWC): WC is an acronym for Word Count. It refers to the word breakdown of a translation project, i.e., the number of words in a project specified by category (new words, repetitions, 100% matches, fuzzies, etc.). WWC is an acronym for Weighted Word Count. The weighted word count is an estimate of the average number of words requiring translation. It serves both the purposes of estimating the time required for a project and the amount to be paid for it. To calculate the WWC, each category of words is assigned a percentage of effort required. E.g., a new word will require 100% of effort to translate but a fuzzy word will require a 50% of effort. A high fuzzy will require a 25% of effort, but a 100% match word will only require a 10% of effort (percentages may vary from one company to another). So if your project consists of 250 100% match words and 250 new words, the weighted word count for the task will be 275 words (250 100% match account for 25 “actual” words, and 250 new words account for 250 actual words).


  • XLF: File extension of XLIFF files. In terms of functionality and content, there is no difference between an XLIFF file and an XLIFF file with an XLF extension.
  • XLIFF: RWS Trados Studio saves all documents in the bilingual XLIFF format (i.e. not in the native format, e.g. DOC or PPT). XLIFF files contain both languages, the source language segments and your translated segments.