The language industry might seem confusing to a newcomer, with lots of unknown terms and scientifically sounding jargon coming at you from all sides. I compiled this short (ha-ha) glossary to make things easier and show you that making sense of translation terms isn’t rocket science. I also hope that the article will be helpful as a quick reference source for industry veterans, too. So, without further ado:
See Exact match.
101% and 102% matches
When going through a translation memory, a CAT tool may identify some segments as having 101% or 102% matches. Although this looks weird, it is not an error: A 102% match means that not only the current segment is identical to the one stored in the TM, but also the segments before and after it. Respectively, a 101% match means that either the preceding or subsequent segments are identical. Together, these two segments before and after any given segment are referred to as its context.
See LSP margin.
In software development, Agile refers to a development process where new features are rolled out continuously, in a “conveyor belt”-like fashion. The old approach is “waterfall” development, where features are developed and delivered in large batches at larger time intervals. Agile approaches are becoming increasingly common, which is having an impact on software localization, because customers now expect new strings to be translated as quickly as they are delivered. This makes continuous localization increasingly in demand.
Application programming interface / API
A “window” that a piece of software opens for other pieces of software so that they are able to communicate with each other. In software and website localization, APIs allow localization engineers to configure automated or continuous localization workflows or develop custom scripts for translation management systems and CAT tools.
People who grew up speaking two languages in equal or comparable proportions or who moved to another country at a sufficiently young age to internalize the second language. Apart from the English-to-Spanish pair, true bilingual speakers are very rare. When a translator claims they are bilingual, make sure they really mean it, because just becoming fluent and especially good at a second language does not constitute bilingualism.
In software localization, the character limit refers to the maximum amount of space a string can take up. Character limits are especially used in mobile app localization, where the screen size is more limited. Some CAT tools, such as Smartcat’s built-in editor, allow project managers to define character limits for specific segments:
Cloud translation tools
An approach to translation that allows several participants to contribute to it at the same time. Examples include several translators’ working simultaneously on different parts of the same content, or translators’ working with editors and proofreaders down the workflow stream at the same time. Collaborative translation is usually a feature only available in online CAT tools.
Further reading: 5 features of a truly collaborative translation platform.
In the context of CAT tools, “comments” refer to bits of text that are added to certain segments within the source text. Comments, along with screenshots, help reduce mistranslations due to lack of context.
Computer-aided translation / computer-assisted translation / CAT
(Not to be confused with machine translation)
A software tool that makes the translation process easier for the translator. This involves breaking up the source text into sentences, or segments, and displaying it in tabular form (sometimes called a “CAT editor”) so that a translator can easily switch from one sentence to another and see their overall progress, like this:
Further reading: CAT tools in the content economy: A necessity, not a luxury.
In computer-aided translation, a segment that a translator has explicitly marked as complete. In a multi-stage workflow, this will allow further contributors down the line, e.g. an editor or proofreader, to start making edits. In a single-stage workflow, confirming a segment will increase the project progress as seen by the translation project manager. Confirming a segment will also insert the translation into the translation memory and make it available for reuse. It is therefore important for a translator to consistently confirm each segment they are working on. In most CAT tools, this is done by pressing Ctrl+Enter or ⌘+Enter.
Example from Smartcat’s built-in editor:
A new language industry paradigm, in which the translation process is “connected”, both internally and externally, at multiple levels. This includes integrations to source content management systems and repositories; collaborative translation workflows; real-time progress tracking; and even automated payments to vendors. The main proponent of the paradigm as of 2019 is Smartcat.
Further reading: Connected Translation: The new paradigm for the language industry.
An increasingly widespread paradigm in which companies and brands no longer compete on products and features but on the ways they communicate with their customers, i.e. through their content. For the language industry, the rise of the content economy brings both new opportunities, as more content needs to be translated than ever, and new challenges, as customers are no longer willing to wait for weeks to have their translations delivered.
Further reading: Embracing the global content economy (slides).
Content management system / CMS
A content management system is a piece of software that allows its users to accumulate, organize, and publish content. As of 2019, the most popular CMSs in the market are WordPress, Joomla, and Drupal. In website localization, CMSs can be integrated with TMS and CAT tools to enable continuous localization.
For the specific, translation memory-related sense, see 101% and 102% matches.
In computer-aided translation, context is what defines the surroundings of any given text or string. For example, depending on the context, “reorder” can mean “change the order [sequence] of something” or “make an order again”. Lack of context is an especially challenging issue in software localization, where strings can come arbitrarily, and one containing nothing but the word “Reorder” is not uncommon. Screenshots and segment-related comments help avoid mistranslations due to lack of context.
A process in which each new piece of original content in a company’s CMS is automatically sent to the CAT tool and/or TMS for translation. Once a translation to a target language is completed, it is automatically pulled back into the CMS. Continuous localization is usually implemented via APIs or pre-built integrations.
Continuing education / CE
Continuing professional development / CPD
Educational courses and programs aimed at improving attendees’ translation skills or subject matter expertise. Some translators associations require their members to accumulate a certain number of CPD “points” to continue membership or renew certification. Such “points” are awarded for attending certain, association-approved CPD courses.
A copywriter creates all kinds of content for companies: landing pages, newsletters, whitepapers, blog articles, SMM posts, among many, many others. Copywriting and general writing skills are especially important for marketing translators and transcreators, because in this line of work literal translation is rarely expected or encouraged.
A crowdsourced translation is one done by several people at once, with or (more often) without any organized splitting of the source content between them. Although modern CAT tools, especially ones in the cloud make it easier to do translations this way, the customer should be very wary about the quality, as it’s much harder to keep the output consistent, let alone ensure it flows well Translation memories, glossaries, and QA checks can be of help, but are no panacea against quality issues.
See Customer references.
Quotes from past customers that highlight a translator’s skill set. Along with a portfolio and credentials, feedback is one of the main factors on which customers select translators, especially when they’re looking for reliability. The best references are written in an informal, relatable style and highlight the translator’s main strength, not everything “in general”.
A date or time by which a translation project must be completed. Deadlines can be “hard” and “soft”. Soft deadlines are indications of customer expectations and can be pushed back a little if necessary, while hard deadlines are not usually flexible, because they are linked to other deadlines down the “production line”. Being able to always meet (hard) deadlines is a fundamental ability for a translator.
For translation agencies, it is important to understand that if you agree to a customer’s deadline request, you should set your vendors’ or employees’ deadlines to an earlier time, especially in a multi-stage workflow.
Default language-based internationalization
An internationalization framework where every string is initially written in one language, usually English. In this case, the translations are matched with the default-language strings by the framework. For example, if there’s a line for “Create an account” in English, it can be directly matched with the German “Konto einrichten”. There are several issues with this approach. First, if you decide to change the source string (to, say, “Create your account”), you will have to re-match every translation in every target language to that new string. Second, this way you are forcing developers to come up with text lines, which you cannot — and should not — expect them to be good at. An alternative and more modern approach is key-based internationalization.
Desktop publishing / DTP
As languages are not identical when it comes to the average number of words per sentence, it is often necessary to hire a desktop publisher after translation — especially in marketing translation. They will check that the layout of the translated document is identical to the original, i.e. that there are no “hanging lines”, “stray letters”, and so on. If you don’t have your own desktop publishers, a good place to start searching is the Smartcat Marketplace.
Desktop translation tools
Unlike online translation tools, desktop tools such as SDL Trados Studio, memoQ, or OmegaT, need to be installed on a PC. Such tools — which preceded the ones online by more than a decade — have the advantage of working slightly faster and not requiring an Internet connection. On the downside, most of them are platform-dependent, and none of them support true, “Google Docs-like” collaboration.
Further reading: CAT tools in the content economy: A necessity, not a luxury.
Dialects are variants of a language that are similar enough to be classified as the same language but different enough to require different translators for each dialect. Differences between some dialects are more pronounced than between others. For example, speakers of US English will easily understand speakers of UK English, while a text written in European Portuguese might at times make little sense to speakers of Brazilian Portuguese. When ordering a translation to a language that has dialects, it is important to know where your target audience is located. Sometimes, e.g. for Spanish, it is possible to request a “dialect-neutral” translation that will try to avoid dialect-specific terms.
Direct clients / direct customers
Customers other than translation agencies. Working with direct clients usually means you can charge more but requires spending more time on sales, marketing, and developing customer relationships. If a company doesn’t have their own localization team, working with them might also require providing additional services such as editing, proofreading, localization engineering, etc.
Editing is a critically important step in any translation work. The editor’s goals are to ensure there are no factual mistakes in the translation, the terminology is consistent, and the target text reads well. Unlike a proofreader, who does minor last-stage edits and weeds out typos, an editor might introduce major changes to the translation and will usually have the final word if there is a linguistic disagreement with the translator. At the same time, despite popular belief, an editor’s goal is not to make a poor translation good, but to make a good translation brilliant.
A piece of text stored in a translation memory that is completely identical to the one contained in the current segment in a CAT tool, either together with the segments immediately surrounding it or without them.
In translation and especially machine translation, “false fluency” refers to a phenomenon when a target text reads as natural, but does not actually convey the same meaning as the source text. Translators and post-editors should be especially on the lookout for false fluency when working with neural MT engines, which, by design, are very good at coming up with fluent-sounding sentences.
Further reading: Neural is the New Black.
“False friends” are words that sound similar in different languages while having different, often conflicting meanings. For example, “gracioso” means “funny” not “gracious” in Spanish. Mistranslating false friends is a common mistake of rookie or unprofessional translators, especially in language pairs with Romance languages (Spanish, French, Portuguese, Italian, etc.).
See Freelance translator.
A translator who gets jobs from different clients without being employed by any specific one. Being a freelancer gives the translator more liberty compared to an in-house position — for example, they can move freely with an Internet connection being the only prerequisite for their work. At the same time, they have to work hard to grow or maintain the number of orders they receive. Tools such as Smartcat allow freelancers to focus on their work while paying less attention to marketing and sales activities, which some freelancers find daunting and/or boring.
See Translator testing.
Free translation software
Translation software that is free for users without any limitations. As with most open-source software, such freedom comes at the cost of having slower or non-existent support and possibly being less reliable than their paid counterparts. Some translation platforms, such as Smartcat, provide basic functionality for free while charging for additional, growth-oriented features, allowing them to combine the best of both worlds.
When a translation memory contains a phrase that is similar, but not identical, to the one being translated currently, this is called a “fuzzy” match. For example, the phrase “the quick brown fox jumps over the lazy dog” is approximately 90% similar to the phrase “the quick brown fox hopped over the lazy dog”. When such a match is encountered, a CAT tool will highlight the difference in the source text in a separate box, making it easier for the translator to edit the stored translation as necessary:
See Plurals and genders.
One of the first and most popular default language-based internationalization frameworks.
Glossaries, also referred to as terminology bases or term bases, are structured dictionaries of terms with their translations into one or more languages. When used together with CAT tools, glossaries allow companies to keep their terminology consistent across their content.
In the translation industry, hourly rates are most often used not for translation work as such, but for other roles such as interpreting, project management, desktop publishing. An exception is transcreation, where charging by the hour is becoming increasingly common. But even when charging by the word, it is important for translators to track their productivity to make sure that the resulting hourly rate is acceptable.
Further reading: 7 simple steps for setting your translation rate.
Translators who choose to work with a specific LSP or in a fixed localization team instead of freelancing. This approach is beneficial for both the professionals and their employers: Translators gain experience and become more confident, while companies know their translators are always available. The drawbacks are that this reduces the freedom of movement for the translator and increases employment expenses for the company.
A rule of thumb for companies is to start thinking about hiring an in-house translator once they have at least 20,000 words to translate every month for at least a year.
In localization, an “integration” refers to a piece of software or script that connects various parts of the localization process, e.g. a CMS and a CAT tool/TMS. For example, Smartcat has pre-built integrations with popular content management systems and software repositories but also allows localization engineers to build their own integrations via an API.
Internationalization / i18n
Whereas “localization” means making a piece of content of software suitable for a specific foreign audience, “internationalization” means thinking ahead to make this content or software easier to localize in the first place. For example, you could write your mobile app with all the strings hard-coded and rebuild it with different strings for every language. But this represents a very poor internationalization system. Instead, most modern internationalization frameworks are either key- or default language-based.
Internationalization framework / system
Verbally communicating someone’s speech in a different language. Interpreting can be consecutive or simultaneous. In the former case, the interpreter waits for the speaker to finish a piece of speech — usually no more than a couple of sentences — and then translates it to the audience. In the latter case, the interpreter speaks at the same time as — but usually at a different location than — the speaker, with their words being delivered to the audience via special equipment. Simultaneous interpreting is considered harder and costs more than consecutive interpreting.
It is important to note that, although translation and interpreting are related occupations, not all translators make good interpreters, and vice versa. If you are looking for interpreters, a good place to start is the Smartcat Marketplace.
A standard for classifying languages with two- (ISO 639-1) or, less commonly, three-letter (ISO 639-2 & ISO 639-3) codes. You can find the full list on Wikipedia.
Translation of documents pertaining to judicial proceedings, including lawsuits and sentences, identity documents and marriage certificates, and the like. The term is sometimes used interchangeably with legal translation. In many cases, judicial translation requires an official translator certification.
An internationalization approach where each piece of text is assigned a “key” that is matched with a different string in each language. For example, there could be a string with the key “create_an_account” that matches “Create an account” in English and “Konto einrichten” in German. While making it somewhat less obvious for developers than default language-based internationalization, it is much more flexible and universal, and rids developers of the need to deal with writing actual content, which can — and ideally should — be handled by other, non-developer roles down the line.
An industry sector comprising activities facilitating inter-lingual communication. This includes translation, interpreting, subtitling, language teaching, language technology development, etc. According to different estimates, the global language industry is valued between $25 billion and $50 billion as of 2019.
Language industry events
Online and offline events the goal of which is to promote, improve, and facilitate communication within the language industry. The most notable yearly events are organized by Locworld, GALA, TAUS, and ATA.
A specialist responsible for assuring quality in a specific language pair. Responsibilities include selectively reviewing completed translations, assessing translator and editor tests, drawing up style guides, and so on. The role can be either in-house or outsourced.
Language quality assurance / LQA
A role responsible for assuring that the translation meets certain quality standards. An LQA specialist can use both automated checks and manual assessment of quality — either by themselves or, in the case of multilingual projects or otherwise, through hired or in-house editors. Some CAT tools such as Smartcat allow LQA specialists to monitor translations that are still underway, allowing them to fix quality issues on the fly.
Language service provider / LSP
In modern usage, any company that provides translation and localization services. Some LSPs employ in-house translators, while others have extensive databases of freelancers. In both cases, the biggest value a customer gets from an LSP is having a single point of contact and responsibility for all translation requests. At the same time, LSPs introduce their own margin into the cost structure, so a relatively tech-savvy customer might want to employ freelancers directly using tools such as the Smartcat Marketplace.
In a broad sense, a language reviewer assesses the quality of translated text. This can be done as part of translator testing or as an ongoing activity to assure language quality. In a narrower sense, a reviewer is the same as an editor.
Any technology created to facilitate and/or automate parts of the language industry. Examples include translation platforms, CAT tools, translation management systems, machine translation engines, and language teaching apps.
Historically, activity that happens directly before or after the deadline when it turns out that the translators cannot deliver their work on time. The best way to avoid a last-minute rush is by proactively sourcing translators that have proven themselves to be reliable vendors. Additionally, modern translation platforms such as Smartcat allow project managers to track progress in real time and thus avoid last-minute rushes.
The translation of any document pertaining to the law. It most commonly refers to translating contracts for companies, but can also encompass judicial and sworn translation. This usually requires a degree of legal subject matter expertise.
The entirety of language technology-related data accumulated over time by a translator, translation agency, or localization team. The most frequently used linguistic assets are translation memories, terminology bases, and custom-trained machine translation engines.
For the data used in the translation process, see Linguistic assets.
“Resources” is a term sometimes used for all contributors of a translation project, including translators, editors, proofreaders, etc. Some people of these professions consider the term derogatory, so it’s better to use it with care, if at all.
(Not to be confused with Literary translation)
In translation theory, “literal translation” refers to translating each word separately. In a broader and more practical sense, “literal translation” refers to a translation done with little care for idioms, flow, or word usage. Literal translation is discouraged in most forms of translation, but may sometimes be intentionally used for factual correctness in technical, medical, and legal translations.
(Not to be confused with Literal translation)
Translation of works of literature, whether prose or poetry. More so than with any other form of translation, literary translation is generally viewed as an art rather than craft. As such, it often defies the usual approaches to translation like the use of CAT tools, per-word charging, or productivity maximization.
Localization / L10n
There are two senses in which the term “localization” is commonly used. The first one is broader than “translation” in that it also involves adapting the message to a local context, for example avoiding phrases that might be culturally insensitive in the target language. In this sense, this term is very close to transcreation.
In the second, and more practical sense, “localization” refers to translating in tech-heavy domains, such as software, video games, or websites. In this sense, it also includes the “adjacent” areas such as connecting source- and target-language repositories to a CAT tool or a TMS, making the necessary format conversions, and so on.
A team within a company that deals specifically with translating and localizing content into one or several foreign languages. It may include one or several translators, a translation project manager, a localization engineer, a language quality assurance person, and other roles.
Long-tail and short-tail languages
Languages that are requested as target languages less or more often, respectively. The names come from a chart that shows the number of global speakers (or some other metric) for different languages, which looks somewhat like this:
Here, the red part shows the “long tail”. Almost every multi-language vendor can translate to the “short-tail” languages. “Long-tail” ones, on the other hand, are often considered a challenge because there are fewer vendors available globally, let alone good ones. Therefore, it is an interesting niche for LSPs and freelancers to consider.
Languages in the “short tail” are the most common languages content gets translated to. The exact makeup differs depending on the metrics and methodology used, but usually includes English, Chinese, Japanese, Spanish, German, French, Portuguese, Italian, Russian, Korean, and Arabic.
Further reading: Talk to the heart: Localizing long-tail languages (TAUS).
The additional amount or percentage a language service provider adds to the price at which it hires freelancers or pays its in-house specialists. Depending on the specific company, the market, and the project complexity, the LSP margin can range anywhere from tens to hundreds per cent. It is important to understand that honest LSPs don’t charge a margin for no reason, because managing translation projects is a complex and demanding activity. Still, as of 2019, working directly with freelancers is still an option, as advanced translation platforms such as Smartcat make it much easier than even a few years ago.
Machine translation / MT
(Not to be confused with TM)
A translation generated by a computer program, a.k.a. a machine translation engine. Nowadays computers are becoming increasingly good at producing accurate translations, especially thanks to neural machine translation. At the same time, false fluency is becoming more of an issue that requires careful handling by a post-editor.
Machine translation engine
See Machine translation.
Translation of marketing content: newsletters, landing pages, ads, social media posts, etc. Unlike technical, medical, or legal translation, marketing translation is much more flexible when it comes to following the original content and favors flow and style over factual uniformity. Translating marketing content word-for-word is one of the most common errors of rookie translators.
See also Transcreation.
See Translation memory.
The translation of texts pertaining to medicine: health records, prescriptions, drug data sheets, etc. This kind of translation arguably requires the highest degree of subject matter expertise and often a translator certification. It is common for entire LSPs to specialize in medical translations alone.
Mobile app localization
A branch of software localization dealing with mobile apps. Although most of the specifics are the same as for the former, it usually features more target languages and tighter deadlines. For this reason, continuous and connected translation processes are especially relevant for this form of translation.
Most popular languages in translation
Multi-language vendor / MLV
A language service provider that offers translations to several target languages at once. Historically, most LSPs were single-language vendors based in specific geographies/locations. Thus, when the first MLVs arrived — and it was no small feat for them — they brought more value for customers, who could now deal with a single point of contact instead of communicating with several vendors at once. Nowadays, however, thanks to translation platforms such as Smartcat, any LSP can be a multi-language vendor without having to make major investments.
A translation project that involves multiple target languages. Depending on the form of translation, a multilingual project can have dozens of target languages. This strongly encourages translation project managers to use translation automation platforms such as Smartcat.
A person who has been raised in a given language-speaking environment from an early age. It is generally accepted that speakers of the target language produce more grammatically correct and naturally sounding translations than speakers of the source language. A pitfall is that they might miss some of the meaning in the original text that a native speaker of the source language would not have missed. Depending on your specific goal, you might want to go either way, or even use a combined workflow where a speaker of the target language edits a translation done by a speaker of the source language. Yet another option is to hire bilingual translators.
Neural machine translation
Machine translation technology that uses artificial neural networks (ANN). Thanks to the exponential growth of computing technology, as well as groundbreaking advances in ANN theory and practice, neural machine translation engines produce output that is sometimes indistinguishable from human translation. At the same time, false fluency is becoming more of an issue that requires careful handling by a post-editor.
New content economy
See Content economy.
Online translation tools
Unlike desktop translation tools, online translation tools, such as Smartcat or MateCat, do not require installation and can work on any computer connected to the Internet. As a result, they can enable true collaboration and work on any platform, whether it’s Windows, MacOS, or Linux.
Further reading: CAT tools in the content economy: A necessity, not a luxury.
Paid translation software
Translation software that users have to pay for. Most translation software is paid, although exactly what the user pays for differs largely from tool to tool. Some translation platforms, such as Smartcat, provide basic functionality for free while charging for additional, growth-oriented features.
As the translation industry is heavily dependent on the work of freelancers, language service providers have to make dozens to hundreds of payments monthly, both within and outside their own country. For this reason, manual payments can quickly become a bottleneck for growth. Payment automation allows LSPs to save time and money on payments and eliminate unnecessary paperwork as well.
Further reading: Payments in the translation industry: 6 things you want to automate.
The price a translator or an LSP charges for translating one source word. In some languages, where source words can be hard to distinguish, such as Chinese or Thai, the translation rate can be set by source character or target words instead. The median per-word rate for most languages is usually around 0.1 USD, although it varies greatly — from a lot less to a lot more — depending on the language pair and other factors like specialization and turnaround time.
(Not to be confused with Tags)
In software localization, placeholders, or variables, are bits of text that are to be replaced by other text at a later stage. For example, in the string “Hey %firstName, you have %count items in your cart”, “%firstName” and “%count” are placeholders. The exact format of placeholders depends on the internationalization system the developer uses, but it is important that the software can distinguish it from the displayable text (so just “Hey first name” won’t do).
Likewise, translators must not translate placeholders, because the software wouldn’t know what to place instead of “%vorname” in the example above. Some CAT tools, such as Smartcat’s built-in editor, allow defining placeholders so that translators can’t accidentally translate or change them.
See Plurals and genders.
Plurals and genders
The differing use of plurals and genders in different languages is one of the biggest challenges in localizing software and websites. In a very blunt example, if you are coding an online store, and an oft-displayed message says “You have %n items in your cart”, this will not be properly translatable to many popular languages, because, for example in Russian, the word for “items” will be different depending on whether there are 2, 5, or 21 items. It is therefore very important to choose an internationalization framework that will suit your expected needs.
Post-editing machine translation / PEMT / MTPE
To post-edit means to edit a machine translation so that it becomes grammatically coherent and factually correct. Usually, a “good enough” quality is expected, where “the text may sound like it was generated by a computer, syntax might be somewhat unusual, grammar may not be perfect but the message is accurate” (from TAUS Guidelines).
In computer-aided translation, to pretranslate a text means to populate the target text in every segment following a predefined rule. The text to insert can either come from a translation memory or be a machine translation of the source text. Depending on the settings, such segments can be confirmed or just inserted for the translator or post-editor to work on.
The amount of text — usually the number of source words — a translator can translate in a given time frame — usually an hour or a day. Normal productivities vary around 500–1000 words per hour and 2000–3000 words a day. Normal productivities vary between 500–1,000 words per hour and 2,000–3,000 words a day. Be cautious if a translator claims to have or has an abnormally high level of productivity — this might mean they are not doing their job with the required level of diligence and self-review.
Until recently, one of the most challenging parts in managing a translation project was monitoring progress. With desktop tools, the only way to do this was to regularly ask everyone involved and trust their words. Online tools such as Smartcat, on the other hand, allow project managers to track the progress in real time, online, and without distracting translators from their work, avoiding last-minute rush and missed deadlines.
This is similar to editing, but focuses more on the target text rather than on whether it matches the original. Also, unlike with editing, the translator will have the final say if they disagree with the proofreader’s edits. The per-word rates for proofreading vary depending on the subject matter, language pair, and the translation quality, but are generally around 30% of the translation rate, other factors being equal.
The automated checks a CAT tool makes to prevent easily detectable errors such as typos, double spacing, or discouraged terms (in conjunction with glossaries). Some CAT tools such as Smartcat do both just-in-time checks and prepare comprehensive post-factum QA reports that can be revisited by an LQA specialist after the job is done.
See Translation rate.
In computer-aided translation, repetitions are segments that appear, in their entirety, several times in a given document or across many documents (cross-file repetitions). A CAT tool will most often insert a translation for all repetitions once the first occurrence is translated, although this behavior can usually be configured.
Repository / Repo
Similar to a content management system, a repository allows storing arbitrary files while making every change and version of them trackable and identifiable. Although repositories are mainly used for software source code, it is also where developers store strings for such software. It makes integrations with repositories especially useful for setting up continuous localization processes.
A piece of text that a CAT tool handles separately from others. In common translation, a segment is most often identical to a sentence. In software localization, it can be either a sentence or a whole string, even if it includes several sentences. This behavior is often configurable. The process of splitting a piece of content into segments is called parsing.
An important part of the translation process, where the translator goes through the work they’ve done in order to weed out any mistakes and improve the overall flow. It is best to do a self-review after completing the translation in its entirety or a major part of it. (Although in the latter case it won’t hurt to do another round of self-review after everything is finished.)
Single-language vendor / SLV
A language service provider that only translates to one target language. Initially, most LSPs were single-language vendors based in specific geographies/locations. Today, most of these have become multi-language vendors, which is becoming easier to do without major investments thanks to translation platforms like Smartcat.
A branch of localization for computer software which deals with challenges such as a heavy use of placeholders, character limits, plurals and genders, and sometimes lack of context. Software localization often requires continuous localization practices, especially for software developed using Agile approaches.
Source and target
In translation, “source” and “target” are adjectives that refer to the original text and its translation(s), respectively. Examples of compound terms are source and target words or source and target languages.
See Language pair.
See Source and target.
A specific subset of translation activities that a translator or an LSP is especially good at. This can be defined broadly, e.g., technical, medical, legal, or marketing translations, or more specifically, e.g., blockchain and cryptocurrency translations. A translator may have several specializations, but it is commonly accepted that the narrower a translator’s specialization is, the better they will generally be at it.
Splitting and merging segments
In computer-aided translation, to split a segment means to break it into two. This can be useful when the CAT tool incorrectly parses two sentences as one segment (e.g., if a space is missing between two sentences, or if you want two sentences in a string to be regarded as two segments). To merge segments means, on the contrary, to join two or more segments into one. Similarly, this can be useful when the CAT tool incorrectly breaks one sentence into several segments (e.g., if it treats the period after an initial in a name as the end of the sentence).
Two things to be wary of when splitting and merging segments:
- The CAT tool will store the segment’s source and target text in the translation memory in their final, confirmed form. If the original, non-split/merged text appears in many documents, pretranslating them with that translation memory can be challenging afterwards.
- In some CAT tools, if you split/merge a segment in a multilingual project, this will be applied to all languages involved. If the translators of other languages have already completed their translation of that segment, these translations will disappear (although they can be retrieved from the translation memory). Some CAT tools, such as the one in Smartcat, allow project managers to prevent individual contributors from merging/splitting segments for that reason.
In localization, a string is a (usually short) piece of text that is displayed at a certain point and place (in a website, app, video game, etc.). Strings are replaced with text in specific languages by the internationalization framework. For example, the “Create an Account” message you see when navigating to www.smartcat.ai is taken from the key “create_an_account” (or something similar) and the language “English” and will be replaced with “Konto einrichten” once you switch to German.
String length constraint
See Character limit
A style guide is a document detailing the approaches and conventions a given company or translation agency requires their translators to adhere to. This may include using formal vs. informal language, spelling preferences (e.g. American vs. British English), avoiding the passive voice, and so on. Having a style guide in place is paramount to ensuring consistency across company content.
Subject matter expertise
A translator’s knowledge of a specific domain. Solid subject matter expertise is critical for technical, legal, and medical translations, where it can sometimes be more important than translation skills.
To subtitle a video means to create a transcript of it, where specific bits of text are linked to specific time points in the video. Subtitles are widely used in video translation. The most popular subtitle format is SRP.
A document translated and sealed by a translator certified by a governing body, i.e. a sworn translator. In most jurisdictions, a sworn translation is regarded as an official document.
(Not to be confused with Placeholders)
In computer-aided translation, tags are non-textual elements that usually represent some formatting or unusual characters present in the original document. For example, in the Smartcat screenshot below, tags — the orange pentagons — represent hyperlinks.
Further reading: Tags — Smartcat Help Center.
See Language pair.
See Source and target.
Translation of texts pertaining to specific areas of engineering, information technology, and/or science. Just like medical and legal translation and unlike marketing translation, it requires a high degree of subject matter expertise and favors factual correctness and precision over style and flow.
In the translation industry, “TEP” stands for “Translation, Editing, Proofreading”, a workflow where three specialists work on a given text, respectively a translator, an editor, and a proofreader. A TEP service usually costs 1.5 to 2 times higher than “mere” translation but results in the highest quality possible. Some CAT tools, such as the one included with Smartcat, allow all the three stages to be done at the same time within the same context, reducing the overall time spent managing the project.
SDL Trados Studio, commonly known as just Trados, is one of the most popular and oldest desktop tools for computer-aided translation. As of the late 2010s, Trados holds a ~70% share in the CAT tool market, although its popularity is increasingly threatened by the advent of newer and more advanced translation platforms. The software is developed by SDL plc, one of the world’s largest language service providers.
A somewhat controversial term that is sometimes used to describe a subform of marketing translation. In transcreation, instead of translating the source text, the translator, or “transcreator”, rewrites it completely to convey the same “subconscious” idea without sticking to the original. Examples include translating slogans like “Just do it” or “I’m lovin’ it”. Proponents of the term consider a “transcreator” to be more of a copywriter than a translator. Opponents argue that any good translator is also a writer and hence no special distinction is required for “transcreation”. Unlike other forms of translation, transcreation is often charged by the hour, not by the word.
Further reading: Transcreation: a fad or the future of marketing translations?.
See Video transcription.
See Language industry.
Translation industry events
Translation job boards
A place, usually a website or a web page, where customers post vacancies for translators. Depending on the platform, a job board can be a simple bulletin-like list, such as the one on ProZ, or an integral part of a larger workflow solution, like on Smartcat. A functional opposite of a job board is a translation marketplace, where customers search translators and request work from some of them.
Translation management system / TMS
(Not to be confused with TM)
A type of software that translation project managers can use to have an overview of existing translation projects, configure integrations with content management systems and CAT tools, and payment automation systems. Some tools, such as Smartcat, combine TMS, CAT, and other functionalities, to host the whole workflow on a single platform.
In a specific, platform-related sense, a translation marketplace is a website or a web app where customers can find freelance translators. Translation marketplaces can be either standalone, such as the one on Fiverr, or integrated in a larger translation workflow, like on Smartcat.
Translation memory / TM
A database within a CAT tool that allows translators to reuse existing translations. When a translator comes across a sentence that is exactly the same or similar to a sentence stored in the TM, this is called a “TM match”. Matches can be exact — where whole phrases had the same exact wording — or fuzzy, where the past and current phrases vary by a word or two. The extent of a match is usually represented by a percentage, which can vary between 50% and 102%.
Translation memory discounts
Some language service providers offer discounts when an order contains a considerable amount of text that has already been translated. This is usually measured by the extent of the fuzzy matches. For example, if a ten-word sentence differs in just one word from one that has already been translated — a 90% match — this might result in as high as a 90% discount. Some translation automation tools such as Smartcat allow for a flexible configuration of TM discounts:
Translation memory matches
See Translation memory.
Translation memory statistics
See Weighted words.
In the broadest possible sense, any web platform that provides functionality fully or partially related to translation. This includes simple online CAT tools such as MateCat or Memsource, translation management systems such as Protemos, as well as all-in-one platforms such as Smartcat.
A set of documents (or other content pieces) to be translated to one or a number of target languages. This usually involves a soft or hard deadline and a set of additional requirements/specifications from the customer.
Translation project management
A person who manages a translation project by moving it through numerous and sometimes intersecting and interlocking steps:
Although these tasks could be divided among several people and roles, the project manager is the one ultimately responsible for delivering the project to the customer on time and at a high standard. For this reason, it is critically important for the project manager to use all the tools available to minimize human errors and increase productivity.
Further reading: Translation Project Management on Autopilot whitepaper.
Translation software licensing
The way in which a translation technology provider measures usage and subsequently charges its users. Originally, the most common method to license translation software was by counting the number of simultaneous users, or so-called “seats”. However, some vendors, such as Smartcat, believe that this approach is detrimental to the language industry, where most contributors are freelancers, which means the number of users can be unstable and hard to predict.
Further reading: Why we don’t count seats: On licenses, monkeys, and cognitive biases.
Translation vendor management
See Vendor management.
“Proof” of a translator’s skills by an agency or association. Although most customers do not require translators to be certified, some do, and many prioritize these. In certain specializations, though, such as judicial translation, certification is mandatory (see also Sworn translation). The most widely known and reputable translator certification is provided by the American Translators Association.
Education to gain translation skills. Although higher education in translation studies does exist in many countries, most  professional translators in business domains (as opposed to literary translators) are self-taught. Another form of translation education is continuing professional development, which can involve learning translation skills in general or focusing on a specific subject matter.
The amount of money you can make as a translator varies depending on your region, experience, and qualifications. In general, you can expect it to be comparable to that of highly skilled professionals, and amounting to mid-five-figures in developed economies.
Further reading: How much can I make as a translator?.
A practice where a customer — usually a translation agency — asks a translator to do a short, 200–300-word translation test for free before adding them as an approved vendor. Although some criticize the practice, arguing that it leads to over-selecting less experienced translators, it is widely adopted in the industry. Tests are usually assessed by in-house or outsourced language leaders.
Useful links: Build Your Own Team — Smartcat.
Associations that have the goal of advancing the interests of translators, either worldwide or in specific countries, while helping customers find high-quality translators from among their members. The biggest translators associations worldwide are the American Translators Association, the British Institute of Translation and Interpreting, and the German Federal Association of Interpreters and Translators.
Details about a translator’s professional background that testify to their skills. This can include their education, work experience, certification, etc. Along with a portfolio and references, credentials are one of the main factors on which customers select translators, especially when they need a high level of experience. A rule of thumb for compiling a list of credentials is to include only those relevant to the translation industry and your subject matter expertise. For example, mentioning a past position in a clothing shop might be useful if you specialize in fashion translation but would be odd for a medical translator.
A collection of past work a translator chooses to show to their potential customers as proof of their skills. Along with credentials and references, the portfolio is one of the main factors on which customers select translators, especially when they need high expertise. A good portfolio is short but informative and only includes examples relevant to the translator’s specialization. For example, it would be a mistake for a technical translator to include translated poetry in their portfolio.
A vendor manager in an LSP or, less frequently, a localization teamis responsible for building and maintaining a database of vendors — freelancers and (smaller) LSPs. Depending on the setting, their work may also involve testing vendors before vetting and onboarding them with company processes.
Further reading: Top 5 barriers to good vendor management in the translation business.
Video game localization
A branch of software localization dealing with video games. It is distinctive in that it requires a translator to be both tech-savvy enough to handle the typical software localization challenges and creative enough to avoid literal translations. It is also challenging because end users of such translations, i.e. gamers, rarely shy away from pinpointing amusing mistranslations.
On the upside, video game localization is a highly rewarding translation specialization, both emotionally and financially.
Further reading: 10 main challenges of video game localization.
To transcribe a video means to create a text document containing the speech in the video. Depending on the goal, transcription can be complete, including filler words, or condensed to the most important things said. The former case is more often used in subtitling, while the latter can be useful when preparing interviews and customer testimonials.
Translation of speech that appears in films, TV shows, commercials, and the like. Due to the growing amount of video content produced, video translation is becoming an increasingly demanded specialization. Video translation most often uses video subtitles as the source material, although it can be done by listening to the audio if no subtitles are available. In the latter case, it is usually charged per minute of source material.
A form of localization that deals specifically with websites. Although it shares many properties with other forms of localization such as software, mobile app, or video game localization, it involves a set of features that require specific approaches such as the heavy use of CMS integrations or a large number of target languages.
Further reading: How to translate your website content to reach the global marketplace.
The reduced word count of the source text based on translation memory discounts. For example, if there are 1,000 words in a text, and there are 75% translation memory matches for 400 of them, the weighted word count will be 600 × 1 + 400 × 0.3 = 720 (assuming a 70% discount for 75% matches). The breakdown of a project by the number of words in each match category is sometimes called translation memory statistics, analysis statistics, an analysis report, etc., depending on the CAT tool you use.
See Literal translation.
Words per hour / words per day
In translation project management, the term “workflow” has two meanings. The first one refers to the sequence of all stages involved in a typical translation project. An example workflow can include pulling content from a content management system into a CAT tool, assigning translators, pushing the completed translation back to the CMS, and making payments to the translators. In the second, more specific sense, “workflow” means the steps of translation itself. Examples include translation alone, machine translation + post-editing, or translation + editing + proofreading. Modern translation platforms such as Smartcat allow for flexible workflow configurations depending on each project’s needs.