Should You Base Translation Quality on Accuracy or Preference?

When it comes to localization, achieving high translation quality is crucial for getting your message across in a clear manner. But what exactly does “quality” mean in this context?

According to the multidimensional quality metrics (MQM) framework, which is the industry standard for measuring translation quality, it includes proper use of terminology, linguistic conventions, locale conventions, and more. Translation accuracy is also a key consideration. Today, we’ll talk about both why that’s the case and how you can ensure that your translations don’t miss the mark.

How translation accuracy is assessed (and the role of preference)

Say you were creating content for legal purposes. Accuracy would be paramount—you couldn’t afford to have incorrect translations, missing punctuation, or grammatical errors that would cause confusion or change the meaning of the text.

The same is true of other translation use cases. Readers must be able to understand the meaning of the source language, whether informational or instructional.

Assessing whether content is easily understandable and actionable starts with knowing the intent of your source text. Is it meant to direct visitors through your website? In that case, a missing comma might not matter much as long as people can still navigate effectively on-site. Whether or not that missing comma constitutes inaccuracy may be left up to the preference of individual project managers, translators, reviewers, and editors.

The opposite would be true, though, if you were translating critical healthcare information or legal documents with which the consequences of errors are severe. Instances like this leave little room for opinions on voice or tone, word choice, or even grammatical preferences (think: Oxford comma).

Ultimately, though, we believe teams should remove opinions from translation review at all costs. There needs to be a balance between individual preference and constant revisions. The goal of translation review isn't to input an individual interpretation into the review. It’s to objectively evaluate the quality of the translation. Knowing what factors impact translation accuracy and overall quality will help you do just that.

3 common factors that impact translation accuracy

All of the following have a bearing on the quality of your translations:

1. Source language and cultural complexities

Some languages are simpler to translate into than others. For example, English to Spanish is easier than English to Chinese. Errors are more likely to happen with complex language pairs that have significant differences in alphabets and writing systems, grammar rules, word order, length, and so on. Not only do languages differ, but the cultures associated with each also vary. This means cultural references, idioms, and slang can be tricky to translate in a way that sounds natural in your target language.

2. Uncommon or industry-specific vocabulary

Is the vocabulary used in your source text uncommon? Is it specific to an industry with a lot of related jargon, such as the legal or medical industries? In either case, translation accuracy can be impacted by one or more of the following:

• How well your translators understand the vocabulary or terminology
• Whether you have comprehensive, up-to-date glossaries and translation memories
• Whether any AI or machine translation tools you choose are well-suited for the industry or domain your text relates to

As with the other factors here, there are several things you can do to limit the negative effects of this challenge, including hiring skilled linguists.

3. Ambiguous source text

Inevitably, you’ll find that some of your source text contains ambiguities. These could include vague pronouns, unstated assumptions, or homonyms (i.e., words with more than one meaning). When such ambiguities are present, both human and AI translators are more likely to misunderstand the original details and intent, making it harder to convey the correct meaning in the target language.

Clarifying potentially unclear text before translating goes a long way in reducing errors.

And these are just a handful of influencing factors. Others include translator skill level, text length and complexity, and quality assurance processes (or lack thereof).

Leveraging technology to maintain translation accuracy

Technology is more useful than ever when it comes to ensuring the accuracy of your translations. To illustrate, here are some ways Smartling customers are benefiting from it.

Smartling Translate

Smartling Translate is a self-service translation portal through which you can instantly and securely translate files and text. It’s powered by artificial intelligence and is designed to improve the accuracy and fluency of translations.

It can be used for both pre- or post-editing of text to improve grammar, spelling, and punctuation, or adjust formality. But it can also reference your glossaries and translation memory to keep terminology, style, and voice consistent and on-brand across your translations.

Neural Machine Translation (NMT) Hub

Our Neural Machine Translation Hub is a secure workspace for managing your translation needs at scale. You can translate text via more than 10 of the best machine translation engines, including Microsoft Translator, DeepL, and CustomMT. Smartling Autoselect will automatically choose the best one for your content. This multi-engine machine translation approach results in up to 350% higher quality translations.

Quality can be even higher if you opt to use a custom engine that’s trained on your business, industry, or domain. This will help ensure proper use of branded and industry-specific terminology, and ensure that translated text retains its original meaning.

Automated quality checks via TMS

Both the NMT Hub and Smartling Translate are part of our end-to-end translation management system (TMS), which has been voted the #1 enterprise TMS by G2 users for years.

One of many ways it enables you to produce accurate translations consistently is via quality checks. Our software can automatically identify spelling, grammar, and punctuation errors, check for glossary compliance, and more. You can even create custom quality checks to monitor and evaluate other aspects of translation that are important to you.

What are the consequences of low translation accuracy?

Why consider using technology and putting other measures in place to combat the various factors that can hurt translation accuracy? There are several reasons:

• Your audience may not understand your content, leading to frustration, inability to accomplish key tasks, and long-term damage to your reputation.
• Even if they can understand it, your message may not resonate with your target audience and, therefore, fail to contribute to goals such as driving sales.
• Your translation costs may skyrocket if you frequently have to redo translations. You’ll also waste intangible resources like time.

Maintaining high accuracy doesn’t have to be a chore

There’s plenty you can do to make sure your translations are clear, correct, and, if need be, actionable. That includes working with professional translators who are native speakers of your target language and, ideally, have expertise in the relevant culture, industry, or content types in question. Clearing up ambiguities before translators get started is also a smart precaution to take.

You can also use technology like Smartling’s to accomplish a couple of things:

• Improve translation accuracy upfront, such as through the NMT Hub, which autoselects the best machine translation engine and references your glossaries and translation memory to deliver on-brand, high-quality translations consistently.
• Make quality assurance processes more efficient and prevent as many issues as possible from slipping through the cracks via quality checks.

To get a better idea of how Smartling can improve not only your workflows but also the accuracy of your translations, watch our quick demo. We’d be happy to answer any questions you have afterward.