Language Quality Assurance

Do you have established processes in place to monitor ongoing translation quality? As you produce new content, are you taking steps to systematically improve quality in the long-term? If you’re asking these questions, it makes sense to explore Language Quality Assurance services.

What Is Language Quality Assurance?

Translation quality can be a challenge not only to assess but to quantify. Language quality services establish processes for evaluating and providing feedback on previous translations based on established criteria. This independent review of your translated content helps you achieve long-term improvements in the overall quality of your translated materials.

Approaches to Assessing Quality

Language Quality Assurance (LQA), also called Language Quality Assessment, is the process of reviewing translated content to ensure it is fit for its intended purpose, error-free, consistent across translated content, and on brand. The scope of a quality assessment can vary. Some organizations opt for a full comprehensive quality review of 100% of their content, while others opt to spot check random samples. LQA can also include the review and maintenance of glossaries, translation memories (TMs), and style guides to ensure that an organization’s language resources are appropriate for the company and audience.

Why Establish a Process for LQA?

There are many reasons why organizations conduct language quality review based on logging and measuring errors:

• Holding Language Services Providers and internal resources accountable for their work by ensuring that translations adhere to established quality standards
• Identifying trends – both success and problem areas
• Systematically improving translation quality based on tangible data
• Establishing long-term consistency across translated materials and language assets

What Is High-Quality Translation?

First, it’s important to define high-quality translation. Translation entails accurately communicating meaning from one language (the source) to another language (the target). Good translation conveys the original tone and intent of a message, while taking into account cultural and regional differences.

High-quality translations should:

• Effectively communicate the meaning of the source text to the target audience in a manner that is both readable and comprehensible
• Follow all grammatical conventions and adhere to rules for line breaks, punctuation, alignment, and capitalization
• Use proper conventions for addresses, dates, and measurements
• Use accurate and consistent terminology
• Adhere to all language assets (TMs, glossaries, style guides)
• Reflect the style of the source material
• Adhere to current acceptable usage of the language and use the appropriate register for the target audience
• Meet all the client’s guidelines and specifications
• Take any cultural factors into consideration and appropriately adapt any cultural-specific elements

Translated text should read as though it was written in the target language. However, a translator should not add information to the source content and the style, tone, and meaning should mirror the original.

What Does LQA Catch?

LQA captures a wide variety of errors, including the following:

• Misspellings, grammatical and syntactic errors, wrong terms, omissions, inconsistencies, inappropriate word structure or agreement, and other miscellaneous mistakes
• Incorrectly localized numbers, dates, currency, units of measurement, and other numeric formats
• Any inaccuracy or inconsistency with linguistic assets, such as a glossary, style guide, or translation memory
• A translation segment that is out of context
• Formatting errors
• Tone of voice that is inconsistent or inappropriate for the intended audience
• Translation that does not adhere to the source material
• Translation that does not adhere to client guidelines or specifications
• Severity of any errors, omissions, or inconsistencies

Analytic Quality Evaluation

While original models for LQA focused on measuring quality as defined by the absence of defects (as listed above), the approach to LQA has advanced significantly. Today, Multidimensional Quality Metrics (MQM) provide a more customizable framework that identifies and catalogs those errors that are defined as relevant.

An Analytic Quality Evaluation reduces subjectivity by collecting structured data. As a first step, a language services provider works with a client to understand their requirements and establish the metrics that matter most for the content type being reviewed. For example, there may be some types of content, such as legal materials, for which style is not important, and therefore style is not used as a factor for evaluation. In other situations, such as marketing content that requires adherence to a specific tone and voice, then style is an important factor to evaluate.

Metrics are established based on content. Even within a single organization, different sets of metrics may be developed to correspond to different content types. Establishing good metrics upfront ensures that content meets business objectives and allows for root cause analysis.

Wholistic Evaluation

Unlike an Analytic Quality Evaluation, a Wholistic Evaluation does not assess specific errors. Instead, it looks at whether the content as a whole does what it is intended to do. Evaluators consider two to four broad criteria. How accurate was the translation? How effective was the layout? Does it meet style requirements? Each category is assigned a rating on a scale of 1-5. An overall passing score could indicate that the translation is good and does not need major attention. On the other hand, a lower score could indicate that the content requires a second step, an Analytic Quality Evaluation, to pinpoint specific errors. An even lower score could show that the translation is subpar and needs a major rework before any more time is spent looking at the specifics.

Ultimately, the Wholistic Evaluation serves to save time and money by doing a cursory evaluation upfront to ascertain whether the content requires additional review, and if so, what type.

Determining the Right Percentage for LQA

Some organizations opt for doing LQA on a portion of their translation, rather than performing LQA on all content. When this approach is taken, there is no hard and fast rule to determine the amount of content to be checked. If an organization is first implementing LQA, it may be beneficial to review a larger percentage of content, and the percentage may decrease over time. It’s important to consider how long spot checks have been performed as well as the consistency of the results.

What Content Types Benefit Most From LQA?

LQA is especially recommended for content that is specialized, client-facing, or subject to compliance review

Here are a few of the applications where LQA is appropriate:

• Human translation that is client-facing
• Highly specialized content
• Content with a high TM leverage
• Text that was rendered using machine translation
• Source content delivered to the translators piecemeal and out of context

Different organizations have different priorities. An LQA plan customized to specific content types provides an effective way to ensure translation quality aligns with larger business goals. At Eriksen, we work with our clients to develop customized systems for identifying and measuring errors in a manner that adheres to their priorities and objectives. Get in touch to discuss your organization’s quality assurance strategies.