High-Level Structure

Feedback on these documents is welcome and comments can be sent to feedback [at] qt21 [dot] eu

Last update: 2014 November 21

This document provides an informative discussion of the high-level structure of MQM. For a full description of MQM, please consult the online MQM definition. In the event of any discrepancy between the online definition of MQM and this document, the content of the online document should be considered authoritative.

High-Level Structure

MQM issue types are divided into five primary categories:

  • 1. Accuracy. Accuracy issue relate to the relationship between the source and target language meaning. For example, if an English text translates the Italian word ape (‘bee’) as monkey in a phrase that reads “A monkey likes honey,” the output is entirely fluent, but it does not accurately reflect the content of the source text. Accuracy issues are ones that change the meaning conveyed by the source text.
  • 2. Fluency. Fluency issues relate to the text itself, regardless of its relation to another text. For example, “I seed two man tomorrow” has a number of grammatical problems, regardless of its relation to another text (unless it is accurately attempting to reflect a problem in another language). Fluency issues are assessed without reference to a source or target text. With the exception of the category Unintelligible, which is used for instances in which a major break-down in fluency makes it impossible to identify the particular nature of a problem, Fluency issues fall into two subcategories:
    • 2.1. Content. Content issues relate to the propositional value of the text, rather than mechanical issues. For example, use of an incorrect term or a stylistically inappropriate formulation are considered Content issues.
    • 2.2. Mechanical. Mechanical issues relate to the form of the content rather than its propositional value. Spelling errors, grammatical errors, use of an incorrect date format, and other similar errors are Mechanical issues.
  • 3. Verity. Verity addresses the relationship of the text to the world. For example, if a text fluently and accurately translates a legal claim that cannot be made for the target locale and audience, it will still be unsuitable for the target audience. Similarly, if an American English text dealing with electrical appliances states that a ground wire will be bare copper, accurately translating this content into German will cause a problem as German ground wires have green and yellow striped insulation. Verity issues are typically not the fault of the translator, but still need to be addressed in the translated product.
  • 4. Design (mono- and/or bi-lingual). Design issues relate to the physical presentation of the text in a situation where design is considered significant. They do not relate to the text itself, but rather to how it is displayed on the page, screen, or other output device. For example, the wrong font or color choice are Design issues. (Note that Design issues may, in some instances, have semantic impact, but if the text itself is fluent and accurately translated, such cases are still treated as Design issues.)
  • 5. Internationalization. This branch deals with issues in the form of the source content that create problems for translation. For example, if graphics have embedded text that cannot be translated, the problem is with the internationalization of the source, not with the translator. At present this branch of MQM does not have any defined sub-types, but it is anticipated that future updates of MQM will expand this branch.

MQM also includes a branch for legacy issues:

  • 6. Compatibility. Compatibility issues are those issues that are retained from previous metrics (notably the LISA QA Model) that do not otherwise fit in MQM, either due to granularity or because they address process issues that would not normally be covered in MQM. For example, Application compatibility, which checks whether a localized product works with other required applications, would not normally be addressed in MQM, but is retained to ensure compatibility with the LISA QA Model. Note: Issues in the Compatibility branch are deprecated and should be used only in cases where compatibility with legacy metrics is required.