The LISA QA Model: A Historical Foundation for Quality Evaluation

Why it matters

The model is part of the history of making localization quality explicit. Instead of relying entirely on an overall impression, teams could discuss recorded findings and a defined assessment approach.

The practical issue remains familiar: a requester, translator and reviewer need a shared way to describe what went wrong and why it matters. Without that, a quality discussion can become a dispute between unrecorded preferences.

The documented connection to MQM

MQM Council says that MQM's structure drew on LISA QA while taking a more modular approach. It also describes examination of other metrics and tools. MQM therefore has a documented intellectual connection to LISA QA, but should not be described as a simple renamed edition or as proof of organizational succession. MQM history.

Reading an inherited scorecard

Suppose a team receives a spreadsheet labelled “LISA QA” with local categories, weights and a pass threshold. Our recommended first step is to identify that spreadsheet's actual provenance and configuration. A familiar label does not establish which historical version it follows or what was changed locally.

An original exercise for the team is to annotate several shared examples and compare decisions. If reviewers disagree, clarify the instructions before interpreting the resulting score as objective evidence.

This guide does not reproduce a definitive historical scorecard, assign universal LISA weights or claim that a locally adapted sheet is an official publication.

Historical value and present-day use

The historical model is useful for understanding how today's evaluation practices developed. For a new programme, begin by examining maintained references and the project's specific needs. MQM and ISO 5060 are appropriate next reading points.

Preserve historical reports if they explain past decisions, but avoid directly comparing them with new results until the assessment settings have been mapped.

AI and continuity of method

Our editorial position is that adding an AI evaluator does not revive the original model or create an official successor. The useful continuity is a commitment to explicit categories, inspectable evidence and recorded decisions. The proposed evaluator still needs validation.