|
In this issue…
Towards Objective QA
Version 2.0 of the LISA Quality Assurance Model
Created by and for leading localization professionals, the LISA QA Model offers a standardized quality assurance model for product localization covering everything from documentation, help and software through to packaging and CBT tutorials. Version 2 of the model, produced by the LISA Quality SIG under the chairmanship of localization service providers L&L Informatie, was recently released. In the following article, based in part on material appearing in a forthcoming book by John Benjamins [1], Siu Ling Koo explains the basis for the model, and how it is applied in practice. IntroductionMany aspects of the localization process - and especially the translation of screens, user manuals and the like - are not easily judged on quality, and subjective judgements influence the final opinion of a product. In practice, this can lead to uncertainty and unnecessary conflict between vendors and clients, as well as potentially substandard results. Explicit quality assurance programs are therefore needed to establish required quality levels and quality assurance procedures.
Figure 1: Sample LISA QA Model Form The LISA QA modelThis was the rationale behind the development of the LISA QA Model. Version 1.0 was published in 1995 by a LISA Special Interest Group headed by Teddy Bengtsson of Oracle and drew on quality metrics provided by a number of major companies (Microsoft, Digital Equipment, Rank Xerox, IDOC Europe, DLS and IBM). The authors of the Model took the best and most appropriate areas of these metrics and produced guidelines as to how they could be applied, both generically and in practice. The aim was for the Model to enable users to consistently provide localized products that met the needs and expectations of customers, and to help vendors achieve the quality of required for each territory or market. The results of their work comprised checklists for formatting QA, functional QA and language QA, along with a reference manual and QA form templates. The newly released Version 2.0 has updated and revised the original documents and expanded the scope of the Model to include specific issues applicable to Asian languages. In addition, it has added a section entitled "case studies and client requirements", which features Microsoft quality requirements, and a second reference manual detailing the changes to the model and describing their development. Instructions for using the new templates (for use in sampling product functionality, documentation and language quality) are also included. Contributors to the second version include Bellcore Learning Support, Hieronymous, International Communications, Language Management International, Microsoft, Nokia, One World, Oracle, Prolangs, SDL, Augur and Compuware. Putting the Model into practiceQuality assurance as defined by the LISA QA Model revolves around the idea of sampling rather than 100% testing ("quality control"). This is necessary given the size of localization products and the severe restrictions on time involved. The LISA QA Model therefore foresees that a test will be performed on a sample, i.e. a certain percentage of the total material involved. This enables feedback to be given early on, when improvements can still be made relatively easily, and also allows error trends to be identified. The Model provides a system for identifying and categorizing errors into three types (critical, major and minor) according to their severity. The existence of even one critical error means that the work is rejected; major and minor errors, on the other hand, are weighted and counted towards the maximum permissible number. This number, the distribution of errors among the different categories and the definition of the error categories themselves can all be customized in the Model, and need to be agreed in advance between supplier and client. The QA procedure applied to the selected sample produces a total number of error points, which in turn translates into in a final grade of "Pass" or "Fail". The result for the sample is then applied to the job (or task) as a whole. In other words, a grade of "Passed" for the sample leads to the acceptance of the entire piece of work, while a grade of "Fail" leads to its rejection. The proof of the puddingResults from those who using the model seem to be encouraging. In addition to the Microsoft case study mentioned in the Version 2.0 documentation, L&L has reported on its experiences in the forthcoming Benjamins book. One of the main benefits has been the ability to define and map expectations between clients and production staff. Order informationThe LISA QA Model is available to LISA members at a special price of USD 1295.00, and to non-members at a price of USD 295.00. Order forms can be obtained from the LISA administration. All registered users are published on the LISA Network. [1] Working title: "Translating into Success: Cutting Edge Strategies for Going Multilingual in a Global Age" Siu Ling Koo
|
LISA Business Data Forum Summaries and Presentations LISA Globalization Consulting Network Webinars and TouchPoint Advisory Calls LISA Forum USA LISA@Chinasoft Fair LISA Forum Asia LISA Forum Europe LISA Forum India Open Standards • TBX • TMX |
||