TAUS Quality Dashboard

From Quality Evaluation to Business Intelligence

March, 2016

1

About this document

This document describes how the TAUS Dynamic Quality Framework (DQF) generates a Quality Dashboard serving the interests of all stakeholders in the global translation industry. Through easy-to-use plug-ins translators and managers share data and reports that give them valuable statistics, benchmarking and analytics.

The Quality Dashboard is developed for and with the TAUS members and users.

COPYRIGHT © TAUS 2016

All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system of any nature, or transmitted or made available in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of TAUS. TAUS will pursue copyright infringements.

In spite of careful preparation and editing, this publication may contain errors and imperfections. Authors, editors, and TAUS do not accept responsibility for the consequences that may result thereof.

Published by TAUS BV, De Rijp, The Netherlands

For further information, please email info@taus.net

2

Table of Contents

1. Background 4

2. TAUS Industry Program 7

3. The Principles Underlying the TAUS Quality Dashboard 8

4. Benefits 10

5. From Quality Evaluation to Business Intelligence 11

6. What is the Quality Dashboard 13

7. Implementing the Quality Dashboard 18

8. How to Get Started 20

9. Appendix 21

3

1. Background

Since it was founded in 2005, TAUS has always followed the industry trends, providing guidance to the stakeholders in the translation industry. Originally started as a Think Tank, TAUS has gradually expanded its services to the general language service industry, first focusing on automated translation and in the last five years on quality evaluation. TAUS is currently working on the development of the Quality Dashboard – an industry-shared platform for measuring productivity and quality and turn data into business intelligence. This section briefly outlines the evolution that TAUS has undergone in the past few years, focusing on quality evaluation, and highlights the main achievements and milestones up to the present days.

2011 Old school: ‘one size fits all’ Translation quality evaluation is problematic. In 2011, TAUS conducted a survey among its members. We found that despite very detailed and strict error-based evaluation models the satisfaction levels with both translation quality and the evaluation process itself were very low. QE models are static, that is, there is a ‘one size fits all’ approach. Little consideration is given to multiple variables such as content type, communications function, end user requirements, context, perishability, or mode of translation generation (whether the translation is created by a qualified human translator, unqualified volunteer, a machine translation system or a combination of these.)

Dynamic Quality Framework In November 2011, TAUS published the foundational report for the Dynamic Quality Framework (DQF). (Download this report.) The solution proposed in this report was a dynamic evaluation model that takes into account the changing landscape of diversification in content types and the adoption of automated translation technologies.

Content Profiling In collaboration with members and researchers from Dublin City University, TAUS designed a model for profiling content types based on three vectors: utility, timeliness and sentiment (“UTS”). The UTS scoring started to give users guidance on methods for translation quality evaluation.

2012 DQF Knowledge Base In 2012, TAUS started developing a Knowledge Base for everyone working in the global translation industry documenting different approaches to translation quality evaluation.

4

The models described in the knowledge base are: adequacy (accuracy), fluency (readability), usability, community feedback, error typology, productivity measurement. The knowledge base contains use cases, templates, metrics and specifications. In consultation with users and members, TAUS developed best practices for post-editing, sampling and different quality evaluation techniques that are all part of the knowledge base. The knowledge base can be accessed from the TAUS Academy section of the website.

2013 DQF Error Typology A vast majority of the providers and buyers of translation services manage their quality program with an error typology template. The LISA QA model which was developed in the nineteen-eighties forms the basis of the error categories being applied in most cases. The SAE J2450 is another metric that is well-known in automotive translations. TAUS worked with Dr. Sharon O’Brien to develop a more up-to-date version of these error typologies and made it available under DQF. In 2014, DFKI published the MQM (Multidimensional Quality Metrics). Under the European funded project QT21, TAUS and DFKI have harmonized the DQF and MQM error typologies.

DQF training program In 2013, the Localization Institute started offering a quality management training program featuring DQF. In cooperation with TAUS, Willem Stoeller has now developed a much improved and more focused DQF training program is offered online through the TAUS web site. Through this course, industry professionals will be trained in the DQF methods and best practices ensuring a consistent usage of the evaluation tools.

2014 DQF Tools On request of its members, TAUS developed various tools for quality evaluation and made them available through the web site. Since early 2014, users can go to https://dqf.taus.net/ to do a number of quality evaluations: adequacy, fluency, error review, productivity measurement, MT ranking and MT comparison.

Free access to DQF for students Since December 2014, TAUS offers translation students at universities around the world access to all TAUS Data and resources, including the DQF knowledge base and tools through the Free Academic Membership Program.

2015 Quality Dashboard and DQF API After the adoption of DQF Tools by over a 100 companies, users expressed interest in accessing DQF from within their translation technology of choice. Therefore, TAUS set forth in developing an API that would enable users to connect directly to DQF. More and more data were sent to DQF which required a way for users to access them. This is how the Quality Dashboard came into existence. The Quality Dashboard represents the platform where users can access their data and gain business intelligence to further improve their processes.

DQF Users and User Group An active user group has formed around DQF, allowing the TAUS team to constantly improve and fine-tune their services. An early user group centered around the DQF tools, which were designed for incidental quality evaluations or for the comparison of the quality of MT engines.

5

After developing the TAUS API and a number of integrations of DQF into existing translation technologies, a new community has been formed around DQF, bringing together all TAUS DQF users to share best practices, give feedback and put forward suggestions for further developments. The DQF User Group can be found on LinkedIn.

DQF as industry reference Quality evaluation is one of the most pressing themes in the translation industry. Operators realize that, unless they start relying on common metrics and using the same methods and tools, they run the risk of comparing apples with oranges and taking uninformed decisions. DQF is more and more referenced as an industry standard and best practice for translation quality evaluation.

6

2. TAUS Industry Program

TAUS is a think tank and platform for industry-shared services and resources for the global translation sector. The TAUS industry program is focused in three main areas:

The Quality Dashboard is an industry platform for statistics on translation, benchmarking translation activity and quality and analysis of translation performance and production. TAUS provides API specifications allowing translation technology providers and users of translation services to plug TAUS DQF into their work environment.

The TAUS Academy is a knowledge center where users can find training courses for post-editing and quality management with DQF, a library of reports and articles on translation automation and quality evaluation, industry best practices, use cases, technology briefings and webinars. The TAUS Academy also contains directories of translation technologies, certified post-editors and profiles of member organizations.

The Data Cloud is the largest industry-shared repository of translation data, containing more than 70 billion words in 2,500 language pairs. Users can download and pool data for the training and customization of translation technologies. Members can choose to share data publicly or limit the sharing in a Private Vault. TAUS provides an API allowing the integration of the Search and Data Pooling functions in other translation work environments.

This document is about the TAUS Quality Dashboard. Translators, project and vendor managers as well as buyers of translation who are interested in using the TAUS Quality Dashboard will find all the relevant information in this document. Developers, interested in using the plug-in to integrate their technology with the TAUS Quality Dashboard can find the API specifications on the TAUS website.

7

3. The Principles Underlying the TAUS Quality Dashboard

The TAUS DQF and Quality Dashboard have been developed with some important principles in mind. These principles guide the current implementation of the Quality Dashboard and help drive the future developments and improvements of the TAUS services.

1. Industry-shared metrics Translation quality is a pressing theme in the translation industry. The diversification in content types and rapid adoption of translation technologies (including machine translation) drives the need for more dynamic and reliable methods and measurements of translation quality evaluation. The vision behind DQF is to standardize the methods and tools of quality evaluation and to aggregate the scores and measurements and make these available through industry-shared metrics. These new metrics will promote standardization within the industry in the area of translation quality evaluation.

2. QE as business intelligence Quality evaluation is costly and troublesome in many translation environments today. Measurements are often subjective and anecdotal. Industry-shared metrics will lead to more reliable measurements that give translation operators and producers (translators) useful insights that help them to adjust and improve processes. Today, for most companies quality evaluation is a necessity without remedy. The industry-shared metrics will turn quality evaluation into business intelligence steering and supporting management decisions.

3. Benchmarking Translation is often perceived as a black box. It is not always clear to the customer what is involved in the production of a translation. Language service providers apply different methods, tools, processes and metrics to assess the quality of translation, which makes it impossible for buyers of translation to make useful comparisons between services and technologies. Adoption of DQF will lead to standardization of metrics and processes. This standardization will allow buyers of translation then to compare and benchmark the productivity and quality of translation, both across their own translation resources, projects and language resources (internal benchmarking) as well as against peers in the industry (external benchmarking).

4. Democratic The Quality Dashboard promotes democratic access to data. Every stakeholder in the supply chain - the procurement officer, vendor manager, project manager, quality manager, translator - can get access to valuable and relevant reports on the Quality Dashboard. Access is regulated by user profiles making sure that everyone’s access is limited to the reports that s/he is authorized to view.

5. Community-Driven Roadmap Just like the original project of the DQF Tools, the Quality Dashboard has been developed following the needs and requirements of the stakeholders in the industry. TAUS constantly collects feedback and input from its members and users to determine the priorities for feature development.

8

6. Open Source API The key to drive industry-wide adoption of the Quality Dashboard is the API that connects the translation workflows and translation tools with DQF: the standard Dynamic Quality Framework. This DQF API defines the agreed metrics - like the harmonized DQF-MQM error typology - and specifies the data points. TAUS is making the DQF API available under an open source license. Every translation operator and technology developer has free access to the API and can easily integrate DQF in their workflow and translation tool.

7. Measure all translation output The Quality Dashboard is designed to report on the productivity and quality of all translation output, not just of machine translated content.

9

4. Benefits

TAUS DQF and the Quality Dashboard have been developed with the whole industry in mind, so that every stakeholder in the language services industry can benefit from adopting and using the DQF integration. This section briefly summarizes the main benefits for DQF users.

Industry benefits DQF is an industry collaborative program that elevates the discussion about quality to an industry level. On an industry scale, using DQF helps towards:

1. Increased credibility of the quality assurance function and process throughout the industry. This will improve the image of the industry in general and attract more customers, allowing the translation sector to grow faster.

2. Lower cost in general for all translation operators as a result of the shared investments in resources and tools.

3. Increased satisfaction with both vendors and clients by referring to standard metrics.

Business benefits Users of DQF can benefit from this industry collaborative program in multiple ways:

1. Sharing knowledge and standard references. Stepping away from the rigid outdated error typology review as the only way to measure translation quality, each individual translation operator does not have to research new methods on his own. Referring to common methods and best practices avoids confusion and disputes and ensures consistency.

2. Reduced costs. Sharing resources, tools and knowledge allows each individual stakeholder to reduce the costs of overhead in quality research and tool development.

3. Increased efficiency. Agreed best practices and shared tools and metrics reduces friction in each client-vendor relationship and makes it easier to collaborate in a multi-vendor environment.

4. Increased flexibility. Industry consensus and benchmarking on quality evaluation methods and scores allow for a more flexible use of often more efficient and less costly methods of quality evaluation.

Professional benefits While the adoption of DQF may start at larger companies, individual translation professionals are also included in the vision behind the Quality Dashboard. In particular, they can:

1. Benchmark their performance against industry averages to assess their translation throughput.

2. Benchmark their performance against their previous projects to track their improvements both with the aid of specific translation tools and without.

Keep track of their performance over time and having an automatic track history of projects which they simply executed for a customer or even create themselves whenever DQF is not already in use.

10

5. From Quality Evaluation to Business Intelligence

The DQF tools were initially designed for incidental quality evaluations of MT projects, for instance for tracking and benchmarking of translators or translation vendors from time to time, or for the comparison of the quality of MT engines. Following the requests from many users, TAUS has developed an API that allows users to make use of DQF while in their translation production mode.

The main function of the DQF integration is to enable process optimization, particularly when it comes to quality evaluation. Most industry sectors are focused on increasing the quality of their services, products and processes. For most buyers of translation and even the providers and the translators themselves, translation is like a black box. They miss the insights, the data and the business intelligence to take well-grounded decisions about processes, tools, resources, budgets.

The Quality Dashboard will help all stakeholders getting the indispensable statistics, benchmarking and analytics to take well-informed decisions. In this chapter, we list examples of the questions that different stakeholders may have and that the Quality Dashboard will be able to answer. Some examples of business-related questions from each stakeholder’s group are listed below.

Buyer’s questions 1. How are we performing compared to industry peers, in terms of production throughput

and quality?The Quality Dashboard provides the benchmarks by several categories such as language, content type, industry.

2. Should we consider using MT for support content?The Quality Dashboard allows to filter data per process (e.g. MT) and per content (e.g. support content) and also shows the average productivity and quality scores.

3. How is our translation technology ranking compared to industry?The Quality Dashboard shows the scores for different technologies, also broken down by language, industry, content types.

4. How good is our MT engine?The Quality Dashboard will show the productivity and quality scores for a given engine/technology.

5. Are we paying our vendor/translators fairly?The Quality Dashboard shows the average productivity scores (number of words per hour) for vendors/translators and for the specific vendors employed for a given project.

6. How does the quality compare between our languages and vendors?

11

The Quality Dashboard shows quality scores by the categories selected by the project manager or quality manager.

7. How efficient are own internal reviewers? The Quality Dashboard not only tracks efficiency of translators and post-editors but also of reviewers and editors.

LSP’s questions 1. What is the correlation between the productivity (number of words per hour) and the

number of edits per hour, on total and per translator? The Quality Dashboard provides the numbers and the benchmarks by vendor, language, content type, industry.

2. How did our scores change since we changed our translation technology platform?The Quality Dashboard shows the productivity and quality scores by any selected time period. It will also enable a timeline mode to monitor the evolution of performance over time.

3. Should we change to a new technology for the translation of our web content?The Quality Dashboard shows the productivity and quality scores by any selected technology and content type.

4. Where did our translators source their translated segment from?The Quality Dashboard shows distribution of segments and presents the percentage of human translation, machine translation, translation memory, post-editing.

Translator’s questions 1. How is my productivity compared to industry averages?

The Quality Dashboard provides the numbers and the benchmarks language, content type, industry.

2. How am I doing on this project compared to that project?The Quality Dashboard shows the productivity and quality scores by any project and also allows filtering by customer.

3. Am I using the best tool for the job?The Quality Dashboard shows the productivity and quality scores by technology in comparison to other technologies so that translators can possibly add more efficient tools to their translation toolbox.

12

6. What is the Quality Dashboard

The Quality Dashboard delivers on the DQF vision described here above. It is an industry collaborative platform for the global translation services sector, allowing every stakeholder within the supply chain to access the reports that are relevant for their activity. There are different levels of stakeholders in a standard supply chain. For example, in a supply chain consisting of four levels, stakeholders are the Buyer, the Language Service Providers (LSPs) and the Translators/Reviewers. In a four-level chain, we further distinguish between larger LSPs (Multi-Language Vendors, MLVs) and smaller LSPs (Single-Language Vendors, SLVs). The lines represent the connections between the different stakeholders, the dotted lines represent other possible, though less likely, connections.

The Quality Dashboard collects and displays data about productivity and quality for all projects and users across the supply chain. In other words, all the people involved in a project will be able to access the reports relevant for their activity. This will happen on the Quality Dashboard interface, where users will find productivity and quality metrics across content types, industries, processes used, technologies applied and by language pairs, among others.

The Quality Dashboard can be accessed by anyone using the URL: https://dqf.taus.net/quality-dashboard

13

1. Productivity

The landing page of the Quality Dashboard displays the average productivity score across the whole industry in terms of words processed per hour as well as the average edit rate score.

Users can then navigate through the available filters using the menu on the left hand-side. Each filter allows to highlight a specific parameter or property of the project such as the language pair, the process used, the technology involved and so forth.

These reports allow users to compare the industry averages for each of these more specific filters in terms of productivity.

Productivity is the throughput or speed expressed in the number of words per hour. This is currently the most common way of measuring translation performance. Productivity is an absolute score. The distinctive feature of the Quality Dashboard is the possibility to measure one’s own performance against these average values. In order to access the benchmarking feature, users need to be logged in and need to have used DQF for their projects.

14

These project-based attributes will be complemented by filters to track the performance at the customer and vendor level as well. These filters will enable each stakeholder to access the relevant information for their role in the supply chain. This means that a translator will only be able to access data about his/her own performance with respect to the customer(s) he/she is working for.

This data will be aggregated per customer but additional filters can apply.

Similarly, an LSP can monitor its own performance towards the companies they are working for. Because an LSP also works with vendors, they will be interested to see how their vendors performed for a specific project or aggregating data for a specific language pair. For an LSP the tracking goes both directions: towards the customers and towards the vendors.

15

2. Quality The Quality reports of the Dashboard include similar filters to the ones under productivity.

The Quality section provides stakeholders with quality-related information on their project. A quality review can consist of two main activities: flagging errors in a segment and/or correcting them. TAUS has adopted an industry-standard error typology which is the DQF-MQM harmonized model. This model is the result of a cooperation between TAUS and the DFKI institute who worked on developing the MQM error typology as a deliverable in the EU project QT21. MQM (Multidimensional Quality Metrics) is a framework for building task-specific translation metrics. It allows users to create custom metrics that can be used for various assessment purposes. By providing a master vocabulary of error types, users can describe metrics in a fully transparent fashion. MQM has been implemented in a variety of commercial and open-source tools and now the harmonized model DQF-MQM is available for quality measurement.

For a detailed overview of the error categories in the DQF-MQM harmonized error typology and severity levels, please refer to the Appendix.

Under the quality sections, users can have an overview of the number and type of errors found in translation. Every error is also associated with a severity level which determines the weight each error has in the text. Based on the final score a pass/fail threshold can be determined. Similarly to the productivity reports, quality reports will also offer a measure of the changes made to the translation in the form of “Correction Distance”.

16

The Quality Dashboard is developed for and with TAUS members and subscribers. QE Summits are organized to bring users together and agree on best practices and new features and support for the Quality Dashboard. They take place twice a year, hosted by TAUS member organizations. TAUS also organizes bi-monthly user group meetings and webinars for new users. On the TAUS website, users can benefit from an extensive knowledge base and community discussion forum.

The Quality Dashboard will follow the needs of the industry and continuously add new tools and features. The release of a Quality Control environment (based on the harmonized DQF and MQM error categorization) with the possibility for companies to do error annotation and correction is on the roadmap for 2016. There will also be an API for the Quality Control tool to allow members and subscribers to plug this service into their common translation management environment. Additional new features will be announced in the course of the year.

17

7. Implementing the Quality Dashboard

The Quality Dashboard and its reporting rely on the data supplied by users of translation technologies. Each technology can be linked to the Quality Dashboard by means of the TAUS API. This API is developed so that the required data can be collected from the translation system and sent to the DQF server for the reporting part. The API is released under and Open Source license so that anyone can use it to connect their tool to the TAUS DQF.

API specifications are available on the TAUS website and additional support for the integrators will be provided directly by the TAUS team or TAUS partners. If you have any integration-related questions, please write to dqfapi@taus.net. There is also an open GitHub repository available for developers. Just contact TAUS to be added to the team.

Once data flow into the DQF Server, the system will recalculate the industry average scores to be displayed on the Quality Dashboard landing pages. These reports are accessible to anyone visiting the Quality Dashboard. However, for more advanced reporting capabilities users need to:

- ensure that their translation technology of choice is properly connected to the DQF API across the project supply chain

- sign up for a TAUS account in order to access more reports on the Quality Dashboard

A TAUS account is required for every user interested in accessing the reports. An account will also provide them with credentials to use their translation technology in combination with the TAUS DQF. The TAUS account can be of two types: a subscription or a membership.

A user-subscriber is generally an individual working as a freelancer or a small-sized company. The first subscription level (called “Basic”) is free and gives access to a limited number of reports which only include industry average scores.

The professional subscription level enables individual users to benchmark their performance on a number of reports. This subscription level is meant for translators, consultants and individual users.

With the Business subscription, companies will have access to the full content of the Quality Dashboard both in terms of reporting and benchmarking.

The paid subscription plans give users access to additional TAUS resources such as DQF Tools, TAUS Data Cloud and reports.

The TAUS Membership offers companies additional benefits compared to the subscription plans, such as discounts for TAUS events and partner events, increased visibility for the company and a more advantageous pooling ratio for the TAUS Data Cloud. The TAUS Membership fee is calculated on the number of user log-ins required. The Basic level comes with 1 user account, but can be upgraded by the Member, depending on how many user log-ins they need.

The subscription plans and the TAUS Membership will determine the types of reports a user is entitled to see and use. It is useful to remember that the reports a user will see via their paid account will still depend on their respective role within the supply chain, as described in the previous section. For example, a small SLV with a business account will see the reports

18

relating to the company’s activity, i.e. the company performance towards their customers (buyers or larger LSPs) and their vendors (translators). Similarly, two translators will only get to access data about their own performance towards their customers but if they have two different subscription levels (i.e. basic and professional), the former will have a more limited access to reports than the other translator with a professional subscription.

19

8. How to Get Started

There are several ways to get started using the TAUS DQF and the Quality Dashboard. Please see below to learn how to generate reports in the TAUS Quality Dashboard or integrate the TAUS DQF API in your translation technology.

For Users You can access the TAUS Quality Dashboard here and obtain free reports with industry average data on productivity. To see more advanced and personalized reports and benchmark your own data against industry average, you need to have a TAUS account. If you do not have an account with TAUS yet, you can subscribe here.

Visit the DQF users page to find your translation tool(s) of choice and follow the instructions on how to get started. Once you start using DQF for your projects, the TAUS Quality Dashboard will generate productivity and quality reports for you.

Should your tool(s) not be listed, please let us know. Our Member Services team is always available to help you and answer any questions you may have at memberservices@taus.net.

For Developers TAUS welcomes any translation technology provider who would like to integrate DQF into their CAT tool or Translation Management System. You can obtain more detailed instructions on how to become an integrator in the "For Developers" section of the TAUS website.

You can obtain the free TAUS DQF API specifications directly from the TAUS website or from Github. Please write an email to dqfapi@taus.net if you have any questions on how to get started. We are happy to welcome you to the developers community.

20

9. Appendix

Theappendixprovidesadetaileddescrip2onofeacherrorcategoryoftheharmonizedDQF-MQMerror-typologyincludingdefini2onsandexamples.Thesecondtableexplainsingreaterdetailthedifferentseveritylevelsthatcanbeappliedtoeacherrorcategory.

Harmonized DQF-MQM Error Typology

ID

High-levelerrortypes

Granularerror-types Defini7on Example

1 Accuracy

Thetargettextdoesnotaccuratelyreflectthesourcetext,allowingforanydifferencesauthorizedbyspecifica2ons.

Transla2ngtheItalianword'canali'intoEnglishas'canals'insteadof'channels'.

11 Addi7onThetargettextincludestextnotpresentinthesource.

Atransla2onincludespor2onsofanothertransla2onthatwereinadvertentlypastedintothedocument.

12 OmissionContentismissingfromthetransla2onthatispresentinthesource.

Aparagraphpresentinthesourceismissinginthetransla2on

13 Mistransla7onThetargetcontentdoesnotaccuratelyrepresentthesourcecontent.

Asourcetextstatesthatamedicineshouldnotbeadministeredindosesgreaterthan200mg,butthetransla2onstatesthatitshouldbeadministeredindosesgreaterthan200mg(i.e.,nega2onhasbeenomiMed).

14 Over-transla7on

Thetargettextismorespecificthanthesourcetext

Thesourcetextreferstoa“boy”butistranslatedwithawordthatappliesonlytoyoungboysratherthanthemoregeneralterm

15 Under-transla7on

Thetargettextislessspecificthanthesourcetext

Thesourcetextuseswordsthatrefertoaspecifictypeofmiltaryofficerbutthetargettextreferstomilitaryofficersingeneral

16 UntranslatedContentthatshouldhavebeentranslatedhasbeenleRuntranslated.

AsentenceinaJapanesedocumenttranslatedintoEnglishisleRinJapanese.

17 ImproperexactTMmatch

Antransla2onisprovidedasanexactmatchfromatransla2onmemory(TM)system,butisactuallyincorrect.

ATMsystemreturns“PresstheStartbuMon”asanexact(100%)match,whenthepropertransla2onshouldbe“PresstheBeginbuMon”.

2 Fluency

Issuesrelatedtotheformorcontentofatext,irrespec2veastowhetheritisatransla2onornot.

Atexthaserrorsinitthatpreventitfrombeingunderstood.

21 Punctua7onPunctua2onisusedincorrectly(forthelocaleorstyle)

AnEnglishtextusesasemicolonwhereacommashouldbeused.

22 Spelling Issuesrelatedtospellingofwords TheGermanwordZustellungisspelledZustetlugn.

23 Grammar

Issuesrelatedtothegrammarorsyntaxofthetext,otherthanspellingandorthography. AnEnglishtextreads“Themanwasseeingthehiswife.”

24 Gramma7calregister

Thecontentusesthewronggramma2calregister,suchasusinginformalpronounsorverbformswhentheirformalcounterpartsarerequired.

AtextusedforahighlyformalannouncementusestheNorwegianduforminsteadoftheexpectedDe.

25 Inconsistency Thetextshowsinternalinconsistency. Atextusesboth“app.”and“approx.”forapproximately.

21

26 Link/cross-reference Linksareinconsistentinthetext

AnHTMLfilecontainsnumerouslinkstootherHTMLfiles;somehavebeenupdatedtoreflecttheappropriatelanguageversionwhilesomepointtothesourcelanguageversion.

27 Characterencoding

3 Terminology

31 Inconsistentwithtermbase

Atermisusedinconsistentlywithaspecifiedtermbase

AtermbasespecifiesthatthetermUSBmemorys7ckshouldbeused,butthetextusesUSBflashdrive.

32 Inconsistentuseofterminology

Terminologyisusedinaninconsistentmannerwithinthetext.

Thetextreferstoacomponentasthe“brakereleaselever”,“brakedisengagementlever”,“manualbrakerelease”,and“manualdisengagementrelease”.

4 Style

41 Awkward

42 CompanystyleThetextviolatescompany/organiza2on-specificstyleguidelines.

Companystylestatesthatpassivesentencesmaynotbeusedbutthetextusespassivesentences.

43 Inconsistentstyle Styleisinconsistentwithinatext

OnepartofatextiswriMeninalightand“terse”stylewhileothersec2onsarewriMeninamorewordystyle.

44 Third-partystyle

45 Unidioma7c

5 Design

Thereisaproblemrela2ngtodesignaspects(vs.linguis2caspects)ofthecontent. AdocumentisformaMedincorrectly

51 Length

Thereisasignificantdiscrepancybetweenthesourceandthetargettextlengths.

AnEnglishsentenceis253characterslongbutitsGermantransla2onis51characterslong.

52 LocalformaSng

Issuesrelatedtolocalformadng(ratherthantooveralllayoutconcerns)

Apor2onofthetextdisplaysa(non-systema2c)formadngproblem(e.g.,asingleheadingisformaMedincorrectly,eventhoughotherheadingsappearproperly).

53 Markup

Issuesrelatedto“markup”(codesusedtorepresentstructureorformadngoftext,alsoknownas“tags”).

Markupisusedincorrectly,resul2nginincorrectformadng.

54 MissingtextExis2ngtextismissinginthefinallaid-outversion

Atransla2oniscomplete,butduringDTPatextboxwasinadvertentlymovedoffthepageandsothetranslatedtextdoesnotappearinarenderedPDFversion.

55 Trunca7on/textexpansion trunca2on-text-expansion

TheGermantransla2onofanEnglishstringinauserinterfacerunsofftheedgeofadialogueboxandcannotberead.

6Localeconven7on

Charactersaregarbledduetoincorrectapplica2onofanencoding.

AtextdocumentinUTF-8encodingisopenedasISOLa2n-1,resul2nginall“upperASCII”charactersbeinggarbled.

61 AddressformatContentusesthewrongformatforaddresses.

AnonlineformtranslatedfromEnglishtoHindirequiresastreetnumbereventhoughmanyaddressesinIndiadonotincludeahousenumber.

62 DateformatAtextusesadateformatinappropriateforitslocale.

AnEnglishtexthas“2012-06-07”insteadoftheexpected“06/07/2012.”

22

Severity levels for error categories

63 Currencyformat

Contentusesthewrongformatforcurrency.

Atextdealingwithbusinesstransac2onsfromEnglishintoHindiassumesthatallcurrencieswillbeexpressedinsimpleunits,whiletheconven2oninIndiaistogivesuchpricesinlakhrupees(100,000rupees)

64 Measurementformat

Atextusesameasurementformatinappropriateforitslocale.

AtextinFranceusesfeetandinchesandFahrenheittemperatures.

65 Shortcutkey

AtranslatedsoRwareproductusesshortcutsthatdonotconformtolocaleexpecta2onsorthatmakenosenseforthelocale

AsoRwareproductusesCTRL-StosaveafileinHungarian,ratherthantheappropriateCTRL-M(formenteni).

66 Telephoneformat

Contentusesthewrongformfortelephonenumbers

AGermantextpresentsatelephonenumberintheformat(xxx)xxx-xxxxinsteadoftheexpected0xxfollowedbyagroupofdigitsseparatedintogroupsbyspaces.

7 Verity Thetextmakesstatementsthatcontradicttheworldofthetext

Thetextstatesthatafeatureispresentonacertainmodelofautomobilewheninfactitisnotavailable.

71 Culture-specificreference

Contentinappropriatelyusesaculture-specificreferencethatwillnotbeunderstandabletotheintendedaudience

AnEnglishtextreferstostepsinaprocessas“Firstbase”,“Secondbase”,and“Thirdbase”,andtosuccessfulcomple2onasa“Homerun”andusesothermetaphorsfrombaseball.TheseprovedifficulttotranslateandconfusethetargetaudienceinGermany.

8 Other Anyotherissues

Severity1 Cri2calErrorsthatmaycarryhealth,safety,legalorfinancialimplica2ons,violategeopoli2calusageguidelines,damagethecompany’sreputa2on,causetheapplica2ontocrashornega2velymodify/misrepresentthefunc2onalityofaproductorservice,orwhichcouldbeseenasoffensive.

Severity2 MajorErrorsthatmayconfuseormisleadtheuserorhinderproperuseoftheproduct/serviceduetosignificantchangeinmeaningorbecauseerrorsappearinavisibleorimportantpartofthecontent.

Severity3 MinorErrorsthatdon'tleadtolossofmeaningandwouldn'tconfuseormisleadtheuserbutwouldbeno2ced,woulddecreasestylis2cquality,fluencyorclarity,orwouldmakethecontentlessappealing.

Severity4 NeutralUsedtologaddi2onalinforma2on,problemsorchangestobemadethatdon´tcountaserrors,e.g.theyreflectareviewer’schoiceorpreferredstyle,theyarerepeatederrorsorinstruc2on/glossarychangesnotyetimplemented,achangetobemadethatthetranslatorisnotawareof.

Kudos Usedtopraiseforexcep2onalachievement.

23