THE INTERNATIONAL JOURNAL FOR LOCALISATION

March 2006

Be recognised as aLocalisation ProfessionalJoin TILPwww.tilponline.org

6TH LRC SUMMER SCHOOL . . . . . . . PAGE 6

FOCUS ON RESEARCH APPLICATION DEVELOPMENT FOR REMOTE COMMUNITES . . . . . . . . . . . . . . . . . . . . . . . . PAGES 7–11

LANGUAGE COLLATION SEQUENCE —THE SINHALA PROJECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . PAGES 13–19

IGNITE IMPLEMENTS WEB SERVICES . . . . . . . . . . . . . . . PAGES 22–23

VOL. 5 Issue 1Issue Sponsored by the recently merged:

€7.50 / $7.50 USISSN 1649-2358

6TH LRC SUMMER SCHOOL . . . . . . . PAGE 6

FOCUS ON RESEARCH APPLICATION DEVELOPMENT FOR REMOTE COMMUNITES . . . . . . . . . . . . . . . . . . . . . . . . PAGES 7–11

LANGUAGE COLLATION SEQUENCE —THE SINHALA PROJECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . PAGES 13–19

IGNITE IMPLEMENTS WEB SERVICES . . . . . . . . . . . . . . . PAGES 22–23

06 March 2006: The Implied Translator, University of Manchester,UK; http://www.llc.manchester.ac.uk/Research/SeminarSeries/(weekly seminars)

06–08 March 2006: Internationalization and Unicode Conference29, San Francisco, California, USA; www.unicodeconference.org/lf

23–25 March 2006: The Third ATISA Conference, San Diego, California,USA; http://appling.kent.edu/ATISA/ATISAConferences.html

28–30 March 2006: Managing Global Websites and eCommerce Conference, San Francisco, California, USA; www.localizationinstitute.com/

04–06 April 2006: BITcon Asia: The Future of Internationalization –Intelligent Design and Evolution, Hong Kong, China; http://internationalization-conference.org/bitconasia/index.html

18–21 April 2006: LISA Forum Asia 2006, Shangai, China;www.lisa.org/events/2006shanghai/

26–28 April 2006: Localization & Globalization CSN Expo 2006,Miami, Florida, USA; www.csnevents.com/

01–05 May 2006: Multidimensional Translation – Audiovisual Translation Scenarios, Copenhagen, Denmark; www.euroconferences.info

11–13 May 2006: Trends in Natural Language Processing, MelbourneBeach, Florida, USA; www.cs.memphis.edu/~vrus/Flairs-06.html

16–19 May 2006: Translation and Cultural Exchange in the age ofGlobalisation, Barcelona, Spain; www.ub.edu/lettere/congres_2006

22–28 May 2006: LREC 2006 Conference, Genoa, Italy; www.lrec-conf.org/lrec2006

30 May—01 June 2006: Localization World: Working — Together!,Barcelona, Spain; http://www.localizationworld.com/

12–15 June 2006: 5th LRC Summer School, Limerick,Ireland; www.localisation.ie

13–14 September 2006 11th Annual LRC Conference,Dublin, Ireland; www.localisation.ie

Subscription NewsDon’t Forget! You can subscribe to

Localisation Focus online atwww.localisationshop.com.

Publisher InformationLocalisation Focus – The International Journal forLocalisation is the publication of the LocalisationResearch Centre (LRC). It is distributed to thelocalisation community and those with an interest in localisation. Please notify the LRC (email: LRC@ul.ie) if you or one of your colleagueswould like to receive Localisation Focus regularly.

Editor: Reinhard Schäler

Assistant Editor: Thomas Keogan

Production: Litho Studios Ltd., Kylemore Road, Ballyfermot, Dublin 10. Tel: +353 1 626 6123Fax: +353 1 626 9369Email: info@cosmon.ie

Published by: Localisation Research Centre, Dept. of Computer Scienceand Information Systems,University of Limerick,Limerick, Ireland.Tel: +353 61 202 881Fax: +353 61 202 734www.localisation.ie

© 2006 Localisation Research Centre

Tim Altanero, Associate Professor of Foreign Languages, AustinCommunity College, Texas, USADonald Barabé, Operations Vice-President, Canadian Bureau ofTranslation, CanadaTeddy Bengtsson, CEO of Idea Factory Languages Inc., BuenosAires, ArgentinaLynne Bowker, Associate Professor, School of Translation andInterpretation, University of Ottawa, CanadaJosé Eduardo De Lucca, Co-ordinator of Centro GeNESS and Lecturerat Universidade Federal de Santa Catarina, BrazilCarla DiFranco, Program Manager, Windows Division, Microsoft, USABert Esselink, Solutions Manager, Lionbridge Technologies,Netherlands and author of A Practical Guide to LocalizationDebbie Folaron, Assistant Professor of Translation andLocalisation, Concordia University, Montreal, Quebec, CanadaPatrick Hall, Professor of Computer Science, Open University, UKJames M. Hogan, Senior Lecturer in Software Engineering at theQueensland University of Technology, Brisbane, AustraliaSarmad Hussain, Associate Professor and Head of the Center forResearch in Urdu Language Processing at NUCES, Lahore, PakistanLisa Moore, Chair of the UNICODE Technical Committee, and IMProducts Globalisation Manager, IBM, San Jose, California, USASharon O’Brien, Lecturer in Translation Studies, Dublin CityUniversity, Dublin, Ireland

Maeve Olohan, Senior Lecturer and Program Director of MA inTranslation Studies, University of Manchester, Manchester, UK

Pat O’Sullivan, Test Architect at IBM’s Dublin Software Laboratory,Dublin, Ireland

Anthony Pym, Director of Translation- and Localisation-relatedPostgraduate Programmes at the Universitat Rovira i Virgili,Tarragona, Spain

Harold Somers, Professor of Language Engineering, University ofManchester, Manchester, UK

Marcel Thelen, Lecturer in Translation and Terminology, ZuydUniversity, Maastricht, Netherlands

Gregor Thurmair, Chief Technology Officer

Kim Wallmach, Freelance Translating and Interpreting ProjectManager and Lecturer in Translation and Interpreting, University ofSouth Africa, Pretoria, South Africa

Sue Ellen Wright, Lecturer in Translation, Kent State University,Ohio, USA

Om Vikas, Senior Director and Head of Computer DevelopmentDivision in the Ministry of Communication & Information Technology(Government of India), New Delhi, India

Angelika Zerfass, Freelance Consultant and Trainer for TranslationTools and Related Processes, Lecturer in same at University of Bonn,Germany

Members of the Localisation Focus Editorial Board

Upcoming Localisation Events

LOCALISATION FOCUSMARCH 2006 3

Cont

ents

CONTENTS.editorial

From the Editor

EGYPT’S MINISTER OFCOMMUNICATIONS AND IT recentlyannounced an agreement reached between

the LRC, Egypt’s Ain Shams University (with174,000 students) and his own Department tojointly develop localisation courses in Egypt. Theminister supports efforts by the industry to develop Egypt as the localisationhub for the Arabic world and North Africa. Following the agreement reachedearlier with the University of Florianopolis (Brazil) and ongoing negotiationswith universities and state-sponsored bodies in Asia and Africa, thisdevelopment highlights the enormous interest in development localisation andthe efforts undertaken by the signatories of the Limerick Declaration,establishing the Global Initiative for Local Computing (GILC).

Much has been reported about the merger of the giants, Lionbridge andBowne Global Solutions. Much more interesting though than the merger is themedium- to long-term effect this merger will have on the rest of the industry. Inthis issue, we report on another merger, on a smaller scale, but no lesssignificant. Welocalize and Connect Global Solutions have decided to combinetheir forces to fill a void left following the giant leap taken by Lionbridge andBowne. In our interview with E. Smith Yewell of Welocalize, Smith pointstowards an important factor, often forgotten by those with an almost exclusivefocus on shareholder value: This merger, he says, was “a merger tailored tomeet a market opportunity but also one tailored to staff and client needs.”

The LRC has just finalised a detailed review of its activities and its positionin the localisation landscape. This review was supported by its IndustrialAdvisory Board, industry leaders, state bodies and distinguished academics, andfacilitated by Dublin-based company Product Innovator Ltd. As a result of thisreview, the LRC has intensified its relationship with worldwide digitalpublishers and their partners who are interested in future technologies andprocesses for GILT. The LRC will continue to provide relevant well-researchedcontent-rich information on future trends and technologies, offering a uniqueplatform for industry and academic collaboration and providing anunparalleled network of expertise. One important tool for this work isLocalisation Focus – The International Journal for Localisation, which has justbeen included in two highly prestigious academic indexes, Ulrichs PeriodicalsDirectory (see http://www.ulrichsweb.com/ulrichsweb/) and INSPEC (seehttp://www.engineeringvillage2.org), highlighting the relevance and quality ofour publication.

We invite you to join us at our 6th Internationalisation and LocalisationSummer School, 12–15 June 2006 (see page 6 for details). In addition to anexciting, hands-on localisation experience, this year's Summer School includesan overnight excursion to one of Ireland's most scenic and traditional beautyspots — the Aran Islands off the west coast of Ireland in the middle of theAtlantic Ocean.

Reinhard Schäler

Opinions expressed by individual authors do not necessarily reflect those of the editor

When you change address, remember to update us at LRC@ul.ie! This way you will be able to enjoy reading

your magazine wherever you go.

SPONSORSHIP AND ADVERTISINGTo advertise in or to sponsor an issue of Localisation Focus,contact the LRC at lrc@ul.ie and find out about the benefits.

GET YOUR COMPANY NOTICEDSend your press releases to lrc@ul.ie and get your company on the

Localisation Focus INDUSTRY.news page

To subscribe to Localisation Focus, visit www.localisationshop.comFor accessing your issue online: please go to www.localisation.ie

and click on “issue online”. Password: m0106 ; Username: locfocsub

CONTENTS2 Upcoming Localisation Events

Sponsor Close-up:

4 Localisation Focus talks with the CEO of Welocalize

Research7 Lessons Learnt in the Development of

Applications for Remote Communities

Tool Review

12 Alchemy Layout Manager™ 1.00Michael Bourke

Research

13 The Sinhala Collation and its Representation in UnicodeRuvan Weerasinghe, Dulip Herath and Kumudu Gamaga

20 News from GALA & TILP

Personal Profile

21 From Programming in Basic to Creating XLIFFTony Jewtushenko

Research

22 Translation Web Services — An Implementation for the IGNITE ProjectKevin Bargary

Country Focus

24 Localisation in the United StatesRenato Beninatto

27 Games LocalisationBarry Kehoe and David Hickey

30 Localisation Research Centre News

The front cover image shows downtown San Francisco

by night, with the Golden Gate Bridge on the left.

THANK YOUThe LRC would like to acknowledge the support of

Alchemy Software, ]project open[ and con[text] for thisyear’s Professional Development Courses which were

held at the University of Limerick 13–16 February 2006.

Thank you also to Arabize who facilitated the LRC’sDirector to give a keynote at the recent LISA Forum Cairo.

Some IAB members, after a recent review of the LRC. L to R: Mary Ryan (Product Innovator Ltd.), James Grealis (Symantec),

Alan Barrett (IAB President), Tony Jewtushenko (Product InnovatorLtd.), Seamus Gallen (Enterprise Ireland), Ian Dunlop (Consultant).

LOCALISATION FOCUS4 LOCALISATION FOCUS MARCH 2006

Issu

e Sp

onso

r — W

eloc

aliz

eSPONSOR.close-up

Localisation Focus (LF): Before we getstarted Smith, can you tell us a little bitabout your background and how youbecame involved in the industry. Smith Yewell (SY): I got the idea forWelocalize while serving as a Field Art-illery Officer in Germany and refinedthe idea after returning from the GulfWar.

My German friends were alwayscomplaining that Americans think thewhole world speaks English. And itstruck me one day while sitting in thedesert after the Gulf War was over, con-templating what I would do next, that

American companies could really usesome help expanding internationally.Our first job was in 1997, and the jobwas to translate one word, ‘Pathfinder’,into 12 languages. It is amazing to methat the business started with just oneword.LF: Your Company has made it knownthat you intend to lead the mid-tier inour industry. How do you define themid-tier?SY: I would define it as companiesbetween US$ 20 million and US$ 100million in revenue. Proportionally, thereare only a small number of companiesin the mid-tier right now — most ofthem have already been acquired. Thiscreates a great opportunity for us to fillthe gap in the marketplace and offer avertically-focused, professional serviceto companies looking for a mid-sizedsolution provider. Welocalize has takena different approach in this area. Ourmodel is changing the way localisationservices are both sold and delivered.LF: What kind of model is this?SY: It is a Professional Services Model.It is the same model you see in consult-ing, accounting and law. In our case,each of our verticals is lead by aPractice Manager, similar to the Partneryou find in the aforementioned indus-tries. Each of our accounts is guided bythis Practice Manager before, during,and after the sale. All of our Practice

Managers are industry veterans withgreat experience; so they add tremen-dous value to their accounts — espe-cially in times of challenge. Projectsmust be both sold well and executedwell in order to meet expectations onboth sides, and our Practice Managersprovide the leadership necessary tomake sure this happens. It is one of thereasons for our success.LF: How does a company become suc-cessful in this industry?SY: I think it begins with understandingthe industry opportunity, and right nowI think that is focused and responsive

LocalisationFocusspeaks withE. SmithYewell, CEO ofWelocalize.

Projects must be both sold well and executed well in order to meet expectations on both sides,

and our Practice Managers provide the leadershipnecessary to make sure this happens.

All of our existing client teams are remaining in place.

Welocalize recently announced the acquisitionof Dublin-based Connect Global Solutions. Wespoke with Smith Yewell to get his perspectiveon the mid-sized sector of the industry and his predictions for the future — as well as thereasons behind this important merger.

E. Smith Yewell

LOCALISATION FOCUS 5LOCALISATION FOCUSMARCH 2006

Issu

e Sp

onso

r — W

eloc

aliz

e

SPONSOR.close-up

service to the mid-market. What fol-lows is the correct model. Our sales andproduction teams are merged into aninnovative Professional Services model.Finally, it is understanding the needs oftargeted clients. We target clients withinspecific industries (eLearning, Storage,Security, Supply Chain Management,Life Sciences and Telecommunications)and add unique value through our expe-rience and knowledge of their industry.LF: Let’s talk about your merger withConnect. Why did you merge?SY: Eugene McGinty (founder of Con-nect) and I spent 15 months evaluatingthe merger, and what we found was thatthe combined company would definitelybe much stronger. Our company cul-tures were very similar, we both utiliseda vertical specialisation approach andboth companies have strong translators,engineers and project managers. Also,we both had loyal and long-term cus-tomers that were looking for increasedscale to meet their growing needs. Now,we can meet those needs comfortably —we are over 100 people and we havefive offices in four countries.LF: Were you expecting to gain efficien-cies in the merger?SY: Efficiency gains, yes. Redundancygains, no. Our companies had no over-lap in either staff or clients, and thatmade the merger all the more com-pelling. All of our existing client teamsare remaining in place.LF: What makes this merger different?SY: Our approach was very different. It

was very careful and deliberate. We putour emphasis on retaining and provid-ing opportunities for our staff; andthose opportunities have been designedaround specific client needs. Thus, itwas a merger tailored to meet a marketopportunity but also one tailored tostaff and client needs.LF: What technology does a Welocalizesize (mid-tier) of company have todayto compete with the ‘big’ guys? SY: We have gone in a completely dif-ferent direction than our competitors.Rather than develop proprietary tech-nology, Welocalize has invested heavily

in best-of-breed applications from lead-ing Enterprise Software providers.

Utilising Resource Management, Portaland Workflow technology plus advancedaccounting software, Welocalize has builta sophisticated, integrated business envi-ronment that allows our clients to easilyand seamlessly do business with us in themost efficient manner.

Frankly, I don't know why this indus-try continues to devote internalresources to technology when this is aservice business. We focus our resourceson providing the best service available.Welocalize does develop point solutionsto improve efficiency in the engineeringprocess, but we leave the enterpriseapplications to Microsoft.

It is because of this enterprise-classinfrastructure that we have been able toscale readily to accommodate our steadygrowth and not drop the ball.LF: What do you predict for the toolscompanies in our industry?SY: Small tool companies which are privately held will do just fine if theyfocus on a specific niche. Larger toolcompanies that are investor-backed aregoing to have a tough time delivering

the growth necessary to produce areturn for investors. In fact, most ofthese are already gone. Transware haspulled the plug on GlobalSight andIdiom is reaching to reinvent itself.Trados had stretched the limits of theirgrowth potential and sold to SDL.There is just not much of a market forglobalisation tools. I’d be surprised ifthe total market is even US$ 50 million.LF: What do you predict for the serviceside of this industry?SY: It is a great time to be in the indus-try. Globalisation is sweeping the plan-et, and it is exciting to be involved insuch a phenomenon. This will meangrowth opportunities for all of us, and Ibelieve the greatest opportunity is in themid-market. Welocalize is offering anew and fresh approach to the business,and I believe this will help us to capi-talise on the mid-market opportunity.

As the industry evolves, it is becomingmore specialised, and I think nichecompanies have the most to gain. Thereare companies specialising in singleindustries, single services and single lan-guages. All of these companies will per-form well by delivering specific value inthe supply chain. There is still moreroom for change in the relationshipbetween MLVs and SLVs to improveefficiency. Technology helps in thisarea, but I believe a model change iscoming.LF: And what might clients expect?SY: Clients can expect more choice.Look at our merger. It created an excit-ing, new mid-tier option. Even moreoptions will come onto the market asother companies specialise; and withspecialisation comes even better service.Clients are driving us to provide betterchoices and better service, andWelocalize is one company who plansto continue to deliver.LF: Thanks for taking the time to speakwith us Smith, and the best of luck withyour new venture.

Smith Yewell can be contacted atsmith.yewell@welocalize.com

It was a merger tailored to meet a market opportunity but also one tailored to staff

and client needs.

Utilising Resource Management, Portal andWorkflow technology plus advanced accountingsoftware, Welocalize has built a sophisticated, integrated business environment that allows

our clients to easily and seamlessly do businesswith us in the most efficient manner.

Eugene McGinty

6th LRC Internationalisation

and Localisation Summer School12–15 June 2006 — University of Limerick, Ireland

The 6th LRC Internationalisation and Localisation Summer School will offer you hands-on experience of localisation — from basic introductions to advanced concepts — and will include an intensive one-day course

in Microsoft .NET localisation.

There will be ample opportunity for learning, networking and discussions. The social highlight of the weekwill be an overnight excursion to the Aran Islands — one of Irelands' most scenic and traditional regions.

Please note that this programme may be subject to change

Who Should Attend?

If you are researching or studying localisation and localisation-related issues then the LRC Summer School is for you!Regardless of your area of expertise, it will give you the opportunity to improve and broaden your expertise.

Experience in computer programming is not required; you only need a working knowledge of computers and the Windows operating system.

Fees

All Four Days: €980 (€ 780 for TILP Members)

Fees include accommodation, tuition, morning and afternoon refreshments, lunch, Summer School booklet and excursion.

Note that places are limited to 15 individuals and will be allocated on a first-come, first-served basis.

CLP CertificationThe University of Limerick is the first third-level institute to become a TILP-accredited Course Provider, allowing it to offer courses that lead delegates to Certified LocalisationProfessional (CLP) certification. CLP certified localisers can apply to become professional members of The Institute of Localisation Professionals (TILP).

Limited AvailabilityAttendance at the Summer School is limited to 15 people so book early to avoid disappointment.

Event updates will be added regularly to www.localisation.ieTo register for this event log on to www.localisationshop.com

Programme OverviewMonday 12 June Tuesday 13 JuneLocalisation—an Introduction Microsoft .NET Localisation • Localisation Process • Introduction and Concepts

• Exercise: A Sample Project • Hands-on Course

Wednesday 14 June Thursday 15 JuneLocalisation—Advanced Concepts 1 Localisation—Advanced Concepts 2• Reverse Localisation • The Localisation Factory

• Cultural Adaptation • Medical Devices Localisation

LOCALISATION FOCUSRESEARCH.locMARCH 2006 7

Appl

icat

ion

Dev

elop

men

t for

Rem

ote

Com

mun

ities

1. IntroductionThere are many projects aimed at bridging the digital dividethat have been deployed all over the world — specifically inrural areas. Such projects have been implemented in the hopeof bringing about the many potential benefits to these com-munities, in particular, to improve their social, economic andcultural well-being. Instances of these benefits include theelimination of the barriers to physical and virtual isolation,providing access to available information, and increasingopportunities to expand businesses to reach new markets.While many of these projects report on the provision ofaccess of Information and Communication Technologies(ICTs) to remote communities, there are few studies thatreport specifically on development and its processes of specif-ic technologies for the rural communities.

Thus, this paper aims to provide a description of anddetails on the lessons learnt from the development andprocesses employed in the implementation of three applica-tions or systems. In the next section, this paper will providedetails of the location of the two remote communities i.e.,Bario and Long Bedian — the communities for which theapplications were developed. In addition, the eBario project,which aims to bridge the digital divide, is also described. Thisproject provides the context within which the three applica-tions were developed. The ensuing section will then describedthe applications that were built — in particular, the rationalefor the development, the methodology employed, and the out-comes of the implementation. The lessons learnt are detailedin the form of an examination from technological, opera-tional, logistical, and strategic aspects.

2. BarioBario is located in the Kelabit Highlands, near the Kalimantanand Sarawak border (see Figure 1). It is the ‘unofficial capital’

Lessons Learnt in theDevelopment of Applications

for Remote CommunitiesBy Alvin W. Yeo, Azman Bujang Masli, Siou-Chin Ong, Peter Songan,

Jayapragas Gnaniah, Khairuddin Ab Hamid, Poline Bala,

Universiti Malaysia Sarawak (UNIMAS)

AbstractIn this paper, we highlight lessons learnt from our experience in the development of three applications fortwo small remote communities in Sarawak, a state of Malaysia, which has been provided with access toinformation and communication technologies. The applications developed include a digital library employedto capture oral traditions of the Kelabits; a website to promote tourism in Bario, and a word processorlocalised to accommodate the Kayan language. The Kelabits and Kayans are two of 27 ethnic groups found inSarawak. The lessons highlighted will be discussed vis-à-vis the technological, operational, logistical, andstrategic aspects of systems development for remote communities.

Keywords: Bridging the digital divide, eBario Project, ICT, Open Source, digital library, word processor,tourism website, community informatics, localisation.

In the world of language solutions

cubes are in fashion!

www.skrivanek.com

Belgium Bulgaria China Czech Republic Estonia Germany Hungary Latvia Lithuania Poland Slovakia Slovenia USA

we do

> translations in all languages

and language combinations

> localizations

> cross-cultural consulting

> linguistic and professional

proofreading by native speakers

> DTP processing and graphic

design

> software solutions

our advantages

> single contact point for

all languages

> a registered supplier for the EU

> ISO 9001:2001 certified

> 12 years in the market

International Management Centre Pragueinfo@skrivanek.com

+420 233 320 560

LOCALISATION FOCUS RESEARCH.loc MARCH 20068

Appl

icat

ion

Dev

elop

men

t for

Rem

ote

Com

mun

ities

of the ‘land’ of the Kelabits, one of the 27ethnic groups in Sarawak. Prior to theintroduction of a daily flight into Bario,the Kelabits only means of communicationwith the closest town was by foot —climbing mountains, following mountainridges, and crossing and re-crossing riversand valleys for several weeks. Today, flying(which takes about an hour) to Bario, themain Kelabit centre, is the only practicalway to get there.

Bario has a number of governmentoffices, and also provides education andhealth services to the Bario communityand surrounding villages.

There are about 1,200 people living inBario. The Bario district is occupied prin-cipally by the Kelabit (78%) (one of thesmallest ethnic groups in Sarawak), withother ethnic groups including Penan,Kenyah, Iban, Bidayuh and Malays,Chinese, as well as some Indonesian immi-grants. The majority are farmers (93%),planting wet rice as their main crop.About 5% of the population work in gov-ernment offices, whereas about 2% oper-ate personal businesses and trading. Inaddition to rice cultivation, the communi-ty also rears livestock such as buffalo, cat-tle, sheep, chicken and pigs. Some mem-bers of the community are also involved inhunting, fishing and forest gathering.

3. eBario: Bridging the Digital DivideThe idea of bringing the Internet to Bariowas conceived as a research project todetermine opportunities for social devel-opment available from the deployment ofinformation and communication technolo-gies (ICT) within remote communities inSarawak. Desirable results from pilot stud-ies in other developing countries haveencouraged the team to work among thosecommunities in Sarawak to have equalaccess to ICTs, specifically, the Internetwhich could provide significant improve-ments in their lives. This was included inthe eBario project. Basically the goals ofthe eBario Project were to:

• Define the extent that contemporaryICTs can deliver sustainable humandevelopment and significant improve-ment to the lives of the community

• Demonstrate how significant and sus-tainable development can be achievedby remote communities through theinnovative use of ICT

The objectives of the eBario projectincluded to:

• Empower the Bario community to beable to employ ICTs to improve theirlivelihood through a people-centred/participatory approach

• Provide the Bario community andschool children with access to ICTsthrough:– a computer laboratory at SMK Bario– a community telecentre at Bario

As part of the eBario project, numerousareas were identified as potential beneficiar-ies from the introduction of ICTs. Theseareas included education, culture, com-merce, agriculture, health, community, tech-nology, and human resource development.

…if you could success-fully implement such a

project in Bario, youcould do so anywhere.

3.1 Why Bario? While there were many communities inSarawak that satisfied the criteria forchoosing a rural remote location, Bariowas selected because of its isolation. Inaddition, it has basic infrastructure (no24-hour electricity supply, gravity-fedwater) and no telecommunication service.This can be considered a real case of ‘digi-tal divide’ and ‘digital poverty’. Lastly, thecommunity’s readiness to participate,given that Universiti Malaysia Sarawak(UNIMAS) had conducted other researchprojects in the area and thus were knownto the local Bario community. Because ofits remoteness, the catch-phrase was that ifyou could successfully implement such aproject in Bario, you could do so any-where.

Figure 1: Map of Sawarak showing the location of Bario.

3.2 Benefits to Bario CommunityNumerous benefits were realised in theareas of education, and commerce. Withthe community’s access to ICTs, there isincreased computer literacy among thestudents, teachers and members of thecommunity. Students from Bario are nolonger disadvantaged when they go to theurban areas to continue their studies —they would be just as adept at using com-puters as their urban counterparts.

The community is able to communicatewith the rest of the world due to the avail-

LOCALISATION FOCUSRESEARCH.locMARCH 2006 9

Appl

icat

ion

Dev

elop

men

t for

Rem

ote

Com

mun

ities

ability of telephones and Internet (via VSATs). The community,especially those involved in tourism, have taken advantage of ICTs— they are able to communicate with potential tourists directly viaemail, and confirm accommodation bookings online.

At the state level, the project has served to sensitise the StateGovernment towards the potential for ICT-induced rural develop-ment. In particular, it has demonstrated the importance of ICTs toisolated communities that are denied other forms of infrastructure.The Government of Malaysia is paying increasing attention to ruraldevelopment; different rural ICT programmes have been run —such as the Pusat InfoDesa, and Medan InfoDesa.

4. Long BedianA second remote community that was involved in the project arethe Kayans, who live in another isolated remote location known asLong Bedian. Long Bedian is located in the Apoh Tutoh region ofthe Baram district, in the Miri Division of Sarawak (see Figure 1).The village comprises 180 houses and has a total population of1,686 people. There are only two ways to get to Long Bedian fromMiri town — either an express boat journey of seven hours fol-lowed by an hour-long drive to Long Bedian, or a four-and-a-halfhour drive (by 4WD) from Miri. The village functions as a tradingcentre for the nearby villages, particularly for the Penan community.It also provides education and health services to the Long Bedianand Penan community.

The Long Bedian community comprises several ethnic groups —such as Kayan, Kelabit, Kenyah, Morek, and Punan. The Kayansare the biggest group in the village making up 95% of the popula-tion, while the Kelabits make up 3.9%. The remaining 1.1% of thetotal population in Long Bedian comprises the Kenyahs, Moreksand Punans. The Long Bedian community are all Christians.

The primary occupation in the Long Bedian community is farm-ing (68.4%) — planting paddy, oil palm trees, pepper, and othercrops. About 5.2% of the people are government servants, with theremainder involved in either small businesses or the private sector.

The next section details the systems developed for the communi-ties in Bario and Bedian.

5. Software Applications DevelopedThere are three applications that will be covered in this paper,namely, the Bario Lakuh Digital Library, a Tourism Website and aWord Processor. These applications were the outcome of three sub-projects funded by the Universiti Malaysia Sarawak.

5.1 Bario Lakuh Digital LibraryThis digital library project was aimed at preserving a Kelabit oraltradition, i.e., the traditional Kelabit songs known as ‘lakuh’. Thelakuh is a means of passing information about significant events onto the next generation, as well as depicting one’s feelings.

Thus, one of the objectives of the Bario Lakuh Digital Library(BLDL) project was to explore the cultural benefits of ICT in stimu-lating the production, protection and popularisation of Sarawakrural communities’ oral traditions, which constitute part of anindigenous knowledge system. This project, in line with the objec-tives of the eBario project, aimed to record and transcribe some ofthese traditional songs, particularly the lakuh songs.

There were three main phases to the project; Data collection,Lakuh Translation and Documentation, and Building the DigitalLibrary.

Data Collection and Translation: Both audio and video record-ings of the lakuh singers were carried out by the researchers. As theKelabit women were only fluent in Kelabit (and spoke little BahasaMelayu or English), it was essential to have a Kelabit speaker pres-ent. During these recording sessions, the singers were also inter-viewed. After recording the lakuh, it was transcribed and translated

into English by Florence Apu, a qualified translator who is fluent inboth written and spoken Kelabit as well as English. This translationwas conducted in Bario.

Building the Digital Library: The next step was to digitise theaudio and video recordings and store them in a digital library usingthe open-source Greenstone Digital Library Software (from theUniversity of Waikato, New Zealand). This software allows morelakuh to be added into the existing library, if required.

Outcome: A prototype of the Bario Lakuh Digital Library wascompleted and has been published on CD-ROM. It contains ninelakuh sung by five Kelabit lakuh singers. The lakuh lyrics are avail-able in Kelabit (with a translated version in English), as well as inaudio and on a video recording (of the singer rendering the song).The background of the singers and details about the lakuh and itsmeaning are also provided.

Through this Digital Library, the cultural heritage can thus bepreserved and the knowledge of the indigenous group can be passeddown to the next generation. Linguists will be interested in the lan-guage used in the lakuh which has evolved over time; the lakuh aresung by women of the older generation, and thus use (untainted)Kelabit.

5.2 Tourism Website The second application produced — a map-based tourism website —was developed as part of eBario to promote Bario as a tourist destina-tion. It is believed that residents of Bario will benefit from eTourism.

The objectives in developing the map-based website were to:• provide comprehensive information on Bario including maps• provide information about lodges and homestays • provide information about tourist guides and enable tourists to

reserve a tourist guide in advance

This website was developed using the web-based system develop-ment life cycle; covering web page design, framework and contentdevelopment. This website also included zoomable and interactivemaps in Scalable Vector Graphic (SVG) format.

The website is now complete (see Figure 3 overleaf and alsohttp://www.ebario.com). SVG was employed to provide maps ofSarawak, Kelabit Highlands, Bario Town, Pa Lungan and Pa Umur(villages in Bario). Key landmarks such as lodges, tourist attractionsand government offices, such as the police station and immigrationoffice, were also included.

Based on anecdotal evidence from visitors from Australia (ontheir way to Bario), the website provides the necessary informationfor visitors. Also, through the website, homestay owners in Bariohave received emails from potential visitors enquiring about Bario.Presently, no data has been collected to determine the economicimpact of the website. However, logged visits to the site show thatthere have been consistent numbers of visitors to the website, andnot only Malaysian visitors (see Figure 3 overleaf). In eBario, thehomestay owners are fully utilising the ICTs. They are using emailsto contact their clients and are keen to use the Internet to promotetheir homestays and Bario itself.

Figure 2: Screen shots of Bario Lakuh Digital Library

LOCALISATION FOCUS RESEARCH.loc MARCH 200610

Appl

icat

ion

Dev

elop

men

t for

Rem

ote

Com

mun

ities

Figure 3: Top 10 countries that visited www.ebario.com(last update: 15 February 2006)

5.3 Word Processor In this project, the word processor which allowed interaction inEnglish was customised to accommodate interactions in Kayan andKelabit. This was implemented as part of a thesis to determine the effi-cacy of the existing software development lifecycle (SDLC); currentSDLC is a Western construct and it was argued that the SDLC mayneed to be adapted to suit local contexts (Azman and Yeo, 2004).

Our first plan was to use the Open Source Software (OSS),OpenOffice. However, obtaining a build environment of OpenOfficein Windows became a major obstacle. Due to time constraints, wedecided to work with a less complex software application, i.e.,Abiword which is an Open Source word processor.

The development was conducted in four stages. In Stage 1, weachieved a build environment to create the software, which couldaccommodate different languages. In Stage 2, we identified the com-puting terms to be translated and to translate these terms intoKayan and Kelabit (localisation phase). Translators were identifiedto conduct the translations; approximately 3,000 terms had to betranslated in total. In Stage 3, we tested whether the Kayan andKelabit language could be added to Abiword. In doing this wefocused mainly on the menus and tooltips, and it was successfullycarried out. The Kayan version was more complete and was evalu-ated by native Kayan speakers.

The results indicate that the usage by the Kayans was similar tothat experienced by first-time users of software in their own lan-guage. Also those who had previously used English word processors

were able to identify the English equivalent first before looking forthe Kayan word. Consequently, it was difficult to measure the func-tionality — as the users had to translate the Kayan commands backto English.

A word processor which can accommodate Kelabit and Kayan(see Figure 4) was achieved. However, the effort in adapting OSS inthe project was underestimated. The team was not aware of difficul-ties and only decided to adopt the less complex word processor inthe middle of the project. As OSS developers are located wordwide,Internet communication was the only way to get feedback. Thisinvolved participation in mailing lists and OSS community discus-

Digital Library (DL) Tourism Website Kayan, Kelabit Word Processor

Goal Capture indigenous oral traditions Promote tourism Localise a word processor

Requirements Analysis Identify goals of DL Identify goals and requirements Language requirements/Identify the singers ◊ of websites idiosyncrasies ◊Identifying translator ◊ Identify people to interview and Identification of applications Tasks to be completed collect data from ◊ available ◊

Design Design interfaces, navigation Website design: interface, Design: language-dependent structure, functionalities ◊ navigation, databases, components are as per

functionalities screen interface

Implementation: Recording audio, video (on-site) ◊ Recording and collection of Translation of the Data collection Interviews (on-site) ◊ information/content: interviews, language-dependent components

Transcription & translation of information about culture & into Kayan-Kelabit songs (on and off-site) ◊ tourist attractions, accommodation, (translators were off-site) ◊

photos, maps (on-site) ◊

System Implementation Building the libraries Development of website: webpages, Modification of code to Digitisation of the songs databases, SVG maps accommodate target languagesIntegrate into the DL CD-ROM Incorporation of the different media produced (after evaluation) and write-ups

Application(s) employed University of Waikato’s Greenstone ASP, MS SQL Server, JavaScript Abiword ◊

Testing and Evaluation Testing of the system with users [◊] System testing & usability testing [◊] Usability evaluation of the word Refinement: Editing of the Accuracy of information ◊ processor: corrections of the translation ◊ translations ◊

Maintenance and OperationsCurrent CD-ROM distributed ◊ UNIMAS currently maintains UNIMAS makes the

the website modificationsFuture More lakuh to be added by local Community makes changes and Community updates the

community after training provided ◊ uploads information themselves ◊ information/translation ◊

Table 1: Overview of the Three Applications DevelopedOn-site: in Bario or Long Bedian; off-site (in Unimas); ◊ locals involved; [◊] Optional involvement of locals

Figure 4: Screen Shots of the Word Processor in both Kelabit and Kayan

LOCALISATION FOCUSRESEARCH.locMARCH 2006 11

Appl

icat

ion

Dev

elop

men

t for

Rem

ote

Com

mun

ities

sions. Delays occurred as these developers are mostly volunteers,which curtails their availability to answer queries. At present stan-dard computing terminology of Kayan and Kelabit does not exist.Thus, provision of a tool in the target language may be a way thecommunity can preserve the language.

6. Lessons Learnt from the Application Development An overview of the three applications is provided in Table 1. Theoverview is organised according to generic software developmentphases and details key activities conducted as well as activities thatrelate to involvement of the local community. The lessons learntfrom our involvement in the development of the applications hereare by no means exhaustive, but do provide guidelines for thoseinterested in developing applications for remote communities.

6.1 Crucial to Form Rapport with Target CommunityForming a rapport with the target community is of immense benefitto both parties; the developers will have access to information oth-erwise not available elsewhere, and the local community contributesto the successful completion of the application. (Referring to Table1, there are numerous areas whereby the locals were involved). Inthe case of Bario Lakuh Digital Library, during the data collectionphase Florence Apu — a Kelabit and former English teacher — wasable to identify with whom, when, where and how each interviewcould be conducted.

6.2 Identify a Local ChampionWhere possible, the project team members should identify a localchampion who provides the necessary information not only to theproject team, but also to those on-site. The local champion wouldact as a motivator, at the grass-roots level, to get things done. In thecase of the Tourism Website, John Tarawe was able to persuade therelated parties to cooperate and provide the necessary informationfor the website.

6.3 Do Not Underestimate Logistical ProblemsTravel to remote areas may impact on the scope as well as the proj-ect schedule and budget. Remoteness, long travel time and infre-quent flights to such areas will increase the project duration. Also,such trips may be affected by inclement weather. For example, aflight delay in Bario due to bad weather could leave you stranded inMiri until the weather clears (which could take days).

The development of software for remotecommunities is not just about technology

or logistical issues. It is about working withand for the people. As long as the needs

of the people are taken into account, the technologies (regardless of what they are) will largely be accepted —

albeit with some modifications to suit the local context.

6.5 System Development Both On- and Off-siteGiven the difficulties of travel to remote areas (on-site), certain partsof the system development could be conducted off-site in order toreduce costs. Off-site implementation may be better since access toinformation/tools is easier than in the remote area. Similarly, usabili-ty tests can be conducted off-site if target users are available there.

6.6 Start SmallWhere possible, applications to be developed/translated should beof a small, manageable size. Knowledge on the development envi-

ronment of the target application is also crucial. Success would actas an impetus for bigger applications/projects.

6.7 Sustainability of Software UseTraining of the target community with the software is necessary toensure maintenance and use of software. In the case of the wordprocessor, besides training to use the tool, members of the commu-nity had to be trained to make minor modifications to the transla-tions (in addition to being able to create the build environment formore involved changes).

7. SummaryIn summary, the development of software for remote communities isnot just about technology or logistical issues. It is about working withand for the people. As long as the needs of the people are taken intoaccount, the technologies (regardless of what they are) will largely beaccepted — albeit with some modifications to suit the local context.

AcknowledgementsThe authors acknowledge the funding provided by UniversitiMalaysia Sarawak for the following Fundamental Research Grantprojects 03/14/365/2002 (102), 01/32/381/2003 (118), and02(01)/277/2002(15). The project team acknowledges the use of theGreenstone Software (http://www.greenstone.org) and AbiWordsoftware (http://www.abisource.com/) and would like to thank allthose involved — in particular, the Bario and Long Bedian commu-nity.

BibliographyAzman B.M. and Yeo, A. W. (2004). Unpublished report of Fundamental

Research Grant 02(01)/277/2002(15), Building Software forMalaysians: An Application of Global-Software Development Life-Cycle.

Bala, P., Harris, R.W., & Songan, P. (2003). E Bario project: In search ofa methodology to provide access to information communication tech-nologies for rural communities in Malaysia. In S. Marshall, W.Taylor, & X. Yu (eds.), Using community informatics to transformregions (pp. 115-131). Hershey, PA: Idea Group Publishing.

Centre for International Development. Readiness for the NetworkedWorld. A Guide for Developing Countries. Retrieved July 1, 2005from http://www.readinessguide.org.

Kano, N. (1995). Developing international software for Windows 95 andWindows NT. Microsoft Press. Redmond, WA.

Gnaniah, J., Yeo, A., Songan, P., Hushairi Z. and Khairuddin A. H.(2004). A Comparison on the Implementation Approaches for the e-Bario and e-Bedian Project. Proceedings of the Seventh Conferenceon Work With Computing Systems, Kuala Lumpur, Malaysia, 29 June-2 July.

Gnaniah, J., Songan, Yeo, A., P., Hushairi Z. and Khairuddin A. H.(2004). Communication Patterns of the Long Bedian Community:Implications for the Development of a telecentre. Proceedings of theSeventh Conference on Work With Computing Systems, KualaLumpur, Malaysia, 29 June-2 July.

Liew, J., Yeo, A., Khairuddin A. H., and Al-Khalid O. (2004).Implementation of Wireless Networks in Rural Areas. Proceedings ofthe Seventh Conference on Work With Computing Systems, KualaLumpur, Malaysia, 29 June-2 July.

Songan, P., Khairuddin A. H., Yeo, A., Gnaniah, J., and Hushairi Z.(2004). Community Informatics: Challenges in Bridging the DigitalDivide. Proceedings of the Seventh Conference on Work WithComputing Systems, Kuala Lumpur, Malaysia, 29 June-2 July.

Unimas. (2004). eBario Web Site. Retrieved July 1, 2005 fromhttp://www.unimas.my/ebario.

Dr Alvin Yeo may be contacted at Alvin@fit.unimas.my

LOCALISATION FOCUS LOCALISATION. tools MARCH 200612

Loca

lisat

ion

Tool

s

APARTICULAR BOTTLENECK INTHE LOCALISATION REENGINEER-ING PROCESS is that of resizing the

elements in application windows to account fortext expansion. As everyone knows, text expan-sion can lead to vital information being truncat-ed, such as button labels, text fields, etc. Thisisn’t simply a matter of stretching boxes andbuttons — the reengineering of complex forms,pop-ups and taskbars is a complex and time-consuming process. When we start localisinginto languages with vertical or right-to-left writ-ing directions, the process becomes even harder.It’s a necessary evil that costs money and thetime of developers: not only is this a laboriouschore for software developers who could bedoing far more productive development work, ittakes a relatively long time, which in turnimpacts lead-time to international markets.

Ensuring that the software is internation-alised as much as possible from the beginning isan effective strategy for reducing the workloadin this area. The user interface should ideallyaccount for the possibility of future text expan-sion. But in some cases, this might simply not bean option — perhaps a software developer didnot have the foresight, perhaps functional con-siderations wouldn’t allow it or perhaps theschedule wouldn’t allow it. As a result, localisa-tion engineers spend approximately 70% oftheir time re-sizing, re-shaping and re-position-ing the various elements in the user interface.

Alchemy claim it will be as sig-nificant to the localisation

process as translation memorywas to the translation process.

So what is a software developer to do? It’s a dirtyjob (and wasteful, and expensive) but somebodyhas got to do it. Alchemy Software claims to havean alternative to the hard graft of manual reengi-neering — Alchemy Layout Manager™ automat-ically creates layouts for translated user inter-faces. Alchemy is so confident in this new tech-nology, they claim it will be as significant to thelocalisation engineering process as translationmemory was to the translation process. In anindustry where automation is increasingly seen asthe way forward, Alchemy Layout Manager cer-tainly has the potential to make a big splash.

I witnessed Alchemy LayoutManager reconfigure the user

interface of a small demo appli-cation in a matter of seconds.

The operation of Alchemy Layout Manager isas simple as selecting the files to be processed andrunning the application. Alchemy LayoutManager — although a separate standalone tech-nology and product — tightly integrates intoAlchemy CATALYST 6.0, the latest version ofAlchemy’s flagship localisation product. AlchemyLayout Manager scans localised files and using anadvanced mathematical analysis, determines thespatial relationships between user-interface ele-ments. These relationships are then used to calcu-late a completely new layout for the translatedapplication files. Alchemy Layout Manager canautomatically reengineer the user interface with-out any manual guidance. Buttons and text boxesare automatically resized but the layout of theinterface remains consistent and aligned. Errorsand elements that couldn’t be automaticallychanged are flagged for manual checking by anengineer. I witnessed Alchemy Layout Managerreconfigure the user interface of a small applica-tion in a matter of seconds. A large, major soft-ware release contains thousands of elements thatneed to be analysed — but even then the timetaken to process is supposedly a few minutes(Alchemy claims 1,000 dialog boxes take justunder seven minutes to process). Compare this tothe several weeks that manual changes (and test-ing) would take and you can get a sense of theefficiency this type of technology can bring to thelocalisation engineering process.

Essentially, Alchemy Software Developmenthave spent two years creating an algorithm that iscapable of automatically resizing elements of theinterface that are affected by elongated textstrings. The speed of Alchemy Layout Manager isquite striking: it automatically makes the changesin a matter of seconds that would normally takea software developer several days to carry out. Inaddition, the software error-checks the correc-tions Layout Manager makes and flags anypotential problems. Alchemy claims that, onaverage, just three percent of changes are flagged.Even if these flagged changes have to be checkedmanually, the overall timesavings are enormous.

Figure 1: BEFORE — a simple dialog box afterits text has been pseudo-translated. Truncation

is clearly evident in the buttons.

Figure 2: AFTER — Layout Manager automatically resizes the buttons and labels

to fix the longer text.

Currently, Alchemy Layout Manager™ is avail-able on all Windows development platforms(16-, 32-, 64-bit, VB.NET and MS .NET 1.x and 2.0). It is tightly integrated with AlchemyCATALYST 6.0, the latest edition of Alchemy’sflagship localisation product. And whileAlchemy Layout Manager is significantly moreexpensive than these other versions of AlchemyCATALYST — and indeed other localisationtools — Alchemy is keen to point out that thereturn-on-investment is almost immediate.

At the moment, there are some drawbacks.The tool’s handling of vertical text expansion(i.e. as a result of translation into Chinese, etc.)is currently being refined for the next version,which is due for release in early 2006. The highcost of the tool also limits ownership to thehigher end of the market but this is hardly thekind of work carried out by freelancers anyway.The tight integration into Alchemy CATALYST6.0 is a big plus however, and is symptomatic ofthe intelligent design and ease-of-use that makesAlchemy CATALYST so popular. Indeed, thenew Alchemy CATALYST has had a makeoverand now looks more like Windows XP’s inter-face while remaining intuitive to use and rela-tively easy to learn.

Automation in the localisation process is thebest way to reduce time to market and lowercosts, and typically localisation tool developershave addressed this by adding support for anincreasing number of standards. Alchemy LayoutManager is a radically different approach tomaking reengineering easier for developers. It willbe interesting to see what kind of impact this toolwill have — on both the competition amongstlocalisation tool developers, and the localisationcommunity as a whole.

Michael Bourke is a research associate with theLocalisation Research Centre. He currently over-sees the Localisation Technology Laboratory(LOTS) based in the University of Limerick andthe ELECT Online web portal, as well as work-ing on the IGNITE project. He can be contactedat michael.j.bourke@ul.ie.

Michael Bourke

Alchemy LayoutManager™ 1.00

Michael Bourke looks at Alchemy Layout Manager and wonders what impact it will have on the industry.

1. Background

THE COLLATION ORDER OF ANY LANGUAGE isone of the most important issues that has to beresolved urgently in the process of standardising such

a language. Since languages have been used for centuries byhumans without worrying about their irregularities, oftentheir constructs are extra logical. The collation sequence ofmost languages faces this non-logical nature. Steps are beingtaken to avoid these ambiguities and irregularities and alsoto formalise the collation sequence as much as possible sinceit plays a major role in the process of standardising the lan-guages. This is particularly so in the case of electronic textsof a language since the computer needs explicit orderinginformation in order to process such a language.

To illustrate simply, even in the apparently well under-stood case of the English Latin-1 character set, the font itselfdoes not encode order. If it did, words beginning with ‘Z’will precede those beginning with ‘a’ since all upper-case let-ters precede all lower-case letters in all encodings (includingUNICODE) of English. The case for Sinhala is no exception.In fact, as will be clear in the ensuing discussion, the Sinhalacollation sequence demands us to take some decisions thusfar not explicitly made for the Sinhala language as a whole.

2. Introduction to Sinhala AlphabetThe Sinhala alphabet consists of characters which representalmost all the sounds that can occur in the language. On theother hand, it is phonetically over-specified in that there are

multiple characters to represent the same sound: for example, (dental la) & < (Alveolar la), k (dental na) & K (Alveolarna), g (voiceless ta) & G (voiced ta).

The commonly accepted Mixed Sinhala Alphabet has a setof sixty characters. This set of characters can be classifiedinto three categories, namely vowels, semi-consonants andconsonants.

Vowels: There are 18 vowels in the Sinhala alphabet, w, wd,we, wE,….´ , T!.Semi-consonants: there are two characters which can occuronly with a vowel: x and #Consonants: there are 20 consonants in the alphabet: l, L,.,….., y, <

In addition to the above characters there is another set ofsymbols called vowel-strokes or ‘pilli’, to represent vowelsound when vowels are combined with consonants. Forexample: l + d -> ld (k + aa) and l + f d -> fld (ka +o)

There is generally no disagreement regarding the order ofcharacters within vowels, semi-consonants and consonants,except for ‘Z’ and ‘f’. The relative order of these charactersets is also well defined, i.e. vowels are followed by semi-consonants which themselves are followed by consonants.

3. MethodologyThe methodology adopted in this study is to first gatherexisting views and traditions of Sinhala language collationfrom scholars, observe the collation sequence adopted by the

LOCALISATION FOCUSRESEARCH.locMARCH 2006 13

Rese

archThe Sinhala Collation Sequence

and its Representation in UNICODEBy Weerasinghe A. R., Herath D. L., Gamage K.

Language Technology Research Lab, University of Colombo School of Computing, Colombo, Sri Lanka.

AbstractThe alphabet of a language is perhaps the first thing we learn as users. The alphabet of our mother tonguewould be the first alphabet we ever learn. And yet, a closer look reveals that there is much about such analphabet that we have not explicitly specified anywhere. The Sinhala alphabet order is a prime example. Weuse it, recite it and yet would be hard pressed to define it explicitly.

Sinhala is spoken in all parts of Sri Lanka except some districts in the north, east and centre by approximately20 million people. It is spoken by an additional 30,000 (1993) people in Canada, Maldives, Singapore, Thailandand United Arab Emirates. Sinhala is classified as an Indo-European language and used as an official language.

The UNICODE Collation Algorithm (UCA) is an attempt to make explicit the collation sequence of anylanguage expressed in the UNICODE (or any other) coding system. In order to express the Sinhala collationsequence (alphabetical order) using UCA, the authors undertook the task of identifying unresolved issuesfacing the unambiguous definition of the order. This paper first describes the issues identified through thisstudy, suggesting alternate solutions and recommending one of them. Finally, it sets out the recommendedcollation sequence for Sinhala in the form of the UNICODE collation specification. The outcome of thisprocess is a unique and unambiguous expression of the Sinhala collation sequence which could be testedusing existing tools and software environments.

Keywords: UNICODE, Sinhala, Collation, UNICODE Collation Algorithm, Localisation, Internationalisation.

LOCALISATION FOCUS RESEARCH.loc MARCH 200614

Rese

arch

major standard dictionaries and to research how collationsequence is determined at various state organisations in per-forming their regular tasks.

It is expected that this kind of study would reveal the mainissues related to the collation order of Sinhala and how theseissues are addressed by scholars, in dictionaries and by organ-isational practice.

4. Issues identified in Collation OrderThe following issues* were identified at the beginning of thestudy. With these issues in mind, prominent dictionaries weresearched, the views and opinions of scholars were obtained,and the procedures followed by state institutions and organi-sations were observed. The following listing identifies fivemajor issues (first five with associated levels of importance)and three less critical ones which need to be resolved in orderto proceed with the specification of an unambiguous colla-tion sequence for Sinhala.

Issue # 1 [Level 2]

The positions of ‘anusvara’ and ‘visarga’ in the Sinhala colla-tion sequence. While this is not really ambiguous as far asmany dictionaries and linguists are concerned, its place atthe beginning of the UNICODE code chart made it an issueto be resolved.

Issue # 2 [Level 3]

The position of the ‘hal’ sign (halant form) of a consonant inthe sequence. Many alphabets of Sinhala do not explicitlyspecify the place of the ‘pure consonant’ form (the so-called‘vowel removed form’) of Sinhala letters. As such, there iscommon confusion as to its rightful place in the alphabeticalorder. For digital representation, this becomes an importantissue to be resolved.

Issue # 3 [Level 3]

The positions of words containing yansaya, rakransaya andrephaya when there are two or more alternative forms avail-able for the same word. While in general there is agreementthat these ‘short forms’ are exactly equivalent to their non-shortened forms, in a digital representation a decision has tobe forced as to which of them precedes the other.

Issue # 4 [Level 2]

Miscellaneous issues such as the archaic way of writingwords such as ld hd,h and the irregular forms k~rHr or k~r$r

and even k^^r. Though rare, the exact function and positionof such words in a sorted list of words needs to be explicitlygiven to facilitate digital processing.

Issue # 5 [Level 3]

Whether ‘Z’ is a ligature of j~ and z or a single letter; andthe right position of ‘Z’. This again is an issue raised by itsexistence as a separate code point in the UNICODE codechart.

Issue # 6 [Level 0]

The position of the letter *. The Latin symbol ‘f’ was super-imposed on the Sinhala letter m to produce the symbol ‘,before the symbol * was introduced into the alphabet. Forthis reason, and the phonetic closeness of the sounds repre-sented by m and *, the letter * has been popularly placedafter the letter m in many contexts. On the other hand, theletter ‘*’ being the newest letter of the Sinhala alphabet, isalso placed at the end of the list of consonants in the alpha-bet.

Issue # 7 [Level 1]

There is a rule in the Sanskrit writing system that the conso-nant that comes after a rephaya is doubled, for example thewords jd;_;d, ud._... The reason for this appears to be todisplay other vowel modifiers clearly when they are used with a consonant that comes after the rephaya, forexample as in ldu_ñl,, ;dl_lsl, Ofu_udakaudoh. The prob-lem arising with this kind of phenomenon when sorting iswhether their positions should be considered based on thisdoubled form, or on their corresponding simplest form (asld ñl, ud ., ;d lsl, O fudakaudoh in the examplesabove).

Issue # 8 [Level 1]

Finally, the sort order of words which consist of intra-wordspaces, for example, the name o is,ajd. The issue here iswhether to consider this as two words, to ignore the spaceand consider the string as a single word, or to consider theentire string including the space as the single full word.

While these issues were the ones pre-identified in the study,the availability of online tools for testing any suggested col-lation sequence expressed in accordance with the UNICODECollation Algorithm specification, allowed us to look for anyother issues which might be ‘thrown up’.

5. Observation made in DictionariesFor the purpose of this study some of the main and populardictionaries were selected from among the various Sinhaladictionaries published. These dictionaries were selected afterconsidering multiple attributes such as their quality, quantityin circulation, real usage and the perceived degree of authori-ty of their compiler(s).

The following were the dictionaries selected for the presentstudy:

[1] Sri Sumangala Shabdakoshaya compiled by Ven.Velivitiye Soratha Thero

[2] Sinhala Shabdakoshaya published by the Department ofCultural Affairs

[3] A Sinhalese – English Dictionary compiled by Rev. CharlesCarter

[4] Prayogika Shabdakoshaya compiled by Dr. HarishchandraWijetunge

[5] Sinhala Vishvakoshaya published by the Department ofCultural Affairs

* N.B : Issues # 6, 7, 8 were not taken into consideration in the first phase of the survey. The Levels indicate the perceived severity of the issue concerned, 3 being themost critical.

The issues identified regarding the collation order of Sinhala were kept in mind while these dictionaries werebeing studied. The information gathered from these diction-aries regarding each issue is summarised in Table 1.

Based on the above, some partial conclusions could be madeas follows:

Issue 1: This appears to be a non-issue as far as dictionarycompilers are concerned. These two semi-conso-nants are placed at the end of the set of vowels inthe Sinhala alphabet.

Issue 2: Apart from the Sri Sumangala dictionary, each ofthe other four had a clear decision that the ‘hal’form comes after all other vowel derivatives. Whilethe justification given by each of these two schoolsmakes sense in their own contexts, the majoritydecision may need to be adopted for our purposes.Section 6 illustrates the difference between the twoschemes.

Issue 3: Interestingly, none of the dictionaries are able toshed light on this issue owing to each only contain-ing a single form – either the short or the non-short.

As such this issue cannot be resolved using thismethodology.

Issue 4: There is wide variation on the treatment of thisissue. As such, a final decision on resolving this isdeferred at this stage.

Issue 5: All dictionaries implicitly consider ‘Z’ as a ligatureby their positioning of words beginning with itappearing soon after those beginning with ‘j~’and ‘z’.

Issue 6: In all dictionaries which includes it, the position ofthe letter * is immediately after the consonants.

Issue 7: The doubling of the reph-modified consonant isgiven as a spelling variant of the simpler form in alldictionaries which contained it.

Issue 8: All dictionaries include words which have intra-word spaces where appropriate.

6. Procedures Followed by Dictionaries to Sort WordsThe procedure followed in the Sri Sumangala Sabdhakoshayato arrange words manually according to the alphabetical orderis best specified by the following algorithm:

LOCALISATION FOCUSRESEARCH.locMARCH 2006 15

Rese

arch

Table 1: The position taken by dictionaries on the eight issues under consideration

IssueSri Sumangala Sinhala Carter’s Sinhala Prayogika Sinhala

Shabdakosaya Shabdakosaya English Dictionary Shabdakosaya Vishvakoshaya

The positions of In all the dictionaries the ‘anusvara’ and ‘visarga’ come at the end of the vowels.

the ‘anusvara’ and

‘visarga’

The position of The criteria followed A letter with ‘hal’ sign comes after that letter’s vowel derivatives.

‘hal’ sign is described in the (The criteria followed in the dictionaries is described in the Section 6)

Section 6

The positions of the Observations could not be made regarding this issue since no dictionary

words containing uses two or more forms of the same word.

‘yansaya’,

‘rakaransaya’ and

‘repaya’…

Irregular forms kY# is used only to kY#=k~r# kY# is used only to Both forms are

represent k~r# not k~rF is written represent k~r# not followed in different

k~rF as it is k~rF places kY#=k~r# or k~rF

Status of ‘Z’ -considered as -considered as -considered as -considered as -considered as

a ligature- a ligature- a ligature- a ligature- different char-

The letter * -not present- * comes at the end -not observed- * comes at the end * comes at the end

of consonants of consonants of consonants

Doubled Reph -given along with -given along with -given along with -given along with -not observed-

the main entry as the main entry as the main entry as the main entry as

alternative spelling- alternative spelling- alternative spelling- alternative spelling-

Intra-word space -space has been -space has been -space has been -space has been -space has been

considered- considered- considered- considered- considered-

LOCALISATION FOCUS RESEARCH.loc MARCH 200616

Rese

arch

1. Identify the syllabic units* of the two wordsLet the two words be w1 and w2

2. Write each syllabic unit of both words as a consonant-vowel pair†

3. i=0

4. Choose the ith character of each wordLet the two characters be w1 (ch (i)) and w2 (ch (i))

4.1. If w1 (ch (i)) = = w2 (ch (i))4.1.1. i=i+14.1.2. go to 4

4.2. Else if w1 (ch (i)) > w2 (ch (i))4.2.1. w1>w24.2.2. break

4.3. Else4.3.1. w1<w24.3.2. break

Dictionaries other than Sri Sumangala Sabdhakoshaya com-pare consonant-vowel pairs in a different manner whichmakes the two approaches different. In this method when twoconsonant-vowel pairs are compared two consonants and twovowels are compared separately. In the cases where vowels are

not present the consonant of the next consonant-vowel pair isnot taken as in the Sri Sumangala Sabdhakoshaya.

7. Views of Scholars/Academics and LinguistsThe following scholars and academics were consulted with aview to acquiring their expert views – often based on theirrespective linguistic persuasions. The aim of the consultationwas to attempt to achieve consensus and not just for docu-menting their independent views.

• Professor Vinee Vitharana (VTH)Chief Editor of the Sinhala Dictionary. FormerProfessor of Sinhala at the University of Ruhuna.

• Professor Wimal G. Balagalle (WBA)Former Chief Editor of the Sinhala Dictionary. EmeritusProfessor of the University of Sri Jayewardenapura.

• Professor W. S. Karunathilake (WSK)Former Professor of Linguistics at the University ofKelaniya

• Professor J.B. Dissanyaka (JBD)Emeritus Professor of the University of Colombo

Table 2: The position taken by Linguists on the eight issues under consideration

* Syllabic unit means entity that contains a consonant and a vowel which is represented with pilli, e.g. kW,@k`, k# are syllabic units and in some contexts k~, a ,a` canalso be syllabic units, in ak~k` a and k~ are syllabic units. The syllabic units of aw~wQk\k` are a, w~, wQ, k\, k`

† If the syllabic unit does not consist of a vowel write ‘null’ in place of vowel (e.g. k~ = k~ null). The ‘null’ is considered as a character and it is the character that hasthe greatest value in the weight space.

Issues VTH WBA WSK JBD SGA HWI RUP

1 At the end of the vowels

2 At the end of the vowels

3 Need to have Need to have Need to have a Need to have a Need to have a Need to have a Need to have a a policy: what a policy: policy policy: what is policy: what is policy: priority policy: priority is simple should whatever simple should non-confusing must be given to should be given come first non-confusing come first what is commonly to the tradition

written

4 kY#=k~r# kY#=k~r# Sanskrit loan kY#= k~rF kY#=k~r# kY#=k~r# kY#=k~r#

k~rF should be k~rF should be words should be k^= k~r# k~rF should be kY#= k~rF k~rF should be written as it is written as it is written in their written as it is written as it is

traditional forms, but English loan words can be written in either form

5 ‘Z’ is a ligature ‘Z’ is a ligature ‘Z’ is a ligature ‘Z’ is a ligature ‘Z’ is a ligature ‘Z’ is a ligature ‘Z’ is a ligature j~+ z j~+ z j~+ z j~+ z j~+ z j~+ z j~+ z

6 * comes at the * comes at the * comes at the * comes at the * comes at the * comes at the * comes at the end of end of end of end of end of end of end of consonants consonants consonants consonants consonants consonants consonants

7 -give along with -give along with -give along with -give along with -give along with -give along with -give along with the simplest the simplest the simplest the simplest the simplest the simplest the simplest form- form- form- form- form- form- form-

8 -space has to -space has to -space has to -space has to -space has to -space has to -space has to be considered- be considered- be considered- be considered- be considered- be considered- be considered-

• Professor Sucharitha Gamlath (SGA)Former Professor of Sinhala at the University ofRuhuna.

• Dr. Harishchandra Wijethunge (HWI)Author of the Prayogika Sinhala Sabdhakoshaya

• Mr. Rupasinghe Perera (RUP)Deputy Director, Pirivena Education Branch, Ministryof Education Secretary, Sri Lanka Oriental LanguagesSociety

Having explained the aims and objectives of this study, a listof lexemes that concretely represents all the possible issueswas carefully designed and given to each consultant – ratherthan posing the issues in their abstract form. This approachforced an explicit response rather than inviting rigorousexpositions of the theoretical basis for same. There weresome issues which some of the linguists could not provide adirect answer to. However, most were able to make their sug-gestions as to how to resolve such issues by relying on theirown linguistic theories. The books written by some of thesescholars were also considered during this study. The com-ments made by each expert regarding the identified issues andtheir suggestions are summarised in Table 2.

Issues 1, 2, 6, 7 and 8 were not disputed by any of theexperts who agreed with the majority (4) of mainstream dic-tionaries. While Issue 3 had no consensus solution, all expertsagreed that there should be a single well-specified standard.Issue 4 too had no clear consensus except for the recommen-dation that ‘k~rF’ should be written as it is. There also appearsto be a majority view that ‘Z’ should be treated as a ligature— with the only dissenting scholar too late arriving at a con-sensus in the interest of arriving at an overall consensus.

8. Procedures followed in State Institutions andOrganisationsThe following government organisations and institutes wereselected for the purpose of identifying the different collationorders adopted by them for their regular work.

• National Library & Documentation Centre (NLDC)

a. An explicit alphabetical order is available at NLDC

b. The Sri Lanka National Bibliography is preparedaccording to this alphabetical order

• National Institute of Education (NIE)

a. The NIE has adopted the alphabetical order given inthe Sri Sumangala Shabdakoshaya.

b. This order is followed when school text books and rec-ommended books for school children are prepared.

c. The specified alphabetical order for government exam-inations (e.g.: GCE (O/L) and GCE (A/L)) is also thesame.

d. Further recommendations of the NIE are given in a separate publication entitled Sinhala LekhanaReethiya.

• Public Library – Colombo (PUB)

a. The alphabetical order given in the Sinhala Encycl-opedia is followed.

• Sinhala Dictionary Office (SDO)

a. The criteria followed by the SDO is the criteria fol-lowed in the Sinhala Sabdhakoshaya

• Sinhala Encyclopedia Office (SEO)

a. – not yet responded –

• Election Commissioner’s Office (ELE)

a. An explicit alphabetical order is available at ELE

b. ‘anusvara’ and ‘visargaya’ comes at the end of vowels

c. ‘hal’ sign comes at the beginning of vowels

d. When there are two or more alternative forms avail-able, the collation order is found according to the sim-plest form and the priority is given to the simplest form(issue #3)

e. The letter ‘Z’ is considered as the conjunction of j~ andz.

f. The recommendations of the NIE given in the SinhalaLekhana Reethiya are also followed by the ELE.

• Library — University of Colombo (UOC)

a. The alphabetical order given in the SinhalaSabdakoshaya is followed.

• Library — University of Peradeniya (PDN)

a. The alphabetical order is the same as that used byUOC.

• Library — University of Kelaniya (KLN)

a. – not yet responded –

• Library — University of Sri Jayewardenepura (SJP)

a. –not yet responded –

• Library — University of Ruhuna (RHU)

a. The alphabetical order given in the SinhalaEncyclopedia is followed.

LOCALISATION FOCUSRESEARCH.locMARCH 2006 17

Rese

arch

Of the above, the NIE and Election Commissioner’s Office(ELE) deserve special attention. The ELE standard is of inter-est to this study because it explicitly addresses the issues athand – Issues 1, 2, 3 and 5. In Issues 1 and 2, the ELE stan-dard tallies with those of the majority of dictionaries and lin-guists. Interestingly, ELE has a definite recommendation forIssue 3, i.e. to locate all such form variations together at therightful place of the simplest form with the simplest formpreceding the other forms in decreasing order of simplicity.Finally, on Issue 5 (6, 7 and 8 also), the ELE standard con-curs with that of the expert consensus.

Since the government recognises the NIE as the primeauthority in setting educational standards the order recom-mended by them becomes of utmost importance. Some of theother reasons for attaching such importance to this recom-mendation include:

(a) The standard specified has been created by representa-tive groups of scholars and linguists including many ofthose consulted in the present study.

(b) Generations of school teachers and students havealready adopted this standard and hence it is the closestto a defacto standard.

(c) Their more recent publication, Sinhala LekhanaReethiya, is widely used by state organisations includingthe Commissioner of Elections.

9. Summary RecommendationsThe status of each of the issues considered in this studytogether with the recommended solution is presented below.

Issue 1: The dictionary survey and ratified by the expertconsultation resolved this issue to the satisfactionof the authors: treat both the ‘anusvara’ and ‘vis-arga’ as appearing in the alphabetical order imme-diately after all the vowels. This is also furtherconfirmed by the ELE and NIE standards whichare in wide practical use.

Issue 2: The dictionary disparity with regard to the correctposition for the ‘hal’ form was resolved by theunanimous opinion of the experts consulted that itshould immediately follow the vowels but precedethe ‘anusvara’ and ‘visarga’.

Issue 3: This was one of the issues on which empirical evidence was scarce. However, the openness of all the linguists for some standard and the simplic-ity rule recommended by some of them and clearlyenshrined in the ELE standard is to order all formsof such words adjacent to each other beginningwith the simplest form and increasing in complexi-ty. This would prescribe the following order on thethree common forms of the work karyalaya: ldhd,h, ld©d,h, ldh?d,h.

Issue 4: This is the issue with the greatest degree of diver-gence in opinion. Three of the dictionaries and fiveof the linguists however concurred that ‘kY# is usedonly to represent k~r# not k~rF’. The latter is repre-

sented as it is. This is in contrast to the originalSinhala UNICODE recommendation where ‘kY#=

k~rF and k^= k~r#’. It seems prudent to adopt themajority opinion.

Issue 5: This was the single main success in the consensus-seeking process. It is thus recommended that ‘Z betreated as the ligature of j~+ z’ so that it does notappear in the order thought to be implied in theUNICODE code chart.

Issue 6: This seemed to be an issue as it was introduced laterto the Sinhala alphabet and the phonetic similarityof the letters m and *. There was confusion withthe symbol ‘ too. According to Sinhala LekhanaReethiya, the book published by the NIE forSinhala, and all the scholars it is accepted that theletter * should come at the end of the consonants.

Issue 7: This form is used merely for representation pur-poses. The underlying meaning of both ldu_ñland ld?ñl is the same. Therefore they occupythe same collation position. In dictionaries theseare given along with the main entry as spellingvariations.

Issue 8: It is important to consider intra-word space whensorting is done in some domains (e.g. directories ofnames). However, this cannot be prescribed in thealphabet — it can only be done at the level of theparticular application.

Based on the above recommendations and extensive testingdone using early versions of the proposed collation sequence,a UNICODE Collation Element table together with itsweights is recommended as the explicit specification of theSinhala alphabet for use in electronic processing of Sinhala.The documents observed at the organisation mentionedabove and other relevant documents including the proposedCollation Element Table can be found at the URLhttp://www.ucsc.cmb.ac.lk/ltrl/public/collationDocs.html.

The proposed Collation Element Table can be found at the URL

http://www.ucsc.cmb.ac.lk/ltrl/public/collationDocs.html

10. ConclusionAt the outset we pointed out that a complete and unambigu-ous specification of the Sinhala alphabet is an essential andurgent requirement for all kinds of electronic processing ofSinhala text. The process of study revealed five major areasunresolved as far as the Sinhala collation order was con-cerned and three other areas which needed clarification. Weoutlined a methodology of arriving at a set of well informedrecommendations based on three sources: widely accepteddictionaries, the most respected Sinhala scholars and the mostwidely adopted official standards on collation sequence.

LOCALISATION FOCUS RESEARCH.loc MARCH 200618

Rese

arch

LOCALISATION FOCUSRESEARCH.locMARCH 2006 19

Rese

arch

Using a consensus-based approach, we have successfullyarrived at a unique collation sequence for the Sinhala languageand expressed it explicitly using the UNICODE CollationAlgorithm specification of the UNICODE Consortium.Testing of this specification for arbitrary lists of words is madepossible by online tools available from InternationalComponents for UNICODE (ICU).

AcknowledgementsThis work was partially supported by IDRC under the PANLocalization Project and sponsored by the ICT Agency of SriLanka. The authors are indebted to all Sinhala Languagescholars, Publishers and other Practitioners who collaborat-ed willingly in the work described in this study. In particularwe wish to acknowledge the contribution of Mr. HarshaWijayawardhana and Mr. Asanka Wasala as well as othercolleagues in the Language Technology Research Centre ofthe University of Colombo School of Computing.

References[1] Karunathilake, W.S , Sinhala Basha Viyakaranaya, M.D.

Gunasena & Company Ltd, 1997.[2] Wijetunge, H , Sinhala Akuru Akaradhie Kireema

Pramitha Kireema, S. Godage Brothers, 2003 [3] Perera, R, Sinhala Vahara Athpotha, Thivira Publishers,

2004[4] Sinhala Lekhana Reethiya, National Institute of

Education, 2001[5] Disanayaka, J.B, Akuru ha Pilli, S. Godage Brothers,

2000

A. Ruvan Weerasinghe obtained afirst degree in Mathematics andStatistics from the University ofColombo, a Masters in ComputerScience in 1990 and a PhD in NaturalLanguage Processing in 1994 (bothfrom the University of Cardiff, UK).He then returned to the University ofColombo and served as a SeniorLecturer — first at the Department ofComputer Science and then theUniversity of Colombo School of

Computing (UCSC). After being appointed Head of theDepartment of Intelligent Systems in 2003, he became Directorof the School in May 2004. Ruvan may be contacted atarw@ucsc.cmb.ac.lk

Dulip L. Herath is a ResearchAssistant at the Language TechnologyResearch Laboratory (LTRL) at theUniversity of Colombo School ofComputing, Sri Lanka. He obtainedhis Bachelor of Science degree inComputer Science in 2004 from theUniversity of Colombo and is current-ly reading for his MPhil in NaturalLanguage Processing. His researchinterests are Natural LanguageParsing, Localisation and Local

Language Processing. As a member of the LTRL he is involvedin localisation activities, namely defining the collation order ofSinhala, Interface Terminology Translation, and local languageweb content development using Unicode. Dulip may be contact-ed at dherath@webmail.cmb.ac.lk

Kumudu Nuyanie Gamage obtained aB.A. (Hons) in Linguistics in 2004 fromthe University of Kelaniya, Sri Lanka.She joined the Language TechnologyResearch Laboratory (LTRL) of theUniversity of Colombo School ofComputing as a Research Assistant thesame year. As a member of a develop-ment team, Kumudu was involved inthe theoretical side of ComputationalLinguistics with special reference to theSinhala language. Presently she is work-ing at the Department of Linguistics of

the University of Kelaniya, as an Assistant Lecturer. Her Researchinterests are in Computational Linguistics, including Syntax andSinhala Grammar, and the Analysis of differences between Sinhalaand Japanese Languages with a special focus on Sinhala-Japanese/Japanese-Sinhala Translation. Kumudu may be contactedat kgamage@webmail.cmb.ac.lk

A. Ruvan Weerasinghe

Dulip L. Herath

Kumudu NuyanieGamage

LOCALISATION.orgThe inside information on localisation

News Update from the Localisation Associations

The Institute of LocalisationProfessionals (TILP)

TILP is a world wide non-profit association forindividuals in the translation, localisation, internationalisation and globalisation industry.TILP is managed and directed by the electedTILP Council. Web: www.tilponline.org

For more information contact info@tilponline.org

News BriefsCLPReview of the first course providers is nearly finished and thefirst three or four will be approved and certified soon. Keep aneye on the TILP website for updates as these, and more courseproviders, are added.

TILP/CLP AdministratorTILP is in the process of hiring a part-time administrator to takecare of all the TILP and CLP activities. This is a turning point inTILP and the CLP as it moves TILP from a fully voluntaryorganisation to having its first paid staff member.

Ask the ExpertsAsk the world’s experts about current issues and trends in locali-sation at one-day sessions — each preceded by an online forum.TILP organises regular Ask the Experts sessions where profes-sionals share their knowledge with TILP members. These one-day events give TILP members the opportunity to ask the world’sexperts about current issues and trends in localisation.

In 2005, TILP organised 5 very successful Ask the Experts ses-sions. The topics were:

- Localisation, and Project Management, by Frank Bergman(Director, ]project open[ )

- The CLP Programme, by Reinhard Schäler, (CEO TILP anddirector of the LRC)

- Exploring Issues in Terminology Management, by Sue EllenWright (chair of the ATA Terminology Committee)

- Microsoft.NET Localisation: Problems, Solutions, Best Practice,by Florian Sachse (PASS Engineering) — two sessions

We expect to have even more topics and sessions in 2006.There are a number of sessions planned for Localization Worldin Barcelona and Translation and the Computer in London.Keep an eye on the TILP website for details.

Are there topics that you would like to have TILP organise?Contact us at info@tilponline.org.

Upcoming Events30 May–01 June, Barcelona

Localization World: Working — Together!It’s going to be about Industry collaboration and how the languageindustry ecosystem is changing. Visit http://www.localizationworld.com/for more information. TILP members can avail of a 10% discounton registration fees — just enter the reference code LWBAR20when registering (applies to both pre-conference and conferenceevents).

TILP will be holding a number of Ask the Experts sessions atLocalization World on topics such as Microsoft .NET Localisation,XLIFF and Terminology Management. For the latest updates onevents, member discounts, CLP, future events, or to join TILP, visitwww.tilponline.org

The Globalization andLocalisation Association

(GALA)In the three years since its creation, GALA hasgrown to an association of more than 170 compa-

nies from 35 countries. GALA is a fully representative industry association runby a board of directors who are elected by the membership. The GALA Boardemphasises open communication and complete transparency. In addition toproviding opportunities to find new clients and expand their business, GALAis helping members to develop better, more profitable, businesses. Membershave already benefited from unique networking opportunities, innovative mar-keting and sales opportunities, group discounts, additional exposure for theircompanies, and more. At the request of our members, we are pursuing indus-try initiatives to influence the future direction of the industry. Membership isopen to companies providing translation, localisation, internationalisation, orglobalisation products or services, including tools developers, training suppli-ers, and consultancies.

GALA Board OfficersThe GALA Board Officers for 2006 are as follows:• Aki Ito, Prisma International (USA): Chair• Arancha Caballero, TSG Global (Spain): Vice Chair • Stephen Ryan, Moravia Worldwide (Czech Republic): Secretary • Christiane Bernier, Lionbridge Technologies (USA): Treasurer • Matthias Caesar, Locatech GmbH (Germany): Board Member

GALA on Technology BlogGALA invites translation and localisation professionals to partici-pate in the new GALA on Technology Blog. Developed by GALAmembers, the blog is intended to be a non-biased open forum for dis-cussion of the technological resources, new developments and on-going challenges that vendors and clients are facing in the localisa-tion and translation industries. Visit www.gala-global.org/blog

Free subscriptions to GALAxyGALA invites readers to subscribe to GALAxy, a free e-newsletter forthe language services community. To subscribe or to read past issues,visit http://www.gala-global.org/index.php?action=view_newsletter

GALA Articles Database — Call for ArticlesGALA invites authors to submit new and previously published articles on globalisation, internationalisation, localisation and translation to the online GALA Articles Database athttp://www.gala-global.org/index.php?action=view_articles&source=resources

GALA Vendor DatabaseIndustry companies are invited to submit their profile to the GALAVendor Database. The database is searchable by company services,subject matter specialties, language combinations, company size andlocation.

Upcoming EventsGALA European Member Meeting in BarcelonaGALA will hold its fourth European member meeting on 30 May inBarcelona. The focus of the meeting will be interaction and input frommembers on key issues for member companies and the association.

Localization World Barcelona30 May–01 June. With the theme Working — Together, LocalizationWorld Barcelona 2006 will focus on industry collaboration. The con-ference will have a special focus on translation industry collaboration,as well as software, automotive, government and non-governmentinstitutional translations. During the conference, eight GALA membercompanies will exhibit in a group booth. GALA is assisting with con-ference content through board member Aki Ito’s role on the programcommittee. In addition, GALA will organise the conference closingceremonies.

LOCALISATION FOCUSPERSONAL. profileMARCH 2006 21

Pers

onal

Pro

fileFrom Programming in

BASIC to Creating XLIFF In describing his career in localisation, Tony Jewtushenko takes us from writing his first BASIC program to chairing

the OASIS XLIFF Technical Committee. Tony Jewtushenko

IN THE 1960’S AND 1970’S, DINNER GUESTS AT THEJEWTUSHENKO FAMILY’S SUBURBAN PHILADELPHIAHOME would likely encounter simultaneous conversations inUkrainian, Portuguese, Spanish and English. Tony’s parents wereboth born in Ukraine. His father emigrated to America in theearly 50’s, and his mother’s family initially moved to Brazil,where they lived for 15 years, before emigrating to America. Likemost immigrants to the USA, over the years the family assimilat-ed to American life and English slowly emerged as the dominantlanguage. None the less, Tony’s multilingual and multiculturalchildhood provided the motivation and context for his subse-quent career in international software product development.

“Remember this day — life will never be the same again”.

Tony’s passion for computing took hold at an early age. In1971, his primary school maths teacher acquired a DECWriterand modem for his classroom, and at the age of 9 Tony wrote hisfirst BASIC program. When the IBM PC made its initial debut in1981, Tony was studying Mathematics (concentration in com-puter science) at Boston University. One day a professor broughtin a full-page Wall Street Journal advertisement that announcedpricing and availability for the IBM PC and proclaimed,“Remember this day — life will never be the same again”. Hewasn’t joking. A few months later, Tony commenced his profes-sional career, providing programming services to a couple ofsmall Boston law firms. Following his university studies, for thefirst ten years of his career he was a COBOL programmer forBoston based ASC, Inc, and later a system analyst for a realestate investment bank, AEW, Inc, where he developed databaseand advanced spreadsheet applications for econometric analysis.

Advanced spreadsheet skills helped Tony land a job withLotus Development Corporation in 1990 as test engineer for the1-2-3 for Windows 1.0 project. Lotus, like many software ven-dors in the early 90’s, was researching methods of reducing thetypical gap of three to six months between release of US andlocalised versions of software. In 1992 Tony transferred toLotus’ International Product Development division to developautomated tests as part of a wider simship strategy. Tony devel-oped automated tests to verify that all UI resources were exter-nalised — a process known as ‘pseudo translation’. Automatedtests identified bugs introduced into language-specific buildsearly in the development cycle. By the mid 90’s Lotus achievedtheir simship goal of seven languages, but the bar would soon beraised much higher.

In 1993, Tony traded in his big comfortable loft apartment inBoston’s North End for a large suitcase and a fistful of airlinetickets. He spent a year implementing localisation and softwaretest best practices in Taipei, Seoul, and Dublin. Later he wasassigned to Beijing for six months to help start up the Chineselocalisation office, and afterwards a year in Singapore imple-menting automated acceptance testing for DBCS-enabled builds.Tony’s love for Asian cuisine increased as his empty passportpages diminished. Even today, he continues to suffer from occa-

sional but acute cravings for Bi Bim Bop, Shabu-Shabu, ChoDofu, Beijing Duck, Bah-Cut-Teh and Chilli Crab.

In 1996 Tony was assigned to Lotus’ bidirectional (Hebrewand Arabic) development team in Dublin. It didn’t take long forTony to fall in love with the beauty and hospitality of Irelandand its people — especially one particularly lovely and brilliantIrish lass named Mary, who eventually became his wife. Nolonger content to live his life out of suitcases and airports, Tonytransferred to Lotus’ Irish division and became a localisationproject manager for the eSuite Java applications team. TheLotus eSuite project was groundbreaking: in 1998 the teamshipped via web 22 language versions simultaneously, includingthe US version, for which it received an IBM Innovation award.This was no small feat given that Java technology was in itsinfancy.

This challenge eventually led to thecreation of the OASIS XLIFF TechnicalCommittee (TC) that Tony chairs today.

In 2000 Tony joined Oracle Corporation in Dublin. His man-ager, Paul Quigley, had been discussing localisation challengeswith management counterparts at Sun and Novell. They allagreed it was imperative to reduce the overhead associated withmany disparate resource formats. This challenge led to the cre-ation of the OASIS XLIFF Technical Committee (TC) that Tonychairs today. XLIFF started as an informal group of mainlyIrish-based localisation professionals seeking to develop a speci-fication influenced by and compatible with the best features ofexisting standards such as OpenTag and TMX.

The XLIFF TC has published 1.0 and 1.1 specifications, andis currently putting finishing touches onto the 1.2 spec, whichsupports segmentation and provides implementation profiles forpopular resource containers. Support for XLIFF by tools ven-dors and the Open Source community is strong and continues toexpand, as evidenced by newly available XLIFF-aware productsand open source projects. Tony also contributes to TranslationWeb Services, an OASIS TC, which defines standard localisationbusiness transactions as web services.

Tony is currently employed by Product Innovator Ltd, aDublin-based management consultancy. Product Innovator Ltdprovides training programmes, coaching and outsourced servicesto technology companies seeking to implement software productmanagement and development best practices. In his spare time,Tony enjoys angling, camping, travel and but most of all authen-tic Asian food.

Tony Jewtushenko is the founding and present chair of theXLIFF TC as well as the director of R&D for Product Innovator,in Dublin, Ireland. During his 23-year career, Tony contributedto dozens of releases of successful commercial software products,including Lotus 1-2-3 and Notes, Oracle JDeveloper, and iDS.He can be reached at tony.jewtushenko@productinnovator.com.

Translation Web Services— an Implementation for the IGNITE Project

Kevin Bargary describes the Localisation Research Centre’s (LRC) involvement in theTranslation Web Services specification implementation, which is part of the IGNITEproject currently being undertaken by the LRC and its partners; VeriTest, Archeptypon,PASS Engineering and Vivendi Universal Games.

IGNITE will pool together linguistic infrastructure resources and provide convenient access and a market-place for them. This will satisfy a pre-requisite for the success of the European digital content industries.

Keywords: Translation Web Services, OASIS, Localisation, Standards, SOAP, WSDL.

IntroductionIGNITE is divided into three phases, the third of which is the‘Performance Analysis’ stage of the project. This third phaseaims to firstly verify standards and their use and implementationin tools and source material, and secondly, enhance the stan-dards themselves by providing feedback from the project find-ings to the various standards committees. With this in mind theLRC became actively involved in the Translation Web Services(TWS) Technical Committee (TC) which operates under theumbrella of OASIS. Our main function as members of the TCwas to provide an independent implementation of the draft spec-ification that was already in place. As this was the first draft ofthe specification, and it had not been previously implemented,some issues were expected — it was our job to identify these andsee how the specification functioned in a practical working sce-nario.

Web ServicesThe basic premise of TWS — and indeed any web servicesimplementation — is that you have a client machine and a servermachine. The server machine will contain a pre-programmed setof methods or functions that the client machine will access usingweb services. An example of this would be an online credit cardvalidation system. One website (the client) connects to a remoteservice (the server) with the details; the service validates thecredit card and returns the result to the website.

Implementation WorkThe first stage was to decide on an implementation platform.The underlying technology in web services is SOAP (SimpleObject Access Protocol). SOAP can be thought of as a message-passing system between two computers using the HypertextTransfer Protocol (HTTP) over the Internet. The TC decided touse the J2EE development platform and the Java programminglanguage. The rationale behind this decision was that the opensource ‘Apache Project’ had a Java-based implementation ofSOAP called AXIS. The AXIS implementation is a reliable andstable base on which to implement Java Web Services. Thisimplementation provides an Application Programming Interface(API) into the SOAP actions that are required for implementingthe TWS specification. The logical choice of web server to com-plement the use of Apache AXIS was Apache Tomcat — so thiswas used to host the web services on our internal LRC server.

The development process was based on the prototype modelof software development. The first stage was to start with one ofthe 18 services currently available in the TWS specification, andfrom there to develop one service at a time and report back tothe TC on any issues or suggestions for improvements that aroseas we progressed. Table 1 shows a list of all of the services avail-able in the current TWS specification. The services are dividedinto three categories; namely ‘Required Services’, ‘OptionalServices’ and ‘Recommended Services’.

Table 1: List of services in the current Translation Web Services(TWS) specification.

As previously mentioned, AXIS provides an API to the SOAPfunctionality. It also provides two command line utilities thatfurther aid the implementation of web services. The‘Java2WSDL’ utility takes pre-existing Java code and creates aWSDL file for that code. A WSDL (Web Services DescriptionLanguage) file indicates how the client can access the service, i.e.what parameters need to be passed to the service in order toevoke a response. The TWS specification already has a WSDL,so this utility was useless for our purposes. However, the second utility, ‘WSDL2Java’, creates the Java stubs (Java files that con-tain the code needed to use SOAP) required by

(a) the server to write and deploy the service and

(b) the client to access the service through its own code.

Figures 1 and 2 show this process.

LOCALISATION FOCUS RESEARCH.loc MARCH 200622

Tran

slat

ion

Web

Ser

vice

s

Kevin Bargary

Required Optional Recommended Services Services Services

submitJob retrieveServiceList rejectJob

retrieveJobInformation requestQuote associateResource

retrieveJob acceptQuote disassociateResource

retrieveActiveJobsList retrieveQuote retrieveResourceInformation

suspendJob retrieveFullJobsList retrieveFullResourceList

resumeJob UploadFile

cancelJob

LOCALISATION FOCUSRESEARCH.locMARCH 2006 23

Tran

slat

ion

Web

Ser

vice

sFirstly the Java stubs are created:

Figure 1: Creating the Java stubs from the WSDL.

Then the Java stubs are used by the client application toaccess the services and by the server to deploy the services.

Figure 2: Connecting client and server (through the Java stubs) overHTTP using SOAP.

The first service that we implemented was the retrieveServiceListservice. This service was chosen because there were no inputparameters required for it. All that was required to invoke theservice was an instance of the retrieveServiceListRequest class.The retrieveServiceList service returns “a complete list of servic-es offered by a particular vendor. This will include the languagesdealt with and services offered by a particular vendor”(Translation Web Services Specification Draft 1.0). After writingthe server-side code to handle a retrieveServiceListRequest, i.e. return all of the appropriate values, the next stage was to create a simple test class that could instantiate aretrieveServiceListRequest and handle the results received backfrom the server in a retrieveServiceListResponse. With bothclasses now ready, we needed to deploy the services to theApache web server. The WSDL file also contains the location ofthe service, i.e. where it can be accessed from. Deployment isnecessary to ensure the service is in the location as defined in theWSDL. When the utility ‘WSDL2Java’ creates the Java stubsneeded for the server-side machine, it also creates two other filesthat are used to deploy and ‘un-deploy’ the service to a webserver (Apache Tomcat). These files are called Web ServiceDeployment Descriptors (‘deploy.wsdd’ and ‘undeploy.wsdd’).

The implementation is currently onwww.electonline.org:8080/index.html

The next stage in the development of the implementation wasto write the code for the rest of the services. While writing thecode we encountered some issues with the specification (includ-ing inconsistencies between the schema and the specificationdocument). These issues were quickly amended by the TC.During the process of coding we also made some suggestions to

the TC about possible improvements to the specification and wewere actively involved in applying these changes. For example,the service ‘retrieveQuote’ in the original specification did notreturn any information about the location of the actual quote.This was deemed to be an important piece of information forthis service and was promptly included in the specification.

With the code for the implementation of the services nowwritten, the initial service deployed (retrieveServiceList) was un-deployed and the full list of services was deployed to the webserver. A JSP (Java Server Pages) client interface was developedto allow for the input of the parameters required for each serviceand also to show the responses from the server — see Figure 3.

ConclusionsWorking on the implementation of the Translation Web Servicesspecification has been very beneficial for the IGNITE project.One valuable lesson learnt during this implementation was theimportance of having a reference implementation of a standard.This is vital to ensure the standard works in a practical environ-ment. Becoming involved in the development of the TWS specifi-cation draft has allowed the IGNITE team to see how a standardcomes together from initial design specification to an industry-accepted standard. The real benefit of the TWS standard will beseen when an implementation exists that incorporates the XLIFFstandard for the exchange of localisable content through thelocalisation process. XLIFF was designed as an interchange fileformat for the loss-less exchange of data for localisation. If TWSis used as a medium for the transport of XLIFF, then a whole newlocalisation process in which automation is a key factor couldevolve. Our work within the TWS TC has allowed us to beingrained at each stage in the development of this standard. Withthe knowledge gained from this work we in the IGNITE projecthope to research the possible marriage of these two importantlocalisation standards in a working automated environment.

For more information on Translation Web Services visithttp://www.igniteweb.org/documents.php to see a presentation byPeter Reynolds entitled Web Services for Translation, given at theIGNITE Working Conference in Dublin on 14 December 2005.

Figure 3: JSP client interface used for inputting service parametersand showing server responses.

The implementation is currently on www.electonline.org:8080/index.htmlKevin Bargary graduated from the University of Limerick in2004 with a B.Sc. in Computer Systems. As well as working as aresearcher on the LRC-coordinated IGNITE project, he is cur-rently pursuing an M.Sc. in the area of open source localisation— also at the University of Limerick. He can be reached atKevin.Bargary@ul.ie

LOCALISATION FOCUS COUNTRY.focus MARCH 200624

Coun

try F

ocus Localisation in the

United StatesRenato Beninatto gives an interesting overview of

the localisation industry in the United States.

FIRST OF ALL, HERE’S A FACTTHAT EVERYONE KNOWS: TheUnited States is by far the biggest

economy in the world. Its GrossDomestic Product (GDP) of almost US$12 trillion dwarfs that of any singlecountry, with only the entire EuropeanUnion coming anywhere close. The U.S.is the headquarters to more than a thirdof the Global 2000 companies. Due toits massive size, many firms both insideand outside the U.S. believe that theywill generate half their revenue in theUnited States and the balance from therest of the world. As a foreigner who haslived in the United States for the last sixyears, I cannot help but be impressed bythe size of this economy.

With so much economic activity concen-trated in one place, it should come as nosurprise that a lot of localisation businessis transacted in the United States. Whatis surprising is the resounding irony

you find when looking at how manyAmericans think about the rest of theworld and the fact that U.S. companiesspend so much on localising products. Inthe aftermath of the 11th September 2001terrorist attacks it seems that manyAmericans want to build FortressAmerica, blocking out the rest of theworld. This leads to the not uncommonattitude that if residents cannot speakEnglish, Americans shouldn’t cater totheir needs — instead, everyone shouldassimilate to the American culture. Anextension of this attitude is that for thosewho don’t happen to live in the UnitedStates, partial translation and localisationshould suffice — foreign buyers of

American products don’t really needaccess to all of the information that U.S.consumers get. Contrast this with the factthat most English-speaking, assimilatedAmericans themselves are of ethnic stock,many are not that far distanced from theirancestral lands, even the smallestAmerican city boasts a wide range ofeateries serving foreign food, and you finda paradox for the localisation industry.

Reflecting this English-first attitude,translation and adaptation tend to showup last in most corporate budgets.However, even with this last place seed,Common Sense Advisory estimates that43 percent of all translation and localisa-tion purchases are made in the UnitedStates, totaling what we estimate will beUS$4 billion in 2006. With so muchactivity happening in the United States,17 of the world’s 20 largest languageservice providers (LSPs) maintain a phys-ical presence in the United States. Eventhose headquartered in Europe and Asiagenerate a significant part of their reven-ues from American companies (see theCommon Sense Advisory website for anarticle entitled “Ranking of Top Trans-lation Companies” from July 2005).

But it’s not just the huge size of theU.S. economy that draws LSPs to theUnited States. The traditionally openAmerican business culture, its compara-tively laissez faire regulatory environ-ment, employer-friendly employmentpolicies, access to credit, and a wide-ranging transportation system make itrelatively easy for companies to set upshop and do business in all 50 states. Infact, over 2,000 companies with five ormore employees in the United Statescompete for language services. However,

this ease of setting up shop leads to avery competitive marketplace, withmany firms bidding for the same busi-ness — often without any real differenti-ation from what their rivals offer.

LSPs are spread widely across theUnited States (see Figure 1). Even though,in the Internet age, physical location isoften irrelevant to the quality of the serv-ices provided, many clients prefer to workwith local suppliers that tend to be moreavailable for rush jobs, can be visited ifcircumstances require face-to-face interac-tions, and offer a reassuring presence tobuyers not yet accustomed to outsourcingor offshoring. Nonetheless, our researchshows most LSPs clustering in denselypopulated states; including California,New York, Massachusetts, Texas, andFlorida. These are all major centres ofeconomic production that boast majorcities, world-class academic institutions,ethnically diverse populations, interestingpolitical climates, and easy access to air-line travel anywhere in the country or theworld. The balance of LSPs are scatteredaround the rest of the country, often serv-ing local industries such as automotive,aerospace, and other region-specificneeds.– Growth Sectors: In the last three years,life sciences (especially medical devices),legal, advertising, defence, and govern-ment have driven language industrygrowth in the United States. Followingthe roll-up of many agencies into thefederal Department of HomelandSecurity, we have seen a marked increasein linguistic service demand for govern-ment, military, and intelligence needs.Finally, for 2006, we expect to see morelocalisation demand for capital goodslike manufacturing equipment and high-tech products, driven by an expecteddecline in the value of the U.S. dollar.

An attitude thateveryone shouldassimilate to the

American culture… and partial translation

and localisation should suffice.

17 of the world’s 20largest languageservice providers(LSPs) maintain a

physical presence inthe United States.

Over 2,000 companieswith five or moreemployees in the

United States competefor language services.

Renato Beninatto

LOCALISATION FOCUSCOUNTRY.focusMARCH 2006 25

Coun

try F

ocus

– Buyer Demographics: With the excep-tion of larger technology companies likeMicrosoft and Symantec that have beenlocalising products for many years, buy-ers of localisation services in the UnitedStates tend to be middle- or lower-levelmanagers with small budgets. They clus-ter around product management, techni-cal publications, and marketing depart-ments in large organisations. The maindifference between the U.S. buyer andthe in-country buyer tends to be thenumber of languages. We expect to seemore American companies translatetheir websites for domestic ethnic needsand for leading international markets.Consumer devices with embedded com-puters will drive everything from iPodsto Jeeps, requiring localisation of abroader array of products (see theCommon Sense Advisory website for anarticle entitled “Real World Enterprise”from January 2004). This expansioncould lead to an expansion of demandfrom marketers, product managers, anddevelopers.– Legally Mandated Translation: Legis-lation, judicial rulings, and presidentialorders require translation and interpreta-tion services for people with LimitedEnglish Proficiency (LEP). Governmentagencies — the firms that provide priva-tised services — and companies thataccept federal funds find themselves sub-ject to these regulations. This means thatsoftware and instruction manuals to geta driver's license need to be translated

into languages such as Tagalog, H'mong,Creole, Ethiopian, Russian, and theubiquitous Spanish (see the CommonSense Advisory website for an articleentitleed “Translation: It's the Law”from April 2004). Smart commercialmarketing to those same people willdrive exponential language servicegrowth for Spanish, Chinese, and otherAsian languages.

– Education and Certification: Local-isation is a growing subject at Americanuniversities, driven by the realisationthat classic approaches to languagelearning — literature or linguistics —don’t lead to employment. Therefore,U.S. colleges have begun merging lan-guage with business and technologystudies to help students become moreemployable. While Kent State University,the Austin Community College, theCalifornia State University at Chico, andthe Monterey Institute of InternationalStudies offer formation programs relatedto the localisation industry, they stillgraduate a small number of students.Unlike other countries, the United Statesdoes not require translators to be univer-sity graduates but some states, likeCalifornia, have fairly stringent regula-tions regarding interpretation. Somegroups are trying to change the practiceof localisation in the States. Fore exam-ple, the American Translators Assoc-iation (ATA) provides a peer certificationprogram that has lost some of its appealdue to the now widespread practice of

outsourcing translations to countries forwhich the translations are intended.Translations into English represent lessthan 20 percent of the language marketin the United States.– The U.S. Lags in Language TechnologyDevelopment: Most language technologyused in the United States is developed inEurope or Canada. A few exceptions aremachine translation suppliers such asIBM and Language Weaver, the latteroffering a statistical-based MT technolo-gy funded by the U.S. government. IdiomTechnologies, with its WorldServer trans-lation workflow management system,remains the last company standing ofseveral similar companies from the late1990s. We expect government attentionto this sector to increase over the comingyears, although the initial funding mech-anisms tend to be secretive intelligenceagencies or the not so secret venture capi-tal arm of the CIA. Given some of thework we’ve seen being carried out in aca-demic settings, the U.S. could become amajor language technology superpower ifthe government invested more money inresearch and commercialisation.

– Final Comment: The United Stateslocalisation market is in constant growthas new players with recent innovationsand creative products enter the marketeveryday. American executives, althoughaware that the world is their playingfield, address localisation as a cost ofdoing business, but still tend to placelanguage services at the bottom of the‘must-have’ list. For LSPs, the market isvery competitive, which means that any-body with good service and sales, as wellas select vertical focus can claim a shareof the market.

About the AuthorRenato Beninatto is the Chief OperatingOfficer at Common Sense Advisory, Inc.,a market research and consulting firmthat specialises in the globalisation, inter-nationalisation, localisation, and transla-tion industry. For more information, visithttp://www.commonsenseadvisory.comor Renato Beninatto can be contacted atrenato@commonsenseadvisory.com

The main differencebetween the U.S. buyer

and the in-countrybuyer tends to be thenumber of languages.

The United States doesnot require translators

to be universitygraduates.

American executivesstill tend to place

language services atthe bottom of the‘must-have’ list.

Figure 1: Translation companies in the United States. (List of 450 companies compiled fromindustry association memberships, advertisements in industry publications, and web searches

plotted into a map.) Source: Common Sense Advisory, Inc

Graduate Diploma/ Master of Science in

Software LocalisationUniversity of Limerick, Ireland

• Develop internationalised applications

• Localise products and manage a localisation team

• Use the widest range of technologies and tools

• Create sophisticated document structures

• Understand the rationale behind localisation

• Work on exciting international research projects

• Become a Member of The Institute of Localisation Professionals

Established in 1997, the Graduate Diploma / MSc in Software Localisation programme has been continuously adapted and expanded in line with the ever-changing needs of localisation.

Students are offered the opportunity to gain practical work experience, for example in the SUN Localisation Laboratory at the University.

The University of Limerick is home to the Localisation Research Centre (LRC), one of the world’s leading intelligence, technology and educational localisation centres.

Lead by its Industrial Advisory Board of international experts from industry and academia, the LRC is the focal point and the research and educational centre for the localisation industry.

Registration is now open for the 2006/2007 Academic YearFor details on registration and fees, visit www.ul.ie

For further programme details, contact Reinhard.Schaler@ul.ie

The University of Limerick is the first university accredited by The Institute of Localisation Professionals (TILP)

to offer courses leading to the Certified LocalisationProfessional (CLP) qualification.

Join the world’s most established postgraduate programme dedicated exclusively to software localisation.

LRC Industrial Advisory BoardAlan Barrett Retired

Gerry Carty General Manager, Vivendi Universal GamesIreland

Tom Connolly Business Development Director, PulseLearning

Ian Dunlop Independent

Mervyn Dyke Managing Director, VistaTec

Seamus Gallen National (Ireland) Informatics Directorate

James Grealis Director, EMEA Localisation, Symantec

Wendy Hamilton Vice-President, Business Development, Bowne Global Solutions

Martin Hynes Director, The Embark Initiative, IRCSET

Brian Kelly Senior Vice President, Bowne Global Solutions

John Malone Independent

Paul McBride Vice-President, VeriTest

David McDonald Management Consultant

Eugene McGinty General Manager, Welocalize Ireland

Michael O’Callaghan Vice President, Oracle Corporation

Seamus Gallen National (Ireland) Informatics Directorate

Terry Landers Head of Corporate Affairs, Microsoft Ireland

Pat O’Sullivan Test Architect, IBM Ireland

Anthony O’Dowd President, Alchemy Software

Chris Pyne Business Partner Manager, SAP

Kevin Ryan Principal Investigator, ISERC

Reinhard Schäler Director, LRC

VIVENDI UNIVERSAL GAMES IS A GLOBAL LEADER INMULTI-PLATFORM INTERACTIVE ENTERTAINMENT.The company develops, publishes and distributes interactive

products across all major video game consoles as well as PC andMac. Vivendi Universal Games’ development studios and publishinglabels include Blizzard Entertainment®, Radical Entertainment,Sierra® Entertainment, Massive Entertainment™, SwordfishStudios and High Moon Studios.

About Vivendi Universal Games IrelandVivendi Universal Games Ireland, based in Dublin, started as alocalisation house for Davidson and over the past ten years hasgone through a number of acquisitions. Built upon ten years ofexperience within the localisation industry and a very enthusiasticteam, the Dublin office has localised some of the biggest games todate such as the Warcraft® and Diablo® series. Vivendi UniversalGames Ireland continues to release high quality localised productsfor current and next generation video game platforms.

The video games industry is a very challenging, yet enjoyable areaof localisation to work in. This article covers some of the chal-lenges faced by Vivendi Universal Games, explaining their businessmodel and the implications this has on their localisation strategy,and outlining some of the reasons for adopting such a strategy. Italso highlights some of the differences between the localisation ofbusiness-type applications and the localisation of video games.

Business StrategyAt Vivendi Universal Games (VUG) the localisation strategy is tosimultaneously ship (sim-ship) the majority of titles into their targetregions. The VUG goal is to have the product on the shelves in thesame week — or even the same day, if possible — in all regions. Thenature of this approach increases the cost of localising a product,both in terms of human resources and the need to purchaselocalised assets from Vendors. Having a sim-ship strategy increasesthe localisation life cycle. When localising a product, the shortestlife cycle is achieved by waiting for the source language (usuallyEnglish) to be declared Gold (not requiring updates to any softwareor localisable assets) before starting the localisation process.However, this approach impacts time to market and makes a simship strategy unachievable. The company strategy of sim-shiprequires over-compensation, with the localisation process beginninglong before product development is finalised. This introduces ineffi-ciencies into the process, and a resultant increase in resource over-heads.

VUG adopts a sim-ship strategy for six reasons: 1. Creates marketing synergies on a global basis2. In a global market, simultaneously shipping product meets

consumer expectations3. Reduces risks of grey imports4. Reduces impact of software piracy5. Reduces impact of negative reviews6. Reduces time to market

Figure 1 shows the top level value chain for VUG on a typical VideoGame. The VUG localisation process starts at point five in Figure 1and the company is heavily involved right through to point ten.

Figure 1: The top level value chain for Vivendi Universal Games on atypical video game.

Vendor StrategyBeing a global publisher of interactive entertainment software, VUGreleases products in all major markets around the world. Unlike a lotof global software companies who have opted for a Multi-LanguageVendor (MLV) solution, VUG has not opted to use this model. Oneof the main differences between localising games compared to, say,business applications is the nature of the content being localised.While an MLV solution has great benefits, the multimedia nature ofthe content that VUG works with requires the services of specialisedvendor companies. To localise a business application, for example,requires a business-oriented translation style, whereas a very differ-ent translation style is required to localise a game. It is of utmostimportance that the scene being set in a game is conveyed throughtranslation in a way that does not take away any of the emotionalimpact to the end user. Added to this is the localisation of audio con-tent. The quality of the spoken audio in a localised game has to be ofa very high standard, making sure the intonation and style of dia-logue matches the character and the situation/conversation takingplace. The complexity and importance of this increases in sceneswhere there is humour, sarcasm or a raised sense of tension.Ensuring lip synchronisation appears correct in the final product isalso very important. We have all probably seen badly dubbedJapanese to English movies that have been released over the years;this effect is something to be avoided at all costs. The goal is to givethe users of the localised game the same level of experience as theusers of the original language version.

David HickeyBarry Kehoe

LOCALISATION FOCUSGAMES.locMARCH 2006 27

Gam

es L

ocal

isat

ionGames Localisation

Barry Kehoe and David Hickey discuss some of the differences between the

localisation of business-type applications and the localisation of video games.

LOCALISATION FOCUS GAMES.loc MARCH 200628

Gam

es L

ocal

isat

ion Internationalisation

The localisation industry in general has benefited from years ofevolution; in the early days internationalisation was a major issue.Over the years, standardisation has featured strongly to improvethe level of internationalisation in applications. This has beenhelped by the creation of generic tools and processes which addgreatly to the efficiency of localising products, enabling companiesto release an increased amount of languages in a reduced time tomarket — and without having to invest heavily to do so.

Unfortunately, at this time this is not the case in games localisa-tion. Where business application developers use Microsoft Devel-opment standards for Windows applications, standard resourcefile types etc., games developers do not. In a video game for PC,the game engine essentially takes the place of the operating system.Games developers wish to maximise the performance of their titleby maintaining high frame rates, increasing the level of processingpower allotted to the Application Interface (AI), and increasing theresolution of textures and polygon counts. They rarely use stan-dard fonts or system True Type Fonts, but usually create their ownor license a font(s) which suits the style of the game they are devel-oping. The end result of this is that games rarely support anyextended characters and have a lot of hard-coded strings with nodynamically resized or scrolling text fields. Although this situationis improving as the games industry matures, in VUG full interna-tionalisation audits are done throughout the games developmentlife cycle and all uncovered issues are addressed before movinginto the QA cycle. The alternative is going into QA with noextended character support (which makes QA’ing the productpretty much useless) and having high numbers of truncations andEnglish text in localised builds (due to hard-coded strings).Another major issue that can arise is the use of concatenated textto build sentences in-game. Although VUG invests in educatingdevelopment teams about the problems of such sentence-buildingsystems, we have at times had to re-write code, or create concate-nation matrices to handle the creation of randomly generated sen-tences at run time. The complexity of this increases with the num-ber of target languages the title is being localised into. To create asystem that can deal with the rules of multiple languages is verychallenging — bordering on impossible. The end result of this dif-ficulty can be that some sentences make little or no sense to thecontext of the game.

Tools and TechnologyBesides using standard file management tools such as Catalyst,Rsync, and also Trados for tracking updates and leveraging, VUGstaff are required to use a wide range of proprietary tools createdinternally and supplied by each development studio. Proprietarytools are used for reviewing components, working with non-stan-dard file formats, internationalisation, compiling resources, andfor compressing audio, text and graphic files. Sound Forge, AdobePremiere, Adobe After Effects and Adobe Audition are used forcreating and editing both audio and video files.

Components such as audio, text, graphics, videos, and combina-tions of audio with text subtitles or lip synching in videos make itvirtually impossible to take advantage of automated testing tools.Various tools have been developed internally for testing compo-nents outside of the game environment, but each and every com-ponent needs to be verified in-game to ensure the end user receivesa high quality product.

The consumers’ expectations for better games drive the constantdevelopment of new technology within the gaming industry. Theintroduction of new gaming technology also increases the demandon PC and console hardware, which in turn results in hardwareupgrades on a regular basis. PC upgrades are required in order tomeet the recommended hardware specification for future projects.The most common PC upgrades would include a variety of

We speak your language.

Everywhere.

www.adapt-localization.comadapt@adapt-localization.com

ADAPTL O C A L I Z A T I O N

LOCALISATION FOCUSGAMES.locMARCH 2006 29

Gam

es L

ocal

isat

ionhigh-end graphic cards, processors and RAM. For our console hard-

ware requirements, we need to ensure we have sufficient debug anddevelopment kits for testing and build creation at all times. Due tothe large amount of console hardware items, all items are checked inand out via an inventory database for tracking and security purposes.

Product AdaptationsAdapting games to meet the requirements of local age ratingsboards and/or government bodies is one of the more challengingareas of games localisation. The challenges faced here are increaseddue to country-specific guidelines being vague, and their interpreta-tion changing based on ‘on the day’ political issues. This requiresVUG to work very closely with its local country offices to reviewand get adaptations approved — so that the product released getsthe target age rating that local marketing teams are looking for.

Much like the movie industry, Video Games have to have anapproved age rating prior to the release of the product. WhileNorth America and most of Europe have similar age ratings criteria,products need to be reviewed and adapted to make them suitablefor release in other countries. The required modifications vary —from reducing the level of violence in a product to modifying char-acter models to changing story lines. It is also necessary to makingchanges to references that may cause issues with local sensitivities incertain countries. Some examples of these would be:

Modifying level of violence in a game: Germany has very strictlaws on violence in video games. These laws relate to how the char-acter is interacting at the time the violence is taking place, e.g. doesthe player character have to kill, and can they complete a missionwithout causing physical harm. It is also necessary to reduce the lev-els of blood and gore that appear in games, as well as reducing thenumber of death screams that happen when a character is killed. Inaddition, the removal of Swastikas (or symbols resemblingSwastikas or Nazi signs, e.g. SS on uniforms) is necessary for theGerman market as it is illegal for a company to sell products inGermany with any of these references visually apparent.

Character Model change: The Japanese games market is one ofthe biggest in the world. However, it is very challenging for westerngames publishers to successfully release titles into Japan. To helpproducts meet Japanese gamers’ expectations it is often necessary tochange character models, e.g. one of VUG’s product IP’s is ‘CrashBandicoot’. In the USA and Europe the Crash Bandicoot model hasthree fingers and a thumb but the Japanese versions of the characterhas four fingers and a thumb — see Figure 2 for a comparison.These changes are required in all graphics, animations, models, boxcovers, POS materials, etc.

Figure 2: ‘Crash Bandicoot’ character modified for the Japanese market(left) and the original for the western market (right).

Story Changes: The majority of issues that occur with stories arespecific to local cultural differences. These tend to be more of anissue for the Asian market. For example, in Warcraft III — whichVUG released several years ago — part of the story involved a sonbetraying his father. Due to the importance of honour and respectattributed to the family unit in Korea, this betrayal aspect of the

Warcract III story had to be modified before the game was allowedto be released in Korea.

Sex/Nudity: Any VUG products containing references of a sexualnature or on-screen nudity, must be carefully reviewed as such con-tent can affect ratings differently in different countries. Consider arecently released product that contains on-screen nudity: for the USmarket ‘censor’ bars had to be used to cover the parts of the charac-ter models that were considered unsuitable or offensive, for theEuropean market no censorship was required and for the Australianmarket the product simply could not be released at all.

The Approval ProcessA major challenge and difference between localising business appli-cations and console games is the approval required from 1st Party(Sony, Nintendo or Microsoft) before the console product can go tomanufacturing. This approval process must be taken into accountwhen creating the localisation schedule. When the game is declared‘Gold’ internally, the Gold version has to be submitted to 1st Partyfor their approval. For VUG, the planned date for submission buildto arrive at the 1st Party’s office must be booked in advance. In turnthis pressurises the VUG team to ensure the Gold date is met. Toassist with achieving on-time delivery, guaranteed overnight couri-ers are regularly used. Occasionally, to ensure the submission buildarrival date is met, VUG team members travel abroad at shortnotice to hand deliver the submission package to 1st Party.

Each 1st Party has individual technical requirements that everygame must adhere to before being released to manufacturing. Themore features (LAN/Online Multiplayer, extra peripherals, etc.) agame contains, the larger the list of requirements. The approvalprocess can take up to eight weeks, depending on the platform. If agame does not meet all the critical requirements, it will fail to beapproved and will have to be resubmitted — complete with expla-nations detailing how each of the issues in the fail report have beenaddressed.

In order for a game to meet its street date, it is extremely impor-tant to ensure the game adheres to all the technical requirementsbefore being submitted for 1st Party approval. Due to the impor-tance of meeting the console requirements prior to submission,Vivendi Universal Games Ireland has a separate group within itsQuality Assurance Department that focuses solely on consolerequirements. With the introduction of new console platforms (PSP,PS3 & Xbox 360) and the demand for online gaming, the approvalprocess continues to present new challenges.

The Dublin office maintains a close relationship with Marketingand Operations. Regular communication becomes paramount whenapproaching the completion of projects. This is because marketingcampaigns and local sales teams acquiring shelf space for a specifictime period are very reliant on the product being released exactly onschedule. Product release is dependent on the project meeting itsGold date and achieving 1st Party approval on schedule, so anypossible delays must be communicated — unforeseen delays canhave a major impact on forecasted sales.

Barry Kehoe is the Localisation Manager at Vivendi UniversalGames Ireland. During his 13-year localisation career Barry hasworked in a number of companies — on both the client and vendor side. He worked briefly in QA and Software LocalisationEngineering, and then Project Management before moving into hiscurrent position. Barry can be contacted at Barry.Kehoe@vugames.ie

David Hickey is the QA Manager at Vivendi Universal GamesIreland. He started working in the gaming industry in 1994 as atechnical assistant in Ironwind Software, Southern California. Hemoved from Ironwind Software to Disney Interactive testing, wherehe assisted in game design for 16-bit game systems, before returningto Ireland in 1998 to work for Vivendi Universal Games. David canbe contacted at david.hickey@vugames.ie

LOCALISATION FOCUS LRC.news MARCH 200630

Loca

lisat

ion

New

sLOCALISATION FOCUS

Localisation Research Centre News

11th Annual LRC Conference13–14 September 2006,

Dublin, Ireland

LRC WebsiteRedesignThe LRC website recently underwent a layout redesign. Check out its improved readability and accessibility at www.localisation.ie

2006 Awards programme AnnouncedDeadline for entries is 01 August 2006

Localisation Focus an Indexed JournalTowards the end of 2005 Localisation Focus underwentan in-depth assessment with a view to it becoming anindexed journal. The LRC is proud to announce thatLocalisation Focus has already been included on twoindexes.

1. Ulrichs Periodicals Directory™ Owned by R.R Bowker Company and named after theeminent librarian Carolyn Ulrich, ‘Ulrichs’ as it is com-monly known, is a global and authoritative bibliographicreference used extensively in libraries, and by publishers,subscription agencies, corporations and researchersworldwide.

2. Inspec Database (belongs to the IEE)Part of a web-based information service entitledEngineering Village 2, Inspec contains over 7 million bibliographic records taken from 3,500 scientific andtechnical journals.

What does this mean for authors?As an author of a paper published in Localisation Focusyou will now gain much greater exposure and recognitionfor your work. Indeed, publication in indexed journals is a pre-requisite for advancement in many academicinstitutes.

Entries are now being accepted for the 2006 edition of the LRC Best Global Website Award—sponsored by Euro RSCG 4D.

For more information and to enter this award visit www.bestglobalwebsiteaward.com

10th LRC Best Thesis Award—sponsored by

For more information and to enter this award visit http://www.localisation.ie/Archive/Awards/Thesis.htm

4th LRC Best Scholar Award—sponsored by

For more information and to enter this award visit http://www.localisation.ie/Archive/Awards/Scholar.htm

Interested in Writing for Localisation Focus?• Send your submission to LRC@ul. • Deadline for next issue is 17 April 2006