clarin seen from the other side of atlanticclarin seen from the other side of atlantic brian...

CLARIN seen fromthe other side ofAtlantic

Brian MacWhinneyCarnegie Mellon UniversityPittsburgh, USA

Iwould like to thank the CLARIN group forallowing me to participate, albeit at a dis-

tance, in the process of developing the compu-tational infrastructure for language studies.Work now being conducted under the CLAR-IN banner both complements and supple-ments work done in North America throughthe Linguistic Data Consortium (LDC,www.ldc.upenn.edu) and the TalkBank Project(talkbank.org ). As the director of the TalkBankProject, I would like to outline ways in whichcollaboration between CLARIN, LDC, andTalkBank can benefit each project and theresearch communities we are serving.It is interesting to compare and contrast theapproaches to infrastructure building taken byCLARIN, LDC, and TalkBank. CLARIN hasworked to bring together widely divergent pro-jects across more than a dozen Europeannations. In this vein, the CLARIN missionstatement notes that its primary goal is, “toturn existing, fragmented technology andresources into accessible and stable services thatany user can share or adapt and repurpose.”Traditionally, work in the European traditionhas focused on the building of tools and ser-vices that can then be deployed across themany European languages. These tools areoften developed through work groups, such asCLARIN, that hammer out consistent stan-dards for metadata descriptions, annotation,and analysis tools. Although this process cansometimes seem ponderous, it has the greatadvantage, particularly in the European con-

text, of developing momentum behind ashared consensus regarding standards. Oneclear example of this progress is the ongoingdevelopment of the IMDI metadata codingsystem and its web-based implementationthrough Arbil and the IMDI Browser. (cor-pus1.mpi.nl). Although this system requiresmore careful coding than OLAC (www.lan-guage-archives.org), it offers greater detail forprojects that wish to use a greater array ofmetadata. Moreover, can be linked in usefulways to further tools for data analysis.

On the North American side, the two majorsites of infrastructure development have beenLDC and TalkBank. Although these two sys-tems have often collaborated on specific pro-jects (SBCSAE, SCOTUS, AG-Toolkit), theytake markedly different approaches to dataencoding. LDC has focused its efforts on thearchiving and publication of a large number ofcorpora that use very different data encodingstandards. TalkBank, on the other hand, hasfocused on the development of a database oftranscriptions that adhere to a tightly-definedannotation system, called CHAT which isaccompanied by a detailed XML specificationand validator (talkbank.org/software). Theadvantage of the approach taken by LDC isthat a wide variety of data can be more readilyingested preserving the original format. Anadvantage of the TalkBank approach is thatdata can be analyzed more uniformly with asingle set of tools. Another advantage of theTalkBank approach is that the standards canallow fields, such as child language(CHILDES), aphasia (AphasiaBank), or sec-ond language acquisition (BilingBank), toenforce standards regarding data quality andanalysis, sometimes using standardized datacollection protocols.

Both LDC and TalkBank have emphasized theimportance of providing maximally openaccess to language data. TalkBank providestotally open access to nearly all of its database.Moreover, TalkBank transcripts linked tomedia can be played back directly from thebrowser interface (talkbank.org/browser). Thestrategy of maximizing open access to materialshas played an important role in generating over

4500 published articles based on use ofTalkBank data. Happily, this policy of promot-ing open access is now being forcefully sup-ported by recent initiatives at the ESF, NSF,NIH, and LSA that support and often mandateopen, public sharing of research data.Given recent advances within CLARIN, LDC,and TalkBank, we can begin to see a variety ofways in which these projects could usefully col-laborate. Let me provide three concrete exam-ples.o I have been working with colleagues inDenmark to build up the DK-CLARIN data-base for Danish spoken and written texts.Because many Danish researchers are particu-larly interested in Conversation Analysis (CA),we extended the CHAT system to provide afull range of support for CA transcription. Toprovide a metadata access to this database, weare using the IMDI system. During the processof integration into IMDI, we developed toolsthat to automatically construct IMDI metada-ta from CHAT files not just for the Danishcorpora, but for all of the TalkBank andCHILDES corpora. Using these tools, we havenow incorporated a complete mirrored copy ofthese corpora inside the MPI repository at(corpus1.mpi.nl).o We have developed a set of programsdesigned to convert to and from CHAT andother popular formats such as ELAN, Praat,EXMARaLDA, WaveSurfer, and Transcriber.o On the higher level of cross-disciplinarycooperation, TalkBank, LDC, and CLARINhave all begun to explore the construction of ashared cyberinfrastructure for the social sci-ences and humanities. In 2009, I chaired anNSF panel that issued a report on this subjectthat can be located at (talkbank.org/dreams).This report is very much in line with the newDASISH initiative in which CLARIN will beparticipating.The methods and goals of LDC, TalkBank,and CLARIN are moving quickly toward con-vergence. Hopefully, we can develop methodsfor building up solid links between these pro-jects, as they seek to integrate themselves intothe wider cyberinfrastructure for the sci-ences. C

Number 11-12, 2010, September-December

22 NNoo 1111--1122 // NNeewwsslleetttteerr CCLLAARRIINN

Editors’Foreword

Marko Tadi}& Dan CristeaCLARIN Newsletter editors

Dear readers,this double issue of CLARIN News-

letter according to the CLARIN projectDescription of Work was planned to be thelast one. It was our intention to have twoseparate issues, number 11 at the end of2010 and number 12 that would step into2011 and thus accomodate the six monthprolongation of the project and yet remainwithin the boundaries of project.Due to a series of unwanted events, it turnedout that our initial editorial plan had to beadjusted, so you have in front of you thedouble issue 11-12 that covers the plannedperiod by dates, but certainly not by the timeof its appearance.Therefore we decided that CLARIN project,its consortium partners and the communitythat has built around them in previous threeyears deserve a proper final and closing issue.We hope that number 13 will not bear anytrace of misfortune. On the contrary. Webelieve that it will pave the way to CLARINinfrastructure in its full form.But about the next issue you will be able toread in the next editorial. Let’s concentrateon this one that is in front of your eyes.

How CLARIN is perceived from the otherside of Atlantic you can read in the frontpage contribution by Brian MacWhinneywhom, we hope, we do not have to intro-duce.How LT scene, after some time, has becomemore vigorous again and how research infra-structures with their expected and plannedsustainability has become important, youcan find out from theMemorandum ofUnderstanding betweenCLARIN and META-NET. We publish it inits entirety because webelieve this gives thelong awaited opportu-nity to LT communityto coordinate its effortsat large.Peter Wittenburg ispresenting several possi-ble offsprings of theprojects CLARIN andDARIAH within the new call for RI pro-jects. We will certainly hear more aboutthem in the issue to come.One of the most notable use cases of usage ofLT in the humanities, i.e. CLARIN-support-ed project about the analysis of folk tales ispresented by Piroska Lendvai and ThierryDeclerck. This topic has been presented atseveral digital humanities conferences and isrising the interest ever since.The demonstrative combination of textualand geographical analysis of 17th centurymanuscript of Romanian Nicolae Milescu'sIter in Chinam, is another case that clearlyshows how digital humanities depend on LTbut also how LT has to be combined withother types of information.

Our middle pages are traditionally orientedto presentation of important events connect-ed to CLARIN. First we have the report by

Hetty Winkel from SDH-NEERI 2010 con-ference that took place in Vienna in Octoberand is to be considered the major event injoint organisation of CLARIN andDARIAH.

It is followed by reports from two LT con-ferences that embraced Europe from twosides, South-East and North. These areFormal Approaches to South-Slavic and

Balkan Languages(FASSBL7) and thefourth Baltic HLTconference. Thesetwo show how LThas spread accros theEurope and is grow-ing mature at theregional and not justnational level.

The first META-FORUM is present-ed by Aljoscha

Burchardt and Georg Rehm in its fullstrenght since after the LREC2010 confer-ence, it has been the LT event in Europe thatcollected the largest number of participants.

Our issues regularly end with reports on thestatus of LR&T from different Europeancountries. There is no need to drop this prac-tice, so we are bringing you the reports fromfour countries in this double issue: Israel,Iceland, Turkey and Slovakia. Each of themdepicts the different situation and level ofdevelopment of LT, but what can be noticedis that all these efforts are oriented to a com-mon goal. If we contributed so far to thiscommon goal with our editorial work on theprevious issues of CLARIN Newsletter, wecertainly hope that we did not disappointyou with this one either.

Enjoy your reading! C

List of nationalcorrespondents

Austria

Gerhard Budin

Belgium – Flanders

Inneke Schuurman

Bulgaria

Svetla Koeva

Croatia

Marko Tadi}

Czech Republic

Karel Pala

DenmarkBente Maegaard

Hanne Fersøe

ELRA/ELDAStelios Piperidis

Khalid Choukri

EstoniaTiit Roosmaa

FinlandKimmo Koskenniemi

FranceWilliam Del Mancino

Bertrand Gaiffe

GermanyLothar Lemnitzer

Greece

Maria Gavrilidou

Hungary

Tamás Váradi

Italy

Valeria Quochi

Latvia

Andrejs Vasiljevs

Malta

Mike Rosner

Netherlands

Peter Wittenburg

Norway

Koenraad De Smedt

Poland

Maciej Piasecki

Portugal

Antonio Branco

Romania

Dan Cristea

Dan Tufis,

Spain

Nuria Bel

Sweden

Sven Strömqvist

UK

Martin Wynne

Call for contributions

Dear readers of theCLARIN Newsletter,If you have ideas, thoughts,comments, additions, corrections,arguments, questions etc. which areconnected to the CLARIN project,even remotely, please feel free tosend them to us as your contributionat [email protected] or directly tothe editors at [email protected] [email protected].

CCLLAARRIINN NNeewwsslleetttteerr // NNoo 1111--1122 33

Introduction

The Network of Excellence “Technologiesfor the Multilingual European InformationSociety”, henceforth META-NET, estab-lished by DG INFSO of the EuropeanCommission (EC) under FrameworkProgramme 7, and the “Common LanguageResources and Technology Infrastructure”,henceforth CLARIN, a consortium estab-lished under the Research InfrastructuresProgramme of the EC, recognise the com-plementarity of their objectives and declaretheir intent for multi-level collaboration.

Scope and Objectives of theParticipants

CLARIN is a research infrastructure in theESFRI framework. It constitutes a large-scale pan- European collaborative effort tofacilitate research by coordinating and mak-ing existing language resources and toolsavailable and readily useable for the SocialSciences and Humanities (SSH) on a sus-tainable basis. CLARIN offers resources(data and tools) and services to allow com-puter-aided language processing, addressingone or more of the multiple roles languageplays (e.g., carrier of cultural content andknowledge, instrument of communication,component of identity and object of study)in the Humanities and Social Sciences andneighbouring disciplines in the broadest pos-sible sense. By doing this CLARIN will pro-vide an advanced and innovative environ-ment for e-Research and e-Science whichwill make language resources and technolo-gies (LRT) visible, accessible and interopera-ble based on portals and standards.

META-NET is a Network of Excellencededicated to the technological foundationsof the European multilingual informationsociety. By developing a long demandedopen resource exchange and sharing facilitycode-named META-SHARE, it will createthe basis for developing the necessary tech-nologies and applications for the multitudeof European and other relevant languages.By forging an alliance of researchers, tech-nology providers, corporate users, language

professionals and other stakeholders and bydeveloping together with these partners ashared vision and a strategic research agenda,META-NET shall prepare an ambitiousjoint effort needed for realizing Europe’sdigital single market and information space.By building bridges to neighbouring tech-nology areas, META-NET will approachopen research problems in collaborationwith other fields such as machine learning,social computing, cognitive systems, knowl-edge technologies and multimedia content.

Orientation and TargetCommunities of the Participants

CLARIN is oriented towards the deploy-ment and adaptation of language resources,tools and technologies in order to offer lan-guage-related services primarily to the SSHresearch community (e.g., scholars andresearchers in history, anthropology, sociolo-gy, linguistics, literary studies, computation-al linguistics, law, etc.). CLARIN aggregateslanguage data and tools relevant to a widerange of languages, including endangeredlanguages, and to content that has its ownproperties and processing requirements, be itin the form of text, speech or other modali-ties. Catering for this community entailsoffering services to researchers who may havelimited IT skills or interest, and thereforerequire highly automatic processing of dataand content. CLARIN anticipates the needof modern research to build interdisciplinaryvirtual communities working on virtual col-lections and building virtual workflows totackle their research questions.

CLARIN is in the process of establishingitself as an ERIC that operates a federationof (interconnected) national CLARINCentres. They offer language resources andservices to the whole European SocialSciences and Humanities communities in apersistent and sustainable way mainly guar-anteed by strong repositories. The right todecide about financial and licensing condi-tions will always remain with the owners,but CLARIN favours and actively promotesa free and open access and open source poli-cy. It builds on existing LRT centres (includ-

ing repositories, service centres and centresof expertise) and infrastructure initiativesand plays a pioneering role in the emergingEuropean ecosystem of infrastructures,thereby significantly strengthening theEuropean Research Area.

META-NET aims at supporting HumanLanguage Technology (HLT) development,i.e., technologies underpinning language-savvy products and services for the digitalinformation society and a single online mar-ket. META-NET focuses primarily on EUlanguages (official, national but also regionalones) and the languages of EU’s major part-ners, i.e., more on contemporary businessand everyday communication language(broadly conceived). All language-enabledcommunication modalities, i.e., text, speech,still and animated images, sign languages,etc. play an important role. Equally impor-tant is the cross-lingual dimension for multi-lingual Europe.

META-NET aims at creating a sustainablenetwork of repositories of language data,tools and related web services documentedwith high-quality metadata, aggregated incentral inventories allowing for uniformsearch and access to resources. Data andtools can be both open and with restrictedaccess rights, free and for-a-fee. This net-work of repositories, META-SHARE, tar-gets existing but also new and emerging lan-guage data, tools and systems required forbuilding and evaluating new technologies,products and services. In this respect, reuse,combination, repurposing and re-engineer-ing of language data and tools, play a crucialrole. META-SHARE will eventually be animportant component of a language technol-ogy marketplace for HLT researchers anddevelopers, language professionals (transla-tors, interpreters, etc.), as well as industrialplayers, especially SMEs, catering for the fulldevelopment cycle of HLT, from researchthrough to innovative products and services.To achieve its goals, META-NET intends toplay a central role in building an ecosystemwith collaborating projects tackling different

Memorandum of Understanding betweenCLARIN and META-NETAn important agreement between two keyinitiatives in LT community

Continued on the next page �

44

facets of language resources, technologydevelopment and evaluation.

The establishment of such a network mayinclude a joint effort between the EuropeanCommission and the Member Statesthrough available adequate instruments.

Conclusion

CLARIN and META-NET are two comple-mentary initiatives, serving different com-munities with different but harmonizablegoals. They have identified shared interestsand approaches with respect to methods andinstruments in the area of language resourceinfrastructures concerning both technicalsolutions and organizational models.

Both CLARIN and META-NET aim atdescribing language resources with appro-priate metadata, as well as their sharing,maintenance and reuse. While the fact thatthe two projects target different user com-munities may call for different metadataprofiles, CLARIN and META-NET envis-age an open metadata domain of data andtools, based on interoperable metadatadescriptions for categories of data and toolsrelevant to their joint user communities.The purpose of interoperable metadatadescriptions is to enable cooperation, thepossibility to link data objects and/or theirdescriptions, the exchange of data of interestto both communities and the way towardsfostering standardisation and interoperabil-ity of data and tools, wherever possible andnecessary. Both parties declare their intentto pursue this cooperation and to buildupon their deliberation on an agreementconcerning existing and emerging stan-dards.

At the time of signature, CLARIN andMETA-NET intend to collaborate in thesetup of language resource repositories,resource identification procedures, metadatainfrastructure, as well as legal issues in shar-ing and distributing language resources.

The parties also agree that while they retaintheir integrity and pursue their goals, theywill explore the possibility to give access todata objects residing in their respectiverepositories. The parties agree to prepare, bySeptember 2010, an initial list of concreteareas and mutual objectives on which to col-laborate, and plans for implementing this.

Signed on behalf of CLARINSteven Krauwer, Coordinator of CLARIN

Signed on behalf of META-NET

Hans Uszkoreit, Coordinator of META-NETC

Peter WittenburgMPI, Nijmegen

The preparatory phase of the ESFRIresearch infrastructure projects is

close to its end and this also holds forCLARIN. As it looks at this verymoment CLARIN seems to be success-ful in so far that about 18 countries havesignaled their interest to sign the memo-randum of understanding to participatein the ERIC and there are a few addi-tional countries from which we assumethat they will participate, but yet did notfinish the national roadmap discussionscompletely. Of course the CLARINExecutive Board and the colleagues fromthe national teams have done theirhomework at several layers. In this arti-cle we do not want to explain theCLARIN internal activities – at EU andnational level – but want to describe theactivities to embed CLARIN in a varietyof activities. Yet we cannot say whetherall these activities will lead to success,but yet we are rather optimistic.

Data Infrastructure

One CLARIN expert was appointed tobecome a member of the high level

expert group of the EC to work out astrategy for the management of ResearchData in the coming decades. The reportwith the name “Riding the Wave”1 washanded over recently to commissionerNeelie Croes. One result of the reportwas the insight that organizing datamanagement and access will be a multi-layer task as illustrated in the figure. Atthe second layer there would be com-munity oriented centers employingexperts with decent knowledge aboutthe workflows, formats, semantics andencoding principles of the respective dis-cipline. At the third level we would havecommon data services which are widelydiscipline independent and given bylarge data centers. It is obvious that datacuration is a task for all actors startingwith the researchers as data creators. Thesmooth functioning of such a multi-layer system can only be guaranteed ifmutual trust has been established. Since CLARIN has a very explicit strat-egy about data and technology centers asthe backbone of a research infrastructurefrom its beginning, CLARIN was a can-

NNoo 1111--1122 // NNeewwsslleetttteerr CCLLAARRIINN

MMaannaaggeemmeenntt ooff ddaattaa iinn tthhee ccoollllaabboorraattiivvee sscceennaarriioo aass ddeessccrriibbeeddbbyy tthhee HHiigghh LLeevveell EExxppeerrtt GGrroouupp oonn SScciieennttiiffiicc DDaattaa

didate community to participate in theEUDAT proposal to the EC. EUDAT isa broad consortium existing of strongcenters from 15 communities from thevarious discipline areas on the one handand 11 of the strongest data centersfrom various countries on the otherhand. By bringing together such a vari-ety of different actors it wants to estab-lish exactly the Collaborative DataInfrastructure as indicated in the figurewhich is, given the heterogeneity, a chal-lenging task. CLARIN will participatein three ways if the proposal will beaccepted: (1) A CLARIN memberwould act as Scientific Coordinator; (2)It would participate to establish a safedata preservation infrastructure; (3) Itwould participate to test out a genericweb services landscape where large datacenters will host the services. Thus with its participation in EUDATCLARIN showed its awareness aboutthe need to adopt generic solutionswhere possible. The following CLARINcenters were accepted to join the consor-tium: CU Prague, U Tübingen, MPI.Partly the already started collaborationbetween CLARIN and DEISA2 in theREPLIX3 project can be continued.EUDAT will be coordinated by CSC4 –the Finnish national computer center.Very much related to the EUDAT pro-ject is the German Radieschen projectwhich aims at working out a roadmaptowards a national data infrastructure.This has already been granted.

Humanities and Social ScienceCluster

The EC launched a call for a joint activ-ity between the 5 existing humanitiesand social sciences research infrastruc-tures initiatives (DARIAH5, CLARIN,CESSDA6, ESS7, SHARE8) to look forcommonalities and to exploit synergies.A board with one representative per ini-tiative was built to work out tasks whichare of interest at least for a few of the ini-

tiatives and to establish principles for theconsortium building. Accepting thatDASISH (Data Services Infrastructurefor the Social Sciences and Humanities)also will contribute to the eco-system ofinfrastructures, we finally worked outthe following major tasks: (1) getting adeep understanding about data organi-zations in the disciplines and designing aroadmap towards a higher degree of har-monization; (2) Improve the quality ofthe social science surveys; (3) offerrobust deposit and long-term preserva-tion services; (4) improve metadataquality and the infrastructure integra-tion with respect to AAI (authenticationand Authorization Infrastructure), PID(persistent identifiers) and metadata; (5)improve the visibility of all tools and ser-vices; (6) develop a cross-disciplinaryannotation platform useful for all SSHresearchers; (7) improve the legal andethics situation and (8) carry out lots ofactivities in education and training.It was agreed that maximally 5 partnersper initiative could join the consortium.

CLARIN is represented by UPFBarcelona, U Tartu, U Copenhagen, UBergen and MPI following a number ofpublished criteria and will participate inall activities except (2) which is only ofinterest for the social scientists. DASISHwill be coordinated by U Gothenburgand MPI will coordinate the researchactivities.

CLARICLE

The EC launched another call devotedto foster the construction work ofCLARIN. Central for CLARICLE is thecreation of the Integrated EuropeanCorpus and the Integrated EuropeanServices making use of the CLARINresearch infrastructure. Thus the CLAR-ICLE intention is to fill the infrastruc-ture with useful data and services and tomake these available to the users withthe help of a user workbench that offerseasy access to all resources, services/tools


Embedding of CLARIN FutureTowards the end of preparatory phaseof the ESFRI research infrastructures


and to other useful infrastructure ser-vices. The major challenges were todefine the content areas for theIntegrated European Corpus and to seewho of the CLARIN centers can con-tribute with accessible resources and ser-vices/tools that fulfill the CLARINrequirements widely. CLARICLE alsohas the intention to act as part of theemerging eco-system of infrastructuresand wants to collaborate on executionwork spaces with strong data centers.

The following major work dimensionshave been identified for CLARICLE:

1) standards,

2) IPR issues;

3) building the integrated corpus;

4) building the integrated servicesdomain;

5) building cross-domain services;

6) adapt the infrastructure;

7) carry out a user survey and launchcalls for small project proposals;

8) develop a workbench, annotation andvisualization tools.

The integrated corpus will address inparticular the needs of the followingcommunities: (a) linguistic variationresearch based on written, spoken, mul-timodal and sign language resources; (b)news corpora relevant for historians,social scientists and linguists; (c) record-

ings and proceedings of parliament ses-sions and (d) oral history.

The consortium includes 19 CLARINmembers from a variety of countries.9

INNET

Another small proposal with 4 partnerswas submitted separately to ask for sup-port for the world-wide activities in doc-umenting and archiving endangered lan-guages. Mostly it is not known even tolinguists that from the 6500 languagesstill spoken every week one is dying andwith it an enormous knowledge aboutlinguistic systems, environmentalaspects, etc. In the realm of the DOBESproject10 the MPI has set up 13 remoterepositories (see stars on the map) at var-ious places world-wide so that a dataexchange can be carried out. We haverequests for 10 additional repositoriesand the wish to have regular and closeinteractions between the various teams.

The proposal thus includes funds for settingup such repositories, for local training cours-es, for regular workshops and conferences tofoster an exchange about methods and tech-nologies and for some educational effort.The intention is to fully adopt all CLARINrequirements in this international exchange.From CLARIN MPI, HAS-RIL and UCologne (CLARIN D) are involved, sincethey have a close relationship to the endan-gered languages topic.

Summary

We can say that CLARIN was very active inplanning its construction phase and inbringing services to users, in participating inlarger European activities to exploit syner-gies and in activities to spread CLARINmethods even worldwide. Synchronizingwith all participants about the many detailsand getting all these proposals finished was avery time consuming job during the last halfyear in 2010. Finally between 25 and 30CLARIN members are involved in theseactivities. We could see that for most of theseactivities the state of the centers and thequality of their services was a crucial criteri-on. We assume that in future similar callswill be launched, since building researchinfrastructures is a strategic goal in Europe.Yet we do not know the outcome of the eval-uation process. C

References1 http://cordis.europa.eu/fp7/ict/e-

infrastructure/docs/hlg-sdi-report.pdf

2 http://www.deisa.eu/

3 http://www.mpi.nl/research/research-projects/the-language-archive/projects/replix-1/replix

4 http://www.csc.fi/english/pages/parade

5 http://www.dariah.eu/

6 http://www.cessda.org/

7 http://www.europeansocialsurvey.org/

8 http://www.share-project.org/

9 During the editing of this issue of CLARINNewsletter we received the announcement fromthe evaluation committe that CLARICLE did willnot be funded (editor’s remark).

10 http://www.mpi.nl/DOBES .

66 NNoo 44 // NNeewwsslleetttteerr CCLLAARRIINN

CCLLAARRIINN NNeewwsslleetttteerr // NNoo 44 77

Piroska LendvaiHASRIL, Budapest, HungaryThierry DeclerckDFKI, Saarbrücken, Germany

In line with one of CLARIN's main goalof deploying language resources and

technology to enable e-Humanities,recently a use case has been established,dedicated to the automated linguistic andsemantic annotation of folk tales.Emerging from the AMICUS project'sresearch and networking activities(Automated Motif Discovery in CulturalHeritage and Scientific CommunicationTexts1), CLARIN has established specialcooperation with D-SPIN2, the Germancontribution to CLARIN, whereby theuse case is currently under implementa-tion.

The Folk tale use case aims at establishinga combination of linguistic and domain-specific content descriptors from the fieldsof Literature, Ethnography, and Folklore,offering semantic annotation to describeand model the complex interdependenciesbetween a tale's characters, function roles,and events, and their designated linguisticvehicles. The basic linguistic processingchain is enabled by the WebLicht web ser-vices framework (D-SPIN project); itsoutput is being mapped onto TEI3 andISO4 standardized annotation structures.The semantic annotation layer is based onvarious generic semantic resources (forexample a family ontology, a temporalontology, WordNet5 and FrameNet6) and aspecialized annotation schema that origi-nates from a proposal by Vladimir Propp,based on literary and ethnographic studies(Propp 1968). The linguistically andsemantically annotated tales are toimprove querying and corpus-basedresearch by literature scientists and other

Humanities specialists, as well as bylaypersons.

Drawing on the connection with theAMICUS project, we established a broad-er set of cooperation partners, prominent-ly the Swedish School of Library andInformation Science — providing exper-

tise on the topic of motifs across variousliterary genres —, the UniversidadComplutense de Madrid — contributingontological resources like ProppOnto andProtoPropp, the Fairy-Tale Generator —,and the University of Pittsburgh: creatorof the Proppian fairy tale Mark-upLanguage (PftML). On the basis of theirresources, the first step taken within theuse case was to develop an extended anno-tation schema for folk tales – APftML(Augmented Proppian fairy tales Mark-upLanguage), see (Scheidel & Declerck2010), which was presented at the AMI-CUS workshop7 (Vienna, 21 October2010), a satellite event to the SDH 2010

conference8, co-organized by CLARINand DARIAH.

A first dissemination of our joint worksucceeded at LREC 2010 (Lendvai et al.2010a), where the motivations andresearch setup were presented to the lan-guage technology community. A more

focused presentation, including recentdevelopments, has been submitted suc-cessfully to the Digital Humanities confer-ence9 (DH2010) in London (Lendvai etal. 2010b), giving us clear indication thatour work in CLARIN is going the rightway: the DH conference is the annualinternational conference for digital schol-arship in the humanities. We demonstrat-ed our data, approach, as well as researchand implementation methods there withina two-hour poster session, exchangingideas with colleagues from differentresearch fields. In the Digital Humanities

The CLARIN Folk Tale Use Caseat DH-2010 and ElsewhereDeploying language resources and technologyto enable e-Humanities

DDiiggiittaall HHuummaanniittiieess 22001100:: tthhee ffuullll aauuddiittoorriiuumm aatt KKiinngg’’ss CCoolllleeggee,, LLoonnddoonn


community there is notable interest forobtaining folktales annotated with fine-grained linguistic and semantic annota-tion, which our use case targets to create.Such systematically grown and automati-cally assigned markup is currently unavail-able to Humanities researchers, except formetadata supporting general textual classi-fication. Our choice of TEI as the basic

annotation scheme for textual informa-tion, likewise suggested by LaurentRomary and Andreas Witt within a D-SPIN meeting, has also been confirmed,since TEI appeared to be frequentlyapplied and cited. We can thus report onencouraging signals received from both theHLT and DH communities with respectto the use case. An interview conducted atthe end of our poster presentation atDH2010, available online, provides ashort audio explanation of our work.10 C

ReferencesScheidel & Declerck (2010) Antonia Scheidel &

Thierry Declerck. APftML – Augmented Proppianfairy tale Markup Language. In: Proceedings of thefirst International AMICUS Workshop.

Lendvai et al. (2010a) Piroska Lendvai, ThierryDeclerck, Sándor Darányi, Pablo Gervás, RaquelHervás, Scott Malec & Federico Peinado.Integration of Linguistic Markup into SemanticModels of Folk Narratives: The Fairy Tale Use Case.In Proceedings of the Seventh conference onInternational Language Resources and Evaluation(LREC 2010).

Lendvai et al. (2010b) Piroska Lendvai, ThierryDeclerck, Sándor Darányi & Scott Malec. ProppRevisited: Integration of Linguistic Markup intoStructured Content Descriptors of Tales. InProceedings of DH 2010.

Propp (1968) Propp, V. A. Morphology of the Folktale.Publications of the American Folklore Society.University of Texas Press, 2nd edition.

Notes1 http://amicus.uvt.nl2 http://weblicht.sfs.uni-

tuebingen.de/englisch/index.shtml3 http://www.tei-c.org/index.xml4 http://www.iso.org/iso/standards_development/

technical_committees/other_bodies/iso_technical_committee.htm?commid=297592

5 http://wordnet.princeton.edu/6 http://framenet.icsi.berkeley.edu/7 http://amicus.uvt.nl/amicus_ws2010.htm8 http://www.dariah.eu/index.php?option=com_con-

tent&view=article&id=135&Itemid=1979 http://dh2010.cch.kcl.ac.uk/10 Downloadable at: http://www.arts-

humanities.net/audio/interview_thierry_declerck_piroska_lendvai_dh2010.


Daniela Dumbravu

aAnamaria CiucanuGeorgiana C

u

aru

aus,uFaculty of Computer Science,Alexandru Ioan CuzaUniversity of Ias,i

CLARIN represents one of the bestEuropean research infrastructures

which aim is to make associationsbetween humanistic disciplines andinformatics, an increasingly visible ten-dency in contemporary interdisciplinarystudies. For humanists themselves, oneof the major challenges is the integratedarticulation of the conceptual languageswithin the various disciplines. There-fore, large scale infrastructures are need-ed for publishing and sharing data; bothinformaticians and humanists are intrin-sically involved in this process. It seemsto us that the main direction of furtherresearch projects will be to place or re-construct knowledge in large scale infra-structures. The project described below– “Nicolae Milescu's Iter in Chinam(1676). Visual, computational andencyclopaedic reconstructions” –embodies an enormous set of data, fromthe 17th century, and dedicated toNorthern Asia and Beijing.

General context

Nicolae Milescu, a prominent figure ofthe European political and intellectualelite in the XVII century, was assignedthe task of rigorously exploring thetranscontinental travelling routesbetween Europe and China. This result-ed in Milescu's iter in Chinam (1676)and his encyclopaedic description of theNorthern Asian space, which he present-ed to the Moscow Parliament in 1678.The purpose of our research is to recon-struct iter in Chinam virtually, employ-ing the Google Earth tools and theNatural Language Processing (NLP)technology, and integrating temporaland spatial information (GIS), as well ascartographic and encyclopaedic data.Furthermore, the digitalization of themaps which incorporate iter in Chinamwill allow us both to analyze them com-paratively and to correlate them withMilescu's description of Northern Asia.Multimedia applications will make itpossible to uncover the connections

between the Asian and the Europeanspace which Milescu envisaged in theXVII century, and which remainunknown to the international commu-nity, despite their scholarly importance.

The NLP Module

NLP is used in the project to extracttemporal information from a corpus oftexts which incorporates geographicaldata. Utilizing shallow-parsing and pat-tern-matching techniques, spatial infor-mation can be extracted from this col-lection of texts. The NLP modulereceives as input a plain text and returnsa set of annotated versions of the text,one for each type or information (tem-poral and spatial); spatial expressionswill be marked using SpaceML1 andtemporal ones using TimeML2. Theentire annotation process is automated.Between different versions of annotationseveral links will be created. Employingthese annotations will make it possibleto extract automatically all those correla-tions which are emphasized in Milescu'sitinerary, if we consider its spatial andtemporal development. Methodolog-ically, one can obtain spatial and tempo-ral annotation with NLP as follows: i)identify the nominalizations of thosetemporal and spatial expressions whichcan be extracted from the text (in thiscase a historical corpus); ii) highlightspatial and temporal relationshipsbetween the spatio-temporal entities evi-dent in those expressions. We hypothe-size that a significant part of these spatialdata and relations between them can bere-elaborated (standardized) and subse-quently located into the GIS database.

The Cartography Module (CM)

From this perspective, iter in Chinam: A.contains a corpus of texts which lendsitself to NLP processing (see infra). B.allows for a Google Earth interface, aninteractive platform which can select,extract and visually represent thedescription of this iter. Within this

Nicolae


framework, the project seeks to: i) createan XML document which wouldemphasize the temporal, spatial, topon-imic and etnonimic references and thenetworks for global communication inthe XVII the century; ii) re-create theiter in Chinam route according to thecharacteristics mentioned at points Aand B in this paragraph. The project uses the Google Earth(GE)/Maps (GM) API's for Javascript,integrated in a C# software which storesall the information that can't be foundin GE/GM. In practical terms: 1) CMreceives a xml file set from the NLPModule, which then changes, depend-ing on its category. XML's should con-tain data on the space, time and culturedescribed by Milescu in his journal; 2)The changes could be, for example,from a list of places with links inbetween, processed by the NLP Module,into a KML (XML read by GE) with theplaces found in the GE database, theircoordinates and a KML with theunknown places related to those previ-ously identified. The distance will bemeasured in versts (Russian unit oflength; 1 verst = 3500 feet or 1,0668km). Therefore, the CM is responsiblewith creating algorithms for detectingthe coordinates of the unknown loca-tions, depending on several pieces ofinformation extracted from text, i.e.17th century cities, distances, directionsand narrator orientation. There is anintrinsic link between space and time, aswell as between rich historical data. Thesecond set of information will be placedinto info balloons, and it will constituteour specialized database. The user canalso activate levels of information forbuildings, vegetation, info balloons,geographical and ethnographical data.Most of this information and detailswhich may depend on the 3D space willalso be stored in a special database, incase the elements do not exist in the GE

environment. Parallel activity betweenthe 2D GM scene and the 3D GE viewport would be preferred.

Expected benefits of the project

A multimedia application which wouldincorporate the information obtainedand processed following the above men-tioned practical stages of this project; aRomanian mini-corpus adnotated tospatiality, based on the aforementionedtextual and cartographic sources; theelaboration of articles with an educa-tional role and the promotion of a localmedia campaign dedicated to iter inChinam and to Nicolae Milescu; last,but not least, the elaboration of a tech-nology of spatio-temporal adnotation oftexts and their linking with GIS visualsoftware.If successful, the innovative technologiesdeveloped by this project create thepremises of further developing extreme-ly diverse applications, all rooted in tex-tual information, such as: educational:e-learning platforms (level: primary andsecondary school, high-school, under-

graduate, adult learning); museology:input in virtual scenarios for the recon-struction of knowledge simultanouslyderived from historical, geographical,ethnological, religious, cartographical,etc. information; editorial: the promova-tion of XVIIth century humanists ofseveral cultures (South-East European,East-Central and Northern Asian).The work described in this article is partial-ly supported by the grant POSDRU89/1.5/S/4944. C

References1 BOUTOURA Cryssoula, LIVIERATOS Evangelos,

“Some fundamentals for the study of the geometryof early maps by comparative methods”, e-Perimetron 1 (1): 60-70; BALLETTI Caterina(2006), “Georeference in the analysis of the geo-metric content of early maps”, in e-Perimetron 1(1): 32- 42; JEANSOULIN Robert, PAPINI Odile,PRADE Henri, SCHOCKAERT Steven (2010),Methods for Handeling Imperfect Spatial Information,Springer Verlag, Berlin-Heidelberg.

2 Marta Guerrero NIETO, M. J. García RO-DRIGUEZ, Adolfo U. ZAMBRANA, WillingtonSIABATO, Miguel A. B. POVEDA (2010),“Incorporating TimeML into a GIS”, IJCLA 1/1-2:269-283.

Milescu's Iter in Chinam (1676)Visual, computational and

encyclopaedic reconstructions


SDH-NEERI, Vienna,October 19-20, 2010

Hetty WinkelUtrecht University

In October 2010 CLARIN organized aconference together with our colleagues

of the DARIAH project: Supporting theDigital Humanities, (SDH2010). Theconference was followed by the secondNetworking Event for ResearchInfrastructures: NEERI2010. The localorganization was in the hands of ourCLARIN colleague Gerhard Budin andhis team of the univer-sity of Vienna and theAustrian Academy ofSciences. The venuewas a centrally locatedbuilding of theTechnical University ofVienna.

SDH and NEERIconferenceformats

SDH2010 consisted of anumber of topical ses-sions where providers and users presentedand discussed results, obstacles and opportu-nities for digitally-supported humanitiesresearch. Whereas the focus of SHD2010was on the types of research made possibleby research computing, NEERI2010addressed the technical, architectural andsocial challenges of building the infrastruc-ture. NEERI focused on what we share andwhat we can learn from each other.Examples of such commonalities are archi-tectural issues, communication with usersand integration of services and tools. The conference was opened by Prof DrWolfgang Dressler. He took the place ofGerhard Budin, who unfortunately had fall-en ill. Prof Dressler is well known to CLAR-IN, being the Austrian member in ourScientific Board. He then introduced Mrs.Barbara Weitgruber, Head of ScientificResearch and International Relations of theFederal Ministry of Science and Research,

who welcomed the participants on behalf ofthe Minister.The keynote speech was delivered by Prof.Neil Fraistat, Director of the MarylandInstitute for Technology in the Humanities(http://mith.umd.edu/) In his speech“Digital Humanities Centers as Cyber infra-structure”, Neil took us on a virtual tourthrough his institute, introducing us to hiscolleagues and their daily tasks. Thus paint-ing a picture of what a DH centre is allabout. A successful center can incubateimportant research, foster a new generationof scholars, devise creative modes of gover-nance, develop a variety of strategy for fund-ing and build digital collections and tools.“Strong local partnerships” and “Centers asnexus for local and global”, were some of hiskey phrases. There are also pitfalls: the insu-

larity of centers, competition among centers,national boundaries, and cultural dividesbetween language communities. He alsostressed the importance of reaching out tothe users: how can digital technologies facil-itate a broader engagement with differentpublics and what are the implications foruniversities in the new digital media, and theimplications for the humanities. Neil men-tioned centerNet: a network of 100 DH cen-ters in 19 countries, and steering committeesin Asia Pacific, Europe, North America, andthe UK& Ireland (http://digitalhumani-ties.org/centernet/). He also referred toCHAIN, the Coalition of Humanities andArts Infrastructures and Networks. The aimof CHAIN is to support and promote theuse of digital technologies in research in thearts and humanities (http://www.arts-humanities.net/chain). An interesting ques-tion from the audience was if Neil sawDigital Humanities as becoming part of the

main stream of humanities research, or thatit might run the risk of standing too muchon its own. This was a risk he acknowledged,and we should do our best to prevent this.Another risk factor could be the pressure onyoung humanist researchers to publish;which would prevent them from working onthese innovative research methodologies.

Spreading out to differentfields

The programme then continued with paral-lel sessions in the fields of Archeology,Manuscript Studies, Endangered Languagesand Language Variation, and LanguageVariation and Digital Technology. The sec-ond day started with a plenary session onNarrative Psychology, followed by four par-allel sessions on Socio-economic history,Musicology, Literary history and the sessionon Language technologies in humanitiesstudies. The last one was convened by TamásVáradi, and presented the results of the Callfor Humanities projects from CLARIN.

It is impossible to describe in more detail allthese sessions here, so I refer to the websitewhere the abstracts of all the presentationscan be downloaded (see below).

I would just like to mention two examples.The presentation by Frans Wiering ofUtrecht University on Musicology was veryinteresting and lively. His area is music infor-mation retrieval (MIR), and he had some

Supporting Digital Humanities at LargeAn Impression

TThhee RReesseeaarrcchh IInnffrraassttrruuccttuurreess ppaanneell ppaarrttiicciippaannttss

TThhee kkeeyynnoottee ssppeeaakkeerr NNeeiill FFrraaiissttaatt


nice examples of music to show us the roleMIR can play in musicology research. Hisconclusion is that the benefits for MIR arenot solely in supplying tools and data, but inhelping the community to formulate itsdemands and convince them that they needthe MIR tools, and more generally, ane-Infrastructure. This is exactly what ourCLARIN colleagues in WP3 in the Call forHumanities projects have been dealing with,the results of which were presented in thesession chaired by Tamás Váradi. The pro-jects that were supported by CLARIN withexpertise and advice, presented themselves inthis session. A conclusion of Koenraad deSmedt, the coordinator of the Humanitiesproject, was that it was a real challenge tobridge the gap between research questionsand the way to apply the technology. This isan issue that keeps coming back, in this ses-sion, but also in the talk of our keynotespeaker and in the panel discussion onWednesday.

What is the message to takehome?

The conference was concluded with a paneldiscussion on the Future of ResearchInfrastructure and Humanities: “'What isthe message to take home”?

Peter Doorn of DARIAH convened this ses-sion, and there were representatives from theEuropean Science Foundation (AriannaCiula), ESFRI SSH (Peter Farago),DG-INFSO (Wim Jansen), HERA (SeanRyder), DG Research (Lorenza Saracco, also

CLARIN project officer), CLARIN (StevenKrauwer), and DARIAH (Laurent Romary).

In this brief report, I can only touch uponsome of the issues that were addressed by thepanel. Not surprisingly, one of them wasabout the users: “catch them when they areyoung”, one of the panelists insisted: involvethem at an early stage, include it the educa-tion at PhD level or sooner. Another issuewas about how difficult it is to measure ourimpact as data infrastructure, and how wehave to find ways to do this. Then there isthe problem of sustainability, also at gover-nance level. Many countries are prepared tosign an ERIC, but governments and policieschange, so this might be a problem if youwant stable funding. Open and sustainableaccess: a key issue in order to make thiswhole thing work. We should not try andsolve this on our own. Wim Jansen men-tioned a recent report of the High levelgroup on scientific data: “Riding the Tide”,which is calling for joint future thinking andaction in the field of integrating data infra-structures. Sean Ryder also mentioned theneed for more coordination of the fundingagencies. In the European planning for newresearch programmes, CLARIN and DARI-AH could play a role in advising on digitalcomponents for new research. Such expertisecould be helpful in preparing new researchprogrammes. In the digital e-Humanitiesyou need to work in teams, but this is notsomething the average humanities researcheris used to. In this respect, Laurent Romarymentioned the example of the musicologyproject (see above). If we are able to find

solutions for historians, archeology, musicol-ogists etc. then we are on the right way. Thenhow do we commit the users? Most of theusers are still language oriented people, wehave to reach out to the others: historians,archeologists etc. Neil Fraistat stressed thathe does not work with an infrastructure pro-ject, if the content is not there. Easy accessand simplicity of use is also crucial for RI's.In CLARIN and DARIAH we want to havea knowledge sharing infrastructure, which isat least as important as the technical infra-structure. Dissemination is not enough, itshould be community based and made sus-tainable. This should not have to cost verymuch. Wim Jansen has a simple approachfor the use of the words RI and data: there isRESEARCH infrastructures, and there isresearch INFRASTRUCTURES. The firstone is about producing the data and thecooperation between researchers and thefacilities that are produced, and the secondone is the technology which is necessary tocreate and develop the 1st one. RI's likeCLARIN and DARIAH are more than justcontent, but also knowledge sharing, andshould also focus on training. LorenzaSaracco again stressed the importance ofopen access policy.

The last conference of CLARINand DARIAH?

The above is just a brief impression of aninteresting and lively conference.Discussions went on in the coffee and lunchbreaks, and at the cocktail on Wednesday.And last but not least at the wonderful andanimated dinner in the Melker Stiftskelleron Tuesday evening. We would again like tothank Gerard Budin, Claudia Resch,Victoria Weber, Barbara Berger and DanielMeyrath for their great help in organizingthe conference. Also our DARIAH col-leagues Peter van Doorn, Milena Piccoli andRené van Horik for the smooth collabora-tion. And of course the other members ofthe JOCO (Joint Organizing Committee),Martin Wynne, Tamás Váradi, and MatthewDriscoll. And last but not least the con-venors of the sessions and their speakers.

Although it was the final conference of theCLARIN and DARIAH projects, we hopeand expect that it will be the start for a seriesof e-Humanities conferences or meetings inthe future. All abstracts of the sessions, bothof SDH and NEERI can be found at theDARIAH website: http://www.dariah.eu/index.php?option=com_docman&task=cat_view&gid=87&Itemid=200. C

TThhee ppaarrttiicciippaannttss ooff tthhee ccoonnffeerreennccee iinn aa pprrooppeerr aammbbiieenntt ffoorr ddiiggiittaall hhuummaanniittiieess

FASSBL7, DubrovnikOctober 4-6, 2010

Kre{imir [ojatUniversity of ZagrebFaculty of Humanities andSocial Sciences

The Seventh International Conference“Formal Approaches to South Slavic

and Balkan Languages” (FASSBL 7) washeld from 4 to 6 October 2010 in thebeautiful Croatian coastal city ofDubrovnik. The conference was co-orga-nized by the Institute of Linguistics(Faculty of Humanities and SocialSciences, University of Zagreb), CroatianLanguage Technologies Society, theDepartment of Computational Linguisticsof the Institute of Bulgarian Language“Prof. Lyubomir Andreychin” (BulgarianAcademy of Sciences) and the NorwegianUniversity of Science and Technology. Thebiannual conference FASSBL aims atbringing together researchers dealing withall aspects concerning formal and compu-tational approaches to South-Slavic andBalkan languages from various institutionsfrom all over the world. This FASSBL is theseventh event in the FASSBL conferenceseries.

Every time getting stronger

This year more than 30 researchers fromAustria, Bulgaria, Croatia, France, Norway,

Romania, Serbia, USA and Turkey partici-pated at the conference. The conferencewas divided into 4 sections. The topics cov-ered by the papers extend from morpholo-gy ([najder & Dalbelo Ba{i},), syntax(Comorovski, Radeva-Bork, Vlahova &Atansov, Vu~kovi} et al.) and categorialgrammar (Mihali~ek) to corpus and com-putational linguistics (Ion et al., Koeva,Koeva et al., Miji} et al., Öztürk et al.,Stoyanova, [ojat et al.) as well as machinetranslation (Iliev). Research infrastructuresare traditionally part of the interest ofmany presenters so CLARIN was men-tioned in quite a number of presentationssince at least four CLARIN partner siteswere represented at the conference.

There were 4 invited key-note speakers:Tamás Váradi, Peter Kosta, Karel Oliva andMaria-Luisa Rivero. Let us mention brieflyjust the first one: Tamás Váradi from TheHungarian Academy of Sciences gave a pre-sentation of the EU sponsored projectCESAR which is to begin in the spring of2011. The project aims at collecting andbringing into the standardised level all rel-evant resources from partner countries inthe project: Poland, Slovakia, Hungary,Croatia, Serbia and Bulgaria (last threebeing covered by the conference main areaof interest). We believe that this project willhave a large impact to the availability oflanguage resources and tools that have beencollected and developed for quite sometime for these languages and thatresearchers will be able to access them soonover the META-SHARE platform.

The FASSBL 7 conference was also animportant event for a number of EU spon-sored projects, like CLARIN, ATLAS,Let'sMT!, ACCURAT and others. Themembers of these project partners werehaving several papers in the proceedingsand gave the presentations. Within the

conference the exhibition of European pro-jects dealing with LT and covering respec-tive languages was organised where partici-pants had the opportunity to get acquaint-ed in more details about these projects anddiscuss their goals and the achieved results.

See you in 2012

The guided city tour and a trip to the mostsouthern Croatian region Konavle wereorganized for the conference participants,as well as a number of other memorablesocial events. More information and theproceedings of the FASSBL 7 conference,(edited by Marko Tadi}, Mila Dimitrova-Vulchanova and Svetla Koeva) is availableat http://hnk.ffzg.hr/fassbl2010. The nextFASSBL conference will be held in 2012 atthe same venue in the end of September. C


LT in the South-East Europe

LLiivveellyy ddiissccuussssiioonn oovveerr LLeettssMMTT!! pprroojjeecctt ppoosstteerr


Fourth Baltic HLT, RigaOctober 7-8, 2010

Inguna Skadin, aInstitute of Mathematics andComputer Science,University of Latvia

It is a tradition that scientists, developersand users of language technologies from

the Baltic region countries meet at theBaltic HLT conference to share new ideasand recent advances in natural languageprocessing, to exchange information andto discuss problems, to find new synergiesand to promote initiatives for internation-al cooperation.

The first larger pan-Baltic event on HLTresearch was the seminar “Language andTechnology 2000” held in Riga in 1994. In2004, ten years later, the first Baltic HTLconference was organized by theCommission of the Official Language of theChancellery of the President of Latvia. Thesecond conference in 2005 in Tallinn wasorganized by the Institute of Cyberneticsand Institute of Estonian Language, andthen the third Baltic HLT conference inKaunas in 2007 was organized by VytautasMagnus University and the Institute ofLithuanian language.

This fourth conference took place in Riga onOctober 7-8, 2010. It was jointly organizedby Institute of Mathematics and Computerscience (University of Latvia) and Tildecompany. More than 70 participants fromEstonia, France, Finland, Germany, Latvia,

Lithuania and Norway enjoyed presenta-tions covering wide range of topics, includ-ing corpus linguistics, machine translation,speech technologies, semantics, and otherareas of HLT research.

The morning session of the conference wasdevoted to CLARIN project and overview oflanguage technologies in Baltic countries.The session started with the invited speechby Steven Krauwer (“CLARIN – how tomake it all fit together?”) who has been par-ticipant and presenter in all Baltic HLT con-ferences. The presentation provided a briefoverview of the current state of CLARINproject and described the steps to be taken,and the strategy to be adopted in order tomake CLARIN happen, not as yet another

project, but as a sustainable facility for theresearch community, based on long termcommitments from the governments.

Since Baltic countries have actively partici-pated in CLARIN project, the overview pre-sentations of representatives from Estonia,Latvia and Lithuania provided not only ananalysis of the current situation in the HLTin Baltic countries, policy, main projects andachievements, but also activities andprogress in respect to CLARIN aims.

The afternoon session started with anotherCLARIN related invited talk by professorKimmo Koskeniemi: “HFST (HelsinkiFinite State Transducer Technology): A newdivision of labour between software industryand linguists”. The presentation proposedan approach where software developers, onthe one hand, can easily use language pro-cessing modules in their products and lin-

guists, on the other hand, can relatively easyproduce such modules for various languagesand tasks.

The rest of afternoon was devoted to presen-tations, demonstrations and posters.Research results, work in progress, and posi-tion papers were presented also on the sec-ond day of the conference that started withthe invited speech by Andreas Eisele: “Fromcorpora to resources and tools – towards aproper treatment of Eastern European lan-guages”. The presentation outlined innova-tive ways how existing techniques can becombined to build the technology requiredfor a proper treatment of Eastern Europeanlanguages. Examples from EuroMatrix Plusand ACCURAT project as well as otherrelated activities were used to show someimportant steps towards these goals. Finally,these activities were shown into a larger con-text of a long-term strategy that will allow todevelop the high-quality language technolo-gy required for a truly multilingualEuropean society.

The next invited speech by Georg Rehm,“META-NET and META-SHARE: AnOverview”, provided an overview of META-NET project – Network of Excellence dedi-cated to fostering the technological founda-tions of a multilingual European informa-tion society. The presentation introducedthe architecture and general principles ofMETA-SHARE, which is an action line oncreation of an open distributed facility forthe sharing and exchange of resources.

The following sessions of presentations weredevoted to semantics, machine translation,

speech technologies and methods for lan-guage processing. The conference endedwith panel discussion “How to foster HLTdevelopment in Baltic countries”.The conference was supported CLARIN,ACCURAT (FP7) and LetsMT! (ICT PSP)projects.More information can be found on-line:http://www.lumii.lv/hlt2010/. C

Language Technology in the Baltic

SStteevveenn KKrraauuwweerr sshhoowwiinngg CCLLAARRIINN aacchhiieevveemmeennttss


Brussels, November 17, 2010

Aljoscha BurchardtGeorg RehmDFKI, Saarbrücken, Germany

Wednesday, November 17, 2010 inBrussels. A cold and foggy autumn

day provided the backdrop for META-FORUM 2010, the first large outreach eventorganised by META-NET, a Network ofExcellence forging the Multilingual EuropeTechnology Alliance (META). META-FORUM 2010 took place in the historic“Theatre” hall of the Hotel Le Plaza andassembled more than 250 participants froma total of 37 countries – a turnout farexceeding our initial expectations. META-FORUM 2010 was the inaugural conferenceof the META-FORUM series of events. Allpresentations are available online athttp://www.meta-forum.eu.As the name indicates, META-FORUM wasmeant to be a gathering point, a hub for diversecommunities of interest to meet and discussdevelopments, problems and opportunities pre-sented by the challenges of a modern and multi-

lingual Europe. In the foyer in front of theTheatre, an exhibition area – META-NETVillage – was dedicated to projects the network iscollaborating with. Ten projects such as, forexample, CLARIN, FLaReNet and PANACEApresented themselves, highlighting their plans ofcollaborating with META-NET. Softwaredemonstrations from within META-NET propercomplemented the picture by showcasing theopen resource exchange infrastructure META-SHARE, the Virtual Information Centre and thefirst results of the Machine Translation researcharm of the initiative.

META-FORUM 2010: Highlightsof the Programme

The welcome address by Algirdas Saudargas(Member of European Parliament and formerforeign minister of Lithuania) conveyed the clearmessage that language must be handled with theutmost care for it is a strong but also fragile bandthat ties together communities, social groups, andnations. Complementary, Roberto Cencioni(European Commission, Luxembourg) in hisgreeting address advised stakeholders fromLanguage Technology research and developmentthat this highly fragmented sector has to joinforces to reach critical mass and that it has toimprove its credibility and visibility. HansUszkoreit (DFKI, Germany), the coordinator ofMETA-NET, took up this topic when he intro-duced the three lines of action the initiative pur-sues to reach exactly these strategic goals: build-ing bridges to neighbouring technology fields,designing and implementing META-SHARE,

the open resource exchange facility, and buildinga homogeneous European LT community with ashared vision and strategic research agenda.

The first invited keynote speech presented acces-sibility and multilingualism as key challenges forthe audiovisual industry in the digital age. YotaGeorgakopoulou (European CaptioningInstitute, UK) left no doubt that in the nearfuture there will be an immense economic needwithin the audiovisual sector to make use of var-

ious Language Technology applications. In thesession “From Shared Visions to a StrategicResearch Agenda” industry representatives andlanguage professionals that had been invited byMETA-NET into three think tanks – VisionGroups – presented the results of their first tworounds of meetings, which were discussed in apanel afterwards. The discussion is now contin-ued in an online forum at http://www.meta-net.eu/forum and will, together with input andfeedback collected through several other means,ultimately lead to the vision paper “EuropeanMultilingual Information Society 2020” and a

Challenges for Multilingual EuropeThe First META-FORUM

HHaannss UUsszzkkoorreeiitt pprreesseennttiinngg pprroojjeeccttssccoollllaabboorraattiinngg wwiitthh MMEETTAA--NNEETT


Strategic Research Agenda. If you are interested inthe needs and visions that have been and will befurther discussed in the Vision Groups, pleasevisit the online forum where you can join andcontribute to the discussion.

Georg Artelsmair (European Patent Office,Germany) in the second invited keynote speechpresented recent developments concerning themachine translation policy at the EuropeanPatent Office, which has become a heavy user ofmultilingual Language Technology. At the end ofthe first day, Bill Dolan (Microsoft Research,USA) accentuated the need for and also opportu-nities of close cooperation between industry andthe research community, especially the impor-tance of sharing both technology and also data.

The second (half ) day of META-FORUM 2010gave companies that either employ or buildLanguage Technologies the opportunity to pre-sent themselves. Seven industry representativesformulated their needs, problems, and wishes forEuropean LT research and development. As anattempt to strengthen the industry community,Jochen Hummel (ESTeam GmbH, Germany)introduced the Language Technology BusinessAssociation (LTBA), which is currently in apreparatory phase and which will be launched atthe beginning of 2011. John Hendrik Weitzmann(Creative Commons Initiative, Germany) and

Prodromos Tsiavos (Creative CommonsInitiative, Greece, UK, Norway) addressed prob-lems and solutions with regard to legal issues inbasic and industrial research on LanguageTechnology and the sharing of LanguageResources through META-SHARE.

In her closing keynote, Swaran Lata (Departmentof Information Technology, Government ofIndia, New Delhi) provided an overview ofnumerous multilingual initiatives in India in thepast 20 years. Indeed, India, with its 22 languages(and various different scripts), can serve as a blue-print to European initiatives.

META-NET and CLARIN

The META-NET Network of Excellence(http://www.meta-net.eu) is forging theMultilingual Europe Technology Alliance(META) through a concerted effort to building astrong European community around LanguageTechnologies. By working together to providevisionary applications and a Strategic ResearchAgenda for Language Technology in Europe,META-NET is reaching out to a large and het-erogeneous community of stakeholders to helpfostering the technological foundations of theEuropean information society. CLARIN andMETA-NET are two complementary initiatives,

serving different communities with different butharmonizable goals. As documented in a com-mon Memorandum of Understanding (availableonline at http://www.meta-net.eu/collabora-tions), signed by their coordinators both initia-tives have identified mutual interests andapproaches with respect to methods and instru-ments. The next step in the collaboration betweenthe two projects will be the preparation of adetailed list of topics to collaborate on, for exam-ple, in the area of metadata descriptions and stan-dards for Language Resources and LanguageTechnologies.

From the overwhelmingly positive feedback wereceived from the participants and also from oursubjective perspective as the organisers of theevent, META-FORUM 2010 was a huge success.Even more so if we take into account that META-NET's kick-off meeting only took place inFebruary 2010 – on a very cold and also foggyday in Berlin.

The next edition of META-FORUM will takeplace in Budapest from June 27-29, 2011. C

Contact

Georg Rehm, Network Manager of META-NET,[email protected]


Alon ItaiDepartment of Computer Science,Israel Institute of Technology,Haifa, IsraelShuly WintnerDepartment of Computer Science,University of Haifa, Israel

Introduction

Hebrew is a morphologically rich languagewhose word formation and inflectional mor-phology are typically Semitic. The standardHebrew orthography, “undotted script”, ishighly ambiguous: Most vowels are omitted,many particles are attached to the word thatfollows them, and even though the Academyfor the Hebrew Language (Gadish, 2001) hasissued standards for transcribing Hebrew theyare not adhered to. All this results in highlyambiguous texts; the average number of analy-ses of a word in a running text is 2.64.Consequently, the first step in processingHebrew is nontrivial and many efforts need tobe dedicated to morphological analysis and dis-ambiguation. In 2003 the Israel Ministry of Science andTechnology established a Knowledge Centerfor processing Hebrew. Its mission was to col-lect and develop tools for processing Hebrewand make them available to Academia andIndustry. Since then the Center, known asMILA, has developed several tools all of whichinteract with one another { they accept andproduce XML data and are available underGPL over the Internet. This note describes themain products of the Center, see (Itai andWintner, 2008) for additional details.

2 Databases

2.1 Corpora MILA distributes a number of Hebrew textcorpora, obtained from various sources: news-paper articles (HaAretz); newswire texts(Arutz 7); parliament proceedings (Knesset);and a small corpora of Spoken Hebrew extract-ed from CoSIH (Izre'el et al., 2001). Corpussizes are listed in Table 1.

TTaabbllee 11:: SSiizzee ooff ccoorrppoorraa((iinn tthhoouussaannddss ooff wwoorrddss))

2.2 Lexicon Computational lexicons are among the mostimportant resources for NLP. In languages withrich morphology, where the lexicon is expectedto provide morphological analyzers withenough information to enable them toprocess intricately inflected forms correctly, a

careful design of the lexicon is crucial. TheMILA Lexicon of Contemporary Hebrew, thebroadest-coverage publicly available lexicon ofHebrew, currently consists of about 25,000entries. Table 2 lists the number of words in thelexicon by main part of speech (POS).

TTaabbllee 22:: SSiizzee ooff tthhee lleexxiiccoonn bbyy ppaarrtt ooff ssppeeeecchh

3 Tools

3.1 Tokenization Partitioning raw Hebrew data into tokens(words) is slightly more involved than inEnglish due to issues of Hebrew encoding,mixed Hebrew/English, numbers, punctuationetc. We developed a tokenization modulewhich operates on raw data (UTF-8 encoded)and produces an XML corpus. The module iscapable of segmenting texts into paragraphs,sentences and tokens.

3.2 Morphological analysis and

generation We initially developed a finite-state morpho-logical analyzer for Hebrew (Yona andWintner, 2008). This solution, however,turned out to be inefficient (Wintner, 2007),and the analyzer was re-implemented in Java.Analysis is performed by generation: First, allthe inflected forms induced by the lexicon (notincluding prepended prefixes) are generatedand stored in a database. Then, analysis is sim-ply a database lookup. At peak performance itis able to analyze 4,000 tokens per second.

3.3 Morphological disambiguation Identifying the correct morphological analysisof a given word in context is an important andnon-trivial task. A single token in Hebrew canactually be a sequence of more than one lexicalitem. For example, the token {bth can be ana-lyzed as {+b+h+th “that+in+the+tea”, {+bth“that+ her daughter”or a single lemma she tookprisoner. The task of segmentation is partitioning thetoken into its prefixes and main lemma, whilemorphological disambiguation in addition deter-mines all the morphological features (tense,person, gender, number etc.) Note that mor-phological disambiguation falls short of fulldisambiguation since it does not distinguishbetween homographs, e.g., spr can be eithersefer = book or sappar = barber. MILA distributes three different disambigua-tion modules. MorphTag (Bar-Haim et al.,2005) is a Hidden Markov Model based tagger.When trained on 4500 annotated sentences, itboasts 97.2% accuracy for segmentation and90.8% accuracy for POS tagging (Bar-haim et

al., 2008). Adler and Elhadad (2006) havedeveloped an unsupervised HMM-basedmethod to morphologically disambiguateHebrew texts. They report results of 92.32%for POS tagging and 88.5% for full morpho-logical disambiguation, i.e., finding the correctlexical entry. Finally, HADAS (Shacham andWintner, 2007) is a morphological disam-biguation module for Hebrew that uses simpleclassifier for each of the attributes that can con-tribute to the disambiguation of the analysesproduced by the analyzer (e.g., POS, tense,state), and then combines the outcomes of thesimple classifiers to produce a consistent rank-ing which induces a linear order on the analy-ses. The results are 91.44% accuracy on the fulldisambiguation task. These disambiguation modules are fully com-patible with the morphological analyzer: theyreceive as input an XML file which is the out-put of the analyzer. The output is a file in thesame format, in which each analysis is associat-ed with a score, reflecting its likelihood in thecontext. This facilitates the use of the output inapplications which may not commit to a singlecorrect analysis in a given context. C

References

Meni Adler and Michael Elhadad. An unsupervisedmorpheme-based hmm for hebrew morphologi-cal disambiguation. In Proceedings of the ACL,pp 665-672, Sydney, Australia, July 2006. ACL(http://www.aclweb.org/anthology/P/P06/P06-1084.)

Roy Bar-Haim, Khalil Sima'an, and Yoad Winter.Choosing an optimal architecture for segmenta-tion and POS-tagging of Modern Hebrew. InProceedings of the ACL Workshop onComputational Approaches to SemiticLanguages, pp 39-46, Ann Arbor, Michigan,June 2005. ACL. (http://www.aclweb.org/anthology/W/W05/W05-0706.)

Roy Bar-haim, Khalil Sima'an and Yoad Winter.Part-of-speech tagging of Modern Hebrew text.Natural Language Engineering, 14(2):223-251,2008.

Ronit Gadish (ed.) Klalei ha-Ktiv Hasar ha-Niqqud.Academy for the Hebrew Language, 4th edition,2001. ISBN 0024-1091. In Hebrew.

Alon Itai and Shuly Wintner. Language resources forHebrew. Language Resources and Evaluation,42:75-98, March 2008.

Shlomo Izre'el, Benjamin Hary, and Giora Rahav.Designing CoSIH: The corpus of Spoken IsraeliHebrew. International Journal of CorpusLinguistics, 6:171-197, 2001.

Danny Shacham and Shuly Wintner. Morphologicaldisambiguation of Hebrew: a case study in clas-sifier combination. In Proceedings of EMNLP-CoNLL 2007, Prague, June 2007. ACL.

Shuly Wintner. Finite-state technology as a pro-gramming environment. In Alexander Gelbukh(ed.) Proceedings of the CICLing2007, pp 97-106, Berlin and Heidelberg, February 2007.Springer.

Shlomo Yona and Shuly Wintner. A finite-state mor-phological grammar of Hebrew. NaturalLanguage Engineering, 14(2):173-190, April2008.

Language Resources for Hebrew


Eiríkur RögnvaldssonUniversity of Iceland, Reykjavík

Introduction

At the turn of the century, Icelandic languagetechnology (henceforth LT) was virtu-ally non-existent. There was a relatively good spellchecker, a not-so-good speech synthesizer, andthat was all. There were no programs or evenindividual courses on language technology orcomputational linguistics at any Icelandic uni-versity, there was no ongoing research in theseareas, and no Icelandic software companieswere working on language technology.In 1998, the Minister of Education, Scienceand Culture appointed a group of experts toinvestigate the situation in language technolo-gy in Iceland and come up withproposals for strengthening thestatus of Icelandic languagetechnology. The group handedits report to the Minister inApril 1999 and in 2000, theGovernment launched a specialLanguage TechnologyProgram, with the aim of sup-porting institutions and com-panies in creating basicresources for Icelandic lan-guage technology work. Thisinitiative resulted in severalprojects which have had pro-found influence on the field.

Icelandic LT Work2000-2010

The main direct products ofthe government-funded LTProgram are the following:

• full-form morphological data-base of Modern Icelandic inflections

• balanced morphosyntactically tagged corpusof 25 million words

• training model for data-driven POS taggers

• text-to-speech system

• speech recognizer

• improved spell checkerMost of these products were developed incooperation between research institutes andcommercial companies. After the LT Programended six years ago, LT researchers from threeresearch institutes (University of Iceland,Reykjavik University, and the Árni MagnússonInstitute for Icelandic Studies), who had beeninvolved in most of the projects funded by theLT Program, decided to join forces in a con-sortium called the Icelandic Centre forLanguage Technology (ICLT), in order to fol-

low up on the tasks of the Program. The mainroles of the ICLT are to• serve as an information centre on Icelandic

LT by running a website (http://iclt.is) • encourage cooperation on LT projects

between universities, institutions and com-mercial companies

• organize and coordinate university educationin LT

• participate in Nordic, European and interna-tional cooperation within LT

• initiate and participate in R&D projects in LT• keep track on resources and products in the

field of Icelandic LT • hold LT conferences with the participation of

researchers, companies and the public • support the growth of Icelandic LT in all pos-

sible manners Over the past six years, the ICLT researchershave initiated several new projects which have

been partly supported by the IcelandicResearch Fund and the Icelandic TechnicalDevelopment Fund. The most importantproducts of these projects are:• linguistic rule-based tagger, IceTagger• shallow parser, IceParser• mixed-method lemmatizer, Lemmald• (prototype of a) context-sensitive spell checkerIn 2009, the ICLT received a relatively largethree year Grant of Excellence from theIcelandic Research Fund for the project“Viable Language Technology beyond English– Icelandic as a test case”. Within that project,three types of LT resources are being devel-oped:• database of semantic relations (a pilot

WordNet)• prototype of a shallow-transfer machine

translation system

• treebank with a historical dimensionThese resources were chosen because they wereconsidered central to current LT work and pre-requisites for further research and developmentin Icelandic LT.

The Prospects of Icelandic LT

Twelve years ago, the LT expert group estimat-ed that it would cost around one billionIcelandic krónas (which then amounted toabout ten million Euros) to make Icelandiclanguage technology self-sustained. After that,the free market should be able to take over,since it would have access to public resourcesthat would have been created by the LanguageTechnology Program, and that would be madeavailable on an equal basis to everyone who wasgoing to use these resources in their commer-cial products.However, the total budget of the government-funded LT program over its lifespan (2000-

2004) was only 133 millionIcelandic krónas – that is, 1/8 of thesum that the expert group estimat-ed would be needed. It shouldtherefore come as no surprise thatwe still have a long way to go. Thereare only 320,000 people speakingIcelandic, and that is not enough tosustain costly development of newproducts. At present, no commer-cial companies are working in theLT area because they don't see it asprofitable. It is thus extremelyimportant to continue public sup-port for Icelandic language technol-ogy for some time, but given thecurrent financial situation, it doesnot seem likely that such supportwill come from the state budget inthe near future.

For a small language communityand a small research environmentlike the Icelandic one, it is vital to

cooperate, not only on the national level butalso internationally. Since 2000, Icelandicresearchers and policy makers have taken anactive part in Nordic cooperation on languagetechnology. This has been of major importancein establishing the field in Iceland. The NordicLanguage Technology Research Programme2000-2004 was instrumental in this respect.

Iceland has just recently entered the CLARINconsortium, and together with the otherNordic and Baltic countries, Iceland also takespart in the META-NORD project which startsFebruary 1st this year. We sincerely hope thatour participation in these projects will help usto develop, standardize and make available sev-eral important LT resources and thus con-tribute to the growth of Icelandic languagetechnology. C

The State of Icelandic LT

TThhee ffrroonntt ppaaggee ooff tthhee IIccccNNLLPP,, aa NNLLPP ttoooollkkiitt ffoorr IIcceellaannddiicc


Güls,en Eryiv

gitFaculty of Computer andInformatics, Istanbul TechnicalUniversity (ITU), Turkey

Introduction

Over the last two decades, we see that theinterest of the Turkish researchers in lan-guage technology research increased notice-ably. Although the interest of private sectorand funding agencies still remains at its earlyphase, the ongoing activity in the field ispromising.

Most of the important Turkish universitiesshowed their interest in the field by offeringcourses in language technology. ITU Facultyof Computer and Informatics, as one of theleaders, offers now three specialized graduatecourses in the field and many supplementarycourses.

The number of language resources such ascorpora, treebanks and online dictionariescontinue to get increased in number. Also,there exist important NLP tools developedfor Turkish.

Turkish Language Resources

Some of the well known data sources forTurkish are as follows:

• online dictionaries of Turkish LanguageAssociation (TDK),

• 2M words corpus by Middle EastTechnical University (METU) and aDiscourse Annotation Bank (whereMETU Turkish Corpus is annotated with

respect to connectives, their senses andarguments)

• morphologically and syntactically annotat-ed 5K sentences dependency treebank byMETU and Sabancr University,

• Turkish Wordnet (part of BalkaNetProject) by Sabancr University,

• large-scale test collection that contains408K documents for information retrievalby Bilkent University,

• 20K documents in 8 different categoriesfor text categorization by ITU,

• 423M words web corpus by BoNaziçiUniversity,

• ongoing project on creating a morphologi-cally annotated 50M words Turkish corpusby Mersin University,

• 160K English-Swedish-Turkish ParallelTreebank by Uppsala University,

• 500K English-Turkish parallel sentences,by Koç University,

• transcribed speech corpus by BoNaziçiUniversity.

Turkish NLP Tools

Turkish, being an agglutinative language,poses interesting challenges for the area oflanguage technologies. The models devel-oped for other well-studied languages do notconform to this language for many NLP lay-ers. On the other hand, research in this lan-guage serves as reference to many other lan-guages with similar features (morphological-ly rich, agglutinative, with scarce dataresources, …) This makes Turkish veryattractive for NLP research not only for

Turkish researchers but also at an interna-tional level. In recent years, Turkish has beenincluded in many international research pro-jects which aim to develop multilingual sys-tems.

The tools which are developed specificallyfor Turkish automatic language processingdates back to the beginning of 1970s startingwith the first morphological analyzer devel-oped at Hacettepe University. Since thattime, many other tools have been developedby different institutions. Many tasks stillremain as open research questions and the per-formance of the tools still needs to get amelio-rated. Some of them may be listed as follows:

• two level morphological analyzer devel-oped at Bilkent and Sabancr Universities,and some replication of the similarapproach at BoNaziçi and Koç Universities.

• morphological disambiguators developedat BoNaziçi University and Koç University,

• dependency parser developed at IstanbulTechnical University,

• CCG grammar and parser at METU,

• LFG grammar at Sabancr University,

• speech recognition and synthesis toolsdeveloped at BoNaziçi University, TubitakUEKAE MTRD and a private companycalled SESTEK.

National Funds for LT

The national funding agencies in Turkey aremainly DPT (State Planning Agency) andTUBITAK (The Scientific andTechnological Research Council of Turkey).Both of the institutions grant projects onlanguage technology. But a national policyon this subject is still lacking. Thus, thefunded projects remain isolated and general-ly at the level of research projects. We hopethat our partnership in CLARIN will speedup the national organization of the Turkishresources and the NLP community inTurkey.

Turkey, being one of the last partners ofCLARIN preparatory phase, has to workhard in order to catch up with the ongoingorganization. Since Turkey is still not a fullmember of EU, it seems that it still needstime to overcome the indefiniteness aboutthe legal regulations of ERICs at the nation-al level. DPT works on the issue. Once thepreparation phase of Turkey is finalized, wehope that the national roadmap will includeCLARIN and provide supports for CLARINactivities in Turkey. C

LR&T situation in Turkey

IIssttaannbbuull TTeecchhnniiccaall UUnniivveerrssiittyy MMaassllaakk CCaammppuuss


Radovan GarabíkMária ŠimkováL’. [túr Institute of Linguistics,Slovak Academy of Sciences,Bratislava, Slovakia

General informationSlovak language belongs to the West Slavic lan-guage group, together with Polish, Czech andSorbian languages. It remained especially close tothe Czech language, due to close historical andlinguistic ties between the languages (especially inthe former common country). Slovak is a typicalSlavic language in retaining complex inflectionaland derivational morphology, with only minorsimplifications compared with neighbouring lan-guages.Slovak language is spoken by about 5 million peo-ple in Slovakia, but also in some other countries.The most numerous Slovak community is in theUSA (1 million people, many of them actuallyspeaking the language), Czech Republic, smallercommunities are present in Romania, Hungary,Serbia and other countries.Slovak orthography is mostly phonemic, withnoticeable etymological and morphological fea-tures. The language is written in the Latin script,with acute accents marking (phonemic) vowellength, há?eks marking palatals and post-alveolarfricatives. The latest substantial orthographyreform has been implemented in 1953, when thelanguage gained contemporary form practically inall of its aspects.

Major linguistic research centres inSlovakiaThe L’. [túr Institute of Linguistics, SlovakAcademy of Sciences, Bratislava is the central lin-guistic institution in Slovakia. Its main area ofresearch is traditional linguistics, with the empha-sis on (but not limited to) Slovak language, itshistory, dialectology, etymology and lexicography,as reflected in a sizeable amount of dictionariesproduced. Traditionally, the Institute was con-nected with the task of regulating the Slovak lan-guage, defining its grammar and orthography andproducing prescriptive dictionaries. However, inthe last decades the main orientation shifted moretowards general linguistic research (not limited tothe Slovak) and descriptive or bilingual dictionar-ies and other popular and scientific publicationsin the field of sociolinguistics, onomastics, lan-guage theory, lexicography and others.Institute of Informatics, Slovak Academy of Sciences,Bratislava is dealing with theoretical and appliedresearch in the field of computer science, infor-mation technologies and artificial intelligence.Their NLP tools include information retrieval,knowledge representation speech synthesis andrecognition. The main research direction ofFaculty of Arts, University of Pre{ov in Pre{ov isphonetics, phonology, derivatology and morphe-matic structure of the Slovak. The results of theirresearch are often used in Slovak NLP. An exten-sive research of child speech is carried out incooperation with the Faculty of Education, espe-

cially in international projects. University of SSCyril and Methodius, Trnava puts special empha-sis on German-Slovak confrontational research oncollocations, valency and corpus lexicography aswell as phraseology and paremiology. Faculty ofHumanities, Matej Bel University Banská Bystricais active in sociolinguistics and research in com-munication and stylistics. Part of their database ofrecordings of spontaneous speech in the city ofBanská Bystrica and surroudings become part ofthe Corpus of Spoken Slovak. The TechnicalUniversity of Ko{ice is active especially in NLP,computer aided lexicography and ontologyresearch.

Slovak language in computer processingThe Slovak National Corpus department was estab-lished as a special project of the Ministry ofCulture and Ministry of Education of the SlovakRepublic and the Slovak Academy of Sciences in2002. This marked rapid expansion of HumanLanguage Technology research of the Slovak lan-guage, with the L’. [túr Institute of Linguisticsbecoming the leading research institution in thefield. The Institute actively leads R&D in com-putational linguistics oriented towards contem-porary written and spoken Slovak language, cov-ering all aspects of language analysis and process-ing. Its strong IT background is demonstrated inthe number of tools and resources that the depart-ment developed during its existence, among themost important are the Slovak National Corpusdatabase and all the necessary associated tools.The Slovak National Corpus is a big, representa-tive corpus of modern written Slovak (since the1953 orthography reform). Currently, the wholecorpus contains about 780 million tokens. Thereare several specialised subcorpora (fiction, profes-sional texts, journalistic texts, original Slovak fic-tion, balanced subcorpus, texts written before1989). The corpus is automatically lemmatisedand morphologically annotated, using its owntagset and morpfology analyser. Access to the cor-pus is provided free of charge, for non commer-cial purposes. The corpus legal status is ratherunusual, if compared with other language corpo-

ra – The L’. [túr Institute of Linguistics didobtain license agreements to use the texts inbuilding the corpus database (for non commer-cial, research and educational purposes), and thecorpus therefore includes the texts with full legalcompliance. New Dictionary of ContemporarySlovak (Slovník sú~asného slovenského jazyka) isbeing compiled with the help of the corpus and isthe first Slovak dictionary based predominantlyon corpus resources.The department actively works on several smaller,but no less important projects. Corpus of SpokenSlovak is a representative corpus of standard spo-ken Slovak as spoken throughout Slovakia, con-sisting of about 160 hours of recordings (1.6 mil-lion words). The recordings are manually tran-scribed on orthographic and phonemic level.Parallel corpora include Slovak-Czech, Slovak-Russian, Slovak-French, Slovak-Bulgarian andSlovak-English corpus. The texts (mostly fictiontranslations) are automatically sentence-alignedand morphosyntactically tagged. Slovak depen-dency treebank contains about 50,000 manuallysyntactically annotated sentences.The Slovak Terminology Database focuses on thefield of law, economy and technology, offeringabout 5000 terminological records that can beclassified by circa 20 EUROVOC descriptors cor-responding to different soft and hard sciences.

International collaborationWith expansion of data and resources, it is moreand more apparent that standardisation and avail-ability of resources, documentation and interop-erability are important for further scientificresearch and technology deployment. Since thebeginning, the Slovak National Corpus depart-ment is committed to making available all theresources and tools under favourable OpenContent and Open Source licensing policies (ifother copyright restrictions permit). Collabor-ation with other partners has proved to be veryfruitful and we expect to extend the ties withother CLARIN partners and to improve the inter-operability and standard compliance of SlovakNLP resources and tools. C

Slovak language in computer processing

WWeebb--bbaasseedd iinntteerrffaaccee ttoo MMaannaattttee//BBoonniittoo ppllaattffoorrmm rruunnnniinngg SSlloovvaakk NNaattiioonnaall CCoorrppuuss


Join CLARINThe CLARIN project is a combination ofCollaborative Projects and Coordinationand Support Actions, registered at the EUunder the number FRA-2007-2.2.1.2. Itstarted with the preparatory phase in 2008that will make the grounds for the nextphases and it will cover the generic, lan-guage independent activities. In order todo our work properly we have to rely on amuch wider circle than just the formal con-sortium partners in the project. For thisreason we have opened up all our projectworking groups for participation by orga-nizations that are not part of the consor-tium.

MembersAAuussttrriiaa ((NNCCPP:: GGeerrhhaarrdd BBuuddiinn)) University of Graz (Graz): AustrianGerman Research Centre (C:Rudolf Muhr); Institut für Romanistik(C:Stefan Schneider)Austrian Academy of Sciences (Vienna): Austrian Academy Corpus

(C:Christoph Benda); Department of Linguistics andCommunication Research (C:Sabine Laaha); Institute ofLexicography of Austrian Dialects and Names (C:Eveline Wandl-Vogt)

Secure Business Austria (Vienna): (C:Edgar R. Weippl)University of Vienna (Vienna): Center for Translation Studies

(C:Gerhard Budin)BBeellggiiuumm ((NNCCPP:: IInneekkee SScchhuuuurrmmaann)) University of Antwerp (Antwerp):

Center for Dutch Language and Speech (C:Walter Daelemans)Vrije Universiteit Brussel (Brussels): Laboratory for Digital Speech and

Audio Processing, Department of Electronics and InformationProcessing (C:Werner Verhelst)

Gent University (Gent): Digital Speech and Signal Processing researchgroup at the Electronics and Information Systems department(C:Jean-Pierre Martens)

University College Ghent (Gent): Faculty of Translation Studies,Language and Translation Technology Team (C:Veronique Hoste)

Katholieke Universiteit Leuven (Leuven): Center for ComputationalLinguistics (C:Frank Van Eynde); ESAT-PSI/Speech (C:PatrickWambacq); Language Intelligence & Information Retrieval(C:Marie-Francine Moens)

Katholieke Universiteit Leuven (Leuven – Kortrijk): iTec(Interdisciplinary research on Technology, Education &Communication) (C:Hans Paulussen)

BBuullggaarriiaa ((NNCCPP:: KKiirriill SSiimmoovv)) University of Plovdiv (Plovdiv): Faculty ofMathematics and Informatics (C:Veska Noncheva)

Bulgarian Academy of Sciences (Sofia): Department of ComputationalLinguistics, Institute for Bulgarian Language (C:Svetla Koeva);Institute for Parallel Processing of Bulgarian Academy of Sciences(Sofia): Linguistic Modelling Department (C:Kiril Simov); Instituteof Mathematics and Informatics, Bulgarian Academy of Sciences(Sofia): Mathematical Linguistics Departement (C:LudmilaDimitrova)

St. Cyril and St. Methodius University (Veliko Turnovo): (C:BoryanaBratanova)

CCrrooaattiiaa ((NNCCPP:: MMaarrkkoo TTaaddii}})) Institute of Croatian Language andLinguistics (Zagreb): (C:Damir ]avar)

University of Zagreb (Zagreb): Department of Linguistics, Faculty ofHumanities and Social Sciences (C:Marko Tadi}); Zagreb UniversityComputing Center (C:Zoran Beki})

CCyypprruuss ((NNCCPP:: --)) Cyprus College (Nicosia): Research Center (C:AntonisTheocharous)

CCzzeecchh RReeppuubblliicc ((NNCCPP:: EEvvaa HHaajjii~~oovváá)) Masaryk University (Brno):Faculty of Informatics (C:Aleš Horák)

Charles University (Prague): Institute of Formal and AppliedLinguistics, Faculty of Mathematics and Physics (C:Eva Haji~ová)

The Institute of the Czech Language, Czech Academy of Sciences(Prague): The Institute of the Czech Language (C:Karel Oliva)

DDeennmmaarrkk ((NNCCPP:: BBeennttee MMaaeeggaaaarrdd)) Copenhagen Business School(Copenhagen): Department of International Language Studies andComputational Linguistics (C:Peter Juel Henrichsen)

Dansk Sprognævn – Danish Language Council (Copenhagen):(C:Sabine Kirchmeier-Andersen)

Society for Danish Language and Literature (Copenhagen): (C:JørgAsmussen)

The National Museum of Denmark (Copenhagen): (C:Birgit Rønne)The Royal Library (Copenhagen): (C:Anders Conrad)University of Copenhagen (Copenhagen): Centre for Language

Technology, Faculty of Humanities (C:Bente Maegaard)University of Southern Denmark (Kolding): Faculty of Humanities

(C:Johannes Wagner)EEssttoonniiaa ((NNCCPP:: TTiiiitt RRoooossmmaaaa)) University of Tartu (Tartu): Institute of

Computer Science (C:Tiit Roosmaa)FFiinnllaanndd ((NNCCPP:: KKiimmmmoo KKoosskkeennnniieemmii)) CSC – the Finnish IT Center for

Science (Espoo): (C:Pirjo-Leena Forsström)Lingsoft Inc. (Helsinki): (C:Juhani Reiman)The Research Institute for the Languages of Finland (Helsinki): (C:Toni

Suutari)University of Helsinki (Helsinki): Department of General Linguistics

(C:Kimmo Koskenniemi)University of Joensuu (Joensuu): Department of Foreign Languages

and Translation Studies (C:Jussi Niemi)

University of Oulu (Oulu): Faculty of Humanities, Finnish Language(C:Marketta Harju-Autti)

University of Tampere (Tampere): Faculty of Information Sciences,Department of Information Studies and Interactive Media (C:EeroSormunen)

FFrraannccee ((NNCCPP:: JJeeaann--MMaarriiee PPiieerrrreell)) Centre de ressources pour la docu-mentation de l'oral (Aix-en-Provence) (C:Bernard Bel)

National Center for Scientific Research (CNRS) (Marseille): Laboratoired'Informatique Fondamentale de Marseille (LIF-CNRS) (C:MichaelZock)

Centre National de Ressources Textuelles et Lexicales (CNTRL) (Nancy):(C:Bertrand Gaiffe)

National Center for Scientific Research (CNRS) (Nancy): Analyse etTraitement Informatique de la Langue Française (ALTIF) (C:Jean-Marie Pierrel)

National Center for Scientific Research (CNRS) (Orsay): Institute forMultilingual and Multimedia Information (IMMI-CNRS) (C:JosephMariani)

Evaluations and Language resources Distribution Agency (ELDA)(Paris): (C:Khalid Choukri)

National Center for Scientific Research (CNRS) (Paris): TraitementÉlectronique des Manuscrits et des Archives (TELMA/DIS)(C:Florence Clavaud)

Université Paris 4 Sorbonne (Paris): Centre de linguistique théoriqueet appliquée (CELTA) (C:Andre Wlodarczyk)

Université de Strasbourg (Strasbourg): Equipe de recherche LiLPa(Linguistique, Langues, Parole) (C:Amalia Todirascu)

National Center for Scientific Research (CNRS) (Vandoeuvre lesNancy): L'Institut de l'Information Scientifique et Technique (INIST-CNRS) (C:Fabrice Lecocq)

University Paris Est/Paris 12 (Vitry Sur Seine): LISSI Laboratory(C:Yacine Amirat)

GGeerrmmaannyy ((NNCCPP:: EErrhhaarrdd HHiinnrriicchhss)) University of Augsburg (Augsburg):Philologisch-Historische Fakultät (C:Ulrike Gut)

Berlin-Brandenburg Academy of Sciences (Berlin): (C:AlexanderGeyken)

Humboldt-University Berlin (Berlin): Institut für deutsche Sprache undLinguistik (C:Anke Lüdeling)

Technische Universität Darmstadt (Darmstadt): Ubiquitous KnowledgeProcessing (UKP) Lab (C:Iryna Gurevych)

TU Dortmund University (Dortmund): Institute for German Languageand Literature (C:Michael Beißwenger)

Universität Duisburg-Essen (Essen): Fakultät Geisteswissenschaften /Germanistik / Linguistik (C:Bernhard Schröder)

University of Frankfurt/Main (Frankfurt/Main): Comparative LinguisticsDepartment (C:Jost Gippert)

University of Giessen (Giessen): Institut für Germanistik (C:HenningLobin)

DFG-Project “Language Variation in Northern Germany“(Sprachvariation in Norddeutschland – SiN) (Hamburg): (C:IngridSchröder)

University of Hamburg (Hamburg): Faculty for Language Literatureand Media, Arbeitsstelle “Computerphilologie“ (C:Cristina Vertan);Fakultät für Geisteswissenschaften, Fachbereich Sprache, Literatur,Medien (C:Angelika Redder); Institute of German Sign Languageand Communication of the Deaf (C:Thomas Hanke); SFB 538Multilingualism (C:Thomas Schmidt)

University of Heidelberg (Heidelberg): Computational LinguisticsDepartment (C:Anette Frank)

University of Cologne (Köln): Institut für Linguistik – Phonetik(C:Dagmar Jung)

Max Planck Institute for Evolutionary Anthropology (Leipzig):Department of Linguistics (C:Hans-Jörg Bibiko)

University of Leipzig (Leipzig): Institut für Informatik, AbteilungAutomatische Sprachverarbeitung (C:Codrina Lauth)

Institut für Deutsche Sprache (Mannheim): (C:Marc Kupietz)Westfälische Wilhelms-Universität Münster (Münster): Institut für

Allgemeine Sprachwissenschaft (C:Gabriele Müller)University of Potsdam (Potsdam): Department of Linguistics

(C:Manfred Stede)German Research Center for Artificial Intelligence (Saarbrücken):

Language Technology Lab (C:Thierry Declerck)University of Stuttgart (Stuttgart): Institut für Maschinelle

Sprachverarbeitung (C:Ulrich Heid)Universität Trier (Trier): Kompetenzzentrum für elektronische

Erschließungs- und Publikationsverfahren in denGeisteswissenschaften (C:Andrea Rapp)

Universität Tübingen (Tübingen): Asien-Orient-Institut (C:Ulrich Apel);Seminar für Sprachwissenschaft (C:Erhard Hinrichs)

GGrreeeeccee ((NNCCPP:: SStteelliiooss PPiippeerriiddiiss)) Institute for Language and SpeechProcessing (Athens): Department of Language TechnologyApplications (C:Stelios Piperidis)

HHuunnggaarryy ((NNCCPP:: TTaammááss VVáárraaddii)) Hungarian Academy of Sciences(Budapest): Research Institute for Linguistics (C:Tamás Váradi);Institute for Psychological Research of the Hungarian Academy ofSciences (Budapest): (C:Bea Ehmann)

Budapest University of Technology and Economics (Budapest):Department of Sociology and Communications, Media ResearchCenter (C:Peter Halacsy)

Department of Telecommunication and Media Informatics, Laboratoryof Speech Acoustics (C:Klára Vicsi)

MorphoLogic Ltd. (Budapest): MorphoLogic Ltd. (C:László Tihanyi)University of Szeged (Szeged): Department of Informatics, Human

Language Technology Group (C:Dóra Csendes)IIcceellaanndd ((NNCCPP:: EEiirrííkkuurr RRööggnnvvaallddssssoonn)) Icelandic Centre for Language

Technology (Reykjavík): (C:Eiríkur Rögnvaldsson)University of Iceland (Reykjavík): Institute of Linguistics (C:Eiríkur

Rögnvaldsson)IIrreellaanndd ((NNCCPP:: --)) National University of Ireland (Galway): Department

of English (C:Sean Ryder)IIssrraaeell ((NNCCPP:: --)) Technion-Israel Institute of Technology (Haifa):

Computer Science Department (C:Alon Itai)

IIttaallyy ((NNCCPP:: NNiiccoolleettttaa CCaallzzoollaarrii)) European Academy Bozen/Bolzano(Bolzano): Institute for Specialised Communication andMultilingualism (C:Andrea Abel)

Università di Pavia (Pavia): Dipartimento di Linguistica Teorica eApplicata (C:Andrea Sansò)

National Research Council (Pisa): Istituto di LinguisticaComputazionale (C:Nicoletta Calzolari)

University of Rome “Tor Vergata“ (Rome): Department of ComputerScience (C:Fabio Massimo Zanzotto)

LLaattvviiaa ((NNCCPP:: IInngguunnaa SSkkaaddiinnaa)) Tilde Language Technologies (Riga):Tilde Language Technologies (C:Andrejs Vasiljevs)

University of Latvia (Riga): Institute of Mathematics and ComputerScience (C:Inguna Skadina)

LLiitthhuuaanniiaa ((NNCCPP:: RRuuttaa MMaarrcciinneekkiieevviicciieennee)) Vytautas Magnus University(Kaunas): Center of Computational Linguistics (C:RutaMarcinkeviciene)

Institute of the Lithuanian Language (Vilnius): (C:Daiva Vaisniene)LLuuxxeemmbboouurrgg ((NNCCPP:: --)) European Language Resources Association

(ELRA) (Luxembourg): (C:S.Piperidis/K.Choukri)MMaallttaa ((NNCCPP:: MMiikkee RRoossnneerr)) University of Malta (Malta): Department of

Computer Science (C:Michael Rosner)NNeetthheerrllaannddss ((NNCCPP:: JJaann OOddiijjkk)) Meertens Institute (Amsterdam):

Meertens Institute (C:H.J. Bennis)Univerity of Amsterdam (Amsterdam): Intelligent Systems Lab

Amsterdam (ISLA) (C:Maarten de Rijke)Vrije Universiteit Amsterdam (Amsterdam): Computational Lexicology,

Faculteit der Letteren (C:Piek Vossen)Data Archiving and Networked Services (Den Haag): (C:Henk

Harmsen)Huygens Instituut KNAW (Den Haag): (C:K.van Dalen-Oskam)University of Twente (Enschede): Human Media Interaction Group,

Department of Electrical Engineering, Mathematics and ComputerScience (C:Roeland Ordelman)

University of Gröningen (Gröningen): Faculty of Arts, Center forLanguage and Cognition (C:Wyke van der Meer)

Digital Library for Dutch Literature (Leiden): (C:C.A. Klapwijk)Institute for Dutch Lexicology (Leiden): Instituut voor Nederlandse

Lexicologie (C:Remco van Veenendaal)Universiteit Leiden (Leiden): Leiden University Centre for Linguistics,

Faculty of Humanities (C:Jeroen van de Weijer)Max Planck Institute for Psycholinguistics (Nijmegen): (C:Peter

Wittenburg)Radboud University (Nijmegen): Centre for Language and Speech

Technology (C:L. Boves / N. Oostdijk); Centre for Language Studies(C:Pieter Muysken)

Tilburg University (Tilburg): ILK Research Group, Department ofCommunication and Information Sciences, Faculty of Humanities(C:Antal van den Bosch)

University of Utrecht (Utrecht): Utrecht Institute of Linguistics OTS,Faculty of Humanities (C:Steven Krauwer)

NNoorrwwaayy ((NNCCPP:: KKooeennrraaaadd DDee SSmmeeddtt)) Norwegian School of Economicsand Business Administration (NHH) (Bergen): (C:Gisle Andersen)

Unifob AS (Bergen): (C:Eli Hagen)University of Bergen (Bergen): Language Models and Resources group

(C:Koenraad de Smedt)SINTEF (Oslo): (C:Diana Santos)The Language Council of Norway (Oslo): (C:Torbjoerg Breivik)The National Library of Norway (Oslo) (C:Kristin Bakken)University of Oslo (Oslo): Department of Linguistics and Nordic

Studies, Faculty of Humanities (C:Janne Bondi Johannessen)University of Tromsø (Tromsø): Det humanistiske fakultet (C:Trond

Trosterud)Norwegian University of Science and Technology (Trondheim):

Department of Electronics and Telecommunications (C:TorbjørnSvendsen)

PPoollaanndd ((NNCCPP:: MMaacciieejj PPiiaasseecckkii)) University of Lodz (Lodz): Institute ofEnglish Language (C:Piotr Pezik)

Polish Academy of Sciences (Warsaw): Institute of Computer Science,Department of Artificial Intelligence (C:Adam Przepiórkowski);Institute of Slavic Studies (C:Violetta Koseska-Toszewa)

Polish-Japanese Institute of Information Technology (Warsaw):(C:Krzysztof Marasek)

University of Wroclaw (Wroclaw): Instytut Informatyki Stosowanej(C:Maciej Piasecki)

Wroclaw University of Technology (Wroclaw): Institute of AppliedInformatics (C:Maciej Piasecki)

PPoorrttuuggaall ((NNCCPP:: AAnnttoonniioo BBrraannccoo)) Universidade Católica Portuguesa(Braga): Centro de Estudos Filosóficos e Humanísticos (C:AugustoSoares da Silva)

University of Minho (Braga): Centro de Estudos Humanísticos (C:PilarBarbosa)

New University of Lisbon (Caparica): Faculdade de Ciências eTecnologia (C:José Gabriel Pereira Lopes)

Instituto de Telecomunicações (Coimbra): Pólo de Coimbra(C:Fernando Perdigão)

University of Coimbra (Coimbra): Centro de Esudos de LinguísticaGeral e Aplicada (CELGA ) (C:Cristina Martins); Centro deInvestigação do Núcleo de Estudos (C:José Augusto SimõesGonçalves Leitão)

Universidade de Évora (Évora): School of Sciences and Technology(C:Paulo Quaresma)

INESC-ID, Instituto de Engenharia de Sistemas e ComputadoresInvestigação e Desenvolvimento em Lisboa (Lisboa) (C:NunoMamede)

Instituto de Linguística Teórica e Computacional (Lisbon): (C:MargaritaCorreia)

New University of Lisbon (Lisbon): Centro de Linguística (C:MariaFrancisca Xavier)

University of Lisbon (Lisbon): Centro de Linguística da Universidade deLisboa (CLUL) (C:Amália Mendes); Natural Language and SpeechGroup (NLX-Group), Department of Informatics (C:António Branco)

University of Azores (Ponta Delgada (Azores)): (C:Luis Mendes Gomes)

University of Porto (Porto): Centro de Linguística (C:Fátima Oliveira);Laboratory of Artificial Intelligence and Computer Science(C:Miguel Filgueiras)

RRoommaanniiaa ((NNCCPP:: DDaann TTuuffiiss,,)) Romanian Academy of Sciences(Bucharest): Research Institute for Artificial Intelligence (C:DanTufis,)

University Babes-Bolyai (Cluj-Napoca): Faculty of Mathematics andComputer Science (C:Tatar Doina)

“Al. I. Cuza“ University of Ias,i (Ias,i): Faculty of Computer Science(C:Dan Cristea)

Romanian Academy of Sciences (Ias,i): Institute of Computer Science(C:Horia-Nicolai Teodorescu)

University of Pites,ti (Pites,ti): Faculty of Letters (C:Mihaela Mitu)West University of Timis,oara (Timis,oara): Faculty of Mathematics and

Informatics (C:Viorel Negru)SSeerrbbiiaa ((NNCCPP:: --)) University of Belgrade (Belgrade): Faculty of

Mathematics (C:Du{ko Vitas)SSlloovvaakkiiaa ((NNCCPP:: --)) Slovak Academy of Sciences (Bratislava): L’. Štúr

Institute of Linguistics (C:Radovan Garabík)SSlloovveenniiaa ((NNCCPP:: TToommaa`̀ EErrjjaavveecc)) Alpineon d.o.o. (Ljubljana): (C:Jerneja

Zganec Gros)Josef Stefan Institute (Ljubljana): Dept. of Knowledge Technologies

(C:Toma` Erjavec)SSppaaiinn ((NNCCPP:: NNúúrriiaa BBeell)) University of Alicante (Alicante):

Departamento de Lenguajes y Sistemas Informáticos (C:PatricioMartínez-Barco)

Institut d'Estudis Catalans (Barcelona) (C:Joan Soler i Bou)Technical University of Catalonia (UPC) (Barcelona): Centro de

Tecnologías y Aplicaciones del Lenguaje y del Habla (TALP)(C:Asuncion Moreno)

Universitat Autònoma de Barcelona (Barcelona): Facultat de Filosofiai Lletres, Dpt. de Filologia Anglesa i de Germanística (C:AnaFernández Montraveta)

Universitat de Barcelona (Barcelona): Departament de LingüisticaGeneral (C:Irene Castellón)

Universitat Oberta de Catalunya (Barcelona): Department ofLanguages and Cultures (C:Salvador Climent)

Universitat Pompeu Fabra (Barcelona): Institut Universitari deLingüística Aplicada (C:Núria Bel)

University of Barcelona (Barcelona): Facultat de Filologia – RamonLlull Documentation Centre (C:Joana Alvarez)

Autonomous University of Barcelona (Bellaterra): Facultad de Letras,Dept. Filología Española (C:Carlos Subirats)

Girona City Council (Girona): Records Management, Archives andPublications Service (C:Joan Boadas i Raset)

University of Jaén (Jaén): Escuela Politécnica Superior, Departamentode Informática, SINAI group (C:María Teresa Martín Valdivia)

University of the Basque Country (Leioa): Computer Science Faculty,Natural Language Processing Group (C:Arantza Diaz de Ilarraza)

University of Lleida (Lleida): Departament d'Angles i Linguistica(C:Gloria Vázquez)

Autonomous University of Madrid (Madrid): Laboratorio de LingüísticaInformática (C:Manuel Alcantara Plá)

University of Málaga (Málaga): Facultad de Filosofía y Letras, Dept.of English, French, and German Philology (C:Antonio Moreno Ortiz)

Universidad Politécnica de Valencia – ITACA (Valencia): Grid andHigh Performance Computing Group (C:Vicente Hernández García)

University of Vigo (Vigo): Facultade de Filoloxía e Tradución,Department of English, Research group LVTC (C:Javier Perez-Guerra); Seminario de Lingüística Informática, Departamento deTradución e Lingüística, TALG Research Group (C:Xavier GómezGuinovart)

University of Zaragoza (Zaragoza): Facultad de Filosofía y Letras(C:Carmen Pérez-Llantada)

SSwweeddeenn ((NNCCPP:: LLaarrss BBoorriinn)) University of Gothenburg (Gothenburg):Department of Linguistics, Faculty of Arts (C:Anders Eriksson);Språkbanken, Dept. of Swedish Language (C:Lars Borin)

Linköping University (Linköping): Department of Computer andInformation Sciences (C:Lars Ahrenberg)

Lund University (Lund): Humanities Laboratory (C:Sven Strömqvist)KTH Royal Institute of Technology (Stockholm): Department of

Speech, Music and Hearing, CSC (C:Rolf Carlson)Language Council of Sweden (Stockholm): (C:Rickard Domeij)Swedish Institute of Computer Science AB (Stockholm): (C:Björn

Gambäck)Umeå University (Umeå): HUMlab (C:Patrik Svensson)Uppsala University (Uppsala): Department of Linguistics and

Philosophy (C:Joakim Nivre)TTuurrkkeeyy ((NNCCPP:: GGüüllss,,eenn EErryyiiggvviitt)) Istanbul Technical University (Istanbul):

Elektrics-Electronics Faculty, Computer Science Department,Natural Language Processing Group (C:Güls,en Eryigvit)

Sabanci University (Istanbul): Human Language and SpeechLaboratory, Faculty of Engineering and Natural Sciences (C:KemalOflazer)

UUnniitteedd KKiinnggddoomm ((NNCCPP:: MMaarrttiinn WWyynnnnee)) Bangor University (Bangor):Language Technologies Unit (C:Briony Williams)

University of Birmingham (Birmingham): Department of English(C:Oliver Mason)

University of Surrey (Guildford): Department of Computing, Faculty ofEngineering and Physical Science (C:Lee Gillam)

Lancaster University (Lancaster): Department of Linguistics andEnglish Langauge (C:Paul Rayson)

National Centre for Text Mining (Manchester): National Centre for TextMining (C:Bill Black)

Oxford Text Archive (Oxford): Oxford University Computing Services(C:Martin Wynne)

University of Sheffield (Sheffield): Natural Language Processinggroup, Department of Computer Science (C:Wim Peters)

University of Wolverhampton (Wolverhampton): Research Institute ofInformation and Language Processing (C:Constantin Orasan)

CCLLAARRIINN nneewwsslleetttteerr •• ppuubblliisshheerr CCLLAARRIINN •• eeddiittoorr--iinn--cchhiieeff MMaarrkkoo TTaaddii}} •• eeddiittoorriiaall bbooaarrdd GGeerrhhaarrdd BBuuddiinn,, DDaann CCrriisstteeaa,, KKooeennrraaaadd DDee SSmmeeddtt,, LLaauurreenntt RRoommaarryy,,

MMaarrttiinn WWyynnnnee •• hhttttpp::////wwwwww..ccllaarriinn..eeuu

clarin seen from the other side of atlanticclarin seen from the other side of atlantic brian...

Documents