Number 1, 2008, May
sent and future needs of our primary target just be sufficient to cover the generic, lan- What is CLARIN? audience, how do we protect the intellectual guage independent activities in the project. property rights of those who have provided In order to do our work properly we will data or tools, and how do we ensure that have to rely on a much wider circle than just whatever the infrastructure that we manage the formal consortium partners in the pro- to build is sustainable. ject. For this reason we have opened up all Following up on the ESFRI action aimed at our project working groups for participation Steven Krauwer identifying potential essential research infra- by organizations that are not part of the con- CLARIN coordinator structures for Europe and the FP7 sortium. Unfortunately, our project budget Infrastructures programme, the EC has does not allow us to support this participa- he easiest way to describe CLARIN in Tnon-technical terms is to say that the goals of the CLARIN enterprise are three- fold. First of all, it aims at uniting existing digital archives in Europe that contain lan- guage based material into a federation that will allow the social sciences and humanities research communities unified access to the content. Second, it wants to make the wealth of language and speech processing tools that have been developed over the recent years available to interested researchers with a view to opening up new research avenues. Third, it wants to provide web based services that will allow non-expert users (especially humanities and social sciences researchers without a technological background) to per- form complex tasks on the materials con- tained in the archives, such as ‘Summarize Le Monde of March 17 2008 — in Polish’. To achieve this a large number of technical challenges will have to be addressed, such as Visit CLARIN web site at http://www.clarin.eu (just to mention a few): how do we unite funded a number of European Infrastructure tion financially, but, fortunately, all partici- existing archives from all over Europe into a initiatives and given them the green light for pating countries (at this moment 22) have single federation, how do we solve the prob- a preparatory phase during which all of the committed themselves to giving additional lem that many language based resources use above issues (and many more) will have to be support at their national level. many different representation conventions, addressed with a view to actual implementa- We envisage that at least part of the national how do we solve the problem that existing tion of the infrastructure. CLARIN is one of CLARIN funding will be used to support tools and applications each have their own them, and in the rest of this newsletter we participation by additional groups in the expectations in terms of input and output will describe our approach and expected project, with a view to ensuring that what is formats, how do we solve the problem that results towards the end of 2010, when (we proposed during the preparatory phase will (in the CLARIN view) all languages are hope) we will be able to embark on the next be adequate to serve the interests of the equally important but that the level of tech- phase: the construction of the CLARIN national research communities and lan- nological coverage is quite uneven, how do infrastructure. guages. we make sure that what we will offer in The EC funding for this preparatory phase is Please contact us if you want to join one of terms of tools and services responds to pre- relatively modest (4.1 million euro) and will our working groups. C CLARIN advances and where they can tell in CLARIN, tries to convince you why a Editors’ their experiences or express opinions. The new infrastructure to deal with resources nice thing about CLARIN is that it facili- and tools is required by humanists and social Foreword tates scientific collaboration among people scientists. We have dedicated page 4 to two that work in domains which were considered categories of scientists dealing with until recently as distant as the Earth’s poles. resources: consumers and producers. In this This is the case of humanities and social sci- issue these two voices belong to János ences, on one hand, and computer science, László, which brings a motivation for IT on the other. They will meet in CLARIN for needs in social psychology, and Max the common enterprise of finding the most Silberztein, the developer of the NooJ tool, appropriate ways in which language which was at the very heart of the work resources, whether reported by János Marko Tadi} they are expressed in László. The central & Dan Cristea speech, in text, or Call for contributions pages are dedicated to CLARIN Newsletter editors multimedia, can be Dear readers the CLARIN kick-off better offered to scien- of the CLARIN Newsletter, meeting, hosted by ear readers, tists and to the public if you have ideas, thoughts, MPI and the Nij- DIn your hands you have the first issue of at large. In order for megen City Hall, CLARIN Newsletter, a publication initiated comments, additions, correc- this to happen, we tions, arguments, questions etc. from 17 to 19 March and supported by CLARIN — an acronym invite you to con- 2008. Some meetings for the Common Language Resources and tribute in this News- which are connected to the in which the CLAR- Technology Infrastructure, which is a com- letter in two ways: CLARIN project even remotely, IN idea took gradual- bination of collaborative projects and coor- either as a correspon- please feel free to send them to ly shape are also dination and support actions, registered at dent editor — there- us as your contribution at briefly described. We the EU under the number FRA-2007- fore sending us peri- [email protected]. have thought that 2.2.1.2. This publication is planned to odically news from many of our readers appear 4 times per year, both electronically your scientific community, being it your would like to know in more details the and on paper, at least for the lifetime of the country, your institute or your group, or as CLARIN structure and how are its activities preparatory activities. The electronic version an occasional author. organized. This is why we have offered the can be accessed by anyone on the project’s following 4 pages to members of the site at www.clarin.eu. On demand, it can be Content e-mailed to subscribers. The subscription Executive Board. This first issue, of a series which we hope to information and forms can be found on the First, Peter Wittemburg makes a presenta- Newsletter section of the CLARIN site. On include 12 of them (the CLARIN prepara- tion of the CLARIN organization, and then paper it will be distributed mainly during tory phase extends from January 2008 till all coordinators of the CLARIN working main language technology and humanities December 2010), aims at offering the first packages WP2-WP8, minus WP4 which and social sciences events, and only occa- contact to scientists with the CLARIN does not exist, describe their specific activi- sionally it will be sent by mail (no budget is world. You have just turn the page on which ties. The last page, now and in all further put aside for this). Steven Krauwer, the project coordinator, issues, is dedicated to a presentation of presented the CLARIN mission in the larg- events, most significant for CLARIN, over a Vision er context of the ESFRI action. Martin period of 6 months in advance. We want this newsletter to be a publication Wynne, who is the port-parole and liaison where people can find the latest news on with the DARIAH (www.dariah.eu) project Enjoy your reading! C
List of national correspondents Estonia Malta Tiit Roosmaa Mike Rosner Austria Finland Netherlands Gerhard Budin Kimmo Koskenniemi Peter Wittenburg Belgium – Flanders France Norway Inneke Schuurman William Del Mancino Koenraad De Smedt Bulgaria Bertrand Gaiffe Svetla Koeva Protugal Germany Antonio Branco Croatia Lothar Lemnitzer Poland Marko Tadi} Greece Maciej Piasecki Czech Republic Maria Gavrilidou Karel Pala Stelios Piperidis Romania Dan Cristea Denmark Hungary Hanne Fersoe Tamás Váradi Dan Tufis, ELRE/ELDA Italy Spain Victoria Arranz Valeria Quochi Nuria Bel UK Latvia Sweden Martin Wynne Andrejs Vasiljevs Sven Strömqvist
2 No 1 / Newsletter CLARIN scholar, researching and publishing alone. As forms of textual representation, metadata, Language Daniel Pitti notes: annotation, and access arrangements. Tools “the most significant impact of information are often developed for ad hoc use within a Resources in technology may be increased collaboration. particular project, or for use by one group of Collaboration, when successful, offers many researchers, or for use with only one text or intellectual, professional, and social bene- set of data, and are not developed sufficient- Humanities and fits...collaboration also presents unfamiliar ly to be deployed as widely-used and sus- Social Sciences Research
Martin Wynne CLARIN EB member
esearch in the humanities and social sci- Rences is changing. The increasingly widespread use of digital resources is chang- ing the types and the amount of data that we use. The use of electronic data means that we need computational tools to create, annotate, process, store, and analyse them. The ways in which we communicate the fre- quency with which we communicate are changing too, and so are the ways that we share out data and our research findings. New requirements for challenges, which require careful attention tainable services. As a result, a large amount infrastructure and time...” (1) of effort has been wasted over many years in many places developing applications with This digital revolution means that the sup- One of the most important challenges for similar functionality. CLARIN offers us an port that is needed to faciliate research is dif- CLARIN is to recognize and cope with the opportunity to address these problems in a ferent today. The traditional support infra- changes in working practices which the systematic, sustainable and global fashion. structure is no longer sufficient. Desktop increasing use of electronic resources will computers, a network, email and web access entail. CLARIN aims to facilitate the Usage of IT in the Humanities deployment of the latest computing tools are now the common currency of academic A Summit on Digital Tools in the and infrastructure so that language resources Humanities in Charlottesville, Virginia in and tools can be used effectively. As well as 2006 estimated that only 6% of scholars in developing computing infrastructure, As well as developing computing the humanities go beyond general purpose CLARIN will be promoting new forms of information technologies (email, web infrastructure, CLARIN will be multi-disciplinary collaborative research for browsing, word processing, spreadsheets and promoting new forms of many researchers in the humanities and presentation slide software), and suggested multi-disciplinary collaborative social sciences. research for many researchers in the that revolutionary change in humanistic There are many language resources and tools research is possible thanks to computational Humanities and Social Sciences... already in use in research across many sub- methods, but that this revolution has not yet ject areas, and there is a clear potential for occurred (2). This is an exciting time for many more to be successfully developed and researchers in the humanities and social sci- work. Use of electronic datasets requires a deployed. If this potential is to be realised, ences, as the introduction of new instru- further support infrastructure, to allow the the developers of the CLARIN infrastruc- ments and datasets make possible new types transfer and processing of large amounts of ture will have to understand the processes of research, but it is clear that a new infra- data, to allow secure shared access to involved in research in the humanities and structure is needed for the potential to be resources and tools, and to allow effective social sciences, to be sensitive to differences realised. C online collaboration and data sharing. in the cultures of the various subject areas, CLARIN aims to build such an infrastruc- and to be responsive to user requirements. (1) Pitti, Daniel V. (2005), ‘Designing ture for the use of electronic language The problems which are faced by CLARIN Sustainable Projects and Publications’ in A resources in the humanities and social sci- are not new ones. Problems of standards for Companion to Digital Humanities, ences. textual representation, interoperability of Blackwell: Oxford (free online at http://www.digitalhumani- One of the most important changes in tools, and problems with licensing, access ties.org/ companion/) research today is the increase in collabora- and preservation have dogged the humani- (2) Summit on Digital Tools for the Humanities: tion. Digital projects entail collaboration, ties since the invention of the digital com- Report on Summit Accomplishments. 2006. but many HSS academic communities are puter. The language resources which are cur- http://www.iath.virginia.edu/dtsummit/Sum more familiar with the model of the lone rently available exhibit a confusing variety of mitText.pdf (Accessed online 25.10.2007)
CLARIN Newsletter / No 1 3 Users’ view Developers’ view What do social psychologists How can expect from CLARIN? NooJ help?
János László Max Silberztein Social Psychologist, Hungarian Academy of Sciences Université de Franche-Comté
urrent research in mainstream social psychology shows that t is my pleasure to give an account of a collaboration with social Clanguage conveys substantial implicit information about inter- Ipsychologists who are investigating latent psychological con- group relations when depicting events. The Linguistic Category cepts in texts. At first blush, tracking down highly abstract concepts Model (Semin and Fiedler, 1988; 1991 and other paradigms all like the individual perception of time and studying group or nation- imply that both wording and compositional patterns of the text al identity with automated methods of text processing seemed a convey information that function similarly to “hidden curricula” daunting task. (Jackson, 1968; Snyder, 1970). They can reinforce or alleviate As it turned out, however, the finite-state linguistic development existing stereotypes; shift blame to out-groups or take in-group NooJ has proved a productive tool in the hands of non-specialist responsibility for negative outcomes; position in-group as active or colleagues, like the social psychologist team of János Lászlo. NooJ passive participant in events. is a freeware corpus processor that allows users to apply queries to Our research group, in collaboration with the Institute of Linguistics large texts and corpora. One special feature of NooJ is that of the HAS, with the Institute of Informatics of the University of queries, which can be very simple or very complex, are based on Szeged and with the Morphologic Ltd, aims at uncovering implicit potentially large linguistic resources such as dictionaries and gram- linguistic and narrative devices in history school books, which medi- mars. NooJ contains large-coverage dictionaries for a dozen lan- ate identification and stereotyping in inter-group relations. guages ranging from French and English to Portuguese, Serbian To explore this hidden dimension of language use, we have and Hungarian. launched an international project with 12 participating countries Beside basic sets of linguistic resources, NooJ users can add their involving historians and social psychologists with the specific aim to study systematically the use of linguistic means in inter-group own specialized dictionaries and grammars, together with their own stereotyping in the history school-books in the participating coun- system of properties, in order to describe complex sets of terms, tries. Results will reflect not only the level of elaboration of so called expressions and their variants. Then, if a user has constructed a dic- “traditional” regional conflicts, but also tendencies of accepting tionary that contains thousands of technical terms, a basic NooJ supranational, e.g. European categories. query such as
4 No 1 / Newsletter CLARIN been dealing with printed scientific publica- providers, CLARIN strongly relies on a dis- The CLARIN tions until now have shown that this is a tributed model fully controlled by the profitable sector. research community. The competition forms Language Resource Recently, Google announced an offer to the an enormous challenge since we need to research world to store primary and sec- achieve a high degree of cost efficiency and ondary research data (1) — in effect, data service quality. However, if CLARIN also Ambition that is essential for carrying out modern wants to guarantee access to the many small data-oriented eResearch. In evaluating this languages and the more complex resource initiative the following points should be types and if CLARIN wants to handle sensi- considered: tive rights management issues with care, ser- 1) Judging from widespread commercial vices will be more expensive. practice, we can certainly expect that the CLARIN — as many other research infra- current business model offering access free structures that want to enable the eScience of charge will be changed when a critical paradigm — needs to be seen as a self-orga- Peter Wittenburg mass of content has been reached. nizing model of dealing with research infor- & Tamás Váradi CLARIN EB members
Nature of Language Resources For a growing number of researchers, including increasingly in the humanities, analysis of vast amount of data has become the key to successful and empirically based research. The nature of the required resources is very heterogeneous and although linear texts still tend to form the largest amount of data it is structured and highly specialized information extracted from the surface which is most often required in processing language material to answer sophisticated research questions. Resource Management The vast majority of this data is currently typically produced by isolated researchers and resides on local computers, unknown and inaccessible to the rest of the communi- ty. The individual researchers do not have Dennis G. Jerz & Phillip Lenssen: Early draft of how Google home page would look like in 1407 any digital access and preservation strategy. (http://jerz.setonhill.edu/weblog/permalink/google-1407/) Data curation and management distracts researchers from the work and imposes 2) The prevalent publishing model in the mation in a sensitive way so that it can pre- requirements they are not properly trained past centuries has been that the researchers sent a viable alternative to the emerging for, a situation that is only aggravated by produce and consume research results and monopolies. lack of budget allocated to the tasks. that the publishers organize the workflow Emphasizing our strengths is important not On the other hand, it is common knowledge process including the dissemination. The just in terms of the competition with com- now that digital collections need continuous Open Access Initiative (2) has made it clear, mercial companies but also to make an curation and management effort to guaran- however, that the research community does impact on our target research community. tee long-term accessibility. Data manage- not readily agree with the way the big pub- The question of incentives is important ment also includes digital rights manage- lishing monopolies handle information in whatever model we are considering. Why ment, which tends to be much neglected. the electronic age. The current model advo- should the individual researchers that we Most research projects, departments and cated and practiced by the companies focus- described earlier turn to either Google or to institutes face almost insoluble challenges to ing on digital content is a centralized one, us with their resources or tools? meet the requirements. i.e. the rules for handling data that is creat- In order to achieve progress from the cur- ed and consumed in a distributed fashion rently fragmented situation CLARIN must Emerging Business will be defined centrally by company execu- clearly make itself attractive to the research Currently, we see large activities from inter- tive boards. Selection, rating and other community and offer easy to use services. ested companies to ensure digital accessibili- important processes will be determined by Highlighting on the wide ranging expertise ty preferably for those resources where a these companies. that CLARIN will also provide we can return of investment can be expected. The demonstrate our usefulness to our target commercial goal is certainly to get control of Challenge for CLARIN audience and at the same time show our advantage over initiatives driven by com- digital content and to establish a business We need to understand that the CLARIN mercial interests only. C model for trading with all sorts of informa- research infrastructure is, to a certain extent, tion, preferably with minimal overhead. in direct competition with proposals such as (1) http://news.bbc.co.uk/2/hi/technolo- Scientific information is just one small slice from Google. In contrast to the model gy/6425975.stm of the cake. But the companies that have offered by Google and other commerical (2) http://www.soros.org/openaccess/index.shtml
CLARIN Newsletter / No 1 5 Kick-off meeting in Nijmegen
Dan Cristea Editor
LARIN had its kick-off meeting in the Ccity of Nijmegen, The Netherlands, between 17 and 19 March 2008. The orga- nization was arranged by the Max Planck Institute for Psycholinguistics, with a gener- ous help from the City Hall of Nijmegen. About 80 participants attended the meeting, representing 56 institutions from 23 CLA- RIN member countries. For most of these people the accommodation was arranged in the small town of Kleve, across the border, in Germany, at about 20 km distance from Nijmegen. The bus brought the participants each morning through the already green country side meadows, but also along a channel and passing a forest in which remains of a frontier check point were still visible. The first day was devoted to bring to the participants detailed information about all
The Round table room of Nijmegen CIty Hall where the plenary sessions took place
The WP5 meeting
CLARIN aspects so that a broad under- standing of the goals could be obtained. It was also meant to give the necessary insight into some details of the work to be realized and to collect comments and recommenda- tions from all CLARIN members. Two con- secutive sessions gathered all CLARIN members, independent of whether they are consortium partners of the EC funded pro- ject or members of national CLARIN groups. These sessions have been hosted by the City Hall, in the impressive round table
mature technology and establishes it as the sultations were organized that led to a com- Previous meetings essential infrastructure for humanities and mon understanding about the general nature social sciences in 21st century. of our project. In the same manner, before the very kick-off The second meeting was the workshop orga- Marko Tadi} meeting in Nijmegen, there has been a series nized within the LREC2008 conference in Editor of founding and preparatory meetings where Genoa, in May 2006. This was the first pre- sentation of ideas behind the CLARIN to a aving in mind the number of different CLARIN was shaped up into the form that it Hlanguage resources and language tech- exhibits today. wider scientific audience. The time and the nologies activites in Europe during the last The first, almost historical one, was in Paris, place was selected well since that conference decade, it should be clear that CLARIN could in February 2006 where initiatives from dif- is always in the focus of researchers, usually not start from the scratch. Moreover, it fol- ferent research groups such as TELRI and around thousand of them, from the field of lowed the line of many previous initiatives EARL gave the necessary wind in the back to language resources and technologies from and projects that, seen from today’s perspec- the whole idea. After that meeting where rep- all around the globe. The series of consulta- tive, were actually preparing the ground for resentatives of more than fifteen countries tive meetings continued until it was peaked in CLARIN as the project that bring into play the and institutions were present, a series of con- a finely organized overall CLARIN preparato-
6 No 1 / Newsletter CLARIN in each of their working packages (for WP2 The WP3 meeting — Peter Wittenburg, for WP3 — Tamás Váradi and Martin Wynne, for WP5 — The third day was dedicated to the Erhard Hinrichs, for WP6 — Dan Cristea, Consortium meeting, addressing the issues for WP7 — Kimmo Koskenniemi and for of the EC founded project, to working WP8 — Bente Maegaard), and by the Vice- groups (which met as needed), and to the Chair of the Scientific Board of CLARIN Executive Board meeting. Again hosted in (Dan Tufis), describing the role and activity the large City Council room of the City of different CLARIN boards. To present the Hall, the Consortium meeting offered the content of the Newsletter and its design possibility to all WP leaders to resume the principles, the WP6 leader invited at the previous day sessions and to participants rep- desk Marko Tadi}, the editor-in-chief of this resenting the partners to express the state of CLARIN publication. Draft copies of this affairs in their countries. Administrative first issue were distributed in the audience. details, such as consortium agreement, fund- ing, travel possibilities, report writing, deliv- Also, Thierry Declerck was asked to shortly erable deadlines etc., acquired also the atten- present the DFKI experience in constructing tion of the full audience. help-desk services, which will be exploited and continued in CLARIN, and Dieter Van The kick-off meeting made very clear the room of the Council of Nijmegen (see goals of CLARIN and the rules of the game photo). Uytvanck made a short on-line presentation of the new CLARIN website. to all parties involved. In particular, it was The opening and the first welcome was stressed to all partners the double funding addressed by the coordinator, Steven The whole second day was organized in par- scheme — national and EC — and to all Krauwer. Then, from the part of the allel sessions. Participants split, generally fol- other members the necessity to urge the for- Municipality, Ms. J. Kunst expressed the lowing the WP leaders, in seminar rooms to mation of their national groups. The attend- honour of hosting this event, and P. Levelt, a discuss relevant issues in their areas (the pic- ing experts interacted with the working former director of the Max Planck Institute ture in the lower left corner shows an aspect package leaders, the working groups begun for Psycholinguistics, in his speech, evoked of the WP2 meeting). The WP2 and WP5 to take shape, the activities inside them were the last 30 years since when humanities start- meetings, touching the very core of the tech- initiated, and ideas and suggestions were ed to be affected by technology. In his vision nical aspects of the project, as the former is taken up from everyone to further improve of future of humanities CLARIN has one of responsible for building the view on the the activity. major roles. This session continued with pre- infrastructure architecture, while the latter The organisers succeeded to create a positive sentations contributed by the coordinator, should provide the access to language work atmosphere, and, most important, the explaining the goals and overall structure of resources and tools, acquired the largest meeting showed that there is a raising enthu- the project, by the work package leaders, audience and had to be scheduled in two ses- siasm in all countries represented to meet the clarifying the aims, deadlines and milestones sions. challenging goals of our project. C
ry meeting in Budapest, on 2006-10-30 and 31 where the CLARIN project, more or less as it is shaped now, was founded. The 28 repre- sentatives from 21 country were present dis- cussing about the issues that had to be solved in order to make an exemplary project pro- posal. The first call for proposals for the Research Infrastructure projects within the FP7 was issued in December 2006 and it was met by the members that formed the consortium of 32 partners between February and April 2007. C
Representatives at the Budapest meeting
CLARIN Newsletter / No 1 7 the national level we have national groups packages. Also all CLARIN’s registered CLARIN funded by national funding agencies. The members, now numbering around 100, have type of organization of the national groups the possibility to directly bring in their wish- Bodies and widely depends on national constraints. es and critic via the members meetings. Similar to the construction of GEANT, the The Boards fast European communication network, Structures CLARIN took care in its organizational The CLARIN work is essentially planned structure that there is a high degree of inter- and carried out by the Executive Board. It action between the EC level and the nation- contains eight members each of which is al level activities. The national groups take responsible for a clearly separable set of tasks care that they are represented in the clustered in work packages. The Executive Peter Wittenburg Board is fully responsible for the success of CLARIN EB member Scientific Board, the Strategic Coordination Board and in the Working Groups that actu- the project and needs to report about its ally carry out the work in the various work plans and decisions to the Scientific Board uropean Research Infrastructures are Eprimarily based on the commitments of the member states. This was clearly expressed in the ESFRI process by the state- ment that the European Commission will only fund the preparatory phase to facilitate the start of initiatives and to achieve a high degree of synchronization between the member states. While the CLARIN propos- al was being prepared, commitment state- ments from 24 member states were received, indicating great interest in its goals. Currently, some member states have already decided about concrete funding schemes for a number of research infrastructure initia- tives, others are busy to decide plotting about national roadmaps and others obvi- ously need more time to decide.
Double scheme CLARIN will need European and national funds already in the preparation phase to achieve meaningful results. This double Overall structure of CLARIN scheme is reflected in the organizational (dashed lines indicate that in general members of the SB and working groups setup. At the European level we have the will be recruited from the national CLARIN groups) CLARIN project funded by the EC and at and the Strategic Coordination Board. The THE IMPORTANCE ies or consortia. On the one hand, the TEI Executive Board has a chairperson who is OF STANDARDS (Text Encoding Initiative; http://www.tei- c.org) has become the main place where coordinating the CLARIN work, reporting guidelines for the representation of primary Laurent Romary to the EC and interacting with the different textual data are issued and maintained. On Editor Scientific and Strategic Coordination Board the other hand, ISO (International Standard which are controlling the EB work. Each EB he creation, maintenance and dissemi- Organization), in continuity with its previous Tnation of data heavily rely on the confor- activities on language coding (ISO 639) and member has to take care that the goals will mance to best practices as well as the exten- terminology representation (ISO 16642), be achieved, i.e. dependent on the require- sive use of relevant standards. It is all the covers the field of language resources with- ments and working groups are being more essential in order to ensure a high in its technical committee TC 37/SC 4, with formed. The Working Groups can be independence between data and tools, too projects encompassing the various levels of formed to create specification documents or complementary topics that will be dealt with linguistic annotation. As a matter of fact sev- by the CLARIN project. In the domain of lin- eral Lirics partners are strongly involved in develop infrastructure frameworks and com- guistic resources, there has been a long- the corresponding activities either as coordi- ponents. standing implication of the research com- nators or experts. Besides, one of the essen- The International Advisory Board will be munity to establish standards for the repre- tial role of Clarin will be to contribute to the formed by international experts and will give sentation and annotation of language struc- wide dissemination of the corresponding tures. Such efforts, carried out in particular results, and contribute to the establishment advice on all matters to ensure that the within previous EU initiatives such as Eagles of a better interoperability scheme in CLARIN work will be done in agreement or Multext, have progressively been integrat- Europe, while ensuring the these standards with international trends and that CLARIN ed and complemented within projects sup- accounts for the specificity of our multiple specifications have a chance to be accepted ported by international standardization bod- languages. C by the international community. C
8 No 1 / Newsletter CLARIN WP2: Technical WP3: Humanities Infrastructure Projects
Peter Wittenburg Tamás Váradi WP2 coordinator WP3 coordinator
ne of the core pillars of ensure that references will upport for humanitites and we expect to have 3-5 at most — OCLARIN is the technical remain stable. In such a federa- Ssocial sciences is the corner- must be funded through inde- infrastructure that is needed to tion services will interact based stone of the CLARIN mission. pendent (i.e. national or other) overcome the large fragmenta- on accepted certificates as a While language technology and sources. Our first task will focus tion that researchers in particu- means to prevent misuse. language resources undoubtedly on defining carefully the criteria lar in the humanities and social Services that will help to over- have great potential in facilitat- for selecting suitable projects. To sciences suffer from when they come interoperability problems ing humanities research, this is the extent that is possible in the want to work with language will range from conversion ser- an area where neither communi- limited framework of this work package CLARIN will then pro- resources and tools. Few are vices that allow users to trans- ties have large-scale experience vide help such as, for example, ready to be used by non-experts form source formats into more in exploiting the obvious poten- advice on how to look for and in general it is impossible generic standardized formats tial benefits. In the light of this appropriate resources and tools for them to easily combine and terminology services that situation, we think it is essential or to convert legacy formats into resources and tools from differ- will help to overcome semantic that the CLARIN project ent projects to tackle new should establish an active inter- formats that can be handled differences, to workflow services seamlessly. The work package research questions. A suitable that will allow users to combine action with the research com- technical infrastructure will rely munities in humanities and the will guide the projects to take tools such as taggers and parsers on a federation of distributed social sciences and should gain care that the implementations to more powerful workflow and results are useful for both service centers that will offer a crucial direct experience about engines. Using existing stan- the selected Humanities projects wide variety of services to users user needs, objectives, data and dards and creating new ones and the CLARIN goals. in a sustainable and unbureau- methods used in research. The where necessary will be required cratic manner. These services most practical means of gaining Another important strand of to facilitate interoperability. In range from repository and this experience is through actual planned activites related to this respect CLARIN can rely archiving services for resources collaboration with colleagues in building links with the humani- on previous standardization and methods for building virtu- some well-chosen areas. ties communities, exploring the efforts by initiatives such as potential stakeholders in the al collections across repositories The purpose of inviting human- EAGLES/ISLE, W3C, TEI and CLARIN mission, rasing aware- to web-services that allow more ities projects to collaborate with ISO TC37/SC4, and on years of ness and promoting the use of advanced users to build complex CLARIN in the preparatory experience in some areas. language resources and tools in applications by combining exist- phase is to enable us to assess the the humanities fields. To reduce ing tools. To achieve a scalable and distrib- technological, methodological, redundancy and optimally The basis of all virtual integra- uted infrastructure that also will organizational etc. requirements take care of sub-discipline spe- exploit available expertise and tion and interoperability is a involved in serving the humani- resources this activity will be cialities we will rely on a service flexible and distributed registry ties in the later phases of CLAR- carried out in close collabora- oriented architecture and flexi- mechanism for all sorts of tools IN. We are committed to the tion with the DARIAH project, ble registry mechanisms. With and resources covering for idea of collaboration with which have similar objectives. respect to the first we can build example schemas, corpora, tools Humanities projects on a proto- Executive Board member, on experiences gathered with as well as ontologies to make type scale as the best means of Martin Wynne, will be in charge them visible and accessible. frameworks such as for example identifying needs and removing of this area of activities. GATE and UIMA for typical Federation middleware will any potential obstance from the Work in WP3 will be coordinat- NLP applications and LAT for guarantee that users can create way of future synergies between ed by Tamás Váradi and will and access virtual collections by typical tasks when working with the two fields. involve the following centres: logging in once using their smaller languages. With respect To this end, we intend to issue Oxford Text Archive, University home identity. Unique and per- to the latter we can refer to the calls to interested humanities of Bergen, Charles University sistent identifiers (PIDs) associ- experience gathered with the projects for collaboration with Prague, University of Lancaster, ated with all resources, resource IMDI, OLAC and DFKI online CLARIN. Unfortunately, the CNR-ILC Pisa, CNRS Nancy fragments and services will catalogues. C work of the selected projects — and University of Zagreb. C
CLARIN Newsletter / No 1 9 organisations and, where necessary, develop one form-based and one lexical-semantic, a WP5: Language new standards within these relevant standardi- manually annotated corpus (treebank) and an sation organisations. automatically annotated large-scale corpus, Resources & Tools Within CLARIN, the term language resource along with standard analysis tools like taggers, subsumes the whole range of linguistic data parsers and speech recognition systems. Overview types such as digitised manuscripts, text, Written and spoken language as well other speech and multimodal corpora, lexica, tree- modalities such as gestures must be included. banks, typological databases, grammars, Potential gaps in the BLARKs of individual ontologies, schemas, and the term language languages will be identified by a coordinated technology covers a wide range of processing action involving the CLARIN members from Erhard Hinrichs and annotation components such as taggers, the involved countries. It will however be left WP5 coordinator parsers, semantic extractors, manual annota- to the national decision boards whether they tion tools, speech alignment tools, etc. In a will fill these gaps. his WP deals with specifying, using and first step a taxonomy of these heterogeneous Another important part of this work package implementing standards for language types of resources will be defined. T concerns the implementation activities as far as resources of all kinds, including e.g. corpora, Parallel to the definition of the taxonomy a they are required in the preparatory phase. lexica, grammars and tools for processing survey of all the existing resources will be initi- This will focus on the following major tasks: them. This is a prerequisite for achieving inter- ated. The survey will include a detailed analy- investigating all aspects that have to do with operability between linguistic resources and sis of the structural and encoding characteris- the integration of resources and tools into the tools. Both will be made available through tics of the resources and the interfaces of the infrastructure; studying usage scenarios, webservices, and workflows integrating several tools that will serve to design a service-orient- including chains of operation in detail and resources, tools, and services will be defined. ed architecture. Based on this broad and integrating selected language resources into Interoperability of language resources is cur- detailed investigation, a comprehensive taxon- the web service-based infrastructure. rently a vital field of research. Several interna- omy of language resources and tools will be tional organisations, including the Inter- specified. Since we assume that at the national level, sev- national Organisation for Standardization In addition to the survey and resource type eral countries will decide to go ahead with lan- (ISO), the World Wide Web Consortium definition, a Basic Language Resource Kit guage resource and with technology creation (W3C), and the Text Encoding Initiative (BLARK) will be specified as a framework for and integration, this work package also needs (TEI), are active in defining standards and a landscape of language resources across lan- to communicate with the national groups to guidelines for best-practice in encoding, lin- guages. The exact shape of the BLARK has to coordinate these efforts and to determine guistic annotation, metadata, and webservices. made be more precise in the course of the pro- whether all work is compliant with the stan- To achieve interoperability we will draw on the ject. For a well-documented language, the dards and practices of the CLARIN infrastruc- use of existing standards defined by these BLARK might consist of two types of lexica, ture. C
posted. Information should be offered both to Helpdesk service WP6: the benefit of partners and members of the Finally, in order to accomplish preparatory consortium during the project’s lifetime, but work for introducing infrastructure services also to people outside the consortium. This Dissemination and able to help researchers in the humanities and site is already operational (www.clarin.eu) and advisory services is hosted by MPI — Nijmegen. We intend to social sciences and to facilitate new types of use it also as the first point of access from research, WP6 will organise a referral help- where resources and tools organized within the desk service. This will have to point interested project will be made visible to the whole people to places (centres), projects, experts, world. documents, web-services or websites where Dan Cristea Newsletter answers to their questions can be found or WP6 coordinator Second, editing and publication of the CLAR- from where modules can be taken and inte- IN Newsletter, with a periodicity of 4 issues, grated into their applications. his work package will co-ordinate the gen- electronically and/or on paper is a WP6 activ- eration of content for publicity and the Based on the experience acquired in similar T ity. We hope that the materials of the dissemination of CLARIN activities. It will initiatives (as, for instance, LT World), this Newsletter will be contributed not only by the also provide advice and support for researchers task will make proposals for ways to set up, in members of the Editorial Board but also by a in the construction and operational phases of the construction phase, of a help-desk able to network of correspondents attached to mem- the project. There are three main activities to fulfil user needs in a more automatic way, for be taken care of in WP6. ber institutions from many states in the CLARIN network. Other dissemination mate- instance, by inferring what resources and what Web site rials, such as brochures, leaflets and posters language processing components could be First, the construction of an online presence will be also in the responsibility of this work combined in a processing architecture that will where information about CLARIN will be package. fulfil the desired task on user’s primary data. C
10 No 1 / Newsletter CLARIN There are problems with existing licenses (Evaluation & Language resources Distrib- WP7: IPR because there are so many language resources ution Agency) make language resources avail- around, created by many authors, published able and identify, classify, collect, validate and Issues by many publishing companies and collected produce the language resources. The task of by many scholars. There may be thousands of this WP is to define the relation between licenses which are somewhat different from ELRA/ELDA and CLARIN, and how they will each other. It is one of the tasks of this work cooperate. package to study the stock of existing licenses Open or commercial Kimmo Koskenniemi and create a set of model licenses to be used resources? WP7 coordinator when making future agreements between the collectors and the authors/publishers, between Traditionally, research use has been free of his work package will deal with intellectu- the collectors and computing centres, and charges but this may not be the only alterna- Tal property rights (IPR) and other legal between the end users and collectors. It is also tive. E.g. the Russian Integrum offers almost issues. The IPRs include copyrights for mate- a task of this work package to find ways for all published Russian newspapers and periodi- rials and patents for software. In addition, migrating existing materials into the framework cals through its commercial on line system licences and many kinds of agreements for of new standard CLARIN licenses. There are which is also used for researchers. The option of authorization and authentication are necessary far more scholars and students than any col- including commercial resources in the CLARIN for the proper handling and use of language lector knows directly. It is the task of this work scheme will be studied in this WP. resources and tools. package to find ways (together with WP2) for CLARIN needs human language technologies managing the licenses and permissions in a rea- What kind of copyright licenses for a wide array of languages as dealt with in we need? sonable way to prevent excessive burden to the WP5 and WP2. users (to get a permission) or the collectors (to The author of written texts and the speaker of The programs and software modules need to grant a permission) for the materials but still oral works have a copyright and often, a com- be compatible, i.e. various parts need to be maintain the confidence of the authors and mercial publisher has acquired some of these combined to form the services. Software publishers. rights. Language resources are often collected licenses may make this very difficult or impos- by scholars. The collectors need an explicit The role of organizations and sible in some cases. license from the author and/or the publisher in agencies This WP will produce recommendations for order to use and let some other scholars use Organizations such as ELRA (European licenses both for open source and commercial the resources. Language Resources Association) and ELDA software. C
the activity to succeed, national agencies will the existing legal forms are appropriate for WP8: be contacted (with the help of the partici- pan-European research infrastructures. Construction and pants) and will form a working group. A It is the intention of the Commission to sub- strong interaction with the Strategic mit a proposal for an ERI framework to the Exploitation Coordination Board and Scientific Board (see Council as soon as possible. We believe that page 8) is required. we can start using the draft proposal in the Agreement development of our ideas already as early as CLARIN Agreement the autumn of 2008.
We will call the document to be created the Participation of national CLARIN Construction and Exploitation agencies Bente Mægaard Agreement (CCEA). The draft CCEA will However, before we start discussions with WP8 coordinator clearly indicate where complete agreement national agencies on their future involvement, between the participating countries has been we first have to understand the landscape well: he problem of building a joint infrastruc- reached and where action needs to be taken at how are the national agencies structured in all Tture has many different dimensions apart higher political or administrative levels. participating countries, how do they handle from the technological and scientific dimen- research infrastructure, how do they manage European Research Infrastructure sions addressed in the rest of the CLARIN funding, in particular funding for internation- project. The main task of this activity is to pre- The European Commission is currently devel- al cooperation, what are the known problems, pare a draft agreement document for national oping a legal framework, European Research how can we describe best practise, etc. This is funding agencies to ensure proper funding of Infrastructure — ERI — for legal entities with the task of the first 12 months. the CLARIN infrastructure in the future — the task of running a pan-European research The fact that the European Commission has after the preparatory phase. infrastructure. The European Commission has started the work on the ERI is a strong sup- Consequently, all countries participating in analyzed existing legal structures under port to our work, as this means that all coun- the CLARIN project have a representative for national law, such as the Dutch foundations, tries will have agreed already to this concept, this activity. However the participants are all as well as existing legal forms under so our task will be only to adapt this general universities or other academic institutions. For Community law, but has found that none of framework to our specific purposes. C
CLARIN Newsletter / No 1 11 May 2008 2008-06-25 to 2008-06-29: Digital Humanities Conference, Oulu, Finland CLARIN calendar 2008-05-27 to 2008-06-01: LREC2008 Conference with pre- and post- 2008-06-26 to 2008-06-28: Stakeholder Workshop “Central and Eastern conference workshops, Marrakesh, Morocco European Scholarship in the Humanities – harnessing the assets”, 2008-05-27: CLARIN Consortium meeting, LREC2008 Conference, European Science Fundation, Sofia, Bulgaria Marrakesh, Morocco of events July 2008 June 2008 2008-07-03 to 2008-07-06: Teaching and Language Corpora Conference, 2008-06-08 to 2008-06-10: NooJ2008 conference, Budapest, Hungary TALC2008 with pre-conference workshops, Lisbon, Portugal Here is the list of CLARIN events and events 2008-06-15 to 2008-06-20: ACL & HLT conference with pre- and post- conference tutorials and workshops, Columbus, Ohio, USA August 2008 from the fields of language resources and lan- 2008-06-19: PID Services talk, Nijmegen, the Netherlands 2008-08-04 to 2008-08-15: ESSLI 2008, Hamburg, Germany guage tools that could be of an interest to 2008-06-23 to 2008-06-26: Information Technology Interfaces (ITI2008), 2008-08-16 to 2008-08-24: COLING conference with pre- and post-con- CLARIN members. Dubrovnik/Cavtat, Croatia ference tutorials and workshops, Manchester, UK C
Evaluations and Language resources Distribution Agency (ELDA); Paris; University of Utrecht/Netherlands Graduate School of Linguistics; Utrecht; Khalid Choukri Jan Odijk Join CLARIN Université Paris 4 Sorbonne / CELTA ; Paris; Andre Wlodarczyk ILK Research Group ; Tilburg; Antal van den Bosch CLARIN project is a combination of LIF-CNRS ; Marseille; Michael Zock Huygens Instituut KNAW ; Den Haag; Karina van Dalen-Oskam Collaborative Projects and Coordination and Germany: Berlin-Brandenburg Academy of Sciences; Berlin; Alexander Norway: Dept. of Culture, Language and Information Technology; Bergen; Geyken Support Actions, registered at the EU under Koenraad de Smedt Deutsches Forschungszentrum für Künstliche Intelligentz; Saarbrücken; the number FRA-2007-2.2.1.2. It started Department of Linguistics and Nordic Studies, University of Oslo; Oslo; Thierry Declerck Janne Bondi Johannessen with the preparatory phase in 2008 that will Institut für Deutsche Sprache; Mannheim; Marc Kupietz Det humanistiske fakultet, Universitetet i Tromsø; Tromsø; Trond Trosterud make the grounds for the next phases and it Max Planck Institute for Evolutionary Anthropology; Leipzig; Hans-Joerg Norwegian University of Science and Technology; Trondheim; Torbjørn will cover the generic, language independent Bibiko Nordgård activities. In order to do our work properly we University of Frankfurt/Main Comparative Linguistics; Frankfurt/Main; Jost Poland: University of Wroclaw ; Wroclaw; Adam Pawlowski will have to rely on a much wider circle than Gippert Institute of Applied Informatics, Wroclaw University of Technology; just the formal consortium partners in the University of Leipzig; Leipzig; Codrina Lauth Wroclaw; Maciej Piasecki project. For this reason we have opened up University of Stuttgart; Stuttgart; Ulrich Heid Institute of Computer Science, Polish Academy of Sciences ; Warsaw; all our project working groups for participa- Universität Tübingen; Tübingen; Erhard Hinrichs Adam Przepiórkowski tion by organizations that are not part of the University of Giessen; Giessen; Henning Lobin Institute of English Language, Univeristy of Lodz; Lodz; Lukasz Drozdz consortium. Computational Linguistics Department, University of Heidelberg; Institute of Slavic Studies, Polish Academy of Sciences ; Warsaw; Violetta Heidelberg; Anette Frank Koseska-Toszewa Members University of Augsburg ; Augsburg; Ulrike Gut Portugal: University of Lisbon, NLX-Natural Language and Speech Group; Country; Institution; Location; Contact person Greece: Institute for Language and Speech Processing; Athens; Stelios Lisbon; António Branco Austria: University of Vienna; Vienna; Gerhard Budin Piperidis Romania: Al.I.Cuza; Iasi; Dan Cristea Belgium: ALT (Acquiring Language through technology); Leuven - Kortrijk; Hungary: Academy of Sciences; Budapest; Tamás Váradi Institute for Computer Science, Romanian Academy of Sciences; Iasi; Hans Paulussen Budapest University of Technology and Economics Media Research (BME Horia-Nicolai Teodorescu Center for Computational Linguistics ; Leuven; Ineke Schuurman MOKK); Budapest; Peter Halacsy Research Institute for Artificial Intelligence, Romanian Academy of Center for Dutch Language and Speech, University of Antwerp; Antwerp; University of Szeged, Department of Informatics, Human Language Sciences; Bucharest; Dan Tufis Walter Daelemans Technology Group; Szeged; Dóra Csendes University Babes-Bolyai; Cluj-Napoca; Doina Tatar ELIS-DSSP; Gent; Jean-Pierre Martens Iceland: Institute of Linguistics, University of Iceland; Reykjavík; Eiríkur Serbia: Faculty of Mathematics, University of Belgrade; Belgrade; Du{ko Legal Informatics and Information Retrieval, Katholieke Universiteit Rögnvaldsson Vitas Leuven; Leuven; Marie-Francine Moens Icelandic Centre for Language Technology; Reykjavík; Eiríkur Rögnvaldsson Slovenia: Josef Stefan Institute; Ljubljana; Toma` Erjavec Laboratory for Digital Speech and Audio Processing - VUB - ETRO/DSSP ; Ireland: National University of Ireland; Galway; Sean Ryder Alpineon d.o.o. ; Ljubljana; Jerneja @ganec Gros Brussels; Werner Verhelst Israel: Technion-Israel Institute of Technology; Haifa; Alon Itai Spain: Institut Universitari de Lingüística Aplicada, Universitat Pompeu ESAT-PSI/Speech; Leuven; Patrick Wambacq Italy: Dipartimento di Linguistica Teorica e Applicata, Università di Pavia; Fabra; Barcelona; Núria Bel Bulgaria: Department of Computational Linguistics, Institute for Bulgarian Pavia; Andrea Sansò Universitat de Lleida ; Lleida; Gloria Vázquez Language, Bulgarian Academy of Sciences; Sofia; Svetla Koeva Istituto di Linguistica Computazionale; Pisa; Nicoletta Calzolari TALG Research Group (University of Vigo) ; Vigo; Xavier Gómez Guinovart Institute for Parallel Processing; Sofia; Kiril Simov Department of Computer Science, University of Rome "Tor Vergata" ; Sweden: Lund University; Lund; Sven Strömqvist Mathematical Linguistics Departement, Institute of Mathematics and Rome; Fabio Massimo Zanzotto Språkbanken, Dept. of Swedish Language, Göteborg University; Informatics, Bulgarian Academy of Sciences; Sofia; Ludmila Dimitrova European Academy Bozen/Bolzano; Bolzano; Andrea Abel Gothenburg; Lars Borin Croatia: University of Zagreb, Faculty of Humanities and Social Sciences; Latvia: Institute of Mathematics and Computer Science, University of Zagreb; Marko Tadi} Latvia; Riga; Inguna Skadina Dept. Speech, Music and Hearing, CSC, KTH ; Stockholm; Kjell Elenius Institute of Croatian Language and Linguistics; Zagreb; Damir ]avar Tilde; Riga; Inguna Skadina Uppsala University, Department of Linguistics and Philosophy; Uppsala; Joakim Nivre Cyprus: Cyprus College / Research Center; Nicosia; Antonis Theocharous Lithuania: Institute of the Lithuanian Language; Vilnius; Daiva Vaisniene Department of Linguistics; Göteborg; Anders Eriksson Czech Republic: Charles University; Prague; Eva Haji~ová Center of Computational Linguistics, Vytautas Magnus University ; Kaunas; Faculty of Informatics, Masaryk University ; Brno; Ale{ Horák Ruta Marcinkeviciene Department of Computer and Information Sciences, Linköping University; Linköping; Lars Ahrenberg The Institute of the Czech Language, Czech Academy of Sciences; Prague; Luxembourg: European Language Resources Association (ELRA); Karel Oliva Luxembourg; Victoria Arranz Swedish Institute of Computer Science AB ; Stockholm; Björn Gambäck Denmark: Center for Sprogteknologi, University of Copenhagen; Malta: University of Malta, Dept. of computer science; Malta; Michael Language council of Sweden ; Stockholm; Rickard Domeij Copenhagen; Bente Maegaard Rosner HUMlab, Umeå University ; Umeå; Patrik Svensson Dansk Sprognævn - Danish Language Council; Copenhagen; Sabine Netherlands: Meertens Institute; Amsterdam; H.J. Bennis Turkey: Sabanci University - Human Language and Speech Laboratory; Kirchmeier-Andersen Data Archiving and Networked Services; Den Haag; Henk Harmsen Istanbul; Kemal Oflazer Society for Danish Language and Literature; Copenhagen; Jørg Asmussen University of Twente, Human Media Interaction Group; Enschede; Roeland UK: Arts and Humanities Data Service; London; Sheila Andersen Estonia: University of Tartu; Tartu; Tiit Roosmaa Ordelman Department of Linguistics and English Langauge, Lancaster University; Finland: CSC - the Finnish IT Center for Science ; Espoo; Tero Aalto Center for Language and Cognition; Groningen; Wyke van der Meer Lancaster; Anna Siewierska University of Helsinki; Helsinki; Kimmo Koskenniemi Digital Library for Dutch Literature; Leiden; C.A. Klapwijk Oxford Text Archive; Oxford; Martin Wynne Department of Foreign Languages and Translation Studies, University of Instituut voor Nederlandse Lexicologie; Leiden; Remco van Veenendaal University of Sheffield; Sheffield; Wim Peters Joensuu; Joensuu; Jussi Niemi Leiden University Centre for Linguistics; Leiden; Jeroen van de Weijer University of Surrey; Guildford; Lee Gillam University of Tampere; Tampere; Eero Sormunen Centre for Language Studies, Radboud University; Nijmegen; Pieter Research Institute of Information and Language Processing at the The Research Institute for the Languages of Finland; Helsinki; Toni Suutari Muysken University of Wolverhampton ; Wolverhampton; Gina Sutherland France: ALTIF; Nancy; Bertrand Gaiffe Centre for Language and Speech Technology, Radboud University; Language Technologies Unit, Bangor University; Bangor; Briony Williams TELMA/DIS CNRS; Paris; Florence Clavaud Nijmegen; L. Boves / N. Oostdijk Department of English, The University of Birmingham ; Birmingham; CNTRL; Nancy; Bertrand Gaiffe Max-Planck-Institute for Psycholinguistics; Nijmegen; Peter Wittenburg Oliver Mason
CLARIN newsletter • publisher CLARIN • editor-in-chief Marko Tadi} • editorial board Gerhard Budin, Dan Cristea, Koenraad De Smedt, Laurent Romary, Martin Wynne • http://www.clarin.eu
12 No 1 / Newsletter CLARIN