Search

The CLARIN Resource Families (CRF) initiative provides manually curated overviews of prominent language resources and technologies deposited across the distributed CLARIN infrastructure (Lenardič and Fišer 2022). The main aim of CRF is to support other core services of CLARIN from the perspective of the FAIR principles (Wilkinson et al. 2016). CRF enhances the findability and accessibility of CLARIN resources by collating them under their most common typological characteristic. The initiative facilitates re-use by providing comprehensive descriptions tailored to the unique technical features of each of the families, as well as their qualitative characteristics. Furthermore, CRF provides a funding instrument for external projects to contribute new overviews. Though originally focused on written corpora (e.g., corpora of parliamentary proceedings, corpora of academic texts), in 2022, CRF was expanded to include corpora of oral history. At present one collection is currently featured – the Ravensbrück corpora (Calamai et al. 2022a) – whose creation was supported by the aforementioned CRF funding instrument. This corpus family contains 8 collections of recorded interviews with survivors of the female concentration camp Ravensbrück, conducted in different languages, such as English, German, Hebrew, and French. See https://www.clarin.eu/resource-families/oral-history-corpora. One collection is available for download (Collection Bruzzone; see Bruzzone and Beccaria Rolfi 1976) while the others can be streamed online. The inclusion of the Ravensbrück corpora in CRF represents an illustrative example of how the CLARIN infrastructure incorporates and provides documentation for complex objects like oral history sources whose provenance and metadata documentation widely differ from standard written corpora and even from contemporary interviews born digitally. The team working on the Ravensbrück resource family (see Calamai et al. 2022b) availed themselves of CLARIN’s Component Metadata Infrastructure (CMDI), which is a framework for metadata description that “supports flexible definitions of metadata structure and semantics” by allowing researchers to “create and use their own [metadata] schema tailored specifically towards the requirements of [their] project” (Windhouwer and Goosen 2022: 194 and 199). All the 8 collections within the Ravensbrück family are accompanied by extensive CMDI metadata, prepared by Calamai et al. (2022a,b). The peculiarity of the interviews in the Ravensbrück family is that they were mostly recorded on an analogue carrier (i.e., audio cassettes), so a new CMDI metadata profile was created that is tailored to such legacy interviews not born digitally. This metadata profile has additional components describing “information about the context in which the interviews were conducted” as well as “information about the process of digitisation” (Calamai et al. 2022a: 3). Being thus digitised, comprehensively described, and carefully curated, the Ravensbrück corpora present a unique opportunity to study and compare these historical interviews. To facilitate their use in research, CLARIN offers through its Speech data and Technology network (Draxler et al. 2020) an open-source web application called TranscriptionPortal (https://speechandtech.eu/transcription-portal), where certain audio recordings (e.g., Collection Bruzzone, United States Holocaust Memorial Museum) can be uploaded and then orthographically transcribed on the fly, with manual phonetic and word alignment for a variety of languages., Funded in the context of the CLARIN Resource Families Project.

The current largest open-source generic automatic speech recognition (ASR) system for Dutch, Kaldi NL, does not include a domain-specific healthcare jargon in the lexicon. Commercial alternatives (e.g., Google ASR system) are also not suitable for this purpose, not only because of the lexicon issue, but they do not safeguard privacy of sensitive data sufficiently and reliably. These reasons motivate that just a small amount of medical staff employs speech technology in the Netherlands. This paper proposes an innovative ASR training method developed within the Homo Medicinalis (HoMed) project. On the semantic level it specifically targets automatic transcription of doctor-patient consultation recordings with a focus on the use of medicines. In the first stage of HoMed, the Kaldi NL language model (LM) is fine-tuned with lists of Dutch medical terms and transcriptions of Dutch online healthcare news bulletins. Despite the acoustic challenges and linguistic complexity of the domain, we reduced the word error rate (WER) by 5.2%. The proposed method could be employed for ASR domain adaptation to other domains with sensitive and special category data. These promising results allow us to apply this methodology on highly sensitive audiovisual recordings of patient consultations at the Netherlands Institute for Health Services Research (Nivel).

Slides for the conference:DH Benelux 2022: RE-MIX. Creation and alteration in DH (Hybrid) Challenges on the Promising Road to Automatic Speech Recognition of Privacy-Sensitive Dutch Doctor-Patient Consultation Recordings Authors:Berrie van der Molen (Utrecht University), Cristian Tejedor-García (Radboud University), Henk van den Heuvel (Radboud University), Roeland Ordelman (Sound & Vision), Toine Pieters (Utrecht University), Sandra van Dulmen (Nivel), Arjan van Hessen (University of Twente)

This white paper provides the necessary basis for understanding the requirements and specifications for remote access to sensitive data (data with potentially harmful effects in the event of their disclosure) in the social sciences and the humanities (SSH). It is result of the work implemented in SSHOC Task 5.4 Remote Access to Sensitive Data. It is intended to provide guidance and recommendations to the EOSC stakeholders for future infrastructure investment for remote access to sensitive data in the SSH. To ensure that this guidance can, in fact, be implemented, the recommendations are based on the knowledge of numerous data professionals who have direct experience planning, implementing, managing, and sustaining diverse forms of remote access and secure facilities. In doing so, our goal has been to maintain the vision of expanding such infrastructure, while remaining grounded in the practicalities of operating such facilities in a sustainable manner. In this domain, it is now recognized that the ideal of “open data” needs to be balanced with privacy and other factors that can require moderating access to sensitive data, as reflected in the EU Commission’s (2016) stance of “as open as possible, as closed as necessary.” Developments in the past five years have advanced data access, primarily through “safe enclaves”, i.e., physical rooms that provide security for data access (see Glossary). This represents a major improvement for data accessibility, but international, comparative, efficient research requires augmenting the research infrastructure by enabling remote access to data from a researcher’s desktop. Solutions have operated for several years (e.g., UK Data Archive Secure Lab, ICPSR Virtual Data Enclave), but most of these still face limitations on the scope of data available, geographic limitations, etc. More recently, new infrastructures are being developed, some spanning several countries. These efforts are commendable and represent major improvements. However, limited resources, and complex legal variations (national implementations of GDPR), as well as other factors, have prevented implementation of a broader solution. As countries across Europe look at the emerging multi-national infrastructures, it is crucial to address the need for a European answer, at scale, with sustainable funding. The recommendations offered here are guided by our observations that most successful infrastructures embody two features: 1) they are human as well as technical, and 2) they are neither purely centralised nor decentralised, but well-crafted hybrids.

What was the workshop about? This workshop was the fifth of a series that started in 2015, and it was the third organized under the CLARIN umbrella. About 30 participants registered and attended the meeting The attendants came from all over Europe amongst others from the Netherlands, Finland, Ireland, Poland, Italy, France, Estonia and the UK, with backgrounds in language and speech pathology, linguistics and phonetics, speech technology, data archiving, ICT, and law. This is exactly the mix that makes DELAD attractive and suited for discussing and sharing CSD. The workshop was organized by the DELAD steering group together with Esther Hoorn from. the CLARIN Legal and Ethical Issues Committee (CLIC) Goal The aim of this workshop was to: Extend DELAD network with new participants; Explore with the participants the potential of the new CLARIN K-Centre on Atypical Communication Expertise (ACE) for hosting CSD of DELAD members; Exchange deeper insights on Data Protection Impact Assessments (DPIAs); Discuss voice conversion as a means to pseudonymize speech. An overview of the workshop can be found here; Day 1 On the first day the workshop started with four presentations: “Croatian written and spoken corpora of speech with communication disorders” (Gordana Hržica) “Oral and written documents of mental health patients” (Silvia Calamai & Rosalba Nodari) “Using electromagnetic articulography for the purpose of studying speaking styles and speech disorders” (Katarzyna Klessa, Anita Lorenc, & Łukasz Mik) “Parkinson’s disease: A French corpus collected using MonPaGe protocol” (Veronique Delvaux) Apart from the discussions about the research itself considerations were made for sharing the resulting CSD. Especially those of Calamai & Nodari and Delvaux, which were still looking for a suitable shelter for the corpus where DELAD could play a role. The afternoon session was exactly devoted to that topic. How can DELAD in cooperation the CLARIN Knowledge Centre for Atypical Communication Expertise assist in realizing the GDPR compliant access to such CSD? “Help from DELAD and CLARIN Centre for Atypical Communication Expertise (ACE) in sharing CSD” (Henk van den Heuvel) “How to access & deposit existing data at CLARIN centres, profiles of metadata, licenses” (Paul Trilsbeek) Day 2 In the morning session of the second day of the workshop a role play was scheduled as devised by Esther Hoorn and her team. The role play led to lively discussion addressing various aspects of the ingredients needed to be taken into account when documenting your considerations in a Data Protection Impact Assessment (DPIA). Such a game is an entertaining way to touch upon various aspects that are relevant when sharing your CSD, leading to various eye openers in the discussions! Here, the breakout rooms in zoom served well to split the group into two. In the afternoon, Rob van Son gave a keynote talk on “Use voice conversion for pseudonymisation?”. The intriguing idea behind the method he presented is on the one hand to retain linguistic & paralinguistic features of the speech, and on the other hand to remove the identity of the speaker. If that could be done successfully then the speech could count as anonymised and won’t be subject to GDPR. Rob van Son presented results from the Voice Privacy Challenge 2020 and concluded that speaker identifying information can be removed from speech, but also noticed that all systems had issues with naturalness and intelligibility. A relevant issue for DELAD of course is if pathological speech, e.g. dysarthric speech, will still be studied after pseudonymisation. And for this further study evidence is needed. During the workshop several participants expressed interest in a case study for part of their material. Lessons learnt, points taken It was great to have the support of CLARIN staff in organizing an online workshop like this via Zoom. CLARIN took care of the zoom addresses and mailings, and the breakout room for coffee breaks and the role play. This relieved us as workshop organizers from a serious organizational burden, so that we could concentrate on the content of the workshop. We were satisfied with the long breaks for the lunch of two hours. It gave participants the opportunity to digest the content of the morning session, and do something (relaxing) in between. We also stopped quite early (at 16:00) to avoid our participants becoming “Zoombies” at the end of the day. This experience let us think about how to take advantage of online meetings in the future in addition to face-to-face meetings. We were happy with the new researchers that subscribed to the workshop. Their presentations were very interesting and diverse. The workshop in general provided a good mix of research and data oriented presentations and presentations focusing on the legal and ICT support that DELAD can offer to share such data. The role play on Data Impact Protection Assessment (DPIA) was a highly valued interactive aspect of the program. Action points As relevant action points for our DELAD network we have identified: Set up a number of case studies on pseudonymization of (multilingual) pathological speech data and organize a workshop around this; Look into ways to promote DELAD and its benefits for sharing CSD on national levels. Mentioned were a slide desk for promotion, social media campaign, folders); Partners are interested in curating and sharing their corpora via DELAD; This will yield further case studies on sharing CSD. DPIAs should be integrated; Share Reference DPIA’s to the community via DELAD website; Share consent form templates via DELAD website, how to structure them, relevant aspects (must haves), checklists with examples. Topics for next workshops: Progress on pseudonymization Sharing data via remote secure access option Experiences on sharing datasets (make it more concrete), If you are interested in becoming a member of DELAD you can subscribe here. Contact for the workshop: Henk van den Heuvel (h.vandenheuvel@let.ru.nl) This event was supported by CLARIN and by the SSHOC Project (Grant Agreement 823782 under H2020)

Since 2016, the tour de CLARIN initiative has been periodically highlighting prominent user involvement activities in the CLARIN network in order to increase the visibility of its members, reveal the richness of the CLARIN landscape, and display the full range of activities that show what CLARIN has to offer to researchers, teachers, students, professionals and the general public interested in using and processing language data in various forms. In 2019, we expanded the initiative to also feature the work of CLARIN Knowledge Centres, which offer knowledge and expertise in specific areas provide to researchers, educators and developers alike. Initially conceived as a series of blog posts published on the CLARIN website, Tour de CLARIN soon proved to be one of our flagship outreach initiatives, which has been released in the form of two printed volumes. this third volume of tour de CLARIN is organized into two parts. In Part 1, we present the six countries which have been featured since January 2020: Norway, the United Kingdom, Germany, France, South Africa, and Iceland. Each national consortium is presented with five chapters: an introduction to the consortium, their members and their work; a description of one of their key resources; the presentation of an outstanding tool; an account of a successful event for the researchers and students in their network; and an interview with a renowned researcher from the Digital Humanities or Social Sciences who has successfully used the consortium’s infrastructure in their work. In Part 2, we present the work of the six Knowledge Centres that have been visited since the publication of the second volume in November 2019: the Impact-CKC K-Centre, the Knowledge Centre for Polish Language technology, the Phonogrammarchiv Knowledge Centre, the Knowledge Centre for Atypical Communication Expertise, the LUND University Humanities Lab Knowledge Centre, and the Spanish Knowledge Centre. Each Knowledge Centre is presented with two chapters: a presentation of what the K-Centre offers to researchers, and an interview with a renowned researcher who has collaborated with the K-Centre.

There is much talk about the need for multidisciplinary approaches to research and the opportunities that have been created by digital technologies. A good example of this is the CLARIN Portal, that promotes and supports such research by offering a large suite of tools for working with textual and audio-visual data. Yet scholars who work with interview material are largely unaware of this resource and are still predominantly oriented towards familiar traditional research methods. To reach out to these scholars and assess the potential for integration of these new technologies a multidisciplinary international community of experts set out to test CLARIN-type approaches and tools on different scholars by eliciting and documenting their feedback. This was done through a series of workshops held from 2016 to 2019, and funded by CLARIN and affiliated EU funding. This paper presents the goals, the tools that were tested and the evaluation of how they were experienced. It concludes by setting out envisioned pathways for a better use of the CLARIN family of approaches and tools in the area of qualitative and oral history data analysis.

The current ‘migration crisis’ in Europe started at the end of 2014, as migrants, primarily from Africa and Syria, and mainly due to war, arrived in Europe in big numbers. Initially, the European Council labelled the situation as tragic (European Council statement 2015, European Commission and its priorities, 2018). With the growing influx of migrants, member states started referring to the situation as a security problem, resulting in a greater reluctance to accept newcomers into their territory. In February 2016, the EU member states imposed a European border, deployed coast guards and implemented a joint Turkey action plan. “Fortress Europe” was gaining momentum. Europe promised aid to Western Balkan countries in handling the massive migration waves. Countries were forced to accept a quota of migrants (Bauerová, 2015)., peer-reviewed

As increasingly sophisticated new technologies for working with numbers, text, sound and images come on stream, there is one type of data that begs to be explored by the wide array of available digital humanities tools, and that is interview data. (De Jong, 2014), (Corti and Fielding, 2016). A stream of concerted activities by the authors of this workshop based on their engagement with interview data from different perspectives, led to a series of 4 workshops from 2016 to 2018, held in Oxford, Utrecht, Arezzo and München, that were funded by CLARIN (Common Language Resources and Technology Infrastructure). This workshop intended to present the motivation for this DH2019-workshop by sketching the complex design and results of these series of multidisciplinary workshops. The premise was that the multimodal character (text, sound and facial expression) and multidisciplinary potential of interview data (history, oral and written language, audio-visual communication) is rarely fully exploited, as most scholars focus on the analysis of the textual representation of the interviews. This might change by getting acquainted with scholarly approaches and conventions from other disciplines.

‘But I don’t know how to work with [name of tool or resource]’ is something one often hears when researchers in Human and Social Sciences (HSS) are confronted with language technology, be it written or spoken, tools or resources. The TTNWW project shows that these researchers do not need to be experts in language or speech technology, or to know all kinds of details about the tools involved. In principle they only need to make clear what they want to achieve. In this chapter we describe a series of tools that are already available as a webservice. Details are not presented — interested readers are referred to the papers mentioned in the References and to the TTNWW website. Kemps-Snijders M., Schuurman I., Daelemans W., Demuynck K., Desplanques B., Hoste V., Huijbregts M., Martens J.-P., Paulussen H., Pelemans J., Reynaert M., Vandeghinste V., van den Bosch A., van den Heuvel H., van Gompel M., van Noord G., Wambacq P., ''TTNWW to the rescue: No need to know how to handle tools and resources'', CLARIN in the Low Countries, chapter 7, pp. 83-93, Jan Odijk and Arjan van Hessen, eds., 2017, Ubiquity Press, London. ispartof: CLARIN in the Low Countries pages:83-93 ispartof: pages:83-93 status: published