Outcomes of the sixth DELAD Workshop held on 22-23 September 2022

What is DELAD? DELAD stands for Database Enterprise for Language And speech Disorders, and is also Swedish for SHARED. DELAD is an initiative to share corpora of speech of individuals with communication disorders (CSD) among researchers. See the DELAD website.

The venue was a zoom room that CLARIN kindly offered and professionally hosted. 

What was the workshop about?

This workshop was the sixth in a row that started in 2015, and it was the fourth organised under the CLARIN umbrella. The workshop was announced via the DELAD website and  via tweets and the CLARIN Newsletter. The workshop was held in the afternoons of 22 and 23 September.

About 30 participants registered and attended the meeting although some of them attended only part time due to other obligations which are difficult to avoid when not physically meeting. In the meeting were 18-20 participants mostly. Participants came from all over Europe amongst others from the Netherlands, Finland, Ireland, Poland, Italy, France, Cyprus, the UK, and Canada, and had backgrounds in language and speech pathology, linguistics and phonetics, speech technology, data archiving, ICT, and law specialists. This is exactly the mix that makes DELAD attractive and suited for discussing and taking care of sharing CSD.

The workshop was organised by the DELAD steering group together with Esther Hoorn from CLARIN’s CLIC. 

The aim of this workshop was to:

• Extend DELAD network with new participants  

• Recent developments at DELAD and CLARIN K-Centre on Atypical Communication Expertise (ACE) 

• Follow up on voice pseudonimisation

• Sharing clinical data: Obtaining clinical data via hospitals vs Obtaining data outside clinical institutes from alternative organisations

• The impact of the Data Governance Act & Data Altruism 

An overview of the workshop’s program can be found here.

Day 1

On the first day we started out with four presentations:

• “Speech STAR - An outline ultrasound-
  tongue-imaging-based curated corpus of disordered and non-disordered speech” (Eleanor Lawson, Joanne Cleland, & Jane Stuart-Smith)

• “Using video‑based pose estimation for automated analysis of interaction” (Satu Saalasti)

• “Rethinking Language: Computational approaches to speech and language assessment for clinical studies” (Chiara Barattieri di San Pietro)

Eleanor Lawson showed the website https://www.seeingspeech.ac.uk/ with UTI and MRI recordings (database). Aimed for teaching and supporting intervention and use of the methods. Speech disorder database is online with search/filter options: https://www.seeingspeech.ac.uk/speechstar/ with all permissions in place to share the data, so they can also be downloaded via Joan Cleland.

Satu Salaasti showed the use of OpenPose for realtime movement estimations. Research into self and other initiated repairs in typical and atypical conversations.

Chiara Barattieri di San Pietro presented the https://www.gatekeeper-project.eu/  They work with a company: ab.acus SRL SWETALY. Mentoring programme: dementia detection.

After the tea break there were three more presentations:

• Recent developments at ACE / DELAD for support of sharing CSD (DELAD Steering group)

• Follow up on voice Pseudonymisation (Rob van Son) (pre-recorded presentation) (Chair: Henk van den Heuvel)

• “Crosslinguistic data in typical and protracted phonological development” (Barbara Bernhardt)

The DELAD steering group highlighted a couple of aspects:

1. CSD annotation tools and techniques

2. Guidelines for consent & storage: https://delad.ruhosting.nl/wordpress/guidelines-consent-storage/ 

3. DELAD DPIA Roleplay material: https://delad.ruhosting.nl/wordpress/dpia-role-play-with-video/ 

The voice pseudonymisation discussion of the previous workshop was continued based on a presentation at Interspeech 2021 by Rob van Son.

In her presentation Barbara Bernhardt formulated some interesting questions to DELAD about the link of DELAD, TLA and Phonbank, and the desired annotation languages for lingual resources.

Day 2

Also on 23 September we started with three research presentations

• “Developments in sensitive data processing at CSC” (Martin Matthiesen)

• “The assumptions for the analysis of speech disorders and primary functions using the CARSTENS AG501 articulograph and an acoustic field distribution analyzer” (Katarzyna Klessa, Anita Lorenc, Łukasz Mik, Agnieszka Borowiec, & Daniel Król)

• “Licensing and distributing language resources via the Language Bank of Finland” (Mietta Lennes)

Martin Matthiesen presented the SD toolbox for sensitive data which ensures that data never leaves the encrypted/safe environment. It is free to use for Finnish institutes and collaborating partners, see https://docs.csc.fi/data/sensitive-data/.

Katarzyna Klessa presented the research goals and data collections for articulatory analysis using the CARSTENS AG501 articulograph, and for acoustic-phonetic analyses using the AFDA camera (Acoustic Field Distribution Analyzer).

Mietta Lennes gave an overview of CLARIN licences used at the Language Bank of Finland: PUB, ACA, RES with extra features for ACA and RES licences. They are now working towards v2.1 of the licences, see https://www.kielipankki.fi/support/clarin-eula/. See also:

https://www.clarin.eu/content/clarin-license-category-calculator

The rest of the afternoon was devoted to legal aspects of data sharing inspired by two presentations:

• The impact of the Data Governance Act & Data Altruism (Pawel Kamocki)

• Lessons learnt from the use of the data sharing template of Dutch Academic hospitals (Anne Jan Sikkema) 

Pawel Kamocki gave a detailed insight into the Data Governance Act which is in force but not in application yet (24 Sep. 2023). He briefly mentioned other interesting acts: AI Act (close to being adopted), Digital services act, Common European Language Data Space. He also spent some time on introducing the 

• Data Intermediation Services (DIS)

• Framework for Data Altruism

• European Data Innovation Board

Anne Jan Sikkema, finally, presented a data sharing template of Dutch Academic hospitals which is a separate controllership agreement: supplying existing data upon request of other party for their research, see 

https://elsi.health-ri.nl/categorieen/verzamelen-en-uitgeven-van-data-en-lichaamsmateriaal/waar-vind-ik-een-voorbeeld-van-0.  Click DTA