What Can I Do with this Data Point? Towards Modeling Legal and Ethical Aspects of Linguistic Data Collection and (Re-)use

Annett Jorschick, Paul T. Schrader, Hendrik BuschmeierCOLING 2024

TL;DR

This paper introduces INF, a novel research data management infrastructure designed to automate and streamline the ethical and legal aspects of linguistic data collection and reuse. By providing researchers with guided tools and customized consent forms, INF aims to foster open science practices while ensuring compliance with complex regulations.

What Problem Does This Paper Address?

The paper tackles a significant challenge in linguistic research: the inherent difficulty in adhering to open science principles due to the sensitive and diverse nature of linguistic data. Unlike many other scientific fields, linguistic data often contains personal information, comes from vulnerable populations, and can be challenging to anonymize without losing crucial research value (e.g., multimodal behaviors like facial expressions or gaze). This complexity creates uncertainty for researchers regarding legal requirements for data collection, consent, storage, and sharing, hindering transparency, reproducibility, and the valuable reuse of linguistic resources. Existing tools for ethics applications and consent forms are often generic and struggle to accommodate the unique needs of linguistic data.

Methods

The core of the proposed solution is the INF (User-oriented research infrastructure assisting linguistic data collection and (re-)use) project. INF is developing a research data management infrastructure with three key modules:

  1. Data Collection Setup Wizard: This tool guides researchers through study planning, automatically generating customized consent forms, assisting with ethics applications, and creating data management plans. It aims to maximize the application of FAIR (Findable, Accessible, Interoperable, Reusable) principles.
  2. Computational Platform: This module centralizes information about studies and their associated data. It converts collected information into standardized metadata formats and integrates it with participant consent to model permissible operations for each data point.
  3. Search Engine: This component allows researchers to query and retrieve available data resources, facilitating data reuse and collaboration. It supports advanced search functionalities based on various criteria like data type, topic, and permissions.

The project emphasizes an interdisciplinary approach, combining linguistic expertise, legal knowledge, and computational science to address methodological, legal, and technical issues. A key aspect is the careful analysis of legal norms (GDPR, BDSG, DSG-NRW) and ethical guidelines (Helsinki Declaration, Belmont Report, DGPs) to create combinable, legally sound, and easy-to-understand text blocks for consent forms, while also resolving redundancies.

Datasets

Not discussed in the paper. The paper focuses on the infrastructure for handling linguistic data, not on specific datasets used for training or evaluation.

Key Insights and Ideas

  • The "Data Point" as a Unit of Control: The concept of a "data point" being characterized by metadata to define permissible operations is crucial for granular control over data usage, especially when consent is complex.
  • Automation as a Solution to Complexity: Recognizing that legal and ethical considerations are often intertwined and challenging to navigate, the paper highlights automation as a powerful tool to support researchers and ensure compliance.
  • Tailored Consent is Essential: Generic consent forms are insufficient for linguistic data. The INF project's focus on generating highly customized, ethically and legally sound consent forms is a significant contribution.
  • Bridging the Gap Between Research and Regulation: The project directly addresses the disconnect between the practicalities of linguistic data collection and the stringent legal and ethical frameworks, aiming to make open science more accessible for linguists.
  • The Importance of Interdisciplinarity: The success of such an infrastructure hinges on the collaboration between linguists, legal experts, and computer scientists, acknowledging that no single discipline can solve this problem alone.

Key Results

The paper describes the development of an infrastructure project (INF) and its conceptual framework. Specific quantitative results regarding the effectiveness or performance of the INF system are not yet presented, as it is in its initial funding phase. The key "results" are the conceptualization and planned implementation of the wizard, computational platform, and search engine, along with the methodology for integrating legal and ethical considerations.

Limitations and Future Work

  • Initial Development Phase: The paper acknowledges that INF is in its first funding phase, focusing on implementation and integration within Bielefeld University.
  • Data Sharing and Internationalization: Data sharing with national and international infrastructures is planned for the second funding phase, contingent on a thorough evaluation of software security.
  • Complexity of Legal Interpretation: While the paper outlines the legal frameworks, the ongoing challenge of interpreting and applying these complex regulations to diverse linguistic research scenarios remains.
  • User Adoption and Usability: Future work will likely involve extensive testing and refinement to ensure the infrastructure is user-friendly and effectively adopted by the linguistic research community.
  • Handling of Novel Data Types: As linguistic research evolves, the infrastructure will need to adapt to new forms of data and emerging ethical considerations.

Summary updated

BibTeX
@inproceedings{jorschick2024what,
  author = {Annett Jorschick and Paul T. Schrader and Hendrik Buschmeier},
  title = {{What Can I Do with this Data Point? Towards Modeling Legal and Ethical Aspects of Linguistic Data Collection and (Re-)use}},
  booktitle = {International Conference on Computational Linguistics},
  year = {2024},
}
Figure