User-Centered Design of Digital Tools for Sociolinguistic Studies in Under-Resourced Languages

Jonas Adler, Carsten Scholle, Daniel Buschek, Nicolo’ Brandizzi, Muhadj AdnanACL 2024

TL;DR

This paper addresses the critical need for better digital tools in sociolinguistic research, particularly for under-resourced languages. By applying a User-Centered Design (UCD) approach, the authors identified key challenges faced by sociolinguists and proposed design requirements for more effective software, aiming to bridge the gap between theoretical research and practical utility.

What Problem Does This Paper Address?

Sociolinguistic research, which studies language variation and change, heavily relies on analyzing linguistic corpora. However, for under-resourced languages, this process is often hampered by a lack of specialized software tools. Researchers spend considerable time on manual tasks like transcription and data annotation, which are prone to errors and inefficiencies. Existing tools, when available, are often not designed with the specific needs of sociolinguists in mind, leading to a steep learning curve and limited adoption. This paper aims to tackle this problem by understanding the workflows and challenges of sociolinguists and proposing a user-centered design roadmap for developing better digital tools.

Methods

The study employed a User-Centered Design (UCD) approach, a methodology from Human-Computer Interaction (HCI). The core steps involved:

  • User Study Execution: Conducting in-depth interviews and Think-Aloud Commentaries (TaCs) with four sociolinguists specializing in Arabic dialects. TaCs were used to observe real-time workflows and identify specific difficulties.
  • Data Organization: Recording, transcribing, and organizing interview and observation data.
  • Data Analysis: Applying Grounded Theory (GT) for qualitative data analysis. This involved open coding (breaking down data into concepts), axial coding (grouping concepts into categories), and identifying a core category that represents the primary challenge.
  • User Requirement Elicitation: Deriving general and specific requirements for software solutions based on the identified user needs and challenges.
  • Design Solution Development: Creating initial sketches and low-fidelity prototypes based on the derived requirements, with iterative user feedback integrated throughout.

Datasets

The paper does not introduce new datasets for training or evaluation. Instead, it focuses on qualitative data collected from interviews and observations of four sociolinguists.

Key Insights and Ideas

  • Manual Task Execution is the Core Bottleneck: The most significant challenge identified across all participants was the heavy reliance on manual processes for tasks like transcription, annotation, and data management. This is time-consuming, error-prone, and psychologically taxing.
  • Data Management is Crucial and Problematic: Inconsistent and disorganized data management practices (using disparate files and locations) lead to difficulties in accessing information, inconsistencies, and decreased research productivity.
  • Existing Software Has Entry Barriers: Current software tools often have steep learning curves and complex interfaces, discouraging researchers, especially those not deeply familiar with programming or advanced software.
  • Under-Resourced Languages are Neglected: The majority of existing tools are designed for well-resourced languages, leaving a significant gap for researchers working with less common languages.
  • UCD is Essential for Tailored Tools: A user-centered approach is vital to ensure that digital tools are not only functional but also intuitive and directly address the specific needs and workflows of sociolinguists.

Key Results

The study identified 14 primary user requirements for digital tools in sociolinguistic research. The most significant ones include:

  1. Intuitive Usability: Software interfaces must be simple, tailored to users' knowledge, and focus on essential features.
  2. Data/Variable-Management System: A system to ensure consistent data management and simplify interaction with data.
  3. Digital Annotation: Tools that facilitate and automate parts of the annotation process, allowing for consistency and efficiency.
  4. Automated Analysis: Features that enable automatic counting and analysis of annotations, supporting complex quantitative and statistical tasks.

The paper also proposes design concepts through sketches and low-fidelity prototypes, illustrating how these requirements can be translated into practical software features.

Limitations and Future Work

The authors acknowledge that their study, while providing valuable insights, may not encompass the needs of all sociolinguistic researchers worldwide, particularly those working with languages that lack formal writing systems or in different academic contexts.

Future work includes:

  • Implementing a Functional Open-Source Tool: The ultimate goal is to develop a fully functional, open-source tool based on the identified requirements and design concepts.
  • Exploring Additional Research Directions: The paper highlights complex requirements like automatic transcription and corpus translation, which require further research and user studies.
  • Stimulating Interdisciplinary Perspectives: The authors hope their work encourages collaboration between sociolinguistics and HCI to address the practical aspects of research that are often overlooked.

Summary updated

BibTeX
@inproceedings{adler2024user,
  author = {Jonas Adler and Carsten Scholle and Daniel Buschek and Nicolo’ Brandizzi and Muhadj Adnan},
  title = {{User-Centered Design of Digital Tools for Sociolinguistic Studies in Under-Resourced Languages}},
  booktitle = {Annual Meeting of the Association for Computational Linguistics},
  year = {2024},
}
Figure