multilingualsr

Screen Reader Interaction with Web Content experiment

Welcome to this repository dedicated to examining the interaction between screen readers and web content. Here, you’ll find insights and findings from this experiment, shedding light on the intricacies of accessibility in digital environments and how screen readers (SRs) navigate them.

Overview

In June 2020, Xurxe Toivo García, an Accessibility Specialist, published the article “The troubled state of screen readers in multilingual situations”. The article clicked with us so much that we decided to dive deeper into his experiments. We shifted our focus to not just see how well or poorly SRs handle different multilingual HTML elements, but to really understand how SRs interact with both the content and the software involved (like browsers and speech synthesizers). We wanted to figure out how they can be set up to work in a multilingual context and why they produce speech output in one language or another for each block of information.

Screen readers, browsers, synthesisers and languages

We chose our SRs and browser combinations based on the latest SR usage survey (WebAIM 2024). To keep things manageable, we only used one browser with the two mobile SRs (TalkBack and VoiceOver) and with VoiceOver for Mac since these are the most common setups. Here’s the full list of combinations we used:

• JAWS with Chrome and Edge (Windows). Vocalizer Expressive synthesiser.
• NVDA with Chrome and Firefox (Windows). Vocalizer Expressive synthesiser.
• VoiceOver with Safari, for Mac OS and iOS. Built-in synthesisers.
• TalkBack with Chrome (Android). Samsung TTS synthesiser.

Methodological Outline

Here’s a quick summary of what we did:

1. Initial Review: We analyzed Toivo García’s article and data, re-testing his experiments with the same SRs.
2. Refocusing the Experiments: We adjusted García’s experiments to fit our research goals, adding elements typical of translation tools and e-dictionaries.
3. Recording Interactions: We recorded how screen readers handled our enhanced test pages, noting translation issues and their causes. We aimed to understand the reasons behind the screen readers’ performances.
4. Results and Conclusions: We identified patterns from our tests, categorized them, and discussed our findings, leading to our conclusions and suggestions for future research.

As a result of this research, we published the article “Multilingual Accessibility in Human–Screen Reader Interaction with Web Content: An Exploratory Study” in Tradumàtica: Tecnologies de la Traducció, a journal published by the Universitat Autònoma de Barcelona. The article provides a comprehensive overview of our methodology, results, and conclusions and is available here.

Testing sites

• Experiment 1
• Experiment 2

Related resources

• Accessible labelling multilanguage experiment, the original experiment by Xurxe Toivo García, which motivated our research and this experiment.
• García, Xurxe T. (2020). “The troubled state of screen readers in multilingual situations”. UX Collective. https://uxdesign.cc/the-troubled-state-of-screen-readers-in-multilingual-situations-f6a9da4ecdf3.
• WebAIM (2020). “Semantic Structure: Regions, Headings, and Lists”. WebAIM. Institute for Disability Research, Policy, and Practice (Utah State University). https://webaim.org/techniques/semanticstructure.

Authors

Álvaro García-Garcinuño (Universidad de Salamanca/Universidad de Puerto Rico)
Jesús Torres-del-Rey (INICO, Universidad de Salamanca)

License

This project is licensed under the MIT License - see the LICENSE file for details.

First published: May 27, 2024
Last updated: September 3, 2024

This site is open source. Improve this page.