Authors: Reeti Mathur | firstname.lastname@example.org, Dr. Erin Brady | email@example.com
Status of Paper: Technical paper under review for W4A, 2019
Link to the paper (not available for distribution until published): W4A Paper under review
Link to previous Poster Paper study: Assets 2018 Poster Paper
Users of social media sites share millions of photographs every day. These posts are mostly inaccessible to people who are blind or visually impaired since they are not accompanied by alternative text descriptions of the image contents which can be read by a screenreader. Although some platforms generate automated image descriptions, these descriptions do not convey the details of the photograph expected by screenreader users. To address this, we designed a mixed methods study involving a collaborative photograph accessibility activity in Facebook groups, and pre and post-activity surveys. We found that online social collaboration between people who are blind or visually impaired and their sighted peers can be beneficial to describe the context of the photograph, provide detailed alt-text, and review the photographs for clarity (a) before the blind or visually impaired user upload their own photographs on Facebook and (b) while interacting with other’s photographs on social media sites.
In order to study the effectiveness of this online collaboration among groups of the BVI people and their Sighted Peers (SP), we applied a mixed-methods approach. Our methods included a pre-activity survey, an in situ collaborative photograph study activity on Facebook, and a post-activity survey.
We recruited two groups to test this online collaboration with each group comprising of one BVI participant and five to ten of their SP. To conduct the study activity, we created two separate closed groups on Facebook for each of the two groups respectively. We maintained the privacy of our participant groups by creating closed groups, which require an invitation from the BVI participant to join and preventing other users from accessing them.
The pre-activity and post-activity surveys were hosted on Qualtrics. We facilitated a mixed-methods approach to analyze the data collected through the pre-activity survey, in situ collaborative photograph study activity on Facebook and the post-activity survey. Quantitative analysis was conducted through Qualtrics to determine the statistics, behaviors and experiences of social media usage of BVI participants and their SP.
We also performed a qualitative analysis by comparing the various styles of alternative text provided by the SP, and charting broader impacts and trends emerging from the Facebook’s study activity pertaining to the details provided by each peer versus the ATDs generated by the AAT feature.
What We Found
From this social collaboration on Facebook, it is inferred that this collaborative practice is a beneficial solution to access inaccessible photographs on social media platforms and to comprehend the context of the photographs clicked by the BVI people themselves. Routing photographs to web volunteers or crowd workers are not recommended since (a) they would not correlate to the context of the photograph as the sighted friends and family would and (b) the privacy concerns with respect to the photographs would restrict the BVI users have them review these photographs for detailed textual descriptions.
“It was a valuable exercise in learning how easy it is to obtain photo content information” – BVI participant 2 (Group 2)
Through a separate post-activity survey that was shared on the two closed Facebook groups for the SP, we learnt that this activity has encouraged the SP to understand the importance of social collaboration to assist their BVI peers get detailed alt-text descriptions in order to better understand the context of the photograph, and the necessity of posting their own photographs with appropriate alternative text for the BVI people to be able to easily interact with media content on social media sites.
“If those of us who can see start adding descriptions like this to our posts it would be incredibly useful to our visually impaired friends” – Sighted Peer 7 (Group 1)
Analyzing these results of individual alternative text descriptions given by the SP of both the groups, we realized that it was an appending process altogether. If one sighted peer commented on the photograph-post with an appropriate alt-text but missed some details that the BVI peer would expect, the other sighted peer adds those details in their comment – thus improving the complete textual description on an iterative basis in a collaborative environment.
“I felt it was better and more accurate when more people commented became small details that one person missed were included by another. It was more involved than I originally thought. I kept finding more small details that needed to be included. It is easier to describe things accurately with multiple people” – Sighted Peer 1 (Group 1)
We aim to further conduct longitudinal schedules with potential collaborative groups on Facebook for a longer duration. In this extended study, we plan to have the BVI participants upload their own photographs themselves directly to their respective collaborative groups to obtain detailed and appropriate alternative text descriptions and photo reviews. We would also have them route any inaccessible photographs they find on their Facebook accounts to their closed groups to get the actual context and content of the photographs.
We believe that this future study would help us determine the manner in which the BVI people route their own photographs and the inaccessible images they find themselves. It also throws light on the captions and instructions they would use to direct their SP to help guide them with detailed descriptions. We potentially plan to recruit five groups in total that represent varying generations of the BVI participants – teens, adults and seniors – and their familiarity with social media usage and image accessibility.