Session 704: Question 1 by

Hello!

2026-08-01 17:30:33 UTC

Leah from GBH Archives. Public television and radio.

2026-08-01 17:31:08 UTC

How to define accessible?

Item Level?

2026-08-01 17:31:27 UTC

More a question of when/if visual material needs to be described item level. My own bias is that archival description ought to be format-agnostic: you can describe anything in aggregate. But certain formats do present unique challenges...

(Not an expert)

2026-08-01 17:31:38 UTC

Ensuring discoverability for the intended audiences is a first/early step.

Fei from Newcomb Art Museum

2026-08-01 17:32:05 UTC

Having thematic groupings, searchable keywords, and images.

2026-08-01 17:32:47 UTC

Visual materials can differ in that there's a hodgepodge of techniques used to create materials. Can be dependent on other materials to be readable. Software with electronic, but physical can have multiple items to form one object.

Hello from me! A Wayne State University MLIS/Archive student…

2026-08-01 17:33:06 UTC

And former accessibility SME:

https://www.w3.org/WAI/tutorials/images/complex/

First...

2026-08-01 17:33:44 UTC

Absolutely necessary? Depends on allotted space/memory, but subject is far more important with visual than with textual archives. Photographs don't usually announce their function.

And hi! I'm Melanie from the same turf as Max and Betts at Harvard.

Hello!

2026-08-01 17:33:48 UTC

Ashley from MIT Office of Campus Planning! Historical context is very important (important in all archives) but when digitization may not be possible visual materials need more physical description as well.

Accessibility

2026-08-01 17:34:37 UTC

how to support low/no vision users?

Context

2026-08-01 17:35:03 UTC

The balance between giving context within a group of visual materials created for a single purpose vs. individual description is tough!

U.S. Federal guidelines for Section 508

2026-08-01 17:36:07 UTC

In a nutshell…

2026-08-01 17:37:13 UTC

https://tlc.sols.asu.edu/teaching/accessibility/alt-text-vs-long-descriptions-whats-difference

Granularity

2026-08-01 17:37:30 UTC

Visual materials often require a bit more information

OH, Canada! Edition

2026-08-01 17:38:08 UTC

Finding Aids

2026-08-01 17:40:16 UTC

Our Finding Aid is a database. The DB is relational so we can give some context through links: Program/Project-Description of Intellectual Content-Format. We are always dealing with the pesky Many to One and One to Many relationships.

EU Edition

2026-08-01 17:40:33 UTC

Appraisal

2026-08-01 17:40:44 UTC

I don't mean for this to sound cheeky, but: appraisal! Can the repository realistically provide access to visual materials being acquired? Photographic negatives, e.g. This gets into "total cost of stewardship."

Selective Digitization

2026-08-01 17:41:36 UTC

combining visual preview (representative sample) with initial description

Owed to researchers?

2026-08-01 17:43:25 UTC

I agree with Max on format agnosticism, ideally, with the caveat that some pieces of data are much harder to establish in visual materials, like date and creator and maybe subject; so do we owe it to researchers to offer our guesswork, if that's all we've got, to establish keyword searchability?

Jean’s Hot Take from Working in the Industry

2026-08-01 17:43:51 UTC

It’s not enough to say “we’re working on it…” anymore.

Web accessibility laws are in effect in the EU and Canada. The U.S. put off their enforcement of the 508 update until May 2027.

You don’t have to go it alone. Find a SME to write the alt text for you if you have a budget. Use the guidance documents from the legal authorities to direct you if you don’t have a budget. (You can always use the W3 guidance because most of the laws are based on the W3 WAI publications)

Hot take part 2

2026-08-01 17:45:54 UTC

Tip: look at the object you’re describing and write down what you see. Include any inscriptions/text on the object

You don’t have to write a novel.

AI and alt text

2026-08-01 17:47:45 UTC

AI may be a tool for photographic collection batch processing. But then you need a human to verify everything AI spits out for each photo. It will be a project requiring significant time either way.