Showing posts with label legibility. Show all posts
Showing posts with label legibility. Show all posts

Sunday, October 14, 2007

It's Not Illegible, It's Asemic

Those marks in the margins of your ancestor's diary may look like scribbles, but they might actually be asemic writing.

It looks like writing, but we can't quite read it.

I call works like this "asemic writing".

Asemic writing seems to be a gigantic, unexplored territory.

Asemic writing has been made by poets, writers, painters, calligraphers, children, and scribblers, all around the world. Most people make asemic writing at some time, possibly when testing a new pen.
If doodles count as asemic writing, here's a fine example from census taker George W. Rand, who left this work of art on a page of the Waterford, Maine, census in 1860:

Fortunately, not all of George's writing lacked semantic content.

Saturday, September 29, 2007

The End of Illegibility?

CMU professor Yang Cai has developed software that can decipher illegible inscriptions on ancient tombstones. He's testing it on stones at Old St. Luke's Church in Carnegie, Pennsylvania.

During the past two weeks, Cai's research team trekked through the church's three-acre cemetery, scanning unreadable gravestones and then storing the images on laptops.

"We are exploring new 3-D reconstruction technology to decipher the gravestone names," said Cai. "Essentially, we reconstruct the tombstone surfaces by applying filtering and detection algorithms for revealing the words on the archaic surfaces," he said.

In addition to discovering who is buried in the church cemetery, Cai is developing a digital cemetery for Old St. Luke's Church. [Link]

Tuesday, February 06, 2007

Decipher Yourself

If you'd like to test your transcribing skills, head over to the Decipher mailing list archives. Almost all of the messages are from KYGenWeb volunteers working on the Kentucky Vital Records Project, which has digitized and indexed more than 126,000 death certificates so far.

I've transcribed tens of thousands of handwritten names over the years, but I still get stumped. The trick is knowing when you should consider yourself stumped, and knowing how to find a "backdoor" solution. I'm glad to see that some respondents on the Decipher list are going beyond the digitized document to confirm their hunches with census and other records.

I'll admit, it's also fun to read the conflicting answers they sometimes give.

Looks like Ase

Joe or Ace

I think it looks like Gee, possibly short for George.

I think it's Lee with an "overly fancy" L.

« Newer Posts       Older Posts »