Wednesday, July 27, 2005

Rapid Serial Visual Presentation

There's been some recent hype about RSVP, or Rapid Serial Visual Presentation, a technique introduced by K.L.Forster in 1970 (see the introduction here for a survey) with a bunch of Stanford researchers trying to make it a commercial product (BuddyBuzz) for delivering e-books to cellphones.

Mark Russell, Marilyn James, and Andrea Cohlmia, ran an experiment with twenty undergraduate students at Wichita State in 2001. [Link]. Regular reading was compared with RSVP at three different speeds - 250, 450 and 650 words per minute. Unsurprisingly, students preferred the regular method. However, reading comprehension of 250wpm-RSVP was almost the same as that of regular reading. Comprehension dropped with faster speeds, but that is unsurprising given the lack of experience of the subjects.

Experienced users claim to be able to handle 700 wpm.

If you want to play with this, try using GnomeRSVP or some other RSVP reader listed here by Matt Vance of MineZone.

Here's a horrifying thought - unless your are the budget director at the New York Times - you are surfing the web in 2008. All advertising banners now use RSVP. There are several banners on each page. People start paying online newspapers so they don't have to read banners.

Tuesday, July 19, 2005

Talking like Dead People

Historical linguists and language evolution researchers have been working on how languages sounded like a long time ago. One method, pioneered by William Labov in the 1960s, is to just listen to old people talking, since people tend to keep pronounciations learnt in childhood.

Of course, that only works for languages several decades ago. If you want to know how a language sounded like centuries ago, a surprising source of information is songs where you know what word has to rhyme with what. Of course, just because you know a bunch of words sound the same doesn't mean you know what they sounded like. That requires additional information, such as pronunciations from slow-changing dialects (e.g. of more isolated communities).

David Crystal has been teaching actors at the Globe how to pronounce Shakespearean English. The BBC has a good report on this:

They say their accents are somewhere between Australian, Cornish, Irish and Scottish, with a dash of Yorkshire - yet bizarrely, completely intelligible if you happen to come from North Carolina.

Somehow, North Carolina doesn't sound like an isolated linguistic community. That must be a coincidence.