Scientists Offers New Way to Read Online Text

Let's face it - the only reason anyone ever reads Scientific American is to see how "shiny" the future is going to be.

Scientists Offers New Way to Read Online Text

Postby TerraFrost » Fri May 11, 2007 5:40 pm

To quote from slashdot.org:

"Scientists at a small startup called Walker Reading Technologies in Minnesota have determined that the human brain is not wired properly to read block text. They have found that our eyes view text as if they're peering through a straw. Not only does your brain see the text on the line you're reading, but it's also uploading superfluous information from the two lines above and the two lines below. This causes your brain to engage in a tug of war as it fights to filter and ignore the noise. The result is slower reading speeds and decreased comprehension. The company has developed a product that automatically re-formats text in a way that your brain can more easily comprehend."


As one slashdotter observes, their FAQ isn't in this format.

Also, one of the examples they give is kinda biased, as another post explains:

http://venturebeat.com/wp-content/uploads/2007/05/ beforeafter1.jpg

I noticed several things that make it difficult for me to actually evaluate the difference. First each uses a different font, then the one that is supposed to be inferior ends with an incomplete sentance "A cell is" - making it gramatically inferior, if you zoom in you'll notice that the inferior sample didn't compress well in the jpg, the fonts are different sizes, and finally live link labeling the new sample as "Section 1:" provides more contextual information making it in fact more informative. While these changes are subtle each by themselves they are all time tested methods for improving text. Don't blur the text, add contextual info, complete your sentances and use standardized grammar. If this is the standard output from their software then this is truly not impressive. Aside from these issues, haven't people used collumns for a long time too?


Here are two other comments that I thought were interesting:

I think you're missing the part about how the positioning of the words are determined. The algorithms used were inspired by spoken syntax: "The prosodic cues in spoken language are more complex than simple pauses at phrase boundaries; subtle variations in pitch, volume, and the duration of word pronunciation have been shown to convey hierarchical structures in syntax (Ferreira & Anes, 1994). When these prosodic-syntactic cues of speech are experimentally stripped away from audiorecordings of sentences, listeners' comprehension drops (Cutler, Dahan, & van Donselaar, 1997). This finding has important implications for reading because, when language is written down, many of these same syntactic cues are similarly stripped away" Also, according to the supporting paper, parsing sentences along these lines help support the goals of the semantic web, helping online readers to parse complex expository writing.

http://science.slashdot.org/comments.pl ... d=19083253

The algorithms used were inspired by spoken syntax:

Which may not be all that relevant to the comprehension of written language.

One aspect the linked article emphasizes is that spoken language is ephemeral, whereas written language is permanent. This is a large difference, as anyone who can read a second language with relative fluency but understand the spoken form hardly at all knows.

For this and many other reasons (no one speaks like a textbook or scientific paper for a reason--writing is far more effective at conveying certain types of information) it is problematic to claim without proof that "making writing more like speech is a good thing." In some cases it is probably true. In lots of other cases it may well be false. It will depend on the nature of the information being conveyed.


http://science.slashdot.org/comments.pl ... d=19083911
TerraFrost
Legendary Guard
 
Posts: 12357
Joined: Wed Dec 04, 2002 6:37 am

Return to All Things Shiny

Who is online

Users browsing this forum: No registered users and 1 guest

cron