productive day so far

woo! i redid these parses to make sure my result was right, and it is. so here's the basic idea. laboratory results have suggested that people have an imaginary voice in their head when they're engaged in silent reading, and that this imaginary voice provides information like intonation and stress to help people get the right structure for the sentences they are reading. So in a laboratory setting, we know that people seem to be sensitive to these imaginary voices (called ``implicit prosody''). But we haven't seen if people produce the right kinds of implicit prosody to actually help much when reading sentences that have not been invented in a laboratory. I have a data set of german sentences that have been basically annotated with implicit prosody (although that's not how the people created the data set looked at it), and found that you actually can get a reasonable increase by using the implicit prosody. w00t!

but there's a lot more to be done.... good thing my MA project actually works...... hee!

-=-raptur-=-