Following another link from fieldmethods.net, I stumbled on Kevin Knight's homepage at ISI (at USC).
I was reminded that my new funding will be on a multi-lab project involving ISI and SSLI at the University of Washington, so I decided to read some of his papers. I am seriously enjoying his paper surveying Machine Translation, partly because of phrases like this one:
[a research system of Knight's] once translated a Japanese sentence as "There is a plan that a baby is happening in her", a pretty reasonable translation, but with a definite Japanese-semantics feel to it. Semantics is not an all-or-nothing proposition in MT any more than in, say, expert systems. [p. 2]
Hee. Thanks, Pat-from-fieldmethods!