From: Christopher Potts
Date: July 9, 2011 08:22:39 MDT
To: Class list
Subject: Computational Pragmatics: quick summary of day 1 + some notes

Computational Pragmaticists!

I think we're off to a good start!  Since class enrollments are still in flux, I'm writing to make sure everyone is up to speed on what we did on day 1.  I also have a few updates stemming from the class discussion.

* We quickly reviewed the overview page (http://compprag.christopherpotts.net/overview.html). The consensus seemed to be that the requirements are reasonable, and no one had any horror stories about failed installs...

* Most of the meeting was devoted to background on conversational implicature (http://compprag.christopherpotts.net/implicature.html). We worked through the examples pretty carefully.  I think it was starting to emerge that Joshian Quality is a useful replacement for Grice's Quality, though its emphasis on the speaker modeling the hearer might be implausible for some aspects of communication.

* I had Hirschberg's definition right; I was just misreading clause 3. I've added page numbers to make it easier to track down the original statements.

* In passing, we discussed embedded implicatures. These mostly come up in the context of addressing whether some (all) implicatures are calculated as part of the compositional semantics. I added a paragraph about this issue to the implicatures page, with references and meta-references:

http://compprag.christopherpotts.net/implicature.html#presumptive

* We also briefly discussed the pragmatics of uncooperative dialogue.  I added a reference to Asher and Lascardes's (2008) paper about exactly this issue, but they seem not to be distributing the manuscript itself.  I was a commentator on the paper at a Michigan workshop, though, so you can get a sense for what they said by looking at my short unpublished commentary: http://www.stanford.edu/~cgpotts/commentaries/potts-umich08-cmts-on-asher-lascarides.pdf. (This also happens to be the event that got me started on IQAPs.)

* We also started to review the new indirect question-answer pairs corpus that will be the basis for our first set of experiments (http://compprag.christopherpotts.net/iqap.html). This is where we'll pick up next time.  I want to be sure we all have a feel for what the data and annotations are like before we go about trying to predict implicatures (and contextual entailments). This is also a good chance to learn some R.

* We agreed that both the weekly assignments and the final project can be group work.  For the weekly assignments, everyone should submit their own write-up (email to this address), with a list of the people they worked with on it. For the final projects, there can be one write-up (code base, experimental data-set, etc.), but the project should include a quick summary of who did what from the research team.

* My office hours are Thursdays and Fridays, 10:30-11:30, and I am available for meetings with people who can't make it then. My office is University Club 216.

* For restricted links at the website:

username: [contact instructor]
password: [contact instructor]

---Chris