No filled pauses at MT3 at Keio

July brought the rainy season to Japan, as usual, but also brought the time for an annual covey of conferences on psycholingustics and closely related fields in Tokyo. Historically, there were several different seminars and workshops held around this time in past years but the various organizers have decided to combine efforts (good idea!) and hold one larger event each summer, now dubbed MT3 (constructed from the acronyms of the former events: MAPLL x TCP x TL x TaLK). This year it was held at Keio University in downtown Tokyo.

There were no filled pauses at the event. Well, OK. there were filled pauses—just not as an object of study. But of course, there were several talks that were very interesting to me. I was quite taken with one talk by Pimrat Fongchamnan about the processing of irony by nonnative speakers of English. One factoid that came out during the talk was the claim that there are 4 occurrences of irony per half hour in US TV shows. I couldn't help wondering if this sounds way too low. Of course, there would be some shows that have very little irony in them, but there are so many comedy shows on TV in which irony is the staple. I would have expected them to bring the average way up. Also, I would expect that the irony average in British TV shows would be even higher than that in the US.

Anyway, the model of irony processing used was based on a notion of gradient salience (a favorite topic of mine since my doctoral dissertation). The idea is that successful irony depends on their being a reasonably salient alternative interpretation. This is a neat idea and something I hadn't quite thought about yet. Definitely worth further work. And I would wonder if this could interact with the use of filled pauses. That is, a filled pause could perhaps make an otherwise less salient interpretation more salient (a la Corley et al 2007).

Another talk of interest was by Douglas Roland—who, by the way, as of this year, is now my colleague at Waseda. He talked about a series of studies he's been doing that looks at subject versus object relative clauses. This is a classic issue that has been much studied in psycholinguistics, but he's working to bring a new perspective to it methodologically. Various methodologies have tended to yield somewhat different results on the question. He's working to try to bring these together. While that work is very interesting, one thing that really stood out to me in his talk was explanation of the various ways of measuring eye-tracking data in terms of first-pass and second-pass eye movements. His explanation as well as the slides to go with it were very clear. I have never understood eye-movement measurements so well!

As for me, my talk had nothing to do with filled pauses. Sorry! I know this is the Filled Pause Research Center, but I'll still go ahead and explain my talk. It may be relevant to what I do with filled pauses later. I've been working on a Javascript project for some time which is designed to enable carrying out self-paced reading experiments in a web browser. In case you don't know, self-paced reading (Just, Carpenter, and Woolley, 1982) is a common experimental task in many psycholinguistic studies in which experimental participants press a button to reveal portions of a text little by little. In this sense, it is self-paced: they control when the text appears. This is a useful technique, then, to locate points at which a reader has difficulty processing a text: when their reading speed slows down, it is presumed that they are having some processing difficulty. This is useful for testing a variety of hypotheses and has proven a useful technique for several decades. My interest was in making it possible to do these experiments online through a web browser. So, I created JESPR (Javascript-Enabled Self-Paced Reading), a Javascript library for managing a self-paced reading experiment.

The experimenter defines an experiment in a json file and decides various parameters for the presentation of text. JESPR then displays the text as the participant presses buttons on the keyboard (or on-screen buttons for touch-screens) and keeps track of the intervals between button presses. The experimenter can then use the interval data for analysis.

Those who are interested in JESPR can find out more information about it at the JESPR site on github. In fact, I have already used it to test one hypothesis regarding filled pauses. But I'll save that for a later post.

[Note: This post was written in September, 2020. However, in order to preserve the chronology of the blog, it has been dated to reflect when the described events actually took place.]