Authors: Matthew Lowder, Adrian Zhou and Peter Gordon
Contact Email(s): matt.lowder@gmail.com, adrian.zhou@richmond.edu, pcg@email.unc.edu
5 thoughts on “The laboratory discovered: Place-for-institution metonyms appearing in subject position are processed as agents”
Leave a Comment or Question Below
You must be logged in to post a comment.
I wonder, you presented your corpus study, but later don’t seem to really use it to further analyze your results. I imagine, indirect vs. direct access might have to do with frequency of use in the literal vs. metonymic sense. Do you plan to perform an analysis on split groups of your stimuli (metonyms of frequent vs. infrequent literal use) or something to this effect?
Hi Ines. Thanks for the question. The corpus data I presented was just collected over the past couple weeks, and we definitely plan to do more with it. Yes, I do think it’s possible that we might find different patterns of effects depending on whether the metonym is more frequently used in its literal or figurative sense. Stay tuned!
This is similar to some of the questions which came up live, but I’m curious if you could say more about your interpretation of these results re: models of access. I’m thinking in particular of the Bott et al. (2016) response to Lowder & Gordon (2013), which argues that the signatures of difficulty found for figurative readings are the result of something other than indirect access (low plausibility, or particularly errorful retrieval for figurative meanings). Do you take any of the results from this talk as useful for resolving that debate?
Hi Jack. Thanks for the question. Yes, I’m familiar with the Bott et al. paper, but I’m not sure how far it gets us toward resolving questions about models of figurative language processing. Specifically, their speed-accuracy-tradeoff data don’t reveal any differences between literal and metonymic expressions until about 1,000 ms, which is much later than the reading-time data we presented in Lowder & Gordon (2013), or that I presented in the talk. In addition, the Bott et al. paper concludes that metonymic sentences are associated with lower retrieval probabilities than literal sentences. However, they also report that participants found the metonymic sentences *more* plausible and *more* sensible than the literal sentences. It’s difficult to reconcile those ratings with the conclusions from the SAT data. In sum, different methods are yielding different results, but I think the eyetracking data that we have presented strongly support the idea that familiar metonyms can be difficult to process if they are in a focused syntactic position, but this difficulty can go away if they are presented in a defocused position.
Thanks for the very clear response! I see. The cross-methodology oddities do seem to be such a constant issue in this literature. I’m presenting a short talk tonight that adds one to the pile, finding what looks like immediate commitment for polysemy/metonymy in the Maze, and I’ve yet to really grasp whether our explanation there, which hinges on properties of the Maze, helps wrestle with why we see these differences in all these other tasks as well. It’s a difficult picture to apprehend all at once.