3 thoughts on “Language Production Under Uncertainty: Advance Planning and Incrementality

  1. Thanks for the interesting talk, Arella and Maryellen! Do you think that the effect you reported is going to be observed only when one of the names is guaranteed to be useful? Say you have three object arrays instead of two, and if two of the three arrays share the same object, would you expect the same (but weaker) effect or no effect at all?

    1. Hi, Shota, this is a great question. Obviously the “guaranteed useful” setup was a good place to start, but I think that it more probabilistic conditions could also yield the effects we observed, though possibly weaker. We have thought about more complex displays to test some other ideas. One limitation is that when the displays get too complex, people don’t necessarily look at everything in the display, and the data gets noisier.

      1. Thanks for your reply! Yeah, there are clearly practical challenges associated with more complex questions. I wonder if stochastically changing some objects in the display in filler trials might be a way to manipulate speakers’ certainty without making the stimuli more complex (like, if the preview pictures are not entirely reliable, speakers would have less certainty about their message before the production prompt even when the preview pictures are not entirely uninformative).

        In any case, I think the theoretical questions behind this question and the question I asked after your talk is something like: (a) do you think the relationship between speakers’ certainty and the probability of advance planning is characterized by a linear function or more like a step function and (b) do you think speakers can do something similar to what you saw beyond word representations (like structure-level representation). (b) might be related to the question of whether the effect you observed is a reflection of conscious strategic processes. If two pictures on a display share some abstract properties (say, thematic structures or transitivity), I can imagine that speakers may not be consciously aware of the parallelism and yet still plan the transitive structures when they can be certain about them!

Leave a Comment or Question Below