You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

 
Poster C44, Thursday, November 9, 10:00 – 11:15 am, Harborview and Loch Raven Ballrooms

ERPs reveal listeners’ sensitivity to discourse history in comprehension

Si On Yoon1, Kara D. Federmeier1;1University of Illinois, Urbana-Champaign

Introduction: When designing referring expressions, speakers typically take into account not only the local context but also the discourse history. For example, speakers produce lexical differentiation: they elaborate their referring expressions with modifiers or use subordinate level nouns (e.g., “the striped shirt” or “the blouse”) if a different exemplar from the same category (“the shirt”) has previously been mentioned in the discourse. Surprisingly, although speakers have shown this sensitivity in prior work, listeners have not. In Yoon and Brown-Schmidt (2014), when participants in a referential communication task encountered expressions describing novel exemplars from a previously experienced category, their eye movement patterns showed no benefit for interpretation of appropriately lexically differentiated expressions. Here, we further examined listeners’ sensitivity to discourse history by testing appropriately differentiated or under/over-differentiated expressions while recording participants’ electrical brain activity. Method and Results: 32 participants performed two tasks alternating in a block. In the first task, participants looked at a series of pictures on a computer screen and listened to picture descriptions. On each trial, one picture was presented in the center of the screen followed 2 seconds later by a pre-recorded audio description. We manipulated the degree of differentiation with respect to the historical context. On critical trials, participants saw either the same or a different exemplar of a particular item type (e.g., dog1 vs. dog2) followed by either the same or a different pre-recorded description (e.g., “dog” vs. “poodle”) compared to the earlier trial. This created four conditions (critical trials are in | |): a. differentiation (pictures: dog1–|dog2|; audio: “dog”–|“poodle”|), b. over-differentiation (pictures: dog2–|dog2|; audio: “dog”–|“poodle”|), c. under-differentiation (pictures: dog1–|dog2|; audio: “poodle”–|“poodle”|), d. repetition (pictures: dog2–|dog2|; audio: “poodle”–|“poodle”|). Pairs of synonyms (e.g., couch-sofa) and basic – subordinate level nouns were used in the pre-recorded descriptions. In the second task, participants were shown 4 pictures on the screen and asked to identify the picture that the audio described. We recorded the continuous electroencephalogram and extracted event-related potentials (ERPs) from critical trials during the first task (from 200ms prior to audio onset to 800ms after). We found electrophysiological effects associated with both types of inappropriate differentiation. Under-differentiation – i.e., having the same expression referring to two different objects – elicited an increase in the P2 compared to appropriate differentiation. In contrast, when the expression was over-differentiated, we saw an increase in the N1 and in the later LPC (Late Positive Complex) compared to simple repetition. Conclusions: Listeners’ brain responses revealed clear sensitivity to historical discourse context. Both under- and over-differentiation were associated with (different) effects on sensory potentials, suggesting modulations of attention associated with infelicitous repetitions (under-differentiation) or with unexpected novel referents (over-differentiation). In the latter case, in addition to attentional capture from the unexpected novel input, enhanced LPC responses point to the possibility that listeners attempt to recollect the prior referential expression. Thus, listeners clearly track the use of referential expressions across a discourse in communicative tasks, even if this does not always manifest in behavioral output (e.g., eye movement patterns).

Topic Area: Meaning: Discourse and Pragmatics

Back to Poster Schedule