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

Poster C10, Thursday, November 9, 10:00 – 11:15 am, Harborview and Loch Raven Ballrooms

The (non-)satiation of P600/SPS effects to distinct grammatical violations

Emma Nguyen1, Jon Sprouse1;1University of Connecticut

The P600/SPS ERP effects that arise to syntactic violations are widely assumed to index the re- analysis processes that the parser deploys to correct those violations. Our goal in this project is to expand the work in the literature that has sought to differentiate subtypes of P600/SPS effects (e.g., Gouvea et al. 2010) in the hopes of constructing a mapping between P600/SPS effects and theories of reanalysis (which could then help unite online measures of reanalysis with offline measures such as acceptability judgments, which are undoubtedly influenced by reanalysis). In this experiment we primarily intended to study the satiation of P600/SPSs. Hahne & Friederici (1999) found that P600/SPSs disappear when the composition of the experiment is 80% violations and 20% gram- matical sentences. Crucially, Hahne & Friederici used only one violation type in their experiment (a phrase structure violation similar to (1) below). This suggests that reanalysis processes satiate, but raises the question of whether distinct violations can cause satiation in each others reanalysis processes. To test this, we created an experiment containing multiple violation types that add up to 80% throughout the experiment, but with no single violation accounting for more than 30% of the items (310 sentences total, with violations roughly evenly spaced). We focused on three distinct violation types, each taken from the P600/SPS literature (Neville et al. 1991, Newman et al. 2007, and Kim & Osterhout 2005, respectively): (1) Phrase structure violation (PSV): *The boys enjoyed Eds about stories the battle. (2) Agreement violation (AGR): *The agents discovers Freds tobacco from Cuba. (3) Semantic P600s (SEM): *The hearty meal was devouring by the kids. We also included 30 case violations and 30 island constraint violations to increase the mix of violation types. For PSVs, we used a pre/post design: 30 violations early in the experiment, and 30 violations late in the experiment, with no PSV violations in between. This allowed us to look for explicit satiation of PSVs by other violations. For AGR and SEM, the 30 violations of each were distributed throughout the middle and end of the experiment, which only allowed us to ask whether the P600/SPSs were present/absent. We used the same experimental parameters as Kim & Osterhout (2005) to ensure that the semantic P600 was potentially present. Crucially, we found P600/SPS-like effects for all three violation types, suggesting that distinct violations cannot satiate each others P600/SPSs. This corroborates the idea that distinct violations trigger distinct reanalysis processes. Furthermore, given the presence of all three P600s, we are able to do a within-participants comparison of their latency and scalp distribution (expanding the findings of Gouvea et al. 2010). We found that PSV has an earlier and longer P600/SPS than the other two; AGR includes a LAN in the 800-1000ms window; and the SEM P600 has a much broader scalp distribution than the other two. In short, these results suggest that the P600/SPS effects to PSV, AGR, and SEM are distinct in all possible dimensions: satiation, latency, and scalp distribution.

Topic Area: Grammar: Syntax

Back to Poster Schedule