Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Manipulating syntax without taxing working memory. MEG correlates of syntactic dependencies in Danish
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster E41 in Poster Session E, Thursday, October 26, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Simone Krogh1, Liina Pylkkänen1; 1New York University
Querying the neural indices of syntax is difficult since syntactic manipulations often tax working memory and introduce semantic confounds. Here we tackle both challenges in a study of Danish syntactic dependencies. First, we utilize two-word Danish yes/no questions, formed from declarative Subject-Verb sentences with a word order swap (Verb-Subject) while keeping the lexical material constant. We also vary argument structure by including verbs hypothesized to trigger argument-movement (unaccusatives and alternating unaccusatives; Perlmutter, 1978), and ones that do not (unergatives). Second, we utilize Rapid Parallel Visual Presentation (RPVP; Snell & Grainger, 2017) as our stimulus delivery method. In this approach, a full sentence is presented at once, for just a few hundred milliseconds. RPVP served to eliminate working memory-related costs and additionally allowed us to investigate the neural bases of the so-called Sentence Superiority Effect (SSE) observed in prior RPVP literature. This effect refers to the advantage in rapid processing observed for structured vis-à-vis unstructured representations. We showed our two-word sentences and control two-verb lists for 300ms in a simple matching task during a magnetoencephalography (MEG) recording (n = 29) as well as in a behavioral-only version (n = 30). Our regions of interest were the left inferior frontal gyrus (LIFG), which has been implicated for long-distance dependencies since Stromswold et al. (1996), and the left posterior temporal lobe (LPTL), an emerging candidate for syntactic processing (Matchin & Hickok, 2020; Flick & Pylkkänen, 2020). The left anterior temporal lobe (LATL; Bemis & Pylkkänen, 2011) and inferior parts of the parietal lobe (IPL; Williams et al., 2017) were also examined given their implication in syntax and/or semantics. Spatio-temporal clustering analyses tested two time windows for all search areas and their right hemisphere homologues: An early time window (100-500ms) targeting the N400-like SSE reported by Wen et al. (2019) and a later window (500-800ms) motivated by the P600 literature on syntactic processing (Kaan et al., 2000; Gouveau et al., 2010). The behavioral results (n = 59) revealed a clear instance of the SSE—consistently faster responses for sentences than lists—and an effect of syntactic frame with faster responses to declaratives than interrogatives. Neurally, we find bilateral SSEs in the IPLs, localizing first in the left (250-290ms) and later in the right hemisphere (420-455ms), as well as a marginal effect in the LATL (435-460ms), broadly conforming with the previously reported N400 effect. A 2(syntactic frame) x 3(argument structure) ANOVA for the sentence stimuli revealed an effect of syntactic frame in the left IPL (510-590ms), with declaratives eliciting more activity than interrogatives, and an effect of argument structure in the right IFG (580-625ms), with unaccusatives eliciting the most activity. To summarize, our working-memory-free paradigm revealed a neural processing cost exclusively for argument movement, while verb-movement in yes/no questions resulted in a decrease in neural activity. Notably, neither effect localized in traditional syntactic hubs like the LIFG or the LPTL. This suggests that the neural processing of displacement may differ significantly from conventional perspectives when dissociated from the confounds typically associated with serial stimulus presentation.
Topic Areas: Syntax and Combinatorial Semantics,