Considering all the stars visible: MEG correlates of English post-nominal modification
Nigel Flower1, Soo-Hwan Lee1, Liina Pylkkänen1,2; 1New York University, 2NYUAD Institute, New York University Abu Dhabi
While adjectival modification is canonically pre-nominal in English, many adjectives describing temporary “stage-level” properties, such as “visible” or “available,” can also occur post-nominally. Since the pre- and post-nominal instances of these adjectives create meanings that are very close to each other, they offer an opportunity to examine neural correlates of syntax. Here we embedded pre- and post-nominal modifications into full sentences, in three different positions within the subjects of the sentences (for generality) as shown below, with the adjectives and nouns serving as critical stimuli for magnetoencephalography (MEG) measurements. The stimuli were presented visually in RSVP. Pre-Early: All visible stars have … Post-Early: All stars visible have … Pre-Middle: All the visible stars have … Post-Middle: All the stars visible have … Pre-Late: All of the visible stars have … Post-Late: All of the stars visible have … While the nouns in both positions are “canonical” (as encountered by an incremental parser), the post-nominal adjective is not. The main question for our 2(pre/post) x 2(noun/adjective) x 3(early/mid/late) ANOVA analysis was to identify neural differences between the pre- and post-nominal adjectives that are absent for the comparison between the nouns in first vs. second position. We conducted both an ROI analysis of regions hypothesized to participate in sentence level combinatory processing (left BA38, BA21, BA20, BA44+45, BA39, BA11 + right hemisphere homologues) as well as a more exploratory full brain analysis. Results from 9 participants (so far) show increased activation for the post-nominal adjectives as compared to their pre-nominal counterparts in the left middle temporal gyrus, inferior temporal gyrus and the angular gyrus, potentially reflecting the higher syntactic integration cost of the post-nominal adjectives. Nouns in pre-adjectival vs. post-adjectival positions did not show any parallel effects. Also, no parallel effects were obtained in the right hemisphere, suggesting strict left laterality. The timing of all activity increases was at ~300-400ms. The left anterior temporal lobe, implicated for conceptual aspects for composition (Pylkkänen, 2019), showed no effects, as predicted by the fact that conceptual combination did not vary in our design. The explorative spatio-temporal clustering analysis showed that our left lateral increase for post-nominal adjectives was clearly centered in posterior middle temporal cortex, consistent with accounts positing this area as a locus of syntactic composition (e.g., Matchin & Hickok, 2020; Flick & Pylkkänen 2020). This work contributes to our understanding of the neural indices of syntax in the absence of any (major) semantic confounds, which is a difficult experimental contrast to create. Our preliminary results suggest a rather strong left-lateral effect of our manipulation, showing that neural activity can be affected by relatively pure manipulations of just structure.
Topic Areas: Syntax, Meaning: Combinatorial Semantics