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

Poster D6, Thursday, November 9, 6:15 – 7:30 pm, Harborview and Loch Raven Ballrooms

In search of syntax: The case of English post-nominal modification

Graham Flick1, Liina Pylkkänen1,2;1New York University Abu Dhabi, 2New York University

A principal challenge for the study of the neural underpinnings of syntax is deconfounding it from semantic composition. Consequently, attempts to isolate syntax typically involve stimuli that are either not representative of natural language (e.g., violation or jabberwocky stimuli) or working memory intensive (making it hard to separate syntactic effects from effects of memory). The perfect test of syntax would involve well-formed lexically identical instances of natural language, which nevertheless differ in syntactic complexity, without semantic consequences. While such cases may be impossible to find in pure form, we developed a study around post-nominal modification in English, which allowed us to come close to meeting these requirements. While the canonical position of adjectival modifiers in English is pre-nominal (wide trail), certain classes of “heavy” modifiers are licensed post-nominally (trail wide enough to allow…). Although the semantic impact of pre- vs. post-nominal modification is very similar, post-nominal modification is hypothesized to be more structurally complex (e.g., Sadler & Arnold, 1994). We compared post-nominal modification to string-identical sequences which instead of post-nominal modification, exemplified predication structures as in "this trail is wide". To make the noun and adjective adjacent, a question form of such expressions was used, resulting in the contrast (a) vs. (b): a. Post-nominal: "There are many trails wide enough to allow…" b. Predication: "Are many trails wide enough to allow…". To test whether any observed differences may be due to the declarative vs. interrogative nature of the expressions, a parallel control experiment was run with pre-nominal modification embedded inside similar questions and statements. Finally, to compare effects of syntactic and semantic composition, we also varied the semantic match between the types of the adjectives and nouns, such that in some cases semantic composition required a coercion operation (e.g., interpreting "difficult trail" requires the insertion of some activity as in difficult TO-HIKE trail whereas an event nominal can compose with difficult directly: "difficult trek"). In all, both the post-nominal main experiment and the pre-nominal control experiment fully crossed the factors Syntactic Frame (statement vs. question), Noun Type (entity vs. event) and Adjective type (entity vs. event). The second word of the adjective-noun or noun-adjective sequences was always the target of MEG analysis. Results: Analysis of the post-nominal main experiment revealed a main effect of Syntactic Frame in the posterior superior temporal lobe, between 205 and 225 ms, while no parallel effect was seen in the control experiment, suggesting the effect indeed tracked the presence of post-nominal modification. While we did not identify any correlate of the hypothesized semantic coercion operation, a semantically modulated pattern of activity was identified in the left anterior temporal cortex, with largest amplitudes for the semantically simplest combinations of adjectives and nouns. This conforms to prior findings implicating the LATL in “simple” cases of conceptual combination (Ziegler & Pylkkänen, 2016). In sum, this study isolates activity in the posterior temporal cortex as a potential correlate of syntactic composition, and offers further evidence for the LATL as a locus of “quick and easy” conceptual combination.

Topic Area: Grammar: Syntax

Back to Poster Schedule