The contribution of semantics to the sentence superiority effect

Stéphanie Massol,Jonathan Grainger,Jonathan Mirault

doi:10.1038/s41598-021-99565-6

Abstract

When a sequence of written words is presented briefly and participants are asked to report the identity of one of the words, identification accuracy is higher when the words form a correct sentence. Here we examined the extent to which this sentence superiority effect can be modulated by semantic content. The central hypothesis guiding this study is that the sentence superiority effect is primarily a syntactic effect. We therefore predicted little or no modulation of the effect by semantics. The influence of semantic content was measured by comparing the sentence superiority effect obtained with semantically regular sentences (e.g., son amie danse bien [her friend dances well]) and semantically anomalous but syntactically correct sentences (e.g., votre sac boit gros [your bag drinks big]), with effects being measured against ungrammatical scrambled versions of the same words in both cases. We found sentence superiority effects with both types of sentences, and a significant interaction, such that the effects were greater with semantically regular sentences compared with semantically anomalous sentences. We conclude that sentence-level semantic information can constrain word identities under parallel word processing, albeit with less impact than that exerted by syntax.

Highlights

When a sequence of written words is presented briefly and participants are asked to report the identity of one of the words, identification accuracy is higher when the words form a correct sentence
The present study was designed to measure the contribution of semantics to the sentence superiority effect
The sentence superiority effect is the observed superior identification of a post-cued word in a briefly presented sequence of words when the sequence is syntactically correct compared with ungrammatical scrambled sequences of the same words. This finding provides support for a cascaded, interactive account sentence comprehension according to which partial word identification processes operating in parallel enable the rapid activation of a sentence-level representation that provides feedback to on-going word identification processes[9,11]

Summary

Introduction

When a sequence of written words is presented briefly and participants are asked to report the identity of one of the words, identification accuracy is higher when the words form a correct sentence. Following this report of a sentence superiority effect using the RPVP procedure it has been shown that: (1) the effect is obtained in primary school children in Grade 3 (average age 8.9 years)[8], and (2) that the effect emerges around 300 ms post-sequence onset in EEG recordings[9] The latter finding points to a key role for interactions between sentence-level structures and on-going word identification processes in driving the sentence superiority effect. Declerck et al.[10] found a robust sentence superiority effect in these conditions and concluded that the effects were driven by the parallel processing of word identities and the association of these word identities with language-independent syntactic categories (e.g., noun, verb) This was hypothesized to enable the rapid construction of a primitive language-independent syntactic structure that would provide feedback to on-going word identification processes

Objectives

Methods

Results

Conclusion