Meaning composition in minimal phrasal contexts: distinct ERP effects of intensionality and denotation

Abstract

A central question in neurolinguistics is how the brain computes the meaning of complex expressions from the meanings of the parts. We investigate adjective-noun composition using EEG, focusing on the effects of the intensional and denotational semantics of the modifying adjective. We used NPs from 4 semantic conditions in Bokmål Norwegian: privative modal (e.g. "fake president"), privative temporal ("former president"), non-privative modal ("real president"), and non-privative temporal ("current president"). In 2 non-semantic conditions, the adjective was replaced by either a non-word or a pseudo-word. The composition contrast (semantic vs non-semantic trials) revealed a larger P600 component after the noun in the semantic trials. The effects of intensionality (modal vs temporal adjective) were found in the N400 time frame, while the effects of denotation (privative vs nonprivative adjective) were found in a post-N400 window. We discuss some implications of these results for neurocognitive theories of compositional language processing. ARTICLE HISTORY

Composition decomposed: Distinct neural mechanisms support processing of nouns in modification and predication contexts.

“…Importantly, it could be the availability of a rich sentential or clausal context, and not adjectival modification per se, which allows semantic processing to by-pass syntax, if that is indeed the case. Composing noun meaning in minimal phrasal contexts may result in different ERP effects than modulations of the N400 component, especially if the noun cannot be predicted, that is, if its associated lexical semantic features cannot be contextually preactivated (Baggio, 2018; Fritz & Baggio, 2020).…”

“…Nonetheless, syntactic composition may still apply to pseudowords (but not to nonwords), if they exhibit phonological or morphological features that render them recognizable as possible words: possible members of a particular grammatical category. Nonwords and pseudowords are therefore best used “in tandem” to identify possibly distinct effects of syntactic and semantic composition (Fritz & Baggio, 2020). Still, it would be beneficial to design more language-like, higher-level control stimuli than nonwords or pseudowords.…”

Generalizing Knowledge of Second Language Collocations: The Roles of Within‐ and Cross‐Language Similarity on Acceptability and Event‐Related Potentials

“…Further, studies that investigated analogical processing (e.g., in metaphor comprehension) have interpreted the LPC (or the ERP patterns observed within the LPC time window) as indexing semantic integration through relational thinking (e.g., Jankowiak et al, 2021;Lai & Curran, 2013). On the basis of these findings, it has been suggested that the attested effects during the LPC time window reflect semantic integration (Fritz & Baggio, 2020;Molinaro et al, 2012), typically in contexts where association with known representations facilitates processing of novel input. As a counterpart to the LPC being associated with successful integration, in some studies, a sustained negativity within the LPC time window has been associated with ongoing difficulty during meaning integration (Tang et al, 2017).…”

Compositionality in a Parallel Architecture for Language Processing

“…The order of these effects, where the LAN precedes the P600, and the N400 precedes the SAN, is compatible with the two‐phase, cascading nature of processing within each stream. Research on the P600 can help answering questions on compositionality (Fritz & Baggio, 2020)—whether a particular expression, given the presence or absence of context, and for a specific experimental setting or task, is processed primarily by the G‐stream. Through systematic study, we may hope to get to grips with the syntactic and semantic structures and operations most likely used by each stream.…”