Dissociable contributions of frontal and temporal brain regions to basic semantic composition

Abstract

Semantic composition is the ability to combine single words to form complex meanings and is an essential component for successful communication. Evidence from neuroimaging studies suggests that semantic composition engages a widely distributed left-hemispheric network, including the anterior temporal lobe, the inferior frontal gyrus and the angular gyrus. To date, the functional relevance of these regions remains unclear. Here, we investigate the impact of lesions to key regions in the semantic network on basic semantic composition. We conducted a multivariate lesion-behaviour mapping study in 36 native German speaking participants with chronic lesions to the language network after left-hemispheric stroke. During the experiment, participants performed a plausibility judgement task on auditorily presented adjective-noun phrases that were either meaningful (‘anxious horse’), anomalous (‘anxious salad’) or had the noun replaced by a pseudoword (‘anxious gufel’), as well as a single-word control condition (‘horse’). We observed that reduced accuracy for anomalous phrases is associated with lesions in left anterior inferior frontal gyrus, whereas increased reaction times for anomalous phrases correlates with lesions in anterior-to-mid temporal lobe. These results indicate that anterior inferior frontal gyrus is relevant for accurate semantic decisions, while anterior-to-mid temporal lobe lesions lead to slowing of the decision for anomalous two-word phrases. These differential effects of lesion location support the notion that anterior inferior frontal gyrus affords executive control for decisions on semantic composition while anterior-to-mid temporal lobe lesions slow the semantic processing of the individual constituents of the phrase.

Syntax through the looking glass: A review on two-word linguistic processing across behavioral, neuroimaging and neurostimulation studies

“…Overall, the reviewed studies support the notion of a neuro-anatomical dissociation between the syntactic and semantic combinatorial processes when building basic two-word structures. In particular, the posterior portion of the left IFG (BA44) and temporal lobe combine abstract syntactic information into constituents, while the ATL, vmPFC (Pylkkänen, 2020), angular gyrus and more anterior portion of the left IFG (Graessner, Zaccarella, Friederici, et al, 2021; are involved in semantic and conceptual composition. Initial evidence exists also for a modality-independent nature of this dissociation (Blanco-Elorrieta et al, 2018;Matchin et al, 2022;Moreno et al, 2018), which should be addressed in future studies.…”

“…Overall, the reviewed studies support a crucial role of bottom-up process in integrating the syntactic features of two words into a constituent. This reliance on basic bottomup operations, with a limited role of top-down predictions, might represent a critical distinction between the syntactic combinatorial system and the semantic one, possibly grounded in differences at the neuro-anatomical level (Graessner, Zaccarella, Friederici, et al, 2021;Pylkkänen, 2020;Zaccarella & Friederici, 2015a;Zaccarella, Schell, et al, 2017). Similarly, this aspect might differentiate syntactic composition, characterised by hierarchical processing, from nonlinguistic domains which might strongly rely on serial topdown predictions (Zaccarella et al, 2021).…”

“…In recent years, a few functional studies have begun to employ simple two-word combinations to more closely focus on the fundamental question of how linguistic features are combined at the most basic level within the language network (Bemis & Pylkkänen, 2011;Zaccarella & Friederici, 2015a). There is general agreement that linguistic processes driven by conceptual-semantic information particularly involve the ATL, the vmPFC, and the AG (Graessner, Zaccarella, Friederici, et al, 2021;, as also discussed in a recent comprehensive review (Pylkkänen, 2020) specifically addressing the so-called "red boat" paradigm (Bemis & Pylkkänen, 2011).…”

“…For example, language composition can be studied and compared in young and old populations (Poulisse et al, 2020;Poulisse et al, 2019), distinguishing its developmental trajectory from the acquisition and decline of general domain processes. Similarly, brain damage can be linked to specific linguistic deficits (Graessner, Zaccarella, Friederici, et al, 2021).…”

Minimal Phrase Composition Revealed by Intracranial Recordings

“…(1) the precise localization of composition effects, addressing a current tension in the literature with respect to frontal, temporal, and parietal contributions to minimal phrase processing (Schell et al, 2017;Flick and Pylkkänen, 2020;Graessner et al, 2021a) and (2) the precise timing and duration of composition effects, addressing the above noted variations documented in the literature.…”