Single‐word semantic judgements in semantic dementia: Do phonology and grammatical class count?

Abstract

Background: Listeners make active use of phonological regularities such as word length to facilitate higher‐level syntactic and semantic processing. For example, nouns are longer than verbs, and abstract words are longer than concrete words. Patients with semantic dementia (SD) experience conceptual loss with preserved syntax and phonology. The extent to which patients with SD exploit phonological regularities to support language processing remains unclear.

Aims: We examined the ability of patients with SD (1) to perceive subtle acoustic–phonetic distinctions in English, and (2) to bootstrap their accuracy of lexical‐semantic and syntactic judgements from regularities in the phonological forms of English nouns and verbs.

Methods and Procedures: Four patients with SD made minimal pair judgements (same/different) for auditorily presented stimuli selectively varied by voice, place, or manner of the initial consonant (e.g., pa –ba). In Experiment 2 patients made forced‐choice semantic judgements (abstract or concrete) for single words varied by (1) concreteness (abstract or concrete); (2) grammatical class (noun or verb); and (3) word length (one‐ or three‐syllable words).

Outcomes and Results: The most semantically impaired patients paradoxically showed the highest accuracy of minimal pair phonologic discrimination. Judgements of word concreteness were less accurate for verbs than nouns. Among verbs, accuracy was worse for concrete than abstract items (e.g., eat was worse than think). Patients were more likely to misclassify longer concrete words (e.g., professor) as abstract, demonstrating sensitivity to an underlying phonologically mediated semantic property in English.

Conclusions: Single‐word semantic judgements were sensitive to both grammatical class and phonological properties of the words being evaluated. Theoretical and clinical implications are addressed in the context of an anatomically constrained model of SD that assumes increasing reliance on phonology as lexical‐semantic knowledge degrades.