Demographic effects on language production: Perspectives from usage-based linguistics

Existing frameworks describing how language abilities change across the lifespan leave important gaps. Firstly, normative language frameworks often fail to account for demographic factors, leading to highly heterogeneous samples in which substantial variance in language outcomes remains unexplained. This heterogeneity makes the identification of reliable linguistic markers of normal or pathological pathways in cognitive ageing more difficult. Secondly, language studies primarily focus on isolated aspects of grammatical (e.g., syntactical complexity) or lexical abilities (e.g., naming), rather than providing a comprehensive account of patterns of linguistic changes. In particular, usage-based measures (frequency and familiarity effects) are not typically considered, despite their potential to capture additional linguistic information apart from that captured by the usual formal language metrics. This project aims to address these two gaps by conducting a comprehensive exploration of language production in old age specifically exploring demographic effects and implementing usage-based measures. It constitutes a secondary data analysis of 107 Cookie Theft picture description transcripts collected at the Dementia Research Centre at University College London. I will implement a method based on usage-based linguistics, and combine formal measures of language (e.g., lexical diversity, syntactic complexity) with usage-based frequentist measures covering conventionality aspects of word combinations (e.g., collocation strength). Overall, measures will cover four linguistic dimensions: lexical, syntactic, semantic and fluency. Three demographic factors will be explored: sex, age, and years of education. Language abilities will also be considered in relation to cognitive abilities (verbal and nonverbal intelligence and working memory).