Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Testing different metrics for syntactic complexity in spontaneous speech: a cross-genre comparison
Poster Session C, Friday, October 25, 4:30 - 6:00 pm, Great Hall 3 and 4
Simona Mancini1,2, Jorge Hidalgo Chagoya3, Roel Jonkers3, Maddi Carrera1, Xabi Ansorena1; 1BCBL, Basque Center on Cognition Brain and Language, 2Ikerbasque, Basque Foundation for Science, 3University of Groningen
A commonly noted characteristic of aphasic discourse is the simplification of syntactic complexity (Ulatowska et al., 1990). Various methods and measures have been designed to evaluate this deficit, making discourse assessment a crucial component of aphasia evaluation, treatment, and research. Four elicitation genres of discourse are typically employed: descriptive, procedural, conversational, and narrative. However, aphasiology studies have shown that linguistic outcomes in discourse are often genre- and task-specific (Stark & Fukuyama, 2021). This research analyzes discourse samples collected during the validation of a digital neurolinguistic battery for the assessment of aphasia in Spanish (Ansorena et al., 2020). The study aims to determine which of the four administered tasks, corresponding to the four discourse genres, is most sensitive in detecting syntactic complexity deficits. Additionally, it investigates which task is most sensitive in detecting morphological complexity deficits, taking advantage of the morphological richness inherent in the Spanish language. Finally, we compare two approaches to measuring the morphosyntactic characteristics of discourse—using syntactic proxies versus directly metrics measuring ‘syntactic structures’ (as defined by Agmon et al., 2024)—to identify the most accurate set of variables for discriminating group membership. The spontaneous speech of twenty native speakers of Spanish with aphasia and fifteen sex- and age-matched, non-brain-damaged individuals was recorded while engaging in four different elicitation genres: the Cookie Theft Picture Description (Goodglass & Kaplan, 1972) and three newly developed tasks in Spanish: the Sandwich procedural task, a semi-structured Interview, and the Hare and the Tortoise Story Retelling. Data were preprocessed using manual transcription, parsing and error tagging, in addition to automated morphological tagging and variables extraction (CLAN; MacWhinney, 2000). We measured morphosyntactic complexity using syntactic proxies such as mean length of utterance (MLU), noun-verb ratio, and verbs per utterance, in addition to the ‘syntactic structures’ metric. Such metric included variables that measure the use of subordination, relative clauses, left-branching clauses, and word integration. Morphological complexity was assessed through an inflectional index based on Bastiaanse and colleagues (1996). Analyses of variance showed that the Cookie Theft Picture Description and the Hare and the Tortoise Story Retelling brought out the most differences across syntactic complexity metrics between the groups: while the descriptive task evoked significant differences in the use of relative clauses between the two groups, the narrative task evidenced stronger differences in the use of left branching clauses. Moreover, the story retelling task effectively captured the morphological deficits present in aphasic speakers. Binomial logistic regression models demonstrated that the ‘syntactic structures’ and morphological metrics best discriminated the output of the two groups compared to syntactic proxies. Overall, the data suggest that a comprehensive assessment of the syntactic complexity simplification in aphasia can be obtained with the administration of both descriptive and narrative discourse samples. These findings provide valuable guidance for clinicians and researchers seeking to assess morphosyntactic complexity in Spanish discourse, contributing to the improvement of the accuracy and effectiveness of language assessment and treatment strategies.
Topic Areas: Disorders: Acquired, Syntax and Combinatorial Semantics