Slide Slam K8
Phonetic similarity facilitates long-distance, but not nested dependency processing in preschoolers: A mismatch response study
Dimitra-Maria Kandia1,2,3, Angela D. Friederici4, Arno Villringer1,3, Jutta L. Mueller5, Claudia Männel4,6; 1Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, 2International Max Planck Research School NeuroCom, 3Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, 4Department of Neuropsychology, Max Planck Institute for Human Cognitive and Brain Sciences, 5Institute of Linguistics, University of Vienna, 6Department of Audiology and Phoniatrics, Charité–University Medicine Berlin
Language comprehension typically involves successful tracking of distant speech units. Tracking of multiple, nested relations between these elements is computationally demanding and mastered only late during preschool years (Fengler, Meyer, & Friederici, 2016; Kidd & Bavin, 2002). While processing of distant relations is generally facilitated when dependent elements are linked by statistical probability and perceptual similarity (Sandoval & Gómez, 2013), this has not yet been explored for phonetic similarity in nested structures. To examine the effect of phonetic similarity on nested dependency processing, we recorded event-related brain responses (ERPs) to syllable sequences in preschoolers aged 2-4 years (n = 38). We employed a passive-listening oddball paradigm containing standard sequences with nested dependencies of the form [A1 [A2 C B2] B1], where each element Ai predicted element Bi. Dependencies were marked such that dependent syllables were phonetically similar (1-2 phonetic feature changes) in three different conditions, randomly presented in the stimulus stream: a) only the outer dependency b) only the inner dependency, or c) both dependencies were phonetically similar. Note that in all conditions, both inner and outer dependencies could be tracked by their repeated occurrence across the standard sequences, yet were additionally highlighted by phonetic similarity according to conditions a-c. For each standard sequence, deviant syllable sequences were included that violated the expected nested dependency by reversing the position of the two final syllables (element Bi). Differential ERP responses to these violations (i.e., mismatch responses) compared to standard sequences would indicate successful processing of the respective dependency. Results revealed positive mismatch responses to dependency violations corresponding to the respective phonetically marked outer dependency (condition a) or inner dependency (condition b) independently of age. Thus, despite both nested dependencies being violated in these sequences, children only showed responses to the phonetically marked ones. In contrast, for sequences where both inner and outer dependencies were marked by phonetic similarity (condition c), children revealed no mismatch responses. Together these findings draw a multifaceted picture of dependency processing during preschool years under passive-listening conditions. Phonetic similarity indeed supports the processing of distant elements, even enabling the processing of long-distance dependencies (spanning three inner elements) that are not tracked without additional marking (see condition b). Similarly, phonetic similarity also enables the processing of the acoustically less salient inner dependency (at non-edge positions) that is not detected without additional marking (see condition a). Yet, when both inner and outer relations are simultaneously phonetically marked, children do not process the nested dependency. Our findings thus mirror previous behavioral work (Fengler, Meyer, & Friederici, 2016; Kidd & Bavin, 2002) showing that only during later preschool years, children reliably process nested dependencies in sentences.