Slide Slam B18
Heterogeneous ERP patterns for different error types in French complex noun phrases
Phaedra Royle1,2, Émilie Courteau1,2, Ariane Brucher2,3, Clara Misirliyan1,2, Karsten Steinhauer2,4; 1School of Speech-language Pathology and Audiology, Université de Montréal, 2Centre for Research on Brain, Language and Music, 3Department of Psychology, Université de Montréal, 4School of Communication Sciences and Disorders, McGill University
INTRODUCTION: The French noun phrase (NP) offers parameters to study morphosyntactic (gender) and syntactic error processing. Gender agreement on adjectives can be irregularly marked (e.g., vert/e [vɛʁ/t], ‘green.M/F’), while singular definite determiners carry predictable transparent vowels for agreement (e.g., le/la, ‘the.M/F’). Furthermore, French lexically constrains adjective position: size adjectives are prenominal while most others are postnominal. Both syntactic and morphosyntactic violations often elicit a biphasic ERP pattern: LAN (left-anterior negativities) or N400s followed by posterior positivities (P600). LANs are linked to rule-based processing, N400s to lexical processing, and P600s to integration and repair (Molinaro et al. 2011; Steinhauer & Connolly 2008). Our picture-sentence matching paradigm tested four error types in French adjective processing. METHODS: 22 adults saw an image, then heard a correct or ungrammatical spoken sentence describing it and completed a judgment task. Six NP conditions each contained an adjective (pre- or postnominal adjective) and a prepositional phrase: 1. Correct: Je vois le soulier vert sur la table ‘I see the.M shoe.M green.M on the table’ 2. Adjective agreement error: Je vois le soulier *verte ‘green.F’... 3. Determiner agreement error: Je vois *la ‘the.F’ soulier vert … 4. Prenominal adjective inversion: Je vois le soulier *grand ‘big.PRE’ … 5. Postnominal adjective inversion: Je vois le vert ‘green.POST’ *soulier … 6. Noun-drop: Je vois le vert __ dans …, I see the green (one) in … Each condition was analyzed time-locked to the adjective or determiner. Correct conditions (1 and 6) served as grammatical controls, and (6) controlled for strategic processing in (5). Agreement errors should elicit N400s on adjectives (2), and LANs on determiners (3), both followed by P600s. Since violations in (4–5) depend on the adjective’s lexical status, we expected N400-P600s, with earlier effects on the adjective in (5–6), as they signal either an unusual word order (6) or a subsequent violation (5). RESULTS: In all but one condition, errors elicited negativities followed by positivities. However, adjective agreement errors (2) elicited LANs (300–500 ms), and large P600s (650–1050 ms), while determiner errors (3) elicited sustained bilateral anterior negativities (400–500 ms) and small P600s (750–1050 ms). Inverted prenominal adjectives (4) elicited biphasic N400-P600s. Postnominal adjective inversion (5) and noun-drop (6) – an unusual but legal word order – both elicited small central positivities at 'vert', followed by a non-significant N400 in (5). CONCLUSION: All P600s were as predicted. The LAN in (2) suggests these violations rely on rule-based processing. The frontal negativity in (3) resembles a phonological mismatch negativity (Connolly & Phillips, 1994) elicited by the mismatch between the image (e.g., shoe.M) and determiner’s gender (the.F). The large N400 in (4) is in line with a salient word-order violation tied to lexical status (prenominal adjectives). Condition (5) illustrates that unusual (grammatical) word orders can initiate syntactic checks (early P600s), while the small N400 on the (ungrammatical) noun conforms with the adjective's postnominal status. Overall, the data suggest that ERPs for adjectival violations in French NPs are modulated by a combination of rule-based, lexical, and phonological factors.