Perception and representation of Bengali nasal vowels

The paper addresses the question of native speakers’ online awareness and perceptual use of phonetic nasalisation by examining surface nasalisation in two types of surface vowels in Bengali: underlying nasal vowels and nasalised vowels before a nasal consonant. In a cross-modal forced-choice experiment, we investigate the hypothesis that only unpredictable nasalisation is represented and that this sparse representation governs listeners’ interpretation of vowel nasality. Auditory primes consisting of CV segments of monosyllabic CVC words containing either nasal vowels ([cã] for cãd), oral vowels ([ca] for cal) or nasalised oral vowels ([ca(n)] for can) preceded visual full-word targets. Faster reaction times and fewer errors are observed after nasal vowel primes compared to both oral and nasalised vowel primes. This indicates that nasal vowels are specified for nasality and lead to faster recognition compared to the oral vowel conditions, which are underspecified and thus cannot be perfectly matched with incoming signals.