Perception of Bengali stops: The role of Markedness

Abstract

Marked segments, by definition, bear additional properties with respect to unmarked ones, in principle requiring more articulatory effort. At the same time, however, these additional properties may make marked sounds more audible and identifiable. This study investigates how voicing and aspiration, as mark properties, affect the perception of Bengali stops, exhibiting the common Indo-Aryan four-way contrast: voiceless unaspirated, voiceless aspirated, voiced unaspirated, voiced aspirated. Specifically, I test whether the properties of voicing and aspiration result in improved perceptibility of stops separately and whether their combination shows an increased cumulative effect.50 native Bengali speakers each listened to 270 CV stimuli, the first syllable extracted from real 3-syllable words. The C was one of the four stop types; the V was /a/. The participants identified the syllable they heard by selecting one of six options (the four stops and two fillers) shown in Bengali script. The results reveal significantly higher perceptual accuracy in the stops marked by voicing vs. voiceless categories (83.78% vs 80.48%, respectively) and higher perceptual accuracy of aspirated versus unaspirated stops ( 84.04% vs 78.49%, respectively). The two “marks” combined in voiced aspirated stops do not increase accuracy more than each alone (83.89%) but overall differ most from plain unaspirated stops (75.8%).