The pronunciation of words in the bat , bet , and bait classes before /ɡ/ (hereafter bag , beg , and bagel ) is well-documented in production studies of Pacific Northwest English and beyond, yet there remains a lack of consensus as to whether the phenomenon involves only phonetic raising, measurable on the acoustic dimension, or whether this raising involves partial reduction in phonemic contrasts, characteristic of merger. We argue that perceptual evidence is crucial to investigating the possibility of phonemic merger and explore the bag-bagel phenomenon through an ambiguous vowel categorization experiment. Listeners from three West Coast cities with different established production patterns with respect to the phenomenon—Seattle and Portland (where the phenomenon is established) and San Francisco (where it is not)—categorized resynthesized vowels on a continuum from bag to bagel . Perceptual differences emerge that mirror production observations. Yet, listener behavior does not support a raising characterization as Seattle and Portland listeners do not show a higher switchpoint on the continuum for bag than San Francisco listeners. Instead, Seattle and Portland listener groups demonstrate less sensitivity to phonemic contrast overall, with more variable categorization for the most canonical tokens at both ends of the continuum. Finally, many individual listeners from Seattle and Portland lack a switchpoint altogether, demonstrating both categorical and highly variable responses. The findings suggest that the phenomenon may involve a conditioned phonemic merger of bag - bagel in the Seattle and Portland communities, a perspective relevant for future studies of bag , beg , and bagel .
Swan et al. (Thu,) studied this question.