Eriko Atagi

Classification of regional dialects, international dialects, and nonnative accents

Abstract Talkers׳ regions of origin and native languages will significantly shape their speech production patterns. Previous results suggest that listeners are highly sensitive to whether a talker is a native or nonnative speaker of the language. Listeners also have some ability to categorize or classify talkers by regional dialect or nonnative accent. However, most previous studies have included variability in only one of these categories (many nonnative accents or many regional dialects). The present study simultaneously examined listeners’ perceptual organization of regional dialects and nonnative accents. Talkers representing six United States regional dialects, six international native dialects, and twelve nonnative accents were included in two tasks: an auditory free classification task—in which listeners grouped talkers based on perceived region of origin—and ladder task—in which listeners arranged talkers based on their perceived distance from standard American English. Listeners were sensitive to the distinction between native and nonnative accents, even when presented with a very wide range of dialects and accents. Further, subgroups within the native and nonnative clusters in the free classification task suggested several organizing factors, including perceived distance from the local standard, specific acoustic-phonetic talker characteristics, and speaking rate.

Auditory free classification of native and nonnative speech by nonnative listeners

Nonnative listeners are less accurate than native listeners at identifying, categorizing, and classifying native talkers’ accents, suggesting a decrement in nonnative listeners’ representation of native variation. Nonnative listeners’ representation of nonnative variation, however, may be as robust as, or more robust than, that of native listeners. To compare native and nonnative listeners’ representations of linguistic variation, the current study examined native English, Korean, and Spanish listeners’ classification of native and nonnative talkers by native language. Overall, the listener groups performed with similar accuracy, but they exhibited different perceptual similarity spaces. Specifically, listener groups demonstrated heightened perceptual sensitivity to talkers with whom they share a native language. The results suggest that listeners’ linguistic experiences significantly shape their perceptual representation of nonnative varieties.

Categorization of regional, international, and nonnative accents

Auditory free classification—a task in which listeners classify auditory samples into unconstrained groups—has provided insight into perceptual representation and categorization for several sources of speech variability including U.S. regional dialects, nonnative accents, and foreign languages. Within these studies, phonological markedness and geography have emerged as central organizing principles. However, previous studies were limited by including only one source of variability. To address this gap, the perception of U.S. regional dialects, international English dialects, and nonnative accents was investigated within one classification task. Listeners categorized talkers based on perceived location of origin. Cluster analysis demonstrated a perceptual divide between native and nonnative talkers. Native talkers were further delimited by geographic proximity into Southern Hemisphere, U.S., and United Kingdom groups. One exception was the consistent grouping of Southern U.S. talkers with talkers from Engl...

Effect of adult aging on pupillary response to auditory stimuli of varying levels of complexity

Pupil dilation as a measure of listening effort has been well documented (Kuchinsky et al., 2013), as has the association between adult aging and increased effort in listening tasks (Wingfield et al., 2015). However, age-related differences in the dynamics of pupillary response over time while attending to auditory stimuli have received less study; nor has the nature of the pupillary response to simple stimuli (tones) versus more complex speech stimuli (words, sentences) been systematically explored. In this study we examined multiple parameters of young adult and older adult listeners’ changes in pupil size, including peak amplitude and latency to peak, elicited while making decisions in response to auditory stimuli that varied in acoustic or linguistic complexity. Results showed that the latency to peak pupil size was slower for older adults and for increasingly complex stimuli. The shape of pupillary response curve also changed dramatically with the nature of the stimuli: attending to briefer stimuli (...

Categorization training for non-native accented word recognition

Non-native accented speech is considered an adverse listening condition for native speakers (Mattys et al., 2012). Previously, accurate categorization of non-native accents was found to relate to accented speech-in-noise recognition (Atagi and Bent, 2015). The goal of the current study is to examine (1) the extent to which native speakers can be trained to categorize the language backgrounds of non-native talkers using feedback, and (2) the extent to which accent categorization training, relative to word and sex identification tasks, improves non-native accented word recognition. Participants were randomly assigned to one of three training tasks: an accent categorization task with trial-by-trial feedback (n = 10), a transcription task (n = 10), and a speaker sex categorization task (n = 10). Critically, all participants were exposed to the same set of training stimuli. Participants completed a transcription task with speech-shaped noise prior to and following training. We assessed keyword accuracy for non...

Foreign-accent discrimination with words and sentences

Native listeners can detect a foreign accent in very short stimuli; however, foreign-accent detection is more accurate with longer stimuli (Park, 2008; Flege, 1984). The current study investigated native listeners’ sensitivity to the characteristics that differentiate between accents—both foreign versus native accents and one foreign accent versus another—in words and sentences. Listeners heard pairs of talkers reading the same word or sentence and indicated whether the talkers had the same or different native language backgrounds. Talkers included two native talkers (Midland dialect) and six nonnative talkers from three native language backgrounds (German, Mandarin, and Korean). Sensitivity varied significantly depending on the specific accent pairings and stimulus type. Listeners were most sensitive when the talker pair included a native talker, but could detect the difference between two nonnative accents. Furthermore, listeners were generally more sensitive with sentences than with words. However, for...

Auditory free classification of nonnative speech by nonnative listeners

Nonnative listeners are less accurate than native listeners at classifying talkers by regional dialect [Clopper and Bradlow (2009)]. This decrement may be due to less robust knowledge about the underlying sound structure of the target language or less extensive experience with socio-cultural phonetic variation in the target language. To disentangle the contribution of these two factors, this study examined native and nonnative listeners’ abilities to classify talkers who varied on another sociophonetic dimension: foreign accent. Unlike regional dialect variation, nonnative listeners typically have more experience with nonnative speech than native listeners, particularly for talkers with the same native language background. Using auditory free classification, native listeners of English and native Korean listeners classified talkers by perceived native language. Talkers consisted of nonnative talkers from six native language backgrounds and native talkers. Results demonstrated that native listeners were ne...