Jeffrey J. Holliday

The perceptual assimilation of Korean obstruents by native Mandarin listenersa)

The current study reports the results of a perception experiment in which 20 naive native Mandarin listeners classified and rated the goodness of Korean stops /p, t, k, p(h), t(h), k(h), p*, t*, k*/, affricates /tɕ, tɕ*, tɕ(h)/, and fricatives /s(h), s*/ in terms of Mandarin segmental categories. It was found that listeners were sensitive to the voice onset time dimension of Korean stops and the presence of aspiration in Korean affricates, but Korean lenis and aspirated obstruents were generally assimilated to a single Mandarin category because the f0 cue differentiating them is not relevant to any Mandarin segmental contrast. The affricates were perceived as alveolopalatal and postalveolar more often than alveolar. The perception of fricatives was strongly influenced by vowel context, as the two fricatives were often perceived as different categories before /a/, but as the same category more often before /i/ and /u/. The results for the affricates and fricatives may be partly explained by Mandarin phonotactic constraints that prohibit alveolar and postalveolar consonants before /i/ and alveolopalatal consonants before /a/ or /u/.

A longitudinal study of the second language acquisition of a three-way stop contrast

Abstract The goal of this paper was to document how native (L1) speakers of a language with a two-way stop contrast acquire a three-way stop contrast in a second language (L2). Mandarin presents a two-way stop contrast cued primarily by VOT, whereas Korean presents a three-way stop contrast cued jointly by VOT and the f0 of the following vowel. Mandarin and Korean stop productions from 12 L1 Mandarin novice L2 learners of Korean were subjected to acoustic analysis. Results revealed a wide variety of production patterns, suggesting that the learning of an L2 contrast may not always be predicted by cross-language acoustic correspondences. Six of the participants were recorded again both six and 12 months later. The longitudinal results showed that some learners were unable to produce the Korean contrast in a native-like way even after one year of intensive L2 instruction. Learners whose initial production strategy was consistent but incorrect fared worse after one year than learners whose productions initially exhibited more variability. These results contribute to our understanding of both the L2 acquisition of “new” and “similar” categories and also how well naive perceptual assimilation can predict L2 production.

Why do adults vary in how categorically they rate the accuracy of children’s speech?

In a recent experiment using continuous visual analog scales (VAS) to examine adults’ perception of children’s speech accuracy, listeners varied in the extent to which they categorically perceived children’s English and non‐English productions [Munson et al., American Speech‐Language‐Hearing Association (2008)]. Some listeners utilized all points on the VAS line for their responses, while others grouped responses around discrete locations on the line. It is hypothesized that differences in categoricity of responses across listeners might relate to listeners attending to either categorical linguistic information (i.e., identifying phonemes in a word, which would promote more categorical labeling) or gradient indexical information (i.e., identifying the child’s sex or age, which would promote more continuous labeling). If this is true, it should be possible to elicit differences in categoricity of fricative “goodness” judgments in individual listeners by priming them to listen to linguistic variables (by in...

Contrastive apical post-alveolar and laminal alveolar click types in Ekoka !Xung

Ekoka !Xung has four contrastive click types—dental, alveolar, lateral, and “retroflex.” We provide acoustic and ultrasound results of five speakers’ productions of the typical alveolar click and the contrastive “retroflex” click. Ultrasound results show that the “alveolar” click is apical post-alveolar and the “retroflex” click is laminal alveolar. The burst duration of the post-alveolar click averages 12 ms which is “abrupt,” while the burst duration of the alveolar click averages 30 ms, which is “noisy.” Mixed effects logistic regression models tested the effects of rise time and burst duration. Burst duration differed significantly among the two clicks (p < 0.001), while the effect of rise time was not significant. The ratio of energy in the click noisebursts below 20 ERB to the energy above 20 ERB is between 1.0 and 1.5 for the post-alveolar click, but between 0.5 and 1.0 for the alveolar click. The ratio was a significant predictor of click type (p = 0.014). The highest concentration of energy for t...

A preliminary investigation of the effect of dialect on the perception of Korean sibilant fricatives

Korean has two sibilant fricatives, /sh/ and /s*/, that are phonologically contrastive in the Seoul dialect but are widely believed to be phonetically neutralized in the Gyeongsang dialects spoken in southeastern South Korea, with both fricatives being acoustically realized as [sh]. The current study investigated the degree to which the perception of these fricatives by Seoul listeners is affected by knowledge of the speaker’s dialect. In the first task, the stimuli were two fricative-initial minimal pairs (i.e., four words) produced by 20 speakers each from Seoul and Gyeongsang. Half of the 18 listeners were told that the speakers were from Seoul, and the other half were told they were from Gyeongsang. Listeners identified the 160 word-initial fricatives and provided a goodness rating for each. It was found that neither the speaker’s actual dialect nor the primed dialect had a significant effect on either identification accuracy or listeners’ goodness ratings. In a second task, listeners identified token...

Modifying speech to children based on perceived developmental level: An acoustic study of adults’ fricatives.

Recent acoustic studies have shown that children learning English gradually differentiate between the sounds /s/ and /∫/ [Li et al. (2009)]. Other studies have shown that adults can perceive different degrees of accuracy in these phonemes when given a non‐categorical response modality, such as a visual‐analog scale [Urberg‐Carlson et al. (2008)]. This presentation combines a VAS rating task with a production task. Adults were presented with children’s productions of target fricatives. They rated the accuracy, after which they were instructed to produce the target sequence as if they were responding to the child whose speech they had just rated. Ongoing analyses examine whether adults respond differently to productions that they have rated as intermediate between /s/ and /∫/ and whether adults produce hyperarticulated version of fricatives in response to productions that they rated as not canonically /s/ or /∫/ , as opposed to ones that they rated as adult‐like. The results of this project will be used in ...

The perceptual acquisition of Korean fricatives by first language Mandarin listeners.

Despite numerous studies, it remains unclear how naive “foreign language” listeners become proficient L2 listeners, particularly in regard to difficult L2 contrasts. In an earlier study in which nine English‐speaking L2 learners of Korean of varying proficiency and history of exposure to Korean identified Korean tense /s*/ versus non‐tense /s/, listeners showed varying perceptual strategies. The results suggested that learners gradually learn to perceive differences in L2 contrasts by re‐weighting useful cues and learning to ignore the “inefficient” cues that are initially relied on when the members of the L2 contrast are assimilated to L1 categories. This paper will report on the results of the same task (identification of CV sequences excised from real words), testing 30 L1 Mandarin speakers who have been in an intensive Korean language program in Seoul for about 2 months. The results of the present study will show whether there are as many inter‐listener differences when the level and type of L2 exposu...