About

Olga Dmitrieva is an Associate Professor at Purdue University, affiliated with the School of Languages and Cultures (SLC) and the Department of Speech, Language, and Communication (SLC). She serves as the Graduate Director within her department. Her research focuses on experimental phonetics and phonology, with particular emphasis on acoustic phonetics, crosslinguistic speech perception, and the acquisition of second language speech. She holds a Ph.D. in Linguistics from Stanford University, obtained in 2012. Her work involves studying the physical properties of speech sounds and how they are perceived across different languages, contributing to a deeper understanding of speech processing and language learning.

Research topics

• Computer Science
• Linguistics
• Psychology
• Speech recognition
• Acoustics
• Physics
• Medicine
• History
• Audiology
• Cognitive psychology

Selected publications

• Comprehensibility and the Acoustic Contrast Between Tense and Lax Vowels in Mandarin-Accented English Speech

  2024-08-16

  articleOpen accessSenior author

  This study investigates the relation between the comprehensibility of Mandarin-accented English speech and the degree of acoustic contrast in the vowel tenseness pairs [i] - [ɪ] and [u] - [ʊ]. We hypothesized that Mandarin speakers under-differentiate the tenseness vowel pairs and that variability in the acoustic contrast between tense and lax vowels correlates with variability in comprehensibility. Twenty Mandarin speakers and ten English speakers produced sentences with the target vowels, and 26 English speakers rated the sentences’ comprehensibility. We found that spectral difference between [u] - [ʊ] was relatively reduced in L2 speech and that L2 productions were overall more variable, compared to American speech. We also found that the degree of spectral contrast correlated with comprehensibility rating, suggesting that the degree of acoustic differentiation between phonological categories contributes to the comprehensibility of L2 speech.

  PublisherOA PDFDOI

• Enhancement of laryngeal contrasts in non-native English clear speech: a comparison between L2-immersed sequential bilinguals and L1-immersed speakers

  Frontiers in Human Neuroscience · 2024-09-13 · 1 citations

  articleOpen accessSenior authorCorresponding

  Clear speech, a speaking style used to mitigate communicative circumstances affecting the transmission or decoding of speech signal, often involves the enhancement of language-specific phonological contrasts, including laryngeal contrasts. This study investigates the role of language dominance in the implementation of language-specific laryngeal contrasts in L2 clear speech. Two groups of Korean-English speakers (L1 Korean) were tested: a relatively less Korean-dominant L2-immersed group of sequential bilinguals ( N = 30) and a strongly Korean-dominant L1-immersed group ( N = 30), with dominance assessed based on the results of the Bilingual Language Profile. Participants read a set of English minimal pairs differing in the voicing of word-initial stops (e.g., tab vs. dab ), and their acoustic enhancement strategies were compared with those of native English speakers ( N = 20). As correlates of the English laryngeal contrast, voice onset time (VOT) and onset f0 were measured. Results showed that both bilingual groups enhanced English laryngeal contrast in clear speech via voiceless VOT lengthening, similarly to native English speakers, but to a smaller extent than native speakers. Both bilingual groups also implemented a greater degree of onset f0 difference between voiced and voiced English stops than native English speakers did, although no enhancement of this parameter was observed in their clear speech. Surprisingly, no significant differences were found between L2- and L1-immersed speakers, suggesting a lack of language immersion effect on the acoustic enhancement strategies in L2 clear speech. We discuss possible explanations for this finding and propose directions for future research.

  PublisherOA PDFDOI

• The Phonetics and Phonology of Adult L2 Learners in the Classroom

  Cambridge University Press eBooks · 2024-11-14

  book-chapter1st authorCorresponding
  PublisherDOI

• Crosslinguistic influence in bilingual production of clear speech

  The Journal of the Acoustical Society of America · 2023-03-01

  articleSenior author

  Clear speech, a listener-oriented speaking style adopted to mitigate communicative barriers, is often characterized by language-specific acoustic enhancement of phonological contrasts (Smiljanic and Bradlow, 2008; Cho et al., 2011). Previous work (e.g., Smiljanic and Bradlow, 2009) suggested a possibility that bilingual speakers’ L1 and L2 clear speech strategies could affect each other. The current study examined English and Korean clear speech produced by late Korean-English bilinguals (n = 30) living in the United States. Their clear speech produced in each of the two languages was compared with clear speech produced by either English (n = 20) or Korean (n = 20) monolinguals. Of specific interest was how English and Korean stop contrasts were realized and enhanced in clear speech with respect to the use of VOT and onset F0. The bilinguals enhanced the English voicing contrast in the same manner as English monolinguals, via asymmetrical lengthening of voiceless stops’ VOT. In Korean clear speech, bilinguals also lengthened VOT of the aspirated and lenis (long-lag) stops, whereas Korean monolinguals did not exhibit the same pattern. These findings suggest that Korean clear speech strategies produced by the bilinguals could be affected by their L2, due to their extensive exposure to the language.

  PublisherDOI

• Non-native talkers and listeners and the perceptual benefits of clear speech

  The Journal of the Acoustical Society of America · 2023 · 5 citations

  Senior authorCorresponding
  • Computer Science
  • Speech recognition
  • Computer Science

  This study examined the intelligibility benefit of native and non-native clear speech for native and non-native listeners when the first language background of non-native talkers and listeners is matched. All four combinations of talkers and listeners were tested: native talker-native listener, non-native talker-native listener, native talker-non-native listener, and non-native talker-non-native listener. Listeners were presented with structurally simple but semantically anomalous English sentences produced clearly or casually and mixed with speech-shaped noise at 0 dB signal-to-noise ratio and asked to write down what they heard. Results showed that native English speech was more intelligible than non-native speech and that native English listeners recovered speech better than non-native listeners did. Clear speech was significantly more intelligible than casual speech. There were no interactions between speaking style and native language background indicating that clear speech intelligibility benefit was not significantly different across distinct combinations of talkers and listeners. In particular, shared first language background among non-native speakers and listeners did not provide an extra benefit in either the overall speech intelligibility or the intelligibility gains of clear speech.

  PublisherDOI

• Crosslinguistic differences in cue primacy affect bilinguals' discrimination performance

  The Journal of the Acoustical Society of America · 2023-03-01

  articleSenior author

  Onset f0 – pitch at the onset of the vowel following the consonant – is an important cue to laryngeal stop distinctions in Korean, especially the lenis-aspirated contrast, while in English onset f0 plays only a secondary role in distinguishing voiced and voiceless stops. The current study investigates to what extent Korean heritage speakers (n = 29) who are English dominant can perceptually discriminate Korean lenis and aspirated stops differing acoustically only in terms of onset f0, but not in terms of VOT. Heritage speakers’ performance is compared to that of Korean-immersed native speakers as a control group (n = 29). We hypothesized that heritage speakers would experience more difficulty than Korean-immersed speakers in using onset f0 as the sole cue to laryngeal contrast, due to the influence from English, where VOT dominates onset f0 as a cue to voicing. The results of a mixed-effects logistic regression model confirmed that heritage speakers were less accurate in the lenis-aspirated discrimination task than Korean-immersed speakers, providing support to the hypothesis that bilinguals’ perception of their first language speech can be subject to crosslinguistic influence from their second language.

  PublisherDOI

• Acoustic convergence of voicing when shadowing an unfamiliar language and a non-native accent

  The Journal of the Acoustical Society of America · 2023-03-01

  articleSenior author

  Previous work has established that speakers are able adjust the acoustic properties of their speech in order to mimic their interlocuter or speech model with whom they share a native language (e.g., Pardo, 2013). However, it is less clear whether convergence to accented speech and/or unfamiliar languages is also a possibility. The goal of the present study is to investigate acoustic convergence between two groups of native English speakers (N = 15 in each group) and a speaker of Russian-accented English versus a speaker of Russian. The study consisted of four experimental phases: baseline, exposure, shadowing, and post-test. Participants in both groups shadowed acoustically identical material, although one group was informed that they were hearing Russian-accented English words (ACC-ENG condition), while the second group was told that they were shadowing Russian words (RUS condition). Initial analysis of the VOT durations of participants’ voiceless stops indicated that only participants in the RUS group converged towards the models’ shorter VOT during shadowing, although this convergence did not generalize to participants’ pronunciation of English words during the post-test. Additional data analysis is ongoing. This presentation will also discuss results with respect to their implications for theories of speech accommodation and non-native speech learning.

  PublisherDOI

• Acoustic properties of non-native clear speech: Korean speakers of English

  Speech Communication · 2023-09-10 · 4 citations

  articleSenior author
  PublisherDOI

• Non-native talkers and listeners and perceptual benefits of clear speech

  The Journal of the Acoustical Society of America · 2022-04-01

  articleSenior author

  Native clear speech aids speech perception for various native populations such as hearing-impaired adults (Picheny et al., 1985). Compared to native speech, little is known about the benefit of non-native clear speech (Smijlanic and Bradlow, 2011). The current study investigates whether non-native clear speech can aid both native and non-native listeners, using every combination of talkers and listeners (native talker/listener, and non-native talker/listener). Non-native participants were L1 Korean speakers, while native participants were L1 American English speakers. Each group had 32 participants listening to semantically anomalous English sentences recorded by four native and four non-native talkers and presented with speech-shaped noise at 0 dB SNR in each speaking style (casual and clear). Each sentence had four keywords, and the number of correct keywords was converted to rationalized arcsine units. The results show significant effects of talker L1, listener L1, and speaking style. Native speech was more intelligible than non-native one and native listeners outperformed non-native listeners. Clear speech was more intelligible than casual speech. However, none of the interactions among the factors reached statistical significance, indicating the two groups of talkers and the two groups of listeners behaved in a parallel way in terms of producing and receiving clear speech benefits.

  PublisherDOI

• The perception of English vowels by native Korean and Mandarin speakers

  The Journal of the Acoustical Society of America · 2022-10-01 · 1 citations

  articleOpen accessSenior author

  Second language learners often experience difficulties discriminating L2 contrasts absent in their L1. Theoretical frameworks, (e.g., SLM, PAM, L2LP), make predictions about the acquisition of L2 phonemes based on the existence of equivalent or similar L1 phonemes, but do not explicitly consider the extent of L1 allophonic variability as a contributing factor. The present study compared the perceptual discrimination of English contrasts /ɪ-i/ and /ɛ-æ/ by learners with Mandarin and Korean L1 backgrounds. Both languages have only one L1 phoneme for the two relevant L2 contrasts. Furthermore, Mandarin [ɛ] is an allophone of the highly variable mid vowel phoneme /E/, while Korean /ɛ/ lacks this variability. Allophonic variability may condition learners to accept diverse vowel qualities as acceptable renditions of the corresponding phoneme, therefore lowering their discriminability. In an AX perceptual discrimination task, 16 Standard Mandarin and 14 Korean speakers were similar in their discrimination of the /ɪ-i/ vowels, but, as predicted, the Mandarin group was less accurate for the /ɛ-æ/ pairs. These findings suggest that allophonic variability in the L1 needs to be considered when making theory-driven predictions concerning the acquisition of specific phonological categories.

  PublisherOA PDFDOI

Frequent coauthors

• Alexander L. Francis

  Purdue University West Lafayette

  10 shared

• Jenna Conklin

  10 shared

• Fernando Llanos

  8 shared

• Ye-Jee Jung

  Purdue University West Lafayette

  7 shared

• Amy Hutchinson

  Purdue University West Lafayette

  6 shared

• Amanda A. Shultz

  Purdue University System

  6 shared

• Wai Ling Law

  4 shared

• А. В. Дмитриев

  4 shared

Education

• PhD, Linguistics

  Stanford University

  2012

• MA, Linguistics

  Stanford University

  2008

• MA, Linguistics

  University of Kansas

  2005

• Specialist Diploma, Romance and Germanic Philology

  Tomsk State University

  2002