Vowel Production Research Articles

Direct articulatory observation reveals phoneme recognition performance characteristics of a self-supervised speech model

Variability in speech pronunciation is widely observed across different linguistic backgrounds, which impacts modern automatic speech recognition performance. Here, we evaluate the performance of a self-supervised speech model in phoneme recognition using direct articulatory evidence. Findings indicate significant differences in phoneme recognition, especially in front vowels, between American English and Indian English speakers. To gain a deeper understanding of these differences, we conduct real-time MRI-based articulatory analysis, revealing distinct velar region patterns during the production of specific front vowels. This underscores the need to deepen the scientific understanding of self-supervised speech model variances to advance robust and inclusive speech technology.

The role of L1-dialect in L2 production

Abstract This study compared the productions of American English (AE) vowels /i/, /u/, /ɛ/, /ɔ/, /ɪ/, /æ/, /ʊ/, /ʌ/, and /ɑ/ between speakers of two Chinese dialects, the Leshan dialect (LSd) and the Xining dialect (XNd). The participants were native speakers of XNd (n = 120) and LSd (n = 120) with advanced English proficiency. Their vowel productions were compared with those of 22 native AE speakers. The results showed that (1) XNd and LSd participants significantly differed in F1 and/or F2 from AE participants when producing /ɪ/, /æ/, /ʌ/, and /ʊ/ and /ɪ/, /ɑ/, /ɔ/, and /ʌ/, respectively; (2) compared with AE participants, XNd participants had significantly longer durations of /ɪ/, /ʊ/, /ʌ/, /ɛ/, and /æ/, and LSd participants had significantly longer durations of /ɪ/, /ʌ/, and /ɛ/. Overall, the vowels that differed in production from that of native speakers were absent in both the L1 dialects and L1 of participants.

First-language interference without bilingualism? Evidence from second language vowel production in international adoptees

Abstract The ability to acquire the speech sounds of a second language has consistently been found to be constrained with increasing age of acquisition. Such constraints have been explained either through cross-linguistic influence in bilingual speakers or as the result of maturational declines in neural plasticity with age. Here, we disentangle these two explanations by investigating speech production in adults who were adopted from China to Sweden as toddlers, lost their first language, and became monolingual speakers of the second language. Although we find support for predictions based on models of bilingual language acquisition, these results cannot be explained by the bilingual status of the learners, indicating instead a long-term influence of early specialization for speech that is independent of bilingual language use. These findings are discussed in light of first-language interference and the theory of maturational constraints for language acquisition.

Velopharyngeal Gap Size During Sustained Vowel Production Correlates With Perceptual Ratings of Hypernasality in Connected Speech.

The purpose of this study was to investigate the relationship between perceptual ratings of hypernasality made during connected speech and velopharyngeal (VP) gap size measured in millimeters in the sagittal plane during sustained vowel production using magnetic resonance imaging (MRI). A retrospective cross-sectional analysis was completed. A subgroup of 110 participants from another study with an Mage of 10.1 years presenting for management of VP insufficiency was included. Perceptual ratings of hypernasality during connected speech and measurement of gap size during sustained /i/ production on MRI were performed by raters blinded to the participants' medical and surgical history. There was a moderate-to-strong, positive correlation (r = .61; p < .001) between hypernasality ratings and VP gap size measured on MRI using sustained /i/. The odds of a higher hypernasality rating increased as the gap size increased (odds ratio = 1.34; 95% CI [1.20, 1.49]; p < .001). The predicted probability for hypernasality ratings of none/minimal/mild steadily decreased as the gap size increased indicating that lower ratings of hypernasality were associated with smaller gap sizes. For the rating of "moderate" hypernasality, the predicted probability of the rating steadily increased up to 8 mm and then decreased as the gap size continued to increase. The predicted probability for a hypernasality rating of "severe" consistently increased as the gap size increased. Hypernasality ratings made at the connected speech level were significantly associated with VP gap size as measured during sustained vowel production. These findings suggest sustained vowel production elicited on MRI may adequately characterize VP gap size in the evaluation of VP insufficiency.

Greek-Cypriot learners’ perception and production of L2 English vowels

Acquiring L2 segments is particularly challenging for L2 learners, especially when the L1 and L2 inventories involve different contrasts and acoustic cues. The present research investigated the perception and production of L2 English vowels by adult Greek-Cypriot learners and examined the effects of orthographic cues and consonantal context in their performance. Perceptual performance was assessed through a forced-choice identification task and production performance through a wordlist-reading and an elicitation task, both analyzed acoustically and through intelligibility ratings. The findings showed the influence of the L1 on both the perception and production of L2 segments, supporting the assumptions of current models of speech perception and production. Learners faced challenges in perceiving the members of L2 contrasts and mostly used their L1 articulatory routines in their productions of L2 vowels. Orthographic cues or consonantal context did not significantly affect learners’ productions or overall perception, although strong contextual effects were observed in individual target vowels. This study is the first to provide an in-depth comparison of CYG and English vowels as well as an examination of the acquisition of L2 English vowels by this population, which can guide EFL teachers and course developers in developing appropriate EFL curricula that incorporate pronunciation instruction.

Consonant and vowel production in children with cochlear implants: acoustic measures and multiple factor analysis

IntroductionThe acoustic limitations of cochlear implants (CIs) can lead to perceptual limitations and consequently to imprecise phonological representations and production difficulties. The aim of the study is to document the phonological and phonetic skills of children with CIs and their typically hearing peers. Phonetically, three types of segments were targeted, each characterized by contrasting acoustic information: nasal/oral vowels, fricative segments, and voiced/voiceless stops.MethodsForty-seven typically hearing children (TH) and 23 children with CIs performed a picture-naming task. Productions were analyzed to obtain phonological measures (percentages of correct phonemes, types of errors), and various acoustic measures were collected to characterize the productions on the three types of segments investigated. Multiple factor analyses were conducted to study productive profiles on the various acoustic measures, and the dimensions were correlated with phonological measures.ResultsThe results showed lower performance in lexical (target word retrieval) and phonological (percentages of correct phonemes) skills among children with CIs (CI group), although with better performances among children exposed to CS. Acoustically, children in the CI group exhibited productions significantly different from those of the TH group in terms of the distinction of fricative consonants, marking nasalization through nasal resonance cues, and in the production of voiceless stops. However, the CI group demonstrated compensatory strategies (lengthening of VOT for voiced stops, marking of nasalization through oropharyngeal configuration cues).ConclusionsThe results indicate that children with CIs are at risk of experiencing difficulties in both phonetic and phonological domains. However, there are opportunities for compensation through the exploitation of acoustic cues better coded by the CI and/or through perceptual means (utilization of visual cues).

The effect of sexual orientation on voice acoustic properties.

Previous research has investigated sexual orientation differences in the acoustic properties of individuals' voices, often theorizing that homosexuals of both sexes would have voice properties mirroring those of heterosexuals of the opposite sex. Findings were mixed, but many of these studies have methodological limitations including small sample sizes, use of recited passages instead of natural speech, or grouping bisexual and homosexual participants together for analyses. To address these shortcomings, the present study examined a wide range of acoustic properties in the natural voices of 142 men and 175 women of varying sexual orientations, with sexual orientation treated as a continuous variable throughout. Homosexual men had less breathy voices (as indicated by a lower harmonics-to-noise ratio) and, contrary to our prediction, a lower voice pitch and narrower pitch range than heterosexual men. Homosexual women had lower F4 formant frequency (vocal tract resonance or so-called overtone) in overall vowel production, and rougher voices (measured via jitter and spectral tilt) than heterosexual women. For those sexual orientation differences that were statistically significant, bisexuals were in-between heterosexuals and homosexuals. No sexual orientation differences were found in formants F1-F3, cepstral peak prominence, shimmer, or speech rate in either sex. Recommendations for future "natural voice" investigations are outlined.

Increased vowel contrast and intelligibility in connected speech induced by sensorimotor adaptation.

Alterations to sensory feedback can drive robust adaptive changes to the production of consonants and vowels, but these changes often have no behavioral relevance or benefit to communication (e.g., making "head" more like "had"). This work aims to align the outcomes of adaptation with changes known to increase speech intelligibility - specifically, adaptations that increase the acoustic contrast between vowels in running speech. To this end, we implemented a vowel centralization feedback perturbation paradigm that pushes all vowels towards the center of vowel space, making them sound less distinct from one another. Speakers across the adult lifespan adapted to the centralization perturbation during sentence production, increasing the global acoustic contrast among vowels and the articulatory excursions for individual vowels. These changes persisted after the perturbation was removed, including after a silent delay, and showed robust transfer to words that were not present in the sentences. Control analyses demonstrated that these effects were unlikely to be due to explicit pronunciation strategies and occurred in the face of increasingly more rapid and less distinct production of familiar sentences. Finally, sentence transcription by crowd-sourced listeners showed that speakers' vowel contrast predicted their baseline intelligibility and that experimentally-induced increases in contrast predicted intelligibility gains. These findings establish the validity of a sensorimotor adaptation paradigm to implicitly increase vowel contrast and intelligibility in connected speech, an outcome that has the potential to enhance rehabilitation in individuals who present with a reduced vowel space due to motor speech disorders, such as the hypokinetic dysarthria associated with Parkinson's disease.

More foreign-accented but more comprehensible: Attrition and amelioration of L1 speech in proficient L2 learners.

There is an increasing interest in cross-linguistic influences of the second language (L2) on the first (L1), but its communicative impact remains to be elucidated. This study investigates how L2 learners' L1 pronunciation is perceived as foreign-accented and (in) comprehensible as a function of their L2 learning experience and proficiency levels. Read speech of 154 L1 Japanese learners of L2 English in the J-AESOP corpus was examined, where approximately one-third of them had lived in English-speaking countries and the rest had never lived outside of Japan. Their L1 speech was rated by another group of native Japanese listeners for accentedness and comprehensibility (from October 25, 2022 to August 20, 2023), while their L2 speech was previously rated by native American English listeners for nativelikeness or proficiency. The speakers' vowel acoustics were also examined. More proficient L2 speakers were perceived as more foreign-accented in their L1, but only if they had lived overseas; their length of residence abroad predicted the degree of perceived accentedness. In contrast, more proficient L2 speakers were consistently perceived as more comprehensible in the L1, regardless of prior overseas experience. Acoustic analyses indicated that perceived accentedness is associated with a clockwise chain shift of all vowel categories in the vowel space. It was also revealed that the dispersion, rather than compactness, of vowel production contributed to perceived comprehensibility, although the degree of L1 vowel dispersion did not predict L2 proficiency. The overall results suggest two main conclusions. First, perceptible L1 foreign accent likely results from L1 disuse rather than L2 interference, thereby L1 pronunciation differs from native norms at a system-wide rather than category-specific level. Second, L2 learning has a positive influence on perceived L1 comprehensibility, rather than individuals with clearer and more comprehensible L1 speech being inherently better L2 learners.

Comparative Formants Analysis of English Vowels Using PRAAT

This research focuses on the comparative formant analysis of English vowels between native and non-native speakers using PRAAT software. The study aims to investigate the comparison of the acoustic properties of English vowels to understand the pronunciation differences encountered by non-native speakers. By analyzing the formant frequencies of monophthongs in English vowels, specifically the first formant (F1), this research seeks to identify any significant differences between English native speaker and five non-native speakers, whose first language is Indonesia. The data collected from speech samples will be descriptively analyzed to determine variations in formant patterns. The finding shows that the F1 values of the vowels show variations among the non-native speakers compared to the native speaker. There are still monophthongs that these participants mispronounce, indicating that the learner's understanding is an obstacle to their fluency in speaking and pronouncing the sounds correctly. The research findings imply that these mispronunciations are linked to differences in tongue height, vowel backness, and vowel quality between native and non-native speakers. The results underscore the analysis of F1 values for 12 words containing vowel monophthongs, highlighting these variations. This suggests that the learner's understanding of these phonetic aspects is crucial for achieving fluency in vowel production and pronunciation.

Learning English vowels: The effects of different phonetic training modes on Arabic learners' production and perceptiona).

This study investigated the effect of different types of phonetic training on potential changes in the production and perception of English vowels by Arabic learners of English. Forty-six Arabic learners of English were randomly assigned to one of three high variability vowel training programs: Perception training (High Variability Phonetic Training), Production training, and a Hybrid Training program (production and perception training). Pre- and post-tests (vowel identification, category discrimination, speech recognition in noise, and vowel production) showed that all training types led to improvements in perception and production. There was some evidence that improvements were linked to training type: learners in the Perception Training condition improved in vowel identification but not vowel production, while those in the Production Training condition showed only small improvements in performance on perceptual tasks, but greater improvement in production. However, the effects of training modality were complicated by proficiency, with high proficiency learners benefitting more from different types of training regardless of training mode than lower proficiency learners.

Yanbian Korean speakers tend to merge /e/ and /ɛ/ when exposed to Seoul Korean

This study examined the vowel merger between the two vowels /e/ and /ɛ/ in Yanbian Korean. This sound change has already spread to Seoul Korean, particularly among speakers born after the 1970s. The aim of this study was to determine whether close exposure to Seoul Korean speakers leads to the neutralization of the distinction between the two vowels /e/ and /ɛ/. We recruited 20 Yanbian Korean speakers and asked them about their frequency of exposure to Seoul Korean. The exposure level of each participant was also recorded using a Likert scale. The results revealed that speakers with limited in-person interactions with Seoul Korean speakers exhibited distinct vowel productions within the vowel space. In contrast, those with frequent in-person interactions with Seoul Korean speakers tended to neutralize the two vowels, displaying considerably overlapping patterns in the vowel space. The relationship between the level of exposure to Seoul Korean and speakers’ vowel production was statistically confirmed by a linear regression analysis. Based on the results of this study, we speculate that the sound change in Yanbian Korean may become more widespread as Yanbian Korean speakers are increasingly exposed to Seoul Korean.

Acoustic Analyses of L1 and L2 Vowel Interactions in Mandarin–Cantonese Late Bilinguals

While the focus of bilingual research is frequently on simultaneous or early bilingualism, the interactions between late bilinguals’ first language (L1) and second language (L2) have rarely been studied previously. To fill this research gap, the aim of the current study was to investigate the production of vowels in the L1 Mandarin and L2 Cantonese of Mandarin–Cantonese late bilinguals in Hong Kong. A production experiment was conducted with 22 Mandarin–Cantonese bilinguals, as well as with 20 native Mandarin speakers and 21 native Cantonese speakers. Acoustic analyses, including formants of and Euclidean distances between the vowels, were performed. Both vowel category assimilation and dissimilation were noted in the Mandarin–Cantonese bilinguals’ L1 and L2 vowel systems, suggesting interactions between the bilinguals’ L1 and L2 vowel categories. In general, the findings are in line with the hypotheses of the Speech Learning Model and its revised version, which state that L1–L2 phonetic interactions are inevitable, as there is a common phonetic space for storing the L1 and L2 phonetic categories, and that learners always have the ability to adapt their phonetic space. Future studies should refine the data elicitation method, increase the sample size and include more language pairs to better understand L1 and L2 phonetic interactions.

Acoustic measurements and intelligibility of English vowels produced by Thai speakers

Effective English communication across different languages and cultures has become increasingly important in a globalised world. To gain a deeper understanding of the segmental features of Thai speakers of English, this study aims to investigate the production of English monophthongs and diphthongs by Thai speakers and examine how intelligible their pronunciations are to others. Data were collected through two tasks: a production task and a listening task. For the production task, three Thai female speakers were chosen as the participants. They were recorded reading a passage consisting of all targeted vowels, and a total of 180 English tokens were analysed. For the listening task, thirty Malaysian listeners were asked to evaluate the speech intelligibility of the English sounds produced by the three speakers. The duration and formant frequencies of English vowels produced were measured to investigate the characteristics of the monophthongs and diphthongs. The findings show that there were significant differences in the durations between long and short vowels of the monophthongs, but there were no significant differences in the vowel quality produced. The ROC (rate of change) of formant frequencies (F1 and F2) indicates that there was no formant movement for the diphthongs /eɪ/, /əʊ/, and /eə/. The three diphthongs showed similar characteristics as other ASEAN Englishes that are monophthongal as /e:/, /o:/, and /ɛ:/, respectively. These findings confirm that speakers’ L1 influences English vowel production. Although there were vowel variations in the characteristics of the vowels produced by the speakers, the intelligibility scores were high.

Phonetic shifts in bilingual vowels: Evidence from intersentential and intrasentential code-switching

Aims and objectives: While previous research on code-switching, defined as the alternation between two languages in a single interaction, has focused on syntactic sociopragmatic patterns, significantly less work has focused on the impact of code-switching on phonetics. Distinct types of code-switching have been previously identified, with the distinction between intersentential and intrasentential code-switches being among the most basic. Yet, existing work on the phonetics of code-switching has largely relied on single feature (i.e., voice onset time) and has exclusively examined one type of code-switch (i.e., intrasentential). The current study investigates the potential for cross-linguistic interference in vowel production and examines the potential role of different code-switch types: intersentential and intrasentential code-switching. Design/Methodology: Thirty-seven Korean–English bilinguals participated in a controlled reading paradigm, producing monolingual English, monolingual Korean, intersentential code-switches, and intrasentential code-switches. Data and analysis: Analysis leveraged differences in the Korean and English vowel inventories and focused on the production of English [æ] and the corresponding Korean vowel [e~ε]. Findings/Conclusions: Results indicated that code-switched English vowels shifted in the direction of Korean vowels with respect to both height and backness. Code-switch type was shown to be significant, with a greater degree of vowel shift for intrasentential than intersentential code-switching. Originality: This study adds to the limited prior literature examining the effect of code-switching on vowel quality and is one of the first studies to systematically examine the role of code-switching typology on cross-linguistic interference in the phonetic domain. Significance/Implications: First, this study adds to the limited prior research that specifically examines the effect of code-switching on phonetic production of vowel quality. Second, this study highlights the nuances of short-term cross-linguistic interference, and notably the possibility for differential outcomes of interference across different linguistic domains (e.g., phonetics and syntax), showing a clear role for code-switching typology at the phonetic level.

Acoustic and Articulatory Visual Feedback in Classroom L2 Vowel Remediation.

This paper presents L2 vowel remediation in a classroom setting via two real-time visual feedback methods: articulatory ultrasound tongue imaging, which shows tongue shape and position, and a newly developed acoustic formant analyzer, which visualizes a point correlating with the combined effect of tongue position and lip rounding in a vowel quadrilateral. Ten Czech students of the Swedish language participated in the study. Swedish vowel production is difficult for Czech speakers since the languages differ significantly in their vowel systems. The students selected the vowel targets on their own and practiced in two classroom groups, with six students receiving two ultrasound training lessons, followed by one acoustic, and four students receiving two acoustic lessons, followed by one ultrasound. Audio data were collected pre-training, after the two sessions employing the first visual feedback method, and at post-training, allowing measuring Euclidean distance among selected groups of vowels and observing the direction of change within the vowel quadrilateral as a result of practice. Perception tests were performed before and after training, revealing that most learners perceived selected vowels correctly already before the practice. The study showed that both feedback methods can be successfully applied to L2 classroom learning, and both lead to the improvement in the pronunciation of the selected vowels, as well as the Swedish vowel set as a whole. However, ultrasound tongue imaging seems to have an advantage as it resulted in a greater number of improved targets.

Movement-related cortical potential and speech-induced suppression during speech production in younger and older adults

With age, the speech system undergoes important changes that render speech production more laborious, slower and often less intelligible. And yet, the neural mechanisms that underlie these age-related changes remain unclear. In this EEG study, we examined two important mechanisms in speech motor control: pre-speech movement-related cortical potential (MRCP), which reflects speech motor planning, and speaking-induced suppression (SIS), which indexes auditory predictions of speech motor commands, in 20 healthy young and 20 healthy older adults. Participants undertook a vowel production task which was followed by passive listening of their own recorded vowels. Our results revealed extensive differences in MRCP in older compared to younger adults. Further, while longer latencies were observed in older adults on N1 and P2, in contrast, the SIS was preserved. The observed reduced MRCP appears as a potential explanatory mechanism for the known age-related slowing of speech production, while preserved SIS suggests intact motor-to-auditory integration.

Improving Second Language Vowel Production With Hand Gestures Encoding Visible Articulation: Evidence From Picture‐Naming and Paragraph‐Reading Tasks

AbstractThis study investigates whether audiovisual phonetic training with hand gestures encoding visible or nonvisible articulation features has a differential impact on learning second language sounds. Ninety‐nine Catalan–Spanish bilingual students were trained to differentiate English /æ/ and /ʌ/, which differ in the visible lip aperture and nonvisible tongue position, with training involving no gestures, gestures representing the lip aperture, or gestures representing the tongue position. Before, immediately after, and 1 week after the training, participants’ perception of the targets was assessed through a word‐identification task, and their production was tested through paragraph‐reading, picture‐naming, and word‐imitation tasks. Although all participants improved in perception and production, the lip hand gesture was more effective in adjusting lip aperture than the other two conditions in the paragraph‐reading and picture‐naming tasks. These results suggest that hand gestures encoding visible rather than nonvisible articulation features are more effective for improving second language pronunciation.

The relationship between L2 learners’ production and perception of English vowels: The role of native-speaker acoustic patterns in production

Despite the abundance of research on the relationship between second language (L2) learners’ production and perception of target-language contrasts, the nature and details of this connection remain unclear. The aim of this study was to extend our understanding of the relationship by investigating whether learners who can produce L2 vowels with the same acoustic properties as those used by native speakers of the target language also perceive the vowels more accurately. To this end, we examined the production and perception of two English vowel contrasts (tense /i/ vs. lax /ɪ/, mid /ε/ vs. low /æ/) in 29 native-speakers of American English and 33 L2 learners of English from three native-language backgrounds: Korean, Portuguese, and Spanish. We found that the L2 learners who produced distinctions between the target vowels using the same acoustic properties as do native speakers of English had significantly better perception scores for these vowels compared to the learners who distinguished the vowels using a pattern of acoustic properties that is not used by native speakers. This was also true when their patterns were compared to the learners who did not make any acoustic distinctions at all. The findings provide compelling evidence that L2 learners’ production patterns are linked to their perception skills.

Sound symbolism in manual and vocal responses: phoneme-response interactions associated with grasping as well as vertical and size dimensions of keypresses

It has been shown that reading the vowel [i] and consonant [t] facilitates precision grip responses, while [ɑ] and [k] are associated with faster power grip responses. A similar effect has been observed when participants perform responses with small or large response keys. The present study investigated whether the vowels and consonants could produce different effects with the grip responses and keypresses when the speech units are read aloud (Experiment 1) or silently (Experiment 2). As a second objective, the study investigated whether the recently observed effect, in which the upper position of a visual stimulus is associated with faster vocalizations of the high vowel and the lower position is associated with the low vowel, can be observed in manual responses linking, for example, the [i] with responses of the upper key and [ɑ] with lower responses. Firstly, the study showed that when the consonants are overtly articulated, the interaction effect can be observed only with the grip responses, while the vowel production was shown to systematically influence small/large keypresses, as well as precision/power grip responses. Secondly, the vowel [i] and consonant [t] were associated with the upper responses, while [ɑ] and [k] were associated with the lower responses, particularly in the overt articulation task. The paper delves into the potential sound-symbolic implications of these phonetic elements, suggesting that their acoustic and articulatory characteristics might implicitly align them with specific response magnitudes, vertical positions, and grip types.