Accent Ratings Research Articles

Converting Foreign Accent Speech Without a Reference

Foreign accent conversion (FAC) is the problem of generating a synthetic voice that has the voice identity of a second-language (L2) learner and the pronunciation patterns of a native (L1) speaker. This synthetic voice has been referred to as a “golden-speaker” in the pronunciation-training literature. FAC is generally achieved by building a voice-conversion model that maps utterances from a source (L1) speaker onto the target (L2) speaker. As such, FAC requires that a reference utterance from the L1 speaker be available at synthesis time. This greatly restricts the application scope of the FAC system. In this work, we propose a “reference-free” FAC system that eliminates the need for reference L1 utterances at synthesis time, and transforms L2 utterances directly. The system is trained in two steps. First, a conventional FAC procedure is used to create a golden-speaker using utterances from a reference L1 speaker (which are then discarded) and the L2 speaker. Second, a pronunciation-correction model is trained to convert L2 utterances to match the golden-speaker utterances obtained in the first step. At synthesis time, the pronunciation-correction model directly transforms a novel L2 utterance into its golden-speaker counterpart. Our results show that the system reduces foreign accents in novel L2 utterances, achieving a 20.5% relative reduction in word-error-rate of an American English automatic speech recognizer and a 19% reduction in perceptual ratings of foreign accentedness obtained through listening tests. Over 73% of the listeners also rated golden-speaker utterances as having the same voice identity as the original L2 utterances.

Short-Term Sources of Cross-Linguistic Phonetic Influence: Examining the Role of Linguistic Environment

While previous research has shown that bilinguals are able to effectively maintain two sets of phonetic norms, these two phonetic systems experience varying degrees of cross-linguistic influence, driven by both long-term (e.g., proficiency, immersion) and short-term (e.g., bilingual language contexts, code-switching, sociolinguistic) factors. This study examines the potential for linguistic environment, or the language norms of the broader community in which an interaction takes place, to serve as a source of short-term cross-linguistic phonetic influence. To investigate the role of linguistic environment, late bilinguals (L1 English—L2 Spanish) produced Spanish utterances in two sessions that differed in their linguistic environments: an English-dominant linguistic environment (Indiana, USA) and a Spanish-dominant linguistic environment (Madrid, Spain). Productions were analyzed at the fine-grained acoustic level, through an acoustic analysis of voice onset time, as well as more holistically through native speaker global accent ratings. Results showed that linguistic environment did not significantly impact either measure of phonetic production, regardless of a speaker’s second language proficiency. These results, in conjunction with previous results on long- and short-term sources of phonetic influence, suggest a possible primacy of the immediate context of an interaction, rather than broader community norms, in determining language mode and cross-linguistic influence.

Does Teaching Your Native Language Abroad Increase L1 Attrition of Speech? The Case of Spaniards in the United Kingdom

The present study examines the perceived L1 accent of two groups of native Spaniards in the United Kingdom, Spanish teachers, and non-teachers, alongside monolingual controls in Spain. While the bilingual groups were carefully matched on a range of background variables, the teachers used Spanish significantly more at work where they constantly need to co-activate it alongside English. This allowed us to test the relative effect of reduced L1 use and dual language activation in first language attrition directly. To obtain global accentedness ratings, monolingual native Spanish listeners living in Spain participated in an online perception experiment in which they rated short speech samples extracted from a picture-based narrative produced by each speaker in terms of their perceived nativeness, and indicated which features they associated with non-nativeness. The results revealed significantly greater foreign-accent ratings for teachers than non-teachers and monolinguals, but no difference between the latter two. Non-native speech was associated with a range of segmental and suprasegmental features. These results suggest that language teachers who teach their L1 in an L2-speaking environment may be particularly prone to L1 attrition since they need to co-activate both their languages in professional settings and are regularly exposed to non-native speech from L2 learners.

Accent familiarity and adverse listening conditions

Our ability to understand spoken language depends on both the speaker and listening environment. Speech intelligibility is higher in more favorable listening environments. Accent familiarity also plays a major role, with familiar accents being more intelligible than unfamiliar accents. Previous research [S. Gittleman and K. J. Van Engen, “Effects of noise and talker intelligibility on judgments of accentedness,” J. Acoust. Soc. Am. 143(5), 3138–3145 (2018)] found a significant relationship between the level of noise and perceived “foreignness” of an accent for Mandarin-accented English and locally accented English speakers. Their results show that when participants listened to Mandarin-accented speakers they rated them as less accented in lower signal-to-noise ratios (SNR). However, when participants listened to local English speakers, they rated the speakers as more accented in lower SNRs. The current study replicates these procedures using locally accented English (n = 5) and Hawaiian Pidgin English (n = 5). In a within-listener design, listeners complete an intelligibility task and an accentedness rating task. In the accent rating task, listeners are presented with sentences in four conditions: clear, + 4 db SNR, 0 db SNR, and −4 db SNR. These results will contribute to our understanding of familiarity in speech processing.

The Impact of Alcohol on L1 versus L2.

Alcohol intoxication is known to affect many aspects of human behavior and cognition; one of such affected systems is articulation during speech production. Although much research has revealed that alcohol negatively impacts pronunciation in a first language (L1), there is only initial evidence suggesting a potential beneficial effect of inebriation on articulation in a non-native language (L2). The aim of this study was thus to compare the effect of alcohol consumption on pronunciation in an L1 and an L2. Participants who had ingested different amounts of alcohol provided speech samples in their L1 (Dutch) and L2 (English), and native speakers of each language subsequently rated the pronunciation of these samples on their intelligibility (for the L1) and accent nativelikeness (for the L2). These data were analyzed with generalized additive mixed modeling. Participants’ blood alcohol concentration indeed negatively affected pronunciation in L1, but it produced no significant effect on the L2 accent ratings. The expected negative impact of alcohol on L1 articulation can be explained by reduction in fine motor control. We present two hypotheses to account for the absence of any effects of intoxication on L2 pronunciation: (1) there may be a reduction in L1 interference on L2 speech due to decreased motor control or (2) alcohol may produce a differential effect on each of the two linguistic subsystems.

Pre-service teachers’ beliefs about second language pronunciation teaching, their experience, and speech assessments

Teacher cognition has attracted increased attention among second language (L2) researchers and practitioners, likely because of its potential consequences for classroom practices, such as teaching and assessment. Prior research has revealed links between teacher beliefs about pronunciation teaching and teachers’ own experience (e.g. amount of teacher training and teaching experience). However, no research has to date focused on how teachers’ experience and their beliefs are intertwined, possibly affecting teacher assessments of L2 speakers’ pronunciation. For this study, 77 Japanese pre-service teachers of English completed an online questionnaire examining their beliefs about the teaching of English pronunciation and eliciting details about their L2 teaching and learning experience. Additionally, pre-service teachers assessed 40 Japanese secondary school students performing an extemporaneous speech task, rating these speakers for comprehensibility, accentedness, and fluency. Results showed that pre-service teachers could be categorized into two distinct profiles, defined by joint contributions of pre-service teachers’ experience (a mixture of language learning/teaching experience and pronunciation-related instruction) and their beliefs (teachability of L2 pronunciation and approaches to its teaching). Pre-service teachers with more experience appeared to be more skeptical about how (easily) L2 pronunciation can be learned and taught and also provided harsher accentedness ratings, compared to pre-service teachers with less experience, revealing links between experience, beliefs, and speech assessments. Taken together, the findings reveal how pre-service teachers’ experience might shape their beliefs and assessments, implying that teacher educators must encourage future teachers to hold positive views about the teachability of L2 pronunciation by shifting their attention toward communicatively oriented dimensions of L2 speech and by providing teachers with pedagogical skills to target these dimensions.

Tuning out tone errors? Native listeners do not down-weight tones when hearing unsystematic tone errors in foreign-accented Mandarin

AbstractListeners can adapt to errors in foreign-accented speech, but not all errors are alike. We investigated whether exposure to unsystematic tone errors in second language Mandarin impacts responses to accurately produced words. Native Mandarin speakers completed a cross-modal priming task with words produced by foreign-accented talkers who either produced consistently correct tones, or frequent tone errors. Facilitation from primes bearing correct tones was unaffected by the presence of tone errors elsewhere in the talker's speech. However, primes bearing tone errors inhibited recognition of real words and elicited stronger accentedness ratings. We consider theoretical implications for tone in foreign-accent adaptation.

CAN CAPTIONED VIDEO BENEFIT SECOND LANGUAGE PRONUNCIATION?

AbstractThe present study investigated the potential benefits of extended exposure to captioned videos for second language pronunciation. We tested 90 L2 adult learners of English on speech processing skills (segmentation, speed of lexical access, and sentence processing) and phonological accuracy in perception (ABX discrimination) and production (accentedness ratings) before and after an 8-week treatment consisting of regular exposure to audiovisual materials. Participants were randomly assigned to four experimental conditions involving two viewing modes (captioned or uncaptioned) and two task focus conditions (focus on phonetic form or focus on meaning). Results showed benefits in speech segmentation and speech processing skills irrespective of viewing mode. No significant benefits were found for phonological accuracy in perception. In production, a focus on phonetic form improved pronunciation only in the absence of captions, whereas captioned viewing led to pronunciation gains as long as there was no focus on phonetic form. These findings suggest that pronunciation improvement can take place with the help of captions or, in the absence of captions, when learners’ attention is directed to pronunciation. Cognitive overload might explain why no benefits were obtained when attention was directed to pronunciation in a captioned viewing mode.

What Makes Spanish-Accented English Sound Spanish-Accented? Acoustic Measures and Listener Cues

This study aimed to understand which acoustic parameters of Spanish-accented English are correlated with listeners’ perception of Spanish-accentedness. Temporal differences were analyzed in multisyllabic target words spoken in sentences by 22 Spanish speakers of English and five native speakers of American English (AE). Recordings were presented to AE listeners who judged the degree of accentedness on a 9-point scale. Spearman rank order correlation showed that the listeners’ ratings of degree of accentedness in sentences correlated strongly (r= +0.82) with those in words. Listeners’ ratings of accentedness correlated in varying degrees with various temporal measures, namely Overall word durations (+0.04 to +0.56), Stressed/unstressed vowel duration ratios (–0.01 to +0.35), Voice Onset Time of stops (+0.26 to +0.36), and, closure duration (+0.29 to +0.59). Results suggest that Spanish-accented English is characterized by systematic temporal differences from native AE, and that these temporal differences contribute to the perception of accentedness. Implications of findings in improving theoretical understanding and applied practices are discussed.

Does personality influence ratings of foreign accents?

Abstract In both research and educational settings, native and non-native speakers are often asked to make foreign accent ratings (FARs) as a measure of second language pronunciation. However, previous research has identified several factors that influence these ratings. The current study investigates one such variable that, to date, has received little attention in the literature: personality. Thirty-six monolingual English speakers completed the Big Five Aspects Scales personality test (DeYoung, Quilty, & Peterson, 2007) and provided accentedness ratings for 15 non-native English speakers (L1 Mandarin) and five native controls. Results show that two of the Big Five personality traits – conscientiousness and extraversion – were significantly correlated with the ratings listeners provided, while another trait – agreeableness – approached significance. These findings further underline the need to interpret FARs with caution, as variables unrelated to foreign accent, in this case listeners’ personality, may be associated with these ratings, as well.

Developing and validating a methodology for crowdsourcing L2 speech ratings in Amazon Mechanical Turk

AbstractResearchers have increasingly turned to Amazon Mechanical Turk (AMT) to crowdsource speech data, predominantly in English. Although AMT and similar platforms are well positioned to enhance the state of the art in L2 research, it is unclear if crowdsourced L2 speech ratings are reliable, particularly in languages other than English. The present study describes the development and deployment of an AMT task to crowdsource comprehensibility, fluency, and accentedness ratings for L2 Spanish speech samples. Fifty-four AMT workers who were native Spanish speakers from 11 countries participated in the ratings. Intraclass correlation coefficients were used to estimate group-level interrater reliability, and Rasch analyses were undertaken to examine individual differences in rater severity and fit. Excellent reliability was observed for the comprehensibility and fluency ratings, but indices were slightly lower for accentedness, leading to recommendations to improve the task for future data collection.

Working memory and second language accent acquisition

SummaryWe explored the proposal that overt repetition of verbal information improves the acquisition of a native accent in a second language. Mandarin‐speaking Chinese learners of English were recorded while repeating and reading out English sentences before and after one of three treatments: (a) repeating native English sentences subvocally, “covert repetition,” (b) repeating sentences out loud, “overt repetition,” and (c) unfilled time of comparable duration. The sentences were rated by English speakers for their nativeness, fluency, and intelligibility. Overt repetition improved accent rating for readout sentences. Covert repetition did not. Neither condition improved accent rating for repeated sentences, suggesting that immediate repetition depends on temporary rather than long‐term representations. Our results provide some support for the use of overt repetition in accent learning. From a theoretical perspective, an interpretation is proposed in terms of a separation between phonological and articulatory coding within the phonological loop component of working memory.

Trinidadian secondary school students’ attitudes toward accents of Standard English

Abstract In the anglophone Caribbean, tendencies of endonormative reorientation have been observed in the development of local standards of English. Situated in the school context, this study adds a language attitude perspective on the question of whether and to what extent an endonormative standard of English is emerging in the island of Trinidad. In an accent rating study, 803 secondary students were asked to evaluate the accents of Trinidadian, other anglophone Caribbean, American, and British teachers and to identify their countries of origin. The results indicate that the respondents’ norm orientation is multidimensional and includes exo- and endonormative influences: first, there is a general coexistence of different standards since no standard serves as a superordinate norm. Second, there is no clear-cut distinction between exo- and endonormative accents, but fine-grained differences in the ratings: British and American voices received slightly higher ratings than local ones, but an American-influenced Trinidadian voice was also highly appreciated. These findings provide some new perspectives for evolutionary models of World Englishes and new insights for the discussion of standards in Trinidad, the wider anglophone Caribbean, and other small postcolonial speech communities where different local and global norms interact.

Effects of pitch contour and speaking rate on perception of foreign-accented speech

Perception of foreign accent is typically studied using an accentedness rating task. For example, native English listeners rate the degree of accentedness in sentences produced by non-native English speakers. However, in the past studies, it has been unclear that on what criteria participants used to judge accentedness. Here, native English speakers rated the accentedness of Korean-accented English sentences on a scale from 1 (strong accent) to 9 (little to no accent). Participants rated sentences that were unmodified or had one acoustic property removed. In one block, pitch contours of sentences were flattened and set to their mean values. In another block, speaking rates were set to the grand mean of all speaking rates (3.8 syllables/second). This way, changes in accentedness ratings across unmodified and modified sentences were attributable to the acoustic property that was removed. Accentedness ratings were not systematically influenced by manipulations of pitch contours but were influenced by speaking rate manipulations. Increasing the speaking rate (to 3.8 syllables/second) made sentences sound less accented than their unmodified versions; decreasing the speaking rate made sentences sound more accented than their unmodified versions. Results suggest that the speaking rate directly contributes to ratings of foreign accentedness.

Italian-German bilinguals: The effects of heritage language use on accent in early-acquired languages

Aims and objectives/purpose/research questions:This study investigates perceived accent in the two early-acquired languages of 21 adult-aged bilinguals with Italian as the heritage language (HL) and German as the majority language (ML). We test the relative ability of ‘age of onset (AoO) in German’ (range = 0–6 years) and ‘Italian use’ to predict perceived nativeness in Italian and German.Design/methodology/approach:Two accent rating experiments were carried out (one in each language) comparing the bilingual speech samples to those of monolingual and second language (L2) control groups. The samples were rated by German and Italian-speaking judges for foreign accent (‘yes’ or ‘no’) and for degree of certainty (‘certain, ‘semi-certain’, ‘uncertain’).Data and analysis:The effects of ‘AoO in German’ and ‘Italian Use’ (operationalized as an Italian Use Score) were analysed using correlational analyses and logistic regression.Findings/conclusions:Our results show that almost all bilinguals were indistinguishable from monolingual controls in German, and that their perceived accent in Italian lay somewhere between that of the monolingual and L2 controls. Based on regression analyses, we conclude that a later introduction of the ML has neither a negative effect on the ML itself, nor does it show up advantages in the HL. Instead, how native-like the heritage speakers (HSs) sound in the HL largely depends on HL use. No negative effects of HL use are found for German.Originality:We test a new population of Italian HSs in South Germany and compare these results to findings from other HS populations in Germany.Significance/implications:This study shows that, although it may be the case for child-aged bilinguals that introducing the ML earlier causes more phonological pressure on the HL, this effect seems to have disappeared by adulthood – at least with regards to perceived global accent. Also, accent in the HL depends on HL use across the lifespan.

Hand-clapping to the rhythm of newly learned words improves L2 pronunciation: Evidence from training Chinese adolescents with French words

Though research has shown that rhythmic training is beneficial for phonological speech processing, little empirical work has been carried out to assess whether rhythmic training in the classroom can help to improve pronunciation in a second language. This study tests the potential benefits of hand-clapping to the rhythm of newly learned French words for the acquisition of pronunciation patterns by Chinese adolescents. In a between-subjects training experiment with a pretest/posttest design, 50 Chinese adolescents either repeated new French words while clapping out their rhythmic structure (clapping condition) or only repeated the words (non-clapping condition). Participants’ oral production before and after training was (1) perceptually rated for accentedness by two French native speakers and (2) acoustically analysed for final syllable duration. While the results showed an only near-significant improvement from pretest to posttest in accentedness ratings for the clapping group, a significant improvement was obtained for acoustic durational measures. Individual musical abilities did not interact significantly in either of the two analyses, and working memory interacted significantly only with accentedness. These results show that a short training session in which clapping is used to highlight the prosodic structure of words can help improve pronunciation in a foreign language.

Exploring the skill of being "good with accents": A study of monolingual English speakers' ability to reproduce a novel foreign accent

We collected production data from monolingual English speakers who were trained to produce a foreign accent of English, specifically Russian English (RE). First, the subjects read a paragraph in their own accent (baseline). Second, they listened to recordings of English sentences produced by a native speaker of Russian; next, the sentences were replayed, and the subjects imitated each sentence after hearing it. Finally, they were instructed to read the baseline sentences again (without audio prompts), trying their best to reproduce the RE accent. Several acoustic analyses are underway, addressing (a) vowel characteristics, (b) realization of stops, and (c) intonation patterns. Preliminary results show considerable differences in speakers' performance. The questions we address based on the data collected are (1) is there a continuous or more categorical change in speakers' ability to reproduce the RE accent? (2) which aspects of RE are most salient to native English speakers?, and (3) is there a correlation in quality and quantity such that speakers who pick up on more cues also produce them in more native-like manner? Our goal is to develop an algorithm to predict a given speaker's accentedness, and compare its performance with native speakers' accentedness ratings in a follow-up perceptual experiment. We collected production data from monolingual English speakers who were trained to produce a foreign accent of English, specifically Russian English (RE). First, the subjects read a paragraph in their own accent (baseline). Second, they listened to recordings of English sentences produced by a native speaker of Russian; next, the sentences were replayed, and the subjects imitated each sentence after hearing it. Finally, they were instructed to read the baseline sentences again (without audio prompts), trying their best to reproduce the RE accent. Several acoustic analyses are underway, addressing (a) vowel characteristics, (b) realization of stops, and (c) intonation patterns. Preliminary results show considerable differences in speakers' performance. The questions we address based on the data collected are (1) is there a continuous or more categorical change in speakers' ability to reproduce the RE accent? (2) which aspects of RE are most salient to native English speakers?, and (3) is there a correlatio...

Implicit biases in monolingual and bilingual speech perception

Accented speech shows a great deal of variation from the “native” norms that can lead to processing difficulty. Previous research on listeners’ performance on accented speech shows that the more varieties of accented speech the listeners are exposed to, the better their processing is (Baese-Berk, Bradlow, & Wright, 2013). Since many bilingualism studies suggest that bilinguals have advantages in allocating different resources such as attention during language processing (Costa, Hernández, & Sebastián-Gallés, 2008), there is a necessity to investigate bilingual listeners' processing of accented speech. In this pilot experiment, we will measure intelligibility and perceived degrees of accentedness of Tamil and British English presented in noise among American English monolinguals, and American English and Spanish Bilinguals. We will also perform different cognitive tests as well as an Implicit Association Test to see if prior implicit biases towards South Asian individuals affect the listeners’ intelligibility and accentedness rating scores. We hypothesize that bilinguals will be more accurate in the intelligibility task compared to monolinguals due to their exposure to more than one language. However, it remains to be seen whether this advantage will be equally realized in both varieties of English accented speech.

Effect of proficiency on perception of English nasal stops by native speakers of Brazilian Portuguese

Brazilian Portuguese (BP) does not contrast /m/ and /n/ in coda position, leading to perceptual difficulties with minimal pairs like lime/line in L2 English. This study examined the effect of acoustic-phonetic and semantic enhancements on word recognition in noise for native English (n = 26) and L1 BP/L2 English (n = 21) listeners. They heard 120 high- and low-predictability sentences produced in clear and conversational speaking styles and mixed with speech-shaped noise. They typed the last word ending in either /m/ or /n/. The L2 listeners were separated into high- and low-proficiency (HP and LP) groups based on the perceived accentedness ratings of speech samples they provided. The results showed that native listeners benefited more from semantic than acoustic-phonetic enhancements alone, and the accuracy was highest when both enhancements were combined. The HP L2 listeners recognized more target words overall than the LP group, and their benefit from semantic over acoustic-phonetic enhancements mirrored that of the native listeners. In contrast, LP listeners could not utilize the higher-level semantic information to the same degree in the absence of signal clarity. These findings show that L2 listeners benefited from acoustic-phonetic, semantic, and combined enhancements when recognizing /m/-/n/ coda pairs, but this effect was mediated by proficiency.

L1 listeners’ perception of L2 pronunciation: effect of prosody on accentedness ratings in Swedish


 
 
 This paper reports on a study in which L1 speakers (n = 53) of Swedish rated the accentedness of read-aloud sentences produced by L2 speakers (n = 6) of Swedish. The aim of the study was to investigate the effect of certain prosodic features on the accentedness ratings. These features include Swedish tone accents and and deviant speech rhythm (e.g. lack of primary stress, and realization of multiple stresses). The results reveal that utterances with deviant speech rhythm yielded perceptions of stronger foreign accent than utterances with target-like rhythm. As for the tone accents, their effect on the scores were non-existent or only marginal. As missing tone accents have been previously found not to compromise intelligibility, the results of the present study reinforce the conclusion according to which emphasis in teaching should be put on prosodic features other than tone accents, especially rhythm.