Accent Recognition Research Articles

An Exploration of Dialect Recognition by researchers through examining non-linguists reactions to Spoken language: Kurdish dialects as an illustration

An Exploration of Dialect Recognition by researchers through examining non-linguists reactions to Spoken language: Kurdish dialects as an illustration

Why are Certain Accents Judged the Way they are? Decoding Qualitative Patterns of Accent Bias

While it is well-established that listeners judge accents different from their own, and that this listener bias has pervasive consequences to the speakers, we have only offered a cursory attention to understand the nature of this accent bias. This paper explores listeners’ judgments, ratings, and qualitative comments associated with psychosocial, personality and behavioral attributes of seven accents. Fifty-five respondents evaluated six different attributes of seven regional and foreign accents on a Likert scale. Means, Standard Deviations, and statistical significance of the difference in the respondent ratings from the Midwestern standard were computed, as well as qualitative analyses of the judgments were conducted. Results showed that perceptual biases for many attributes of the regional and foreign accents were significant at the 95% level. Results also showed that the accent recognition was consistent across a group of people from the same region, and the identification of the region of the accent was largely accurate. Qualitative analyses revealed that the underlying bases of listener judgments are typically borne out of the influence of media, TV, and movies. Besides character traits, the listeners also frequently inferred occupations/professions from the accents. Accent judgments are not reserved for foreign accents alone; regional accents are also subject to judgments and stereotypes. Findings provide important insights for clinicians and teachers working in the area of accent interventions, specifically in improving understanding about how listeners judge accents and how those messages need to be addressed in developing awareness and empowerment for clients in accent interventions. Other implications in the field of speech language pathology, education, social science, communication and business are also discussed.

Telugu Accent Recognition Using MFCC and GMM

Telugu Accent Recognition Using MFCC and GMM

The Role of Acoustic Distance and Sociolinguistic Knowledge in Dialect Identification.

Listeners are able to quite accurately distinguish between different dialects of their native language, but little is known about the process of dialect identification and the phonetic cues listeners use to identify someone’s regional origin. This study examines how different segments, acoustic between-dialect distance, and the listeners’ knowledge about a dialect contribute to this process. Native speakers of Grison and Zurich German were asked to categorise isolated words spoken by eight speakers of Grison and eight speakers of Zurich German. Stimuli contained either none, one, or two segmental cues to regional origin. The presence of one dialect-specific segment was enough to allow for an identification rate well above chance. Sensitivity measures and analysis of reaction time showed that the two dialect groups largely relied on the same segmental cues. Acoustic distance to the other dialect, quantified as Euclidean distance in the F1 × F2 vowel space, generally facilitated dialect identification, but interacted with native speakers’ knowledge about the dialects: in segments which listeners explicitly associated with one of the two dialects, acoustic distance facilitated dialect recognition to a larger extent than in segments in which listeners were not aware of dialectal variation. The results suggest that, depending on the listener’s prior knowledge about a dialect, acoustic variation is weighted differently. Further analysis showed that Zurich listeners were more sensitive to the dialect differences, responded faster, and presented a more marked own-dialect response bias than Grison listeners. These findings are in line with the status of Grison German as a marked dialect and Zurich German as a neutral dialect, and suggest that, depending on their own dialect’s status, listeners used different decision strategies.

Dialect recognition for low resource language using an adaptive filter bank

Dialect recognition of low resource languages is the next stage in the technological advancement in speech recognition. Traditional methods for dialects recognition such as mel frequency cesptral coefficients (MFCC) and discrete wavelet transform (DWT) work better for high resource languages, however, the performance is low when applied in low resource languages. This paper presents a new approach for Pashto dialects recognition using an adaptive filter bank with MFCC and DWT. In this approach, features are extracted using adaptive filter bank in MFCC and DWT followed by classification through hidden Markov model (HMM), support vector machines (SVM) and K-nearest neighbors (KNN). The results obtained from the proposed method are very satisfactory with an overall dialect recognition accuracy of [Formula: see text].

Simulation and detection of tamil speech accent using modified mel frequency cepstral coefficient algorithm

Automatic Speech reconstruction system is a topic of interest of many researchers. Since many online courses are come into the picture, so recent researchers are concentrating on speech accent recognition. Many works have been done in this field. In this paper speech accent recognition of Tamil speech from different zones of Tamilnadu is addressed. Hidden Markov Model (HMM) and Viterbi algorithms are very popularly used algorithms. Researchers have worked with Mel Frequency Cepstral Coefficients (MFCC) to identify speech as well as speech accent. In this paper speech accent features are identified by modified MFCC algorithm. The classification of features is done by back propagation algorithm.

Turkish Regional Dialect Recognition Using Acoustic Features of Voiced Segments

Turkish Regional Dialect Recognition Using Acoustic Features of Voiced Segments

Accent Identification in RP, CE, and GA By Native and Non-Native English Speakers

This study was conducted in order to determine various groups’ accuracy in identifying three major standard English accents. The main purpose of the experiment that was performed was to determine how well speakers familiar with these accents could tell them apart from other accents. It focused on comparing the test subjects’ recognition of Canadian English pronunciation with General American pronunciations. Received Pronunciation was used as a control as it is generally considered to differ much more from the standard North American varieties than they differ from one another.Within North America, the 'standard' accents of Canada and the US are quite similar. So similar, that one of the experiment’s hypotheses is that that despite being identified by linguists as different, many native speakers of the two dialects would have difficulty telling the difference themselves. The differences in the features of Canadian English (CE) and General American (GA) have been identified and studied by linguists before, but what this experiment sought to determine was the degree to which speakers of these dialects could tell them apart purely through listening.Canada is given some degree of stigma from the United States for its dialect, and has had fun poked at it for such Canadianisms as the use of 'eh?' and Canadian Raising- Americans will exaggerate the difference when illustrating it, saying “aboot” [ə'but] for about [əˈbʌʊt]. But how well can they perceive the difference when not already informed about the speaker's origins?An online survey was prepared, with audio clips or words in isolation and sentences, spoken by speakers of GA, CE, and RP, specifically using words that exhibited features that vary between the accents. This allowed us to examine subjects’ degree of recognition with, and without prosody, and to analyze the degree to which prosody affects accent recognition. In order to better determine how prior exposure influences accent recognition, the subjects were broken down into three groups: Native speakers of Canadian English, native speakers of American English and ESL speakers who had had prior exposure to Canadian English.One of the main findings of this experiment is that more than 80% of American respondents recognized their own national standard accent, and around 67% recognized the Canadian accent; while only 62% of the Canadian respondents recognized the Canadian accent accurately. Compared to Canadians, Americans were better at telling the North American varieties of spoken English apart.

THE PREFERRED MODEL OF ENGLISH FOR PERSONAL, NATIONAL, AND GLOBAL COMMUNICATION IN ELF CONTEXT: A COMPARATIVE STUDY BETWEEN FILIPINOS AND INDIANS

An experimental investigation was conducted to compare preferred models of English for personal, national, and global communication in ELF context between Filipinos and Indians. Drawing upon the data gathered through questionnaire and accent recognition survey from over 200 undergraduate students in these two nations, it was revealed that both nationals prefer British English for personal communication. But for national communication, Indians prefer their own standardized variety of English, i.e. Indian English, while the majority of Filipinos chose American English. British English and American English are considered to be the most ideal English varieties to express global identity by Indians and Filipinos, respectively. Finally, the degree of familiarity with the native English varieties which was higher among Filipinos than Indians is inversely proportional to level of awareness of their standard local English variety. Implications of the results and future research directions were also presented.

Dialect Identification Using Spectral and Prosodic Features on Single and Ensemble Classifiers

In this paper, investigation of the significance of spectral and prosodic behaviors of speech signal has been carried out for dialect identification. Spectral features such as cepstral coefficients, spectral flux, and entropy are extracted from shorter frames. Prosodic attributes such as pitch, energy, and duration are derived from longer frames. IViE (Intonational Variations in English) speech corpus covering nine dialectal regions of British Isles has been considered, to evaluate the proposed approach. Since corpus is available in both read and semi-spontaneous modes, the influence of spectral and prosodic behavior over these datasets is distinguishably articulated. Further, two distinct classification algorithms, namely support vector machine (SVM) and an ensemble of decision trees along with the SVM are used for identification of nine dialects. Dialect discriminating information captured from both features are used for constructing feature vectors. Experiments have been conducted on individual and combinations of features. A better dialect recognition performance is observed with ensemble methods over a single independent SVM.

American or British? L2 speakers’ recognition and evaluations of accent features in English

ABSTRACTRecent language attitude research has attended to the processes involved in identifying and evaluating spoken language varieties. This article investigates the ability of second-language learners of English in Spain (N = 71) to identify Received Pronunciation (RP) and General American (GenAm) speech and their perceptions of linguistic variation between these speech varieties. Data were gathered using a verbal-guise experiment in which respondents identified speakers’ places of origin and stated the reasons for their categorisations. Quantitative data analysis demonstrated high recognition rates for RP speakers, more often correctly identified than GenAm speakers. Qualitative data analysis showed that respondents’ knowledge of phonological variation informed the identification process and they often stated which linguistic features formed part of their mental representations of RP and GenAm. Additional resources informed accent recognition, including perceived linguistic quality, intelligibility, familiarity, and cultural associations. Patterns of misidentification revealed that, when GenAm was inaccurately identified as RP, it was ascribed high status. The findings provide an insight into the strategies, conceptual frameworks, and linguistic features which inform the accent identification process as performed by English-language learners in Spain. The results also highlight the usefulness of variety recognition items in interpreting attitudinal evaluations, especially with regard to patterns of misidentification.

“Co-certification”: a close encounter with ELF for an international examining board

Abstract Over the last decade ELF has become a reality in European universities, but this is not reflected in the major international language tests designed for access to higher education and for university students. In this paper I describe an experiment in “co-certification,” a test of English set at level C1 of the Common European Framework of Reference (CEFR), jointly developed by Trinity College London and the University of Ca’ Foscari Venice, in which an international version of the test was adapted locally to include locally relevant components and enhance validity and task authenticity. One feature of the co-certification was the introduction of an independent listening task which took the form of understanding an extract from a lecture given in English by a non-native speaker. Possible problems we had anticipated ranged from the kind of discourse features they should include, and whether or not there was a fairness issue related to accent recognition. In fact, the overall results showed no significant difference from the previous version of the exam, while the listening part had a higher pass rate than the overall oral test which involved interaction with a native speaker. As far as we know, the co-certification is the first example of an internationally recognized board engaging systematically with characteristic features of ELF communication, notably non-native phonology and intonation patterns, in a test of listening. The findings so far suggest that, for the test takers at least, this can be unproblematic and uncontroversial.

Automatic sociophonetics: Exploring corpora with a forensic accent recognition system.

This paper demonstrates how the Y-ACCDIST system, the York ACCDIST-based automatic accent recognition system [Brown (2015). Proceedings of the International Congress of Phonetic Sciences, Glasgow, UK], can be used to inspect sociophonetic corpora as a preliminary "screening" tool. Although Y-ACCDIST's intended application is to assist with forensic casework, the system can also be exploited in sociophonetic research to begin unpacking variation. Using a subset of the PEBL (Panjabi-English in Bradford and Leicester) corpus, the outputs of Y-ACCDIST are explored, which, it is argued, efficiently and objectively assess speaker similarities across different linguistic varieties. The ways these outputs corroborate with a phonetic analysis of the data are also discovered. First, Y-ACCDIST is used to classify speakers from the corpus based on language background and region. A Y-ACCDIST cluster analysis is then implemented, which groups speakers in ways consistent with more localised networks, providing a means of identifying potential communities of practice. Additionally, the results of a Y-ACCDIST feature selection task that indicates which specific phonemes are most valuable in distinguishing between speaker groups are presented. How Y-ACCDIST outputs can be used to reinforce more traditional sociophonetic analyses and support qualitative interpretations of the data is demonstrated.

Kurdish Dialects and Neighbor Languages Automatic Recognition

Dialect recognition is one of the most hot topics in the speech analysis area. In this study a system for dialect and language recognition is developed using phonetic and a style based features. The study suggests a new set of feature using one-dimensional LBP feature. The results show that the proposed LBP set of feature is useful to improve dialect and language recognition accuracy. The acquired data involved in this study are three Kurdish dialects (Sorani, Badini and Hawrami) with three neighbor languages (Arabic, Persian and Turkish). The study proposed a new method to interpret the closeness of the Kurdish dialects and their neighbor languages using confusion matrix and a non-metric multi-dimensional visualization technique. The result shows that the Kurdish dialects can be clustered and linearly separated from the neighbor languages.

Text Independent Amharic Language Dialect Recognition: A Hybrid Approach of VQ and GM

Text Independent Amharic Language Dialect Recognition: A Hybrid Approach of VQ and GM

Speech Processing for Text Independent Amharic Language Dialect Recognition

<span style="color: #666666; font-family: Verdana, Arial, Helvetica, sans-serif; font-size: 11.2px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: normal; letter-spacing: normal; orphans: 2; text-align: left; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; background-color: #ffffff; display: inline !important; float: none;">Dialect is a difference of verbal communication spoken by people from a particular society or geographic area so the paper focuses on Amharic language dialect recognition. In this paper, the authors have used backpropagation artificial neural network, VQ(vector quantization), (Gaussian Mixture Models) and a combination of GMM and backpropagation artificial neural network for classifying dialects of Amharic language speakers. In this research, a total of 100 speakers for each group of dialects are considered each having about 10 seconds duration is collected. The feature vectors of Mel frequency cepstral coefficients (MFCC) had been used to recognize the dialects of speakers. In this research paper the recognition model that uses a tanh activation function have a better result instead of using the Logistic Sigmoid activation function in backpropagation artificial neural network. After conducting the above experiments 95.7% accuracy achieved when GMM and backpropagation artificial neural network with tanh activation function are combined.</span>

Maghrebian dialect recognition based on support vector machines and neural network classifiers

This paper investigates the feed forward back propagation neural network (FFBPNN) and the support vector machine (SVM) for the classification of two Maghrebian dialects: Tunisian and Moroccan. The dialect used by the Moroccan speakers is called "La Darijja" and that of Tunisians is called "Darija". An Automatic Speech Recognition System is implemented in order to identify ten Arabic digits (from zero to nine). The implementation of our present system consists of two phases: The features extraction using a variety of popular hybrid techniques and the classification phase using separately the FFBPNN and the SVM. The experimental results showed that the recognition rates with both approaches have reached 98.3 % with FFBPNN and 97.5 % with SVM.

Perceptions of World Englishes Accents in English Phonetics Instruction of China

With the notion of World Englishes (WE) accepted in the academia, the past decade has witnessed ten studies or so conducted to scrutinize how English learners of China perceptually evaluate English varieties. Those studies, however, seldom explore the juncture between learners’ attitudes toward accent varieties of English and their expectations of English phonetics instruction. Additionally, they primarily draw quantitative data from surveys of large samples without providing in-depth viewpoints from learners themselves. This study aimed to investigate Chinese tertiary-level English learners’ attitudes toward WE accents and how they view accents situated in English phonetics instruction with cross-validated research methods including an accent recognition test, an attitudinal experiment, a questionnaire survey, and semi-structured interviews. Its participants were 64 sophomore English majors at a Chinese university; the accents examined were American English (AmE), British English (BrE), China English (ChE), Australian English, Indian English, and Korean English. Results indicate that students could recognize AmE and BrE better than other accents and that they identified with BrE most, but preferred to imitate AmE and welcome Americans as their phonetics teachers. Results also show students’ propensity for setting a native speaker norm, rather than adopting ChE as a model, at phonetics class. Based on the findings, the study concluded that nowadays teachers for English majors in China still need to adhere to the native speaker model in English phonetics instruction, and meanwhile, expose their students to various WE accents in order to facilitate their English phonetics learning and cultivate their awareness of WE.

Kin-directed food sharing promotes lifetime natal philopatry of both sexes in a population of fish-eating killer whales, Orcinus orca

The vast majority of social animals exhibit sex-biased dispersal as a strategy to reduce kin competition and avoid inbreeding. Piscivorous ‘resident’ killer whales, Orcinus orca, of the eastern North Pacific, however, are unusual in that both sexes remain philopatric throughout life, forming highly stable, multigeneration matrilines that are closed to immigration. We conducted a 12-year study documenting extensive cooperative prey sharing within these matrilines, and hypothesized that extreme natal philopatry in resident killer whales arose due to inclusive fitness benefits gained by provisioning maternal kin. We found that prey sharing was nonreciprocal, and even though whales routinely foraged in mixed associations containing multiple matrilines, prey sharing among individuals belonging to different matrilines was very infrequent. Furthermore, maternal relatedness was a significant predictor of the frequency of prey sharing between individuals, with close maternal kin sharing more often than distant relatives or nonkin. Adult females were much more likely to share prey than adult males or subadults, probably because they mainly provisioned their offspring. However, food sharing was not limited solely to maternal care; all age–sex classes engaged in this behaviour by sharing with close maternal relatives, such as siblings and mothers. We also investigated the frequency of prey sharing between mothers and their offspring as a function of offspring sex and age, and found that maternal food sharing with daughters declined after daughters reached reproductive maturity, which could help to explain matriline fission events. The evolution of kin-directed food sharing requires the ability to reliably discriminate kin, which resident killer whales likely achieve through social familiarity and vocal dialect recognition. We propose that lifetime philopatry of both sexes has been selectively favoured in this population due to the inclusive fitness benefits of kin-directed food sharing, a cooperative behaviour that may also inhibit dispersal by reducing resource competition among kin.

Perceptual dialectology in northern England: Accent recognition, geographical proximity and cultural prominence

Work in perceptual dialectology has argued that listeners’ successful identification of accent areas is facilitated by their geographical proximity to a particular region. Montgomery (, ) argues that this proximity effect is mediated by a ‘cultural prominence’ effect, where localities of high cultural salience seem less distant. We explore these claims using an online survey in which listeners were asked to identify the regional origin of speakers of five accents from England's ‘linguistic north’ (Liverpool, Manchester, Crewe, Stoke‐on‐Trent and Macclesfield) from audio clips of four different sentences. We show that the proximity and cultural prominence effects are partially supported, and that the proximity effect is driven by listeners’ likely contact with/experience of the dialect in question. We also show that listeners react differently to different sentences in the same accent. While some sentences are identified correctly very often, others, even for culturally prominent locations with distinctive accents (i.e. Liverpool), are hardly ever correctly identified. We connect this to the geographical distribution of the linguistic features in the clips, and we argue that the presence, absence and range of individual linguistic features should be systematically considered in all perceptual dialectology work which uses audio stimuli.