Spectral moment analysis has helped further our understanding of the spectral properties of obstruent speech production; however, the physiologic correlates of these spectral measures are not well understood. The aim of the present study was to examine the possible correlations between the linguapalatal contact patterns used to produce the stops /t/ and /k/ and the resulting spectral characteristics. Using spectral moment analysis and electropalatography, the real-word productions of eight speakers of American English were investigated. The spectral measures for the stop consonant tokens in the present study were found to be similar to data reported in previous research with adult speakers. The majority of the correlations examined in this study were found to be statistically insignificant, although significant correlations were found between the anterior vertical and posterior vertical indices with spectral variance and spectral skewness, respectively. Despite the significance of these correlations, this did not account for a large proportion of variance in the data. Further analysis using curve estimates revealed significant curvilinear relationships among the data. These findings may indicate that although the anterior-posterior tongue placement and symmetry of linguapalatal contact contribute to the spectral signature of /t/ and /k/ stop consonants, this articulatory movement is only part of a more complex process that may involve aerodynamic factors and the overall shape of the vocal tract.

No description available.

Electropalatography (EPG) has proven to be a useful clinical and research tool for measuring tongue-to-palate contact. Research has shown sensorimotor adaptation to an EPG device may be possible following a short period of speech practice. This study was developed in order to better understand how a listeners' perception of speech clarity is effected by the presence of a relatively thin artificial pseudopalate in the speakers' oral cavity. Twenty listeners rated 220 speech stimuli on a visual analog scale ranging from normal to very distorted speech clarity. The stimuli included two different American English sentences. Speech clarity ratings were looked at as a function of the gender of the listener, the gender of the speaker, the type of speech sounds being heard, and the ability of the speakers to adapt their articulatory patterns over a period of 20 minutes. The results indicated that with the pseudopalate in place male speakers were generally rated by the listeners as having more distorted speech articulation than female speakers, especially for stop-loaded sentences. Overall, fricative-loaded sentences received higher articulation ratings than stop-loaded sentences. Finally, an adaptation period of 20 minutes showed significant improvement in speech articulation in comparison to ratings immediately following pseudopalate placement, however speech remained significantly distorted.

No description available.

Second language (L2) learners of English must learn to produce English phonemes, words, and sentences. These L2 learners make many errors when learning English; they may change the place or manner of articulation, insert vowels, or delete consonants. Obstruent sounds, such as fricatives, affricates, and stops, can be especially difficult for L2 learners. This study analyzed native English speakers<'> perception of the quality of obstruents produced by native Mandarin Chinese and Korean speakers. Target words containing obstruents had been produced in three different tasks: in a carrier phrase, in a paragraph, and in a spontaneous speech sample. Obstruents were produced in word-initial position and word-final position. Raters with differing levels of expertise listened to these words and rated the perceptual quality of the obstruents within the words. This study found that overall, English obstruent productions by native Mandarin and Korean L2 speakers learning English were rated most clear when produced in word-initial position in a carrier phrase or a paragraph. The lowest ratings given were of obstruents in word-final position in spontaneous speech. No significant differences were found for listener expertise level. Combined with future research, results from this study will help educate the field of second language instruction as to how the speech of Korean and Mandarin learners of English is perceived. It also provides additional information on the effect that listener expertise has on the judgment of L2 speech production.

This dissertation investigates the role of grammatical versus methodological influences in the production and perception of final devoicing in experimentally-elicited speech from Russian. It addresses the question of how the partial preservation of the phonological voicing contrast in word-final obstruents is affected by (i) task-independent factors that reflect phonological and lexical properties of stimuli words (underlying voicing, word length, lexical competition) and (ii) task-dependent biases that arise due to the nature of the experimental task performed by the speaker (availability of orthographic inputs, presence of minimal pairs among the stimuli). Results of a series of acoustic production and perceptual identification tasks reveal that task-dependent factors account for the presence of robust and perceptually salient differences in the parameter of phonetic voicing. Several types of stimuli items also show limited but statistically significant differences in closure/frication duration and release duration that are independent of the presence of orthography or inclusion of full minimal pairs among test items. Taken together, these findings indicate that non-grammatical factors can play a prominent biasing role in both production and perception of the voicing contrast in experimentally-elicited speech, such that certain voicing-dependent cues are maintained only in the presence of task-dependent pressures. However, not all incompletely neutralized differences between phonologically voiced versus voiceless final obstruents can be attributed to the effects of orthography or inclusion of minimal pairs among the stimuli. In the theoretical domain, these results are argued to favour a less restrictive definition of neutralization and a model of phonology that views devoicing as a loss of the primary acoustic cue to the underlying voicing contrast rather than complete identity of the [voiced] feature.

This dissertation investigates the role of grammatical versus methodological influences in the production and perception of final devoicing in experimentally-elicited speech from Russian. It addresses the question of how the partial preservation of the phonological voicing contrast in word-final obstruents is affected by (i) task-independent factors that reflect phonological and lexical properties of stimuli words (underlying voicing, word length, lexical competition) and (ii) task-dependent biases that arise due to the nature of the experimental task performed by the speaker (availability of orthographic inputs, presence of minimal pairs among the stimuli). Results of a series of acoustic production and perceptual identification tasks reveal that task-dependent factors account for the presence of robust and perceptually salient differences in the parameter of phonetic voicing. Several types of stimuli items also show limited but statistically significant differences in closure/frication duration and release duration that are independent of the presence of orthography or inclusion of full minimal pairs among test items. Taken together, these findings indicate that non-grammatical factors can play a prominent biasing role in both production and perception of the voicing contrast in experimentally-elicited speech, such that certain voicing-dependent cues are maintained only in the presence of task-dependent pressures. However, not all incompletely neutralized differences between phonologically voiced versus voiceless final obstruents can be attributed to the effects of orthography or inclusion of minimal pairs among the stimuli. In the theoretical domain, these results are argued to favour a less restrictive definition of neutralization and a model of phonology that views devoicing as a loss of the primary acoustic cue to the underlying voicing contrast rather than complete identity of the [voiced] feature.

Orientador: Eleonora Cavalcante Albano / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Estudos da Linguagem / Made available in DSpace on 2018-08-16T22:07:52Z (GMT). No. of bitstreams: 1 Rinaldi_LarissaMary_M.pdf: 5524182 bytes, checksum: 491c806e0e4d3ade9213ecaf37e85131 (MD5) Previous issue date: 2010 / Resumo: Esta dissertação pretende criar uma base de dados de referência da fala infantil, para o estudo de vogais e obstruintes do Português Brasileiro (doravante PB). Os sujeitos têm de 5 a 7 anos de idade, sem história de patologias de fala. Para isso, desenvolveu-se uma ferramenta lúdica para a coleta da fala dos sujeitos. A metodologia é centrada em uma história infantil, intitulada "Deu a louca nos Contos de Fadas", e usa um jogo de percurso de tabuleiro para gerar a emissão de 57 palavras criadas para o instrumento. O total de palavras alvo é 36. Elas foram criadas pondo-se em correspondência cada obstruinte do PB (fricativas e oclusivas) com as vogais [a], [i] e [u]. A metodologia permite a coleta de todos os sons do PB na posição de ataque inicial e medial. Este estudo utilizou-o apenas para observar obstruintes e vogais em posição tônica inicial. Foram coletados dados de 9 crianças (5 meninas e 4 meninos) na fase final de aquisição da linguagem. Para as gravações, o pesquisador solicitou às crianças que dissessem as palavras alvo inseridas em uma frase veículo que se mantinha dentro do tema lúdico. A frase foi "Digo ____ volte atrás". Essa dizia respeito a desfazer uma suposta confusão gerada pelos protagonistas da história infantil. Era a frase "mágica". Apenas uma sessão de coleta de dados (para cada criança) foi suficiente para gravar todas as palavras-alvo. Em análise preliminar, notou-se que os parâmetros acústicos estáticos nem sempre eram capazes de descrever a variação e a dinâmica dos fenômenos da fala infantil. Por isso usamos dois tipos de análise: uma de parâmetros acústicos estáticos e uma de parâmetros acústicos dinâmicos, baseada em inspeção de forma de onda e espectrogramas. Para a análise estatística optou-se pelo Modelo Linear Geral (GLM) com uma Análise de Variância de Medidas Repetidas. O alpha foi estabelecido em 0,05. Como resultado das análises de parâmetros acústicos de vogais obtivemos que F1 diferenciou [a] de [i] e [u], F2 diferenciou as três vogais e F3 diferenciou [i] de [a] e [u]. As freqüências dos três formantes é mais alta do que o esperado para adultos. A análise dinâmica de forma de onda e espectrogramas parece indicar demora para a estabilização da trajetória formântica, especialmente em F2 para cerca de 9% dos casos. Como resultado das análises de parâmetros estáticos de consoantes fricativas, obtivemos que a assimetria e o centróide foram eficazes para diferenciar os três locais de constrição, o vozeamento e a interação entre local e vozeamento. A curtose não distinguiu as fricativas em nenhum parâmetro. A variância somente diferenciou local de constrição. A análise dinâmica parece indicar que ocorre desvozeamento parcial em fricativas vozeadas e vozeamento parcial em fricativas desvozeadas. Como resultado das análises de parâmetros estáticos de consoantes oclusivas, obtivemos que a assimetria e o centróide foram eficazes para diferenciar os três locais de constrição, o vozeamento e a interação entre local e vozeamento. A curtose e a variância foram eficazes para diferenciar os três locais de constrição e o vozeamento, porém não para sua associação. O VOT foi eficaz para diferenciar não somente vozeamento, mas também o local de constrição. A análise dinâmica indica que ocorre desvozeamento parcial em oclusivas vozeadas e vozeamento parcial em oclusivas desvozeadas. E que vogais infantis tem freqüências mais altas do que da adulta, porém seus espaços vocálicos são semelhantes. Esses fenômenos podem ser iluminados pela fonologia Gestual (Browman & Goldstein, 1992; Ball & Kent, 1997; Scobbie, 1998; Albano, 2001; Kent & Read , 2002; Shadle, 2006; Goldstein, Byrd & Saltzman 2006). Esperamos que essa abordagem possa trazer ainda mais luz à este trabalho / Abstract : This master's thesis study aims at creating a reference database for the study of vowels and obstruents in child speakers of Brazilian Portuguese (henceforth BP). The subjects are between 5 and 7 years old, with no history of speech disorders. Special tool was developed for data collection, centered on a child story named "Fairy Tales Go Crazy" with 57 words created for the instrument with all sounds of BP in initial and medial onset position. To prompt word utterance a board game was used. The total number of target words is only 36: total of words which were created by matching each BP obstruent (fricatives and plosives) with the vowels [a], [i] and [u]. The tool was used in this study to observe obstruents and vowels only in initial stressed position. The data was collected from 9 children (5 girls and 4 boys) in the final stages of language acquisition, selected as randomly as possible. Researcher asks each child to say the target words embedded in a carrier sentence. A preliminary analysis showed that acoustic phonetic parameters were not always sufficient to describe the variation and dynamics of the speech signal in child language. Therefore, we have used two types of analysis: a Static Acoustic Analysis and a Dynamic Acoustic Analysis. The static acoustic analysis made the following measurements: for vowels, relative duration (a percent obtained of segment absolute duration divided per word duration), absolute duration and the first three formants; for fricatives, relative and absolute duration and the four spectral moments (Jongman, 2000). The fricative spectral moments were measured at their center in a 40 millisecond window; for plosives, relative and absolute duration, voice onset time (VOT) and the burst spectral moments (Forrest et al., 1988). Statistical analysis was performed by a General Linear Model (GLM) with a Repeated Measures Analysis of Variance. Alpha was set at 0.05. For the vowels the dependent variables were the formants and the independent ones were the vowels. As expected, F1 was effective in differentiating the vowel [a] from [i] and [u], F2 was effective for differentiating the three vowels, F3 was effective only to differentiate the vowel [i] from [u] and [a]. For the fricatives the independent variables were the articulation places (lip-dental, alveolar or post-alveolar) associated with the voicing (voiced or devoiced) - total of six independent variables - and the dependent variables were the first four spectral moments. The skewness and centre of gravity were effective in differentiating the three places of articulation, and the association between the voicing and place. The kurtosis did not distinguish fricatives in any parameter. The standard deviation differed only place of articulation. For plosives the independent variables were places of articulation (bilabial, dental/alveolar and velar) associated with the voicing (voiced or devoiced) - total of six independent variables - and the dependent variables were the four spectral moments and VOT. The skewness and centre of gravity were effective in differentiating the three places of articulation, the voicing and the association between voicing and place. The kurtosis and variance were effective in differentiating the three places of articulation and voicing separately, but not for their association. As expected, the VOT could distinguish the voicing, but was also effective for differentiating the place of articulation as well as the association between voicing and point. The dynamic acoustic analysis showed that voiced plosives may devoice and devoiced plosives may voice. It showed that, often, normal children's strategies are similar to those with language pathologies described by Berti (2006), Freitas (2007) and Rodrigues (2007). We should note that none of these phenomena would seem strange to the adult listener, being auditorily imperceptible. In spite of deviating from adults in acoustic trajectories as seen on spectrograms, normal children seem to produce something which sounds close to what is expected by adults. These phenomena may be illuminated by Gestural Phonology (Browman & Goldstein, 1992; Ball & Kent, 1997; Scobbie, 1998; Albano, 2001; Kent & Read , 2002; Shadle, 2006; Goldstein, Byrd & Saltzman 2006). We hope this approach may shed further light on this work / Mestrado / Linguistica / Mestre em Linguística

This dissertation investigates the role of grammatical versus methodological influences in the production and perception of final devoicing in experimentally-elicited speech from Russian. It addresses the question of how the partial preservation of the phonological voicing contrast in word-final obstruents is affected by (i) task-independent factors that reflect phonological and lexical properties of stimuli words (underlying voicing, word length, lexical competition) and (ii) task-dependent biases that arise due to the nature of the experimental task performed by the speaker (availability of orthographic inputs, presence of minimal pairs among the stimuli). Results of a series of acoustic production and perceptual identification tasks reveal that task-dependent factors account for the presence of robust and perceptually salient differences in the parameter of phonetic voicing. Several types of stimuli items also show limited but statistically significant differences in closure/frication duration and release duration that are independent of the presence of orthography or inclusion of full minimal pairs among test items. Taken together, these findings indicate that non-grammatical factors can play a prominent biasing role in both production and perception of the voicing contrast in experimentally-elicited speech, such that certain voicing-dependent cues are maintained only in the presence of task-dependent pressures. However, not all incompletely neutralized differences between phonologically voiced versus voiceless final obstruents can be attributed to the effects of orthography or inclusion of minimal pairs among the stimuli. In the theoretical domain, these results are argued to favour a less restrictive definition of neutralization and a model of phonology that views devoicing as a loss of the primary acoustic cue to the underlying voicing contrast rather than complete identity of the [voiced] feature.

Second language (L2) learners of English must learn to produce English phonemes, words, and sentences. These L2 learners make many errors when learning English; they may change the place or manner of articulation, insert vowels, or delete consonants. Obstruent sounds, such as fricatives, affricates, and stops, can be especially difficult for L2 learners. This study analyzed native English speakers’ perception of the quality of obstruents produced by native Mandarin Chinese and Korean speakers. Target words containing obstruents had been produced in three different tasks: in a carrier phrase, in a paragraph, and in a spontaneous speech sample. Obstruents were produced in word-initial position and word-final position. Raters with differing levels of expertise listened to these words and rated the perceptual quality of the obstruents within the words. This study found that overall, English obstruent productions by native Mandarin and Korean L2 speakers learning English were rated most clear when produced in word-initial position in a carrier phrase or a paragraph. The lowest ratings given were of obstruents in word-final position in spontaneous speech. No significant differences were found for listener expertise level. Combined with future research, results from this study will help educate the field of second language instruction as to how the speech of Korean and Mandarin learners of English is perceived. It also provides additional information on the effect that listener expertise has on the judgment of L2 speech production.