Este trabalho apresenta um estudo acerca da fala cantada com o português brasileiro inserido no contexto da canção popular. Utilizando como base o processo de sincronização entre os pares de quatro duplas de cantores, elaboramos um experimento fatorial com dois fatores, denominados Ritmo da Canção e Ponto de Observação de Sincronização. Com o fator Ritmo da Canção, buscamos observar se a fala cantada inserida no contexto sincopado, no caso o da bossa-nova, é mais difícil de ser realizada quando comparada a mesma inserida no contexto não sincopado, aqui o do rock. Com o fator Ponto de Observação de Sincronização buscamos observar se os sujeitos têm a mesma capacidade de sincronização em inícios e em finais de frases. Ambos os cálculos foram realizados através das medidas das lags, diferença temporal entre os Pontos de Observação de Sincronização dos cantores de uma mesma dupla, ou seja, ora separamos as lags por canções (fator Ritmo da Canção), ora separamos por ponto (fator Ponto de Observação de Sincronização). Em paralelo, e de forma a complementar o estudo fonético-acústico, observamos o desenvolvimento métrico-acentual das canções, comparando os acentos lingüísticos e os musicais das canções utilizadas. O conjunto de resultados obtidos por meio de cálculos estatísticos e as observações métricoacentuais indicam que os cantores apresentam maior facilidade em realizar a tarefa pedida cantar o mais sincrônico possível no contexto no qual o ritmo de acompanhamento das canções não era sincopado, assim como indica maior facilidade em sincronizar nos finais de sentenças. / This paper presents a study about the Brazilian Portuguese placed in the context of the song. Using as a base the entrainment process between pairs of four pairs of singers, we designed a factorial experiment with two factors, called Rhythm of the Song and Entrainment Observation Point. With the factor Rhythm of the Song, we observe whether the speech into the context of singing bossa-nova is more difficult to achieve when compared to the same inserted in the context of rock. With the factor Entrainment Observation Point we seek to observe whether individuals have the same ability to sync at the beginning and end of sentences. Both calculations were performed by measuring the lag, time difference between entrainment observation points of the singers of the same pair, i.e., separated the lag by songs (factor Rhythm of the Song), and separated by Entrainment Observation Point (factor Entrainment Observation Point). In parallel, and to complement the acousticphonetic study, we observed the development of accentual metric-songs comparing the linguistic and musical accents of the songs used. The set of results obtained by statistical calculation and accentual-metric observations indicates that the singers have it easier to perform the requested task - to sing as synchronous as possible - in the context in which the rhythm accompaniment of songs was not syncopated, so ease as indicated to sync at the end of sentences.

Consoantes oclusivas sonoras e nasais apresentam similaridades articulatórias. Estes grupos de fones, em algumas línguas indígenas brasileiras, são alofones de um mesmo fonema. Nesses sistemas, há alofones intermediários que apresentam contorno oral-nasal. A língua Dâw, embora descrita com fonemas distintos para as classes oclusivas e nasais, apresenta consoantes de contorno como alofones em situação muito restrita: coda seguindo vogal oral. Este trabalho tem como objetivo descrever e analisar os contextos de nasalização da língua Dâw, através da elicitação de dados originais. Foram realizados três trabalhos de campo, nos quais fizemos gravações de dados com falantes nativos. Obtivemos dados acústicos, através de gravadores digitais, e aerodinâmicos, através do equipamento EVA2 que apresenta transdutores diferenciados para a medição de fluxo de ar oral e nasal. Utilizamos o conceito de distribuição para analisar os dados obtidos, devido à ausência de pares mínimos, pois a língua é tipologicamente isolante-analítica. Corroboramos a descrição anteriormente realizada (Martins, 2004) sobre a categorização de nasais como fonemas distintos, tanto consonantais como vocálicos. Verificamos também a ocorrência de espalhamento de nasalização de aproximantes tautossilábicas a partir de vogais nasais, como descrito, e acrescentamos à descrição o processo de espalhamento para a fricativa glotal surda /h/ quando esta se encontra na mesma sílaba que uma vogal nasal. Conseguimos determinar que o ambiente prosódico de espalhamento de nasalização é a sílaba, já que esse fenômeno não ocorre entre sílabas. Analisamos também se o contorno oral de consoantes nasais poderia ser um processo de longo alcance. Entretanto, os dados demonstraram seu alcance local, também restrito à estrutura da sílaba. As consoantes nasais de contorno oral resgatam, possivelmente, um estado antigo da língua, que pode ser verificado nas línguas-irmãs Hup e Yuhup, de restrição a adjacências mistas oral e nasal. Por ocorrer somente em posição de coda, atribuímos que o contato com o português-brasileiro (PB) manteve esse alofone nesta posição, pois no PB ocorre espalhamento de nasalização regressivo, o que seria indesejável para a língua Dâw, que possui distinção fonêmica entre vogais orais e nasais. Essa dessincronização do gesto velar causa o contorno devido às similaridades articulatórias entre oclusivas sonoras e nasais. Houve, ainda, dados em que a aerodinâmica não correspondeu à percepção acústica, ou seja, escutamos uma nasalização, mas não havia fluxo de ar correspondente. Achamos que essa discrepância deve-se a alguma manobra articulatória não compreendida. Quanto aos processos analisados através do método da Fonologia Prosódica, concluímos que ambos os processos não ocorrem em constituintes prosódicos hierarquicamente superiores / Stop voiced and nasal consonants have articulatory similarities. In some indigenous Brazilian languages, these groups of phones are allophones of the same phoneme. In such systems, there are intermediary allophones that have an oral-nasal contour. Dâw language, although described with distinct phonemes for the stop and nasal classes, has contour consonants as allophones in a very restricted situation: coda after an oral vowel. This dissertation aims to describe and analyze the contexts of nasalization in Dâw language through elicitation of original data. We undertook three fieldwork studies in which we made recordings of data with native speakers. We obtained acoustic data using a digital recorder and aerodynamic data using EVA2 equipment that has separate sensitive transducers for oral and nasal airflow measurement. We used the distribution concept to analyze the data, due to the absence of minimal pairs, since the language is typologically isolating-analytic. We corroborated the previous description (Martins, 2004) on the categorization of both consonant and vowel nasals as distinct phonemes. We also noticed the occurrence of nasal spreading from approximant tautosyllabic to nasal vowels, as described, and added to the description the spreading process for the voiceless glottal fricative /h/ when it is in the same syllable as a nasal vowel. We were able to determine that the prosodic environment of the nasal spreading is the syllable, because this phenomenon does not occur between syllables. We also analyzed whether the oral contour of nasal consonants could be a long-range process. However, the data proved it to be local range, also restricted to the syllable and not the adjacency. Oral-contour nasal consonants hark back to a former state of the language, which can also be seen in its sister languages Hup and Yuhup, with the restriction of mixed oral and nasal adjacencies. As it occurs only in the coda, we attribute the fact that this allophone has maintained this position due to contact with Brazilian Portuguese (BP), because regressive nasal spreading occurs in BP, which would be undesirable for Dâw language, which has phonemic distinction between oral and nasal vowels. This desynchronization of the velar gesture causes the contour due to articulatory similarities between stop voiced and nasal consonants. There were data where the aerodynamics did not match the acoustics, i.e., we heard nasalization, but there was no corresponding nasal airflow. We believe that this discrepancy is due to some articulatory maneuver that is not understood. As regards processes analyzed by Prosodic Phonology, we concluded that both processes do not occur in hierarchically superior prosodic constituents

Este estudo investigou a diferenciação das emoções por meio dos aspectos acústicos da fala na representação teatral das emoções no Português Brasileiro (PB), sobretudo o comportamento da frequência fundamental (F0) e os aspectos entonacionais das quatro emoções: alegria, raiva, medo e tristeza, e da fala neutra. Embora o conceito de emoções seja de difícil definição, buscaram-se subsídios teóricos em consagrados autores como SCHERER (1985), EKMAN (1984), DESCARTES (1998), e alguns representantes da Semiótica da Paixão como GREIMAS; Fontanille (1993). Os primeiros relatos sobre as emoções trazidas por Descartes (1998), conhecidas como paixões, mostravam inferências fisiológicas ainda imaturas, no entanto de grande valor descritivo. A F0, a intensidade de voz, a velocidade de fala (LAUKKA, 1994) e a energia espectral (BANSE; SCHERER (1996) são os parâmetros mais relacionados com as emoções, já que são valores que expressam as mudanças respiratórias, fonatórias e articulatórias do indivíduo. A entonação da fala, representada acusticamente pela variação da F0, é também utilizada frequentemente na diferenciação das emoções, pois está relacionada com a expressão vocal do falante. A amostra de fala foi produzida por três atrizes profissionais com idade entre 40 e 50 anos, com média de 20 anos de profissão. As vozes foram gravadas no laboratório Estúdio Multimeios do Centro de Computação Eletrônica da Universidade de São Paulo (CCE-USP), captadas pelo software de edição de som Sound Forge 9. As atrizes leram o texto (corpus) com fala neutra e interpretando as emoções alegria, raiva, medo e tristeza, com repetição de cinco vezes para cada interpretação e para a fala neutra, totalizando 25 gravações para cada atriz. Buscou-se apoio na teoria dos simulacros, sustentada pela Semiótica das Paixões, para fundamentar a escolha da emoção representada por atores, em vez da fala espontânea e da emoção evocada. O corpus - texto escolhido para ser lido pelas atrizes nessa pesquisa - já fora utilizado por FIGUEIREDO (1993), e uma das motivações da escolha se deve ao fato de se tratar de um texto árido, que não evoca emoções apenas com a sua leitura. Foi necessário segmentar o texto para unidades linguísticas menores (três sentenças) com a finalidade de analisar os dados de forma mais eficaz. O embasamento teórico para essa etapa da pesquisa, a segmentação do texto, é a Fonologia Prosódica proposta por Nespor e Vogel (1986). A fim de obtermos os valores de F0, as sentenças foram segmentadas como unidades ainda menores (sílabas). Para obtenção dos valores da F0 e da curva entonacional, utilizou-se o software Praat. 5.1.23 e seus scripts correlacionados. Os dados foram submetidos à estatística descritiva. As atrizes SL e KK interpretaram a alegria na faixa de frequência mais alta, em seguida, a raiva. Apenas a atriz AA interpretou o medo com valores de média de F0 mais altos e depois a alegria. Em geral, a tristeza manteve-se com valores de F0 mais baixos que todas as emoções, inclusive da fala neutra. O comportamento da F0 corrobora a literatura pesquisada. Por meio do estudo da entonação das sentenças nas quatro emoções e na fala neutra, foi possível identificar a interferência da emoção nos aspectos linguísticos da sentença, principalmente no medo, já que o esperado para as sentenças declarativas do português brasileiro seriam curvas descendentes e não ascendentes como ocorreu nesta emoção. Os resultados destes estudos contribuíram para a diferenciação das emoções, no entanto é necessário incluir mais sujeitos e analisá-los, detalhadamente, do ponto de vista estatístico. / This study investigated emotional differences by means of acoustical aspects of speech in theatrical representations of emotions in Brazilian Portuguese (PB), above all the behavior of the fundamental frequency (F0 ) and the aspects of intonation of the four emotions: happiness, anger, fear and sadness and of neutral speech. Although the concept of emotion is difficult to define, supporting theories have been sought in consecrated authors such as SCHERER (1985) EKMAN (1984), DESCARTES (1988) and other representatives of Semiotic Passion such as GREIMAS; FONTANILLE (1998) who are still known as passions showing immature physiological inference although of a great descriptive value. The F0, the intensity of the voice, the rate of speech (LAUKKA, 1994) and spectral energy (BANSE; SCHERER, 1996) are the parameters most closely related to emotions as they are values that express the changes in breathing, phonetics and articulations of the individual. Intonation, represented acoustically by the variation of F0 is also frequently used to differentiate emotions as these are related to the vocal expression of the speaker. A speech sample was produced by three professional actresses aged between 40 and 50 with an average of 20 years in the profession. The voices were recorded in the laboratory of Studio Multimeios in the Electronic Computation Center of the University of São Paulo (CCE-USP) using Sound Forge 9 software for sound editing. The actresses read the text (corpus) in a neutral voice and then interpreted the emotions of happiness, anger, fear and sadness, repeating them five times for each interpretation including the neutral speech, making a total of 25 recordings for each actress. Support was sought for similarities sustained by Semiotic Passion, to justify the choice of emotion presented by the actors instead of spontaneous speech and the emotion evoked. The corpus used in this research was a text already used in FIGUEIREDO (1993), one of the reasons for this choice being the fact that it is an arid text which does not evoke emotions when reading. It was necessary to divide the text into smaller linguistic units (three sentences) in order to analyze the data more efficiently. The theoretical basis for this stage of research of dividing up the text is prosodic phonology as proposed by Nespor and Vogel (1986). In order to obtain the values of F0, the sentences were divided into still smaller units (syllables). In order to obtain the values of F0 and an intonation curve, the software Praat.5.1.23 was used and the scripts correlated. The data was submitted to a descriptive statistic. The actresses SL and KK interpreted happiness at a higher frequency range, followed by anger. Only the actress AA interpreted fear with average F0 values, lower than all the other emotions including neutral speech. The behavior of F0 corroborates the researched literature. By means of the study of intonation of the four emotional sentences and the neutral speech, it was possible to identify the interference of emotion in the linguistic aspects of the sentence, mainly fear as the declarative sentences of Brazilian Portuguese were expected to be descending curves and not ascending as was the case for this emotion. The results of these studies contribute to the differentiation of emotions; however it is necessary to include more subjects and to analyze them with more detail from a statistical point of view.

No presente trabalho, elaboramos um experimento fonético-acústico que nos permitiu observar como se realizam as diferentes tonicidades das vogais no canto. Estudos a respeito da produção das vogais do português brasileiro e o correlato acústico do acento apontam a duração como principal correlato e, portanto, responsável pela a distinção entre tônicas, pre-tônicas e pós-tônicas, sendo as tônicas as mais longas (Fernandes, 1976; Massini-Cagliari, 1992). Ao escolhermos para o experimento uma canção cujas notas das frases melódicas eram representadas na partitura com uma mesma figura rítmica, indicando, pois, que devem ser realizadas com durações iguais entre si, perguntamo-nos como as vogais da letra dessa canção se comportariam no que diz respeito a relação entre tonicidade e duração. Assumimos a hipótese de que a métrica musical e capaz de influenciar a duração das vogais, ou seja, que a relação temporal entre as vogais tônicas e as átonas seja diferente no canto. O experimento foi elaborado de modo a testar todas as vogais tônicas, [a, ɛ, e, i, ɔ, o, u], pre-tônicas, [a, e, i, o, u] e pós-tônicas, [ɐ, ɪ, ʊ], faladas e cantadas, em logatomas inseridos no texto da canção. As análises descritivas (duração media, desvio padrão e coeficiente de variação) e os testes estatísticos apontaram (i) que as vogais cantadas são mais longas que as vogais faladas; (ii) as pre-tônicas e as pós-tônicas cantadas não diferem; e (iii) no canto, as vogais de um mesmo grupo acentual não se diferem quanto a duração. Sobre o padrão formântico das vogais, as análises de F₁ e F₂ indicaram que as vogais cantadas não diferem das faladas. Foi observado também que, no canto, não ocorrem vogais pós-tônicas ensurdecidas, como na fala. / In the present work, we elaborated an experiment that allowed us to observe the interplay between stressed and unstressed vowels in singing. Studies on stressed and unstressed vowels in Brazilian Portuguese show that segment duration is the main acoustic correlate of stress, therefore allowing distinction between stressed vowels, pre-stressed vowels and post-stressed vowels (Fernandes, 1976; Massini-Cagliari, 1992).When we chose a target-sentence formed with only quarter notes that should be performed with equal duration, as previewed in the score, we asked ourselves about the relationship between vowel stress and vowel length in singing. We hypothesize that the musical metric is able to influence the vowel length, that is, that the temporal relationship between the stressed and the unstressed vowels is different in singing. The experiment was designed in order to test all stressed vowels, [a, ɛ, e, i, ɔ, o, u], all pre-stressed [a, e, i, o, u] and all post-stressed [ɐ, ɪ, ʊ], in speech and in singing, into nonsense words inserted in the song text. The descriptive analyzes (average vowel length, standard deviation and coefficient of variation) and statistical tests indicated (i) that sung vowels are longer than the spoken vowels; (ii) the pre-stressed and post-stressed vowels in singing are not different; and (iii) in the singing, the vowels of the same accentual group do not differ in length. The analyzes of F₁ and F₂ indicated that the sung vowels did not differ from those spoken ones. It was also observed that, post-tonics sung vowels are not devoiced, as in speech.

O objetivo deste trabalho é determinar os limiares de diferenciação tonal (LDTs) do Português Brasileiro, no que se refere à entoação. LDTs são definidos como valores a partir dos quais a percepção de uma determinada grandeza passa a ser perceptualmente relevante. Consoni (2011) determinou que tais valores seriam de +3 e -4 semitons para palavras manipuladas em uma frase. Nosso objetivo é rever tais valores para o nível frasal. Nossa hipótese é a de que a extensão temporal afeta a percepção, graças ao efeito tau (SHIGENO, 1986). Para este fim foram aplicados dois experimento com 13 pares de uma mesma oração falada por uma voz masculina não-marcada. Os pares consistiam de uma sentença neutra (com tom zero) pareada à outras doze com F0 manipulados (6 semitons ascendentes e 6 semitons descendentes, a partir do zero), mais o par zero-zero. O primeiro teste possuía apenas duas alternativas sim e nãoe foi chamado de teste 2AFC, ao passo que no segundo havia três opções: sim, não e talvez, chamado de 3MFC. Aos participantes foi requisitado que respondessem se notavam ou não qualquer diferença entre os pares apresentados. Os testes foram aplicado em 16 pessoas. O teste3 MFC teve seus resultados desconsiderados, seus dados foram utilizados como um padrão qualitativo para o teste de nossa hipótese. A seguir foi aplicada uma ANOVA entre as variáveis do teste 2AFC a fim de se determinar se havia algum efeito, obtivemos um p < 0, 001, para n = 16, Fo > Fe. Para determinar os limiares de diferenciação tonal foi aplicada uma análise de componentes principais, a qual retornou como limiares os valores de -3 e +3 semitons. / The aim of this study is to determine the differential treshold of pitch (DTPs) for Brazilian Portuguese, with regard to intonation. DTPs are defined as values ??from which the perception of a certain greatness becomes perceptually relevant. Consoni et (2011) determined that such values ??would be of +3 and -4 semitones for manipulated words in a sentence. Our goal is to review these values ??from the phrasal level. Our hypothesis is that the temporal extension affects perception, due to the effect tau (SHIGENO, 1986). For this purpose two experiments were applied using 13 pairs of the same sentence, spoken by a male voice unmarked. The pairs consisted of a neutral setence (with zero tone) paired with the others twelve sentences with manipulated F0 (6 semitones ascending semitones and 6 descendants, from zero), plus the zero-zero pair. The first test had only two alternatives yes and no and was named 2AFC test, while the second one had three options: yes, no and maybe and was named 3MFC. Participants were asked to answer whether or not noticed any difference between the presented pairs. The tests were applied to 16 people. The results from 3MFC test were disregarded, the 3MFC data were used as a qualitative standard for testing our hypothesis. Following, ANOVA was applied in the 2AFC test in order to determine whether there was any effect, we obtained a p < 0.001 to n = 16, Fo > Fe. To determine the differential treshold of pitch was applied principal component analysis, which returned as threshold values ??of -3 and +3 semitones.

Harrison, Alissa May. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (leaves 184-194). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Intonation´ؤlexical tone transfer --- p.3 / Chapter 1.2 --- Unresolved issues of intonation-lexical tone transfer --- p.8 / Chapter 1.3 --- Objectives of this study --- p.10 / Chapter 1.4 --- Hypotheses of intonation-lexical tone transfer in L2 Mandarin --- p.12 / Chapter 1.5 --- Methodology and experimental results --- p.13 / Chapter 1.6 --- Major findings and conclusions --- p.15 / Chapter 1.7 --- Outline of the thesis --- p.16 / Chapter 2 --- "Background of Mandarin, English, and Japanese prosody" --- p.18 / Chapter 2.1 --- Prosodic phonology --- p.19 / Chapter 2.1.1 --- Defining the components of prosody --- p.19 / Chapter 2.1.2 --- Theoretical basis for intonation and lexical tone inter- action --- p.23 / Chapter 2.2 --- Mandarin prosody --- p.32 / Chapter 2.3 --- English prosody --- p.41 / Chapter 2.4 --- Japanese prosody --- p.49 / Chapter 2.5 --- Summary of prosodic similarities and differences --- p.54 / Chapter 3 --- Previous studies of intonation and lexical tone acquisition --- p.61 / Chapter 3.1 --- Second language acquisition --- p.62 / Chapter 3.1.1 --- Interlanguage --- p.62 / Chapter 3.1.2 --- Defining transfer --- p.63 / Chapter 3.1.3 --- Theories of L2 phonological acquisition --- p.66 / Chapter 3.2 --- Comparison of first and second language acquisition --- p.71 / Chapter 3.2.1 --- Lexical tone --- p.71 / Chapter 3.2.2 --- Intonation --- p.74 / Chapter 3.3 --- L1 intonation transfer in Mandarin L2 lexical tone acquisition --- p.77 / Chapter 3.3.1 --- Initial hypotheses of intonation-lexical tone transfer --- p.77 / Chapter 3.3.2 --- Experimental studies claiming intonation-lexical tone transfer --- p.85 / Chapter 3.3.3 --- Unaddressed issues of previous studies --- p.91 / Chapter 4 --- Methodology of production and perception experiments --- p.94 / Chapter 4.1 --- Hypotheses --- p.94 / Chapter 4.2 --- Design --- p.97 / Chapter 4.3 --- Subjects --- p.101 / Chapter 4.4 --- Procedures --- p.102 / Chapter 4.5 --- Data transcription and statistical analysis --- p.105 / Chapter 5 --- Results of production and perception experiments --- p.108 / Chapter 5.1 --- Production experiment results --- p.108 / Chapter 5.1.1 --- Effect of lexical tone category on pitch production --- p.109 / Chapter 5.1.2 --- Effect of sentence type on pitch production --- p.119 / Chapter 5.1.3 --- Summary of tone production results --- p.126 / Chapter 5.2 --- Perception experiment results --- p.128 / Chapter 5.2.1 --- Effect of lexical tone category on lexical tone identifi- cation --- p.128 / Chapter 5.2.2 --- Effect of sentence type on lexical tone identification --- p.136 / Chapter 5.2.3 --- Summary of tone perception results --- p.143 / Chapter 6 --- Discussion of experimental results and intonation-lexical tone transfer hypothesis --- p.146 / Chapter 6.1 --- Analysis of experimental results --- p.147 / Chapter 6.1.1 --- Lack of postlexical tone transfer --- p.148 / Chapter 6.1.2 --- Postlexical tone prior to lexical tone acquisition --- p.156 / Chapter 6.1.3 --- Factors in lexical tone errors --- p.166 / Chapter 6.2 --- Limitations of methodological design --- p.170 / Chapter 6.2.1 --- Size of experimental data sample --- p.170 / Chapter 6.2.2 --- Proficiency levels and testing --- p.172 / Chapter 6.2.3 --- Method of elicitation --- p.173 / Chapter 7 --- Conclusions --- p.177 / Chapter 7.1 --- Summary --- p.177 / Chapter 7.2 --- Implications --- p.178 / Chapter 7.2.1 --- Prosodic phonology --- p.178 / Chapter 7.2.2 --- Second language acquisition --- p.180 / Chapter 7.3 --- Future Work --- p.181 / Chapter A --- Production Experiment Materials --- p.195 / Chapter A.l --- Wordlist --- p.195 / Chapter A.2 --- Sentences --- p.196 / Chapter B --- Perception Experiment Materials --- p.199 / Chapter B.l --- Wordlist --- p.199 / Chapter B.2 --- Sentences --- p.200 / Chapter C --- F0 contours of target word productions --- p.203

Jia, Shiwei. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2011. / Includes bibliographical references (leaves 106-114). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

The main objective of this thesis is to investigate acoustic cues for the voicing contrast in stops in Russian for effects of speaking rate and phonetic environment. Although the laryngeal contrast in Russian is assumed to be a [voice] contrast, very few experimental studies have looked at the acoustic properties of Russian voiced and voiceless stops. Most claims about acoustic properties of stops and phonological processes that affect them (voice assimilation and final devoicing) have been made based on impressionistic transcriptions. The present study provides evidence that (1) voicing in voiced stops is affected by speaking rate manipulation, (2) stops in Russian retain underlying voicing contrast in presonorant position and voice assimilation occurs only in obstruent clusters, and (3) phonological processes of voice assimilation and final devoicing do not result in complete neutralization. The target of the investigation is voiced and voiceless intervocalic stops, stops in clusters, and final stops in different prosodic positions within a word and at the phrase level. The acoustic cues to voicing (duration of voicing, stop closure duration, vowel duration, f0, and F1) were measured from the production data of 14 monolingual speakers of Russian recorded in Russia. Speakers produced words and phrases with target stops in three speaking rate conditions: list reading, slow rate and fast rate. The data were analyzed in 5 blocks focusing on (1) word-internal stops, (2) voice assimilation in stops in prepositions, (3) cases of so-called "sonorant transparency", (4) voice assimilation in stops before /v/, and (5) voicing processes across a word boundary. The results of the study present a challenge to the widely-held assumption that phonological processes precede phonetic processes at the phonology-phonetics interface. It is shown that the underlying contrast leaves traces on assimilated and devoiced stops. To account for the findings, a phonology-phonetics interface that allows interaction between the modules is required. In addition, the results show that temporal cues are affected by speaking rate manipulation, but the effect of rate on voicing is found only in voiced stops. Duration of voicing and VOT in voiceless stops are not affected by speaking rate. The results also show that no effect of C2 is obtained on voicing in C1 stops in in obstruent-sonorant-obstruent clusters, thus no "phonological sonorant transparency to voice assimilation" is found in Russian. Rather, the study provides evidence that there is variation in production of voicing in stops in prepositions, and that voice assimilation in stops before /v/ followed by a voiced obstruent is optional for some speakers.

In the present study, the speech of twenty-four normally speaking toddlers and twenty-eight late talking toddlers was analyzed with respect to the syllable structures produced during a speech sample. The groups were matched with regard to age, sex, and socio-economic status, all passed a hearing screening, and all scored at least 85 on the Bayley Scales of Infant Development.

Data-Driven Pronunciation Generation for ASR Maria Obedkova In ASR systems, dictionaries are usually used to describe pronunciations of words in a language. These dictionaries are typically hand-crafted by linguists. One of the most significant drawbacks of dictionaries created this way is that linguistically motivated pronunciations are not necessarily the optimal ones for ASR. The goal of this research was to explore approaches of data-driven pro- nunciation generation for ASR. We investigated several approaches of lexicon generation and implemented the completely new data-driven solution based on the pronunciation clustering. We proposed an approach for feature extraction and researched different unsupervised methods for pronunciation clustering. We evaluated the proposed approach and compared it with the current hand-crafted dictionary. The proposed data-driven approach could beat the established base- lines but underperformed in comparison to the hand-crafted dictionary which could be due to unsatisfactory features extracted from data or insufficient fine tuning. 1