Dissertations / Theses on the topic 'Prosody'

Fundamental frequency (F0) is the most important feature among the components of prosody in a language, tone and intonation languages alike. In a tone language, how lexical tone and sentence intonation can both use F0 as their main acoustic cues has long been an intriguing question. Will the lexical tones be so resistant to modification that no elaborate intonation is possible in the language? How much of the surface sentential F0 is attributable to lexical tones and intonation? If F0 modification is kept minimal, how will prosodic focus be realized? And will the lack of focus-related F0 change be a disadvantage in terms of focus perception? In this dissertation, experimental studies have been made on Cantonese in some less well-understood aspects of its prosody. Firstly, the tonal characteristics of sentence-final particles (SFPs) in Cantonese as a special case of the interaction between tone and intonation are examined. Secondly, the acoustic correlates of prosodic focus in Cantonese are explored. SFPs are a class of words known to have functions similar to intonation. It is not yet clear, however, whether the F0 contours of SFPs are derived purely from lexical tones, purely intonational, or a combination of tone and intonation. As an attempt to offer a solution, a production experiment was designed in which sentences in Hong Kong Cantonese with ten different SFPs were recorded and detailed analyses of their F0 contours, final F0, final F0 velocity and duration were performed. The results show that most of these SFPs are very similar to the lexical tones in terms of F0 contours, but there are significant differences in durations in more than half the cases. In addition, the occurrence of an SFP does not give rise to differences in F0 and duration in the syllables preceding the SFP in most cases. But differences can be seen in sentences with question SFPs, which indicates that the prosody of the SFPs may be partly due to intonational meanings. One of the SFPs, however, exhibits a component F0 contour that seems to be sequentially attached to the end of the lexical tonal component. These findings suggest that Cantonese SFPs have underlying tonal targets just like those of lexical tones, but they also carry intonational meanings by modifying the lexical tonal contours. Previous research has shown that Beijing Mandarin, a tone language, marks focus not only by on-focus prosodic expansion like many other languages, but also by post-focus compression of pitch range and intensity (PFC). However, recently it is found that PFC is absent in Taiwanese and Taiwan Mandarin, two languages closely related to Beijing Mandarin. This finding both highlights the non-universality of PFC and raises questions about its origin. The present study explores these issues by investigating focus production in Cantonese by native Cantonese speakers born and raised in Hong Kong, and in English and Cantonese by bilingual speakers who were born and raised in Southern England. Results from the Hong Kong speakers show that, just as in Taiwanese and Taiwan Mandarin, PFC is absent in Cantonese, and mean F0, duration, intensity and excursion size were found to be higher in on-focus words. Results from the bilingual speakers show that their Cantonese also lacks PFC. More remarkably, out of the fifteen bilinguals tested, only one-third show PFC in all their English test sentences. These findings suggest that PFC is hard to transmit across languages through bilingualism. Moreover, the differential prosodic patterns among the bilingual speakers suggest that in the bilingual community, PFC may be subject to gradual loss. Although tone-intonation relationship in SFPs and the acoustic correlates have previously been studied, most of the discussion lacked supporting evidence from phonetic experiments. The present study is distinguished in its systematic experimental design and detailed acoustic analyses, and it is hoped that the results will lay the foundations for future investigations into Cantonese phonetics.

Thesis (M.A.)--University of North Carolina at Chapel Hill, 2009.
Title from electronic title page (viewed Jun. 26, 2009). "... in partial fulfillment of the requirements for the degree of Master of Arts in the Department of Psychology." Discipline: Psychology; Department/School: Psychology.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Linguistics and Philosophy, 2005.
Includes bibliographical references (p. 317-341).
This thesis proposes a recursive mapping of syntactic derivations to prosodic representations. I argue that the prosody of an expression, just like its meaning, is determined compositionally, as originally proposed in Chomsky et al. (1957), Chomsky and Halle (1968). Syntactic structure are cyclically spelled out and assigned a semantic and phonological interpretation. The cyclic approach is motivated based on data from the prosody of coordinate structures, integrating insights from syntax, combinatorics, and semantics. The algorithm distinguishes two ways of prosodically relating the output of cyclic domains: they can either be mapped to prosodic domains that are on a par and match in prosodic status: PROSODIC MATCHING; or the output of one cycle can be prosodically subordinated to another cycle: PROSODIC SUBORDINATION. Together, they derive a metrical structure that encodes information about phrasing, accent placement, and prominence. Scope relations, argument structure, and information structure affect prosodic phrasing indirectly by determining which of the two principles applies and when a syntactic cycle is spelled out. The derived metrical representation is a relational grid (Liberman, 1975).
(cont.) It encodes syntactic structure and also the derivational history of how it was assembled. The theory attempts to incorporate insights from recent work on stress and prominence (Cinque, 1993, Arregi, 2002) and prosodic recursion Ladd (1988), Dresher (1994), as well as insights from the research on prosodic phrasing and phrasal phonology (Gussenhoven, 1984, Selkirk, 1986, Truckenbrodt, 1995). Phonetic evidence from on-line production is presented to show that speakers implement the predicted metrical relations and scale boundaries later in the utterance relative to boundaries already produced, a phenomenon dubbed BOUNDARY STRENGTH SCALING.
by Michael Wagner.
Ph.D.

It is universally recognised that humans process speech and language in chunks, each meaningful in itself. Any two renditions or assimilations of a given sentence will exhibit similarities and discrepancies in chunking, where speakers and readers use pauses and inflections to mark phrase breaks. This thesis reviews deterministic and stochastic approaches to phrase break prediction, plus datasets, evaluation metrics and feature sets. Early rule-based experimental work with a chunk parser gives rise to motivational insights, namely: the limitations of traditional features (syntax and punctuation) and deficiency of prosody in current phrasing models, and the problem of evaluating performance when the training set only represents one phrasing variant. Such insights inform resource creation in the form of ProPOSEL, a prosody and part-of-speech English lexicon, to create a domain-independent knowledge source, plus prosodic annotation and text analytics tool for corpus-based research, supported by a comprehensive software tutorial. Future applications of ProPOSEL include prosody-motivated speech-to-viseme generation for "talking heads" and expressive avatar creation. Here, ProPOSEL is used to build the ProPOSEC dataset by merging and annotating two versions of the Spoken English Corpus. Linguistic data arrays in this dataset are first mined for prosodic boundary correlates and later re-conceptualised as training instances for supervised machine learning. This thesis contends that native English speakers use certain sound patterns (e.g. diphthongs and triphthongs) as linguistic signs for phrase breaks, having observed these same patterns at rhythmic junctures in poetry. Pre-boundary lexical items bearing these complex vowels and gold-standard boundary annotations are found to be highly correlated via the chi-squared statistic in different genres, including seventeenth century English verse, and for multiple speakers. Complex vowels and other symbolic prosodic features are then implemented in a phrasing model to evaluate efficacy for phrase break prediction. The ultimate challenge is to better understand how sound and rhythm, as components of the linguistic sign, inform psycholinguistic chunking even during silent reading.

My analysis of question-word questions in conversational question-answer sequences results in the decomposition of the conversational question into three systems of constitutive cues, which signal and contextualize the particular activity type in conversational interaction: (1) syntactic structure, (2) semantic relation to prior turn, and (3) prosody. These components are used and combined by interlocutors to distinguish between different activity types which (4) sequentially implicate different types of answers by the recipient in the next turn. Prosody is only one cooccurring cue, but in some cases it is the only distinctive one. It is shown that prosody, and in particular intonation, cannot be determined or even systematically related to syntactic sentence structure type or other sentence-grammatical principles, as most former and current theories of intonation postulate. Instead, prosody is an independent, autonomous signalling system, which is used as a contextualization device for the constitution of interactively relevant activity types in conversation.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Linguistics and Philosophy, 2001.
Includes bibliographical references (leaves 167-171).
In this dissertation, I investigate the grammatical effects of focus and the inseparable phenomenon of givenness. As Schwarzschild (1999) has proposed, a proper understanding of givenness eliminates the need for a separate concept of focus, which is notoriously hard to define, either semantically, syntactically, or phonologically. I propose a semantic constraint, the Givenness Interpretation Principle based on Rooth's (1992) Focus Interpretation Principle, that accounts, in part, for the semantic effects of givenness and focus. I also propose a phonological constraint, *GIVEN, that accounts for the prosodic effects of givenness and focus in Chichewa, Japanese, Hungarian, and Italian. Givenness and focus are represented in the syntax by a functional head G which takes a given constituent in its complement and a focussed constituent in its specifier. This is demonstrably the correct representation in Hungarian, and I propose that this is the representation of givenness and focus in Universal Grammar. A phrase may raise out of the complement of G to its specifier, either overtly as in Hungarian, or covertly at LF.
(cont.) Givenness has demonstrable phonological effects that, as I show, cannot be ascribed to a FOcus constraint (Truckenbrodt 1995) requiring focussed constituents to be the most prominent in their domains of focus. The constraint *GIVEN bars given constituents from being metrically prominent. Since the effects of FOcus and *GIVEN are sometimes difficult to tease apart, I present an in-depth study of the phrasal phonology of Italian, showing how phonological and intonational phrases are formed in Italian, with the aid of the segmental phenomena of raddoppiamento sintattico and gorgia toscana. Once the constraints governing these phenomena are established, I present a rigorous, controlled comparison of the effects of *GIVEN and FOcus in Italian, showing that it is *GIVEN, not FOCUS, that gives the correct results.
by Christopher Bader.
Ph.D.

Semantic prosody is an important concept and has become a primary research interest in corpus linguistics. This thesis undertakes the groundwork of fundamental research into semantic prosody in Thai, a language which has not been subject to studies of semantic prosody before, to set out the parameters for subsequent research in this area. In particular, it addresses these three research questions: 1. What are the advantages and disadvantages of the major approaches to semantic prosody proposed in the literature for describing semantic prosody in Thai? 2. What variation in semantic prosodies across genres can be identified for Thai words? 3. To what extent are the semantic prosodies of words identified as translation-equivalents in widely-used bilingual dictionaries in Thai and English similar or different? The datasets employed in the analysis are the Thai National Corpus and the British National Corpus. To address each research question, a small number of Thai words are selected for the analysis. Two primary approaches, the polarity-oriented approach and the EUM-oriented approach, are employed to identify semantic prosody. Within the polarity-oriented approach, which is founded in work by Louw, Stubbs, and Partington, semantic prosody is identified based on collocates, and is restricted to the positive vs. negative opposition. Within the EUM-oriented approach, which is based in the studies of Sinclair, semantic prosody is identified by examining concordance lines for a pragmatic function or meaning that is spread across an extended unit of meaning. The results of the analysis show that the two primary approaches to semantic prosody do operate successfully with the Thai data. A range of semantic prosodies are identified for /kreeŋcay/ ‘considerate’, /kɔ̀ ɔhâykə̀ ət/ ‘cause’, and /chɔ̂ɔp/ ‘like’, the objects under study, by the two approaches. The discussion of these semantic prosodies shows that the two approaches are useful for different purposes. The polarity-oriented approach is useful when one’s aim is to investigate a word’s tendency to co-occur with positive or negative words. Particularly, it reveals the hidden evaluative potential of words whose evaluation is not obvious from their core semantics. The EUM-oriented approach is, by contrast, suitable for the examination of an extended unit of meaning and its pragmatic function in the Sinclairian sense. They both also have some advantages and disadvantages in terms of practicality. On the issue of variation in semantic prosodies across genres, some variation is indeed found to exist. From the concordance analysis of 19 verbs, each in four different genres, namely academic writing, fiction, newspaper stories, and non-academic non-fiction, 21 different extended units of meaning are identified from 14 of the verbs. The level of variation in the use of these extended units of meaning across genres, which implies variation in semantic prosodies, is considerable with some extended units of meaning, but is limited with others. In particular, a notable contrast is identified between academic and fiction genres in terms of which extended units (and semantic prosodies) are common. Finally, the majority of the translationequivalent pairs under study (36 out of 48) show the same semantic prosody; of these, most present a neutral semantic prosody. In cases where the pairs show different semantic prosodies, there are not any cases where one word in the pair shows a positive semantic prosody, and the other shows a negative semantic prosody, and vice versa. It is thus arguable that there is a relationship between semantic prosody in Thai and English – not a genetic or areal relationship, but one that arises from a functional basis, that is, the meanings that the pairs of words under study express in both languages.

This dissertation takes an integrated approach to the study of audiovisual cues to verbal irony. While pragmatic studies have mainly focused on the role of the discourse context in irony detection, little is known about the role of prosodic and gestural cues in this process. The thesis includes four experimental studies—each one described in a separate chapter—addressing a set of questions using a variety of experimental designs. The first one is a case study of a professional comedian and reveals (a) that ironic utterances display a higher density of prosodic and gestural markers than non-ironic utterances; and (b) that gestural markers can appear both temporally aligned with prosodic prominence but can also appear independently, as gestural codas. The second study includes two experiments: (a) a production experiment eliciting spontaneous ironic speech which reveals that in non-professional spontaneous speech, too, speakers employ a higher density of prosodic and gestural markers in ironic compared to non-ironic utterances; and (b) a perception experiment on the contribution of gestural codas to the detection of verbal irony, which shows that speakers detect ironic intent significantly better when post-utterance gestural codas are present than when they are not. Following up on this idea, the third study presents three perception experiments on the relative contribution of contextual vs. prosodic vs. gestural cues to verbal irony understanding. Overall, results of the three experiments emphasize the role of contrast effects in irony perception. The first experiment shows that (a) listeners detect irony more accurately when they have access to both prosodic and gestural cues than when they just rely on prosodic information, (b) that listeners rely more strongly on gestural information than on prosodic information, and (c) that listeners rely more heavily on gestural cues than on prosodic or contextual ones for detecting irony. Finally, the fourth study addresses the contribution of prosodic and gestural cues to children’s early understanding of verbal irony, showing that mismatched multimodal cues of emotion facilitate the detection of irony by 5-year-old children. Altogether, this dissertation shows that both prosodic and gestural markers of irony aid in guiding the hearer in the interpretation of an utterance by providing overt clues about the assumptions, emotions and attitudes held by the speaker. Together with recent studies on the general pragmatic effects of prosody and gesture, the claim is that audiovisual markers of irony are strong triggers of implicature strength which help decode speech intentions in interaction. In addition, the dissertation presents novel empirical evidence of the stronger effects of multimodal—and especially gestural—cues in comparison with contextual cues, both in adult and child populations. This crucial finding leads us to claim that the study of prosodic and gestural cues to verbal irony should be at the core of any pragmatic or psycholinguistic account of verbal irony production and comprehension.
Esta tesis aborda el estudio de las marcas audiovisuales de la ironía verbal desde una perspectiva integral. Los estudios pragmáticos se han centrado principalmente en investigar el papel del contexto discursivo en la detección de la ironía, pero poco se sabe sobre el rol que desempeñan las marcas prosódicas y gestuales en este proceso. La presente tesis incluye cuatro estudios experimentales —cada uno de ellos incluido en un capítulo separado— que abordan diferentes preguntas de investigación utilizando varios diseños experimentales. El primero es un estudio de caso sobre un cómico profesional y muestra, en primer lugar, que los enunciados irónicos presentan una mayor densidad de marcadores prosódicos y gestuales que los enunciados no irónicos y, segundo, que los marcadores gestuales pueden aparecer alineados temporalmente con la prominencia prosódica, pero también de forma independiente, como codas gestuales. El segundo estudio incluye dos experimentos. Uno de producción —diseñado para obtener discurso irónico espontáneo—, cuyos resultados confirman que también en habla espontánea los hablantes no profesionales emplean una mayor densidad de marcadores prosódicos y gestuales cuando son irónicos en comparación con cuando no lo son; y, en segundo lugar, un experimento de percepción que investiga la contribución de las codas gestuales a la detección de la ironía verbal y cuyos resultados muestran claramente cómo la intención irónica se detecta significativamente mejor cuando los hablantes tienen acceso a codas gestuales que cuando no la tienen. El tercer estudio de esta tesis contiene tres experimentos de percepción que examinan cómo el contexto, las marcas prosódicas y las marcas gestuales contribuyen a la comprensión de la ironía verbal. En general, los resultados de los tres experimentos subrayan la importancia que los “efectos de contraste” tienen en el proceso de detección de la ironía. El primero muestra que los oyentes detectan la ironía con más precisión cuando tienen acceso a las marcas prosódicas y gestuales de manera conjunta que cuando solo tienen acceso a la información prosódica; el segundo, que la información visual resulta más convincente que la información prosódica a la hora de detectar la ironía; y, por último, el tercer experimento muestra que los oyentes emplean preferentemente las marcas gestuales por encima de las prosódicas e incluso de las contextuales a la hora de detectar la ironía. Finalmente, el cuarto estudio investiga cómo los niños desarrollan la capacidad de detectar la ironía verbal a través de las marcas prosódicas y gestuales, y los resultados muestran que las marcas multimodales facilitan la detección de la ironía en niños desde los 5 años de edad. En conjunto, esta tesis muestra que tanto los marcadores prosódicos como los gestuales contribuyen de manera significativa a la comprensión de la ironía verbal, guiando al oyente en la interpretación del enunciado mediante el suministro de pistas sobre las suposiciones, las emociones y las actitudes del ironizador. Siguiendo la línea de algunos estudios recientes sobre los efectos pragmáticos de la prosodia y el gesto, los resultados de los experimentos de esta tesis muestran que los marcadores audiovisuales de la ironía son potentes desencadenadores del proceso inferencial necesario para decodificar las intenciones del hablante en las interacciones comunicativas. Además, esta tesis presenta evidencia empírica de la gran incidencia que tienen las marcas multimodales —y especialmente de las gestuales— en la detección de la ironía verbal en comparación con las marcas contextuales, tanto en la población adulta como en la infantil. Este hallazgo fundamental nos lleva a afirmar que el estudio de las señales prosódicas y gestuales de la ironía debería considerarse una parte integral fundamental de cualquier explicación pragmática o psicolingüística sobre la producción y comprensión de la ironía verbal.
Aquesta tesi adopta una perspectiva integral a l'estudi de les marques audiovisuals en la ironia verbal. Els estudis pragmàtics s'han centrat principalment a investigar el paper del context discursiu en la detecció de la ironia, però se sap poc sobre el rol que juguen les marques prosòdiques i gestuals en aquest procés. La tesi inclou quatre estudis experimentals —cadascun d’ells descrit en un capítol separat— que aborden diferents preguntes de recerca i fan servir diversos dissenys experimentals. El primer és un estudi de cas sobre que analitza el discurs d’un còmic professional i mostra, en primer lloc, que els enunciats irònics presenten una major densitat de marcadors prosòdics i gestuals que els enunciats no irònics; i, en segon lloc, que els marcadors gestuals poden aparèixer temporalment alineats amb la prominència prosòdica, però també de forma independent, en el que anomenem “codes gestuals”. El segon estudi inclou dos experiments. Un de producció, dissenyat per obtenir discurs irònic espontani i els resultats del qual confirmen que en parla espontània els parlants no professionals també empren una major densitat de marcadors prosòdics i gestuals quan són irònics en comparació a quan no ho són; i, en segon lloc, un experiment de percepció sobre la contribució de les codes gestuals a la detecció de la ironia verbal, els resultats del qual demostren que la ironia es detecta millor quan els parlants tenen accés a codes gestuals que quan no en tenen. El tercer estudi presenta tres experiments de percepció que examinen com el context, les marques prosòdiques i les marques gestuals contribueixen a la comprensió de la ironia verbal. En general, els resultats dels tres experiments subratllen la importància dels efectes de contrast en el procés de detecció de la ironia. El primer experiment mostra que els oients detecten la ironia amb més precisió quan tenen accés a les marques prosòdiques i gestuals alhora, comparat amb quan només tenen accés a la informació prosòdica; el segon, que la informació visual és més poderosa que la informació prosòdica a l’hora de detectar la ironia, i, finalment, el tercer experiment mostra que els oients empren preferentment les marques gestuals per sobre de les prosòdiques o les contextuals a l’hora de detectar la ironia. Finalment, el quart estudi investiga com els infants aprenen a comprendre la ironia verbal a través de les marques prosòdiques i gestuals, i els resultats mostren que les marques multimodals faciliten la detecció de la ironia des dels 5 anys d'edat. En conjunt, aquesta tesi mostra que tant els marcadors prosòdics com els gestuals contribueixen a la comprensió de la ironia verbal, tot guiant l'oient en la interpretació de l’enunciat mitjançant el subministrament de pistes sobre els supòsits, les emocions i les actituds del parlant irònic. Seguint la línia d’estudis recents sobre els efectes pragmàtics de la prosòdia i el gest, els resultats dels experiments d’aquesta tesi mostren que els marcadors audiovisuals de la ironia són potents factors que desencadenen el procés inferencial necessari per a descodificar les intencions del parlant en les interaccions comunicatives. A més, aquesta tesi presenta evidència empírica de la gran incidència que tenen les marques multimodals —i especialment de les gestuals— en la detecció de la ironia verbal, en comparació amb les marques contextuals, tant en població adulta com en població infantil. Aquesta troballa fonamental reforça la idea que l'estudi dels aspectes prosòdics i gestuals hauria de ser una part integral fonamental de qualsevol explicació pragmàtica o psicolingüística sobre la producció i comprensió de la ironia verbal.

The purpose of this thesis is to establish a theoretical framework of Korean prosody within which the correlation of the four major components stress/accent, rhythm, intonation and vowel length -- and their relationship with grammar, pragmatics, information structure and attitude are best described and explained. It is hoped that this thesis will contribute to the analysis of the prosody of other languages and eventually to the typology of prosody. In chapter 1, it is argued that Korean is a fixed stress language. The Korean Stress Rule is set up and syllable structure and stress shift in Korean are also discussed. In chapter 2, it is argued that Korean rhythm has a strong tendency towards stress-timing, and that the possible rhythmic patterns of a sentence are determined by the interaction between the rhythmic structure of the sentence, the scope of focus, and the speech tempo and style. The rhythmic structure of a sentence is assumed to be predictable by eight prosodic phrase structure rules. In chapter 3, the intonation system of Korean is established. It is argued that a tune consists of zero or more phrasal tones followed by one obligatory boundary tone, the latter conveying the greater part of the information conveyed by the tune. Nine boundary tones and four phrasal tones are set up. Intonation group boundary placement and the functions of the boundary tone are also discussed. In chapter 4, vowel shortening, both morphophonemic and phonetic, and compensatory vowel lengthening are investigated. It is argued that phonetic vowel shortening can be best described and explained in terms of accent placement. Finally, phonetic variations of vowel length are also discussed.

This dissertation studies the prosodic structure of the variety of Jordanian Arabic that is spoken in the rural areas of the Governorate of Irbid (IA) by investigating the role of syntactic structure in the formation of prosodic domains. It empirically explores the word-level, phrase-level and clause-level prosody of IA and attempts to account for these empirical results in a framework based on the standard syntactic-prosodic interface principles developed in Match Theory (Selkirk 2011) and formulated as OT constraints (Prince & Smolensky 1993). The basic hypotheses in this dissertation are that the prosodic word (ω), phonological phrase (Φ) and intonational phrase (ι) are present in IA, and that they are anchored in syntactic constituents. Relying on hypotheses derived from the MATCH constraints (Selkirk 2011) that ensure the syntactic-prosodic correspondence, ω, Φ and ι should respectively match the grammatical word, syntactic phrase and clause and should recursively match embedded syntactic constituents. A series of experiments was designed to test the hypotheses. Twenty native speakers (ten males and ten females) of Jordanian Arabic living in Irbid participated in the tasks. Each pair of participants performed several tasks in one session. Two game-based tasks were designed to explore intonational and temporal cues to Φ and ι boundaries and examine their relation to XPs and clauses, respectively. Two additional reading tasks were designed to determine the application domain of post-lexical segmental processes in IA (the coarticulation of pharyngealization and vowel hiatus resolution). The collected tokens were submitted to acoustic and statistical analyses. Based on the results of these experiments, the existence of the ω, Φ and ι is confirmed and our understanding of their segmental and suprasegmental cues is refined. ω’s match grammatical words and are the domain of stress, realization of the feminine -t suffix and coarticulation of pharyngealization. Φ`s match syntactic phrases and are cued suprasegmentally: their right boundaries are marked by low phrase accents (L-) and pre-boundary syllable lengthening. As for ι`s, they match clauses and are cued by additional final lengthening, boundary tones (H% or L%) and resistance to vowel reduction. There is also ample evidence that syntactic nesting motivates prosodic recursion. At the ω level, the primary/secondary status of genitive constructs of stress mirrors syntactic nesting. At the Φ level, recursion is evidenced by gradient pre-boundary syllable lengthening, which is greater at the right boundaries of higher prosodic subcategories that match larger syntactic domains. As for recursion at the ι level, it is not only cued by gradient pre-boundary syllable lengthening, but also by boundary tones: continuative H% are used at sentence-internal ι boundaries, but L% tones are cues to boundaries of larger ι’s. However, prosodic recursion is not unconstrained in IA: prosodic domains can only consist of two subcategories, i.e. a minimal and maximal layers. In this way, prosodic recursion is neither prohibited as proposed in the early version of Strict Layer Hypothesis (Nespor &Vogel 1986, Selkirk 1986), nor free to perfectly mirror syntactic nesting. As in most previous case studies, it is proposed that the one-to-one correspondence constraints of Match Theory (Selkirk 2011) account for the prosodic patterns in IA, but have to be complemented with language-specific markedness constraints on phonological weight, exhaustivity and recursion. It is also shown that these explanatory principles can, with minor reorganization, account for the prosodic patterns described in other Arabic dialects.

Lexical access, the matching of auditory information onto lexical representations in the brain, is a crucial component of online language processing. To understand the nature of lexical access, it is important to identify the kind of acoustic information that is stored in the long-term memory and to study how the brain uses such information. This dissertation investigates the contribution of prosodic information to lexical access and examines language-specific processing mechanisms by studying three typologically distinct languages: English, Turkish, and Swedish. The main research objective is to demonstrate the activation of long-term memory traces for words on the sole basis of prosodic information and to test the accuracy of typological phonological descriptions suggested in the literature by studying electrophysiological measurements of brain activation. A secondary research objective is to evaluate three distinct electrophysiological recording systems. The dissertation is based on three papers, each examining neural responses to prosodic changes in one of the three languages with a different recording system. The first two papers deal directly with the interplay between prosody and the lexicon, and investigate whether prosodic changes activate memory traces associated with segmentally identical but prosodically different words; the third paper introduces morphology to this process and investigates whether prosodic changes activate memory traces associated with potential lexical derivations. Neural responses demonstrate that prosodic information indeed activates memory traces associated with words and their potential derivations without any given context. Strongly connected neural networks are argued to guarantee neural activation and implementation of long-term memory traces. Regardless of differences in prosodic typology, all languages exploit prosodic information for lexical processing, although to different extents. The amount of neural activation elicited by a particular piece of prosodic information is positively correlated with the strength of its lexical representation in the brain, which is called lexical specification. This dissertation could serve as a first step towards building an electrophysiological-perceptual taxonomy of prosodic processing based on lexical specification.

Thesis (M.A.) -- University of Maryland, College Park, 2008.
Thesis research directed by: Dept. of Hearing and Speech Sciences. Title from t.p. of PDF. Includes bibliographical references. Published by UMI Dissertation Services, Ann Arbor, Mich. Also available in paper.

This PhD thesis presents the development of a prosody system for European Portuguese (EP) for text-to-speech (TTS) applications. Basically, TTS systems carry out the automatic utterance of a text and consist in a sequence of several modules. Those modules implement the pre-processing of the text input, the phonetic transcription and the supra-segmental processing that consists in the inclusion of prosodic patterns. The prosody is responsible for a communicative intention and guarantees some naturalness in the uttered speech. The prosodic features consist in the imposition of the timing, characterized by the segmental durations and pauses, the intonation, characterized by the fundamental frequency (F0) curve, and by the intensity curve. The preparatory work that was fundamental for modelling and testing purposes is presented in the beginning. It starts with a preliminary study about the stressed syllable. This study identifies the variation range of F0, duration and intensity features in stressed syllable along contexts. Then the FEUP-IPB EP speech database that was used in following studies is presented. The database is labelled at the levels of the phoneme, word, sentence and F0. The thesis follows on with the presentation of two algorithms to provide the syllabic splitting of the text and of the phoneme sequences. This chapter ends with a proposed set of rules for the automatic phonetic transcription of the most problematic graphemes in EP. The proposed prosody model consists of several sub-models, namely, the duration model to predict the segmental durations and the model to predict the F0 pattern. Two proposals, based in artificial neural networks (ANNs), to predict the segmental durations are presented. The first proposal consists of one ANN carefully selected concerning its architecture and type as well as the set of input features with the objective of minimizing the error between predicted and measured durations. The second proposal, entitled alternative model, is based on same considerations of the first proposal but uses one dedicated ANN for each phoneme, in a total of 44 ANNs. The alternative model, with dedicated ANNs, improved the final performance. A model of insertion and prediction of durations of the pauses is proposed, based on a preliminary study over the FEUP-IPB database. The proposed model to predict the F0 contour is based on the Fujisaki model and consists of two sub-models. One predicts the Phrase Commands’ (PCs) parameters and the other predicts the Accent Commands’ (ACs) parameters. The PCs and the ACs were manually estimated in 101 paragraphs of the database under the criterion of the minimization of the error between estimated and measured F0 contours. The prediction of the PCs is performed in two stages. The first stage is carried out by an algorithm responsible for the insertion of the PCs connected to the text and based on a mathematical model obtained from experimental observations. The second stage of the model predicts the PCs amplitude, Ap, and anticipation, T0a, relatively to the initial position. The anticipation allows the determination of the exact position in the speech signal. The two parameters are predicted with ANNs. A strong connection between ACs and syllables was found in the database. This strong connection justified the adopted methodology of predicting ACs associated with syllables. Therefore, the ACs model consists of one ANN to predict the existence of AC associated with the syllable and other three ANNs to predict the parameter’s amplitude (Aa) and anticipation of the onset (T1a) and offset (T2a) instants. The final perceptual test using the category-judgment method and the MOS scale resulted in a classification of 4.6 for the natural speech, 4.4 for the estimated F0, 4.2 for predicted durations, 3.1 for the predicted F0 and 2.9 for the complete proposed model (duration and F0 models). The MOS for the complete model is at the ‘Fair’ level.

Prosody and prosodic modeling in trainable Speech Synthesis systems are often based on large corpora of automatically annotated training data; however, these annotations are often incorrect. In practice, this has been either addressed through labor intensive manual annotation or simply ignored. In order to overcome this problem and improve prosodic realization, an iterative model-based method is proposed for improving linguistic structure, segmentation, and prosodic annotations that correspond to the delivery of each utterance as regularized across the data. For each iteration, the training utterances are resynthesized according to the existing symbolic annotation. Values of various features and subgraph structures are "twiddled:" each is perturbed based on the features and constraints of the model. Twiddled utterances are evaluated using an objective function appropriate to the type of perturbation and compared with the unmodified, resynthesized utterance. The instance with least error is assigned as the current annotation, and the entire process is repeated. At each iteration, the model is re-estimated, and the distributions and annotations regularize across the corpus. As a result, the annotations have more accurate and effective distributions, which leads to improved control and expressiveness given the features of the model.

no
The majority of studies have demonstrated a right hemisphere (RH) advantage for the perception of emotions. Other studies have found that the involvement of each hemisphere is valence specific, with the RH better at perceiving negative emotions and the LH better at perceiving positive emotions [Reuter-Lorenz, P., & Davidson, R.J. (1981) Differential contributions of the 2 cerebral hemispheres to the perception of happy and sad faces.Neuropsychologia, 19, 609¿613]. To account for valence laterality effects in emotion perception we propose an `expectancy¿ hypothesis which suggests that valence effects are obtained when the top-down expectancy to perceive an emotion outweighs the strength of bottom-up perceptual information enabling the discrimination of an emotion. A dichotic listening task was used to examine alternative explanations of valence effects in emotion perception. Emotional sentences (spoken in a happy or sad tone of voice), and morphed-happy and morphed-sad sentences (which blended a neutral version of the sentence with the pitch of the emotion sentence) were paired with neutral versions of each sentence and presented dichotically. A control condition was also used, consisting of two identical neutral sentences presented dichotically, with one channel arriving before the other by 7 ms. In support of the RH hypothesis there was a left ear advantage for the perception of sad and happy emotional sentences. However, morphed sentences showed no ear advantage, suggesting that the RH is specialised for the perception of genuine emotions and that a laterality effect may be a useful tool for the detection of fake emotion. Finally, for the control condition we obtained an interaction between the expected emotion and the effect of ear lead. Participants tended to select the ear that received the sentence first, when they expected a `sad¿ sentence, but not when they expected a `happy¿ sentence. The results are discussed in relation to the different theoretical explanations of valence laterality effects in emotion perception.

Thesis (Ph. D.)--Ohio State University, 2004.
Title from first page of PDF file. Document formatted into pages; contains xx, 229 p. : ill. Advisor: Shari Speer, Dept. of Linguistics. Includes bibliographical references (p. 226-229).

Thesis (Ph. D.)--Ohio State University, 2005.
Title from first page of PDF file. Document formatted into pages; contains xvii, 190 p.; also includes graphics. Includes bibliographical references (p. 181-190). Available online via OhioLINK's ETD Center

This thesis is about the linguistic expression of focus in Ktunaxa. It describes forms for expressing focus using word order and prosody, and describes the function of several focus-sensitive operators in the language. The methodologies used to examine these topics are respectively i) an experiment in which Ktunaxa speakers answer questions about pictures, and ii) classical fieldwork with a fluent Ktunaxa speaker. These methodologies enable different types of research into focus. The experimental data speaks to the form of focus in Ktunaxa; assuming that answers to questions require the expression of focus, unscripted answers provide insight into how Ktunaxa speakers highlight information in an utterance (i.e. focus-mark the relevant constituents). Meanwhile, the classical fieldwork speaks to the functioning of a set of focus-sensitive operators in Ktunaxa; these operators are known to be sensitive to context, but their precise semantics have not been described before. Several theories underlie this project: first, the tradition of Chomskyan generative linguistics provides a framework for describing linguistic structures and relationships; second, the theory of Alternative Semantics (Rooth 1985) formalizes focus as a way of invoking one particular member of a set of alternatives; and thirdly, the autosegmental metrical theory of intonational phonology (Ladd 2008, i.a.), maps prosodic components (pitch accents, lengthening, and stress) onto phonological components (syllables and words). The key findings are as follows: 1. Ktunaxa answers have a default word order of Subject-Verb-Object, contrary to patterns emerging in texts; 2. Word order changes relative to the type of question asked: foci are slightly more likely to be sentence-initial; 3. Ktunaxa employs prosody to mark focus in answers: foci are louder and higher in pitch; 4. Some focus-sensitive operators also trigger prosodic cues in their associates. Further work is needed to fully describe Ktunaxa prosody, and to confirm whether these patterns hold under other experimental conditions, for different types of focus, and for constituents other than nominals. Nevertheless, this study contributes to the documentation of Ktunaxa, and more generally expands the knowledge of how focus is expressed cross-linguistically, particularly in the languages of the Pacific Northwest.
Arts, Faculty of
Linguistics, Department of
Graduate

THESIS ABSTRACT Emotional prosody processing impairment is proposed to be a main contributing factor for the formation of auditory verbal hallucinations in patients with schizophrenia. In order to evaluate such assumption, five experiments in healthy, highly schizotypal and schizophrenia populations are presented. The first part of the thesis seeks to reveal the neural underpinnings of emotional prosody comprehension (EPC) in a non-clinical population as well as the modulation of prosodic abilities by hallucination traits. By revealing the brain representation of EPC, an overlap at the neural level between EPC and auditory verbal hallucinations (AVH) was strongly suggested. By assessing the influence of hallucinatory traits on EPC abilities, a continuum in the schizophrenia spectrum in which high schizotypal population mirrors the neurocognitive profile of schizophrenia patients was established. Moreover, by studying the relation between AVH and EPC in non-clinical population, potential confounding effects of medication influencing the findings were minimized. The second part of the thesis assessed two EPC related abilities in schizophrenia patients with and without hallucinations. Firstly, voice identity recognition, a skill which relies on the analysis of some of the same acoustical features as EPC, has been evaluated in patients and controls. Finally, the last study presented in the current thesis, assessed the influence that implicit processing of emotional prosody has on selective attention in patients and controls. Both patients studies demonstrate that voice identity recognition deficits as well as abnormal modulation of selective attention by implicit emotion prosody are related to hallucinations exclusively and not to schizophrenia in general. In the final discussion, a model in which EPC deficits are a crucial factor in the formation of AVH is evaluated. Experimental findings presented in the previous chapters strongly suggests that the perception of prosodic features is impaired in patients with AVH, resulting in aberrant perception of irrelevant auditory objects with emotional prosody salience which captures the attention of the hearer and which sources (speaker identity) cannot be recognized. Such impairments may be due to structural and functional abnormalities in a network which comprises the superior temporal gyrus as a central element.

It has long been argued that prosodic cues may facilitate syntax acquisition (e.g., Morgan, 1986). Previous studies have shown that infants are sensitive to violations of typical correlations between clause-final prosodic cues (Hirsh-Pasek et al., 1987) and that prosody facilitates memory for strings of words (Soderstrom et al., 2005). This dissertation broaches the question of whether children can use this information in syntax acquisition by asking if learners can use the prosodic correlates of clauses to locate syntactic constituents. One property of certain syntactic constituents in natural languages is that they can move, so learning of constituency was inferred if participants treated prosodically-grouped words as cohesive, moveable chunks. In Experiment 1, 19-month-olds were familiarized with sentences from an artificial grammar with either 1-clause or 2-clause prosody. The infants from the 2-clause group later recognized the prosodically-marked clauses when they had moved to a new position in the sentence and had a new acoustic contour. Adults in Experiment 2 showed similar learning, although their judgments also rely on recognition of perceptually-salient words at prosodic boundaries. Subsequent experiments explored the mechanisms underlying this prosodic bootstrapping by testing Japanese-acquiring infants on English-based stimuli (Experiment 3) and English-acquiring infants on Japanese-based stimuli (Experiment 4). Infants were able to locate constituent-like groups of words with both native and non-native prosody, suggesting that the acoustic correlates of prosody are sufficiently robust across languages that they can be used in early syntax acquisition without extensive exposure to language-specific prosodic features. On the other hand, adults (Experiment 5) are less flexible, and are only able to use prosody consistent with their native language, suggesting an age- or experience-related tuning of the prosodic perceptual mechanism. This dissertation supports prosody as an important cue that allows infants and young children to break into syntax even before they understand many words, and helps explain the rapid rate of syntax acquisition.

Extract: [...]It is true that scholars concentrate on a certain linguistic level in order to reach the greatest depth in their research. But this general stance should not lead to a complete neglect of other levels. When considering a multi-level phenomenon such as language contact and shift, concentration on a single linguistic level can have the unintended and unfortunate consequence of missing linguistically significant generalisations. This is especially true of the main division of linguistic research into a phonological and a grammatical camp, where syntacticians miss phonological generalisations and phonologists syntactic ones. In the present paper the interrelationship of syntax and prosody is investigated with a view to explaining how and why certain transfer structures from Irish became established in Irish English. In this context, the consideration of prosody can be helpful in explaining the precise form of transfer structures in the target variety, here vernacular Irish English. The data for the investigation will consider well-known features of this variety, such as unbound reflexives, non-standard comparatives and tag questions. Furthermore, the paper points out that, taking prosodic patterns into account, can help in extrapolating from individual transfer to the community- wide establishment of transfer structures. In sum, prosody is an essential element in any holistic account of language contact and shift.[...]

This thesis presents an analysis of certain punctuation devices such as parenthesis, italics and emphatic spellings with respect to their acoustic correlates in read speech. The class of punctuation devices under investigation are referred to as prosodic markers. The thesis therefore presents an analysis of features of the spoken language which are represented symbolically in text. Hence it is a characterization of aspects of the spoken language which have been transcribed or symbolized in the written medium and then translated back into a spoken form by a reader. The thesis focuses in particular on the analysis of parenthesis, the examination of encoded prominence and emphasis, and also addresses the use of paralinguistic markers which signal attitude or emotion. In an effort to avoid the use of self constructed or artificial material containing arbitrary symbolic or prosodic encodings, all material used for empirical analysis was taken from examples of electronic written exchanges on the Internet, such as from electronic mail messages and from articles posted on electronic newsgroups and news bulletins. This medium of language, which is referred to here as written conversation, provides a rich source of material containing encoded prosodic markers. These occur in the form of 'smiley faces' expressing attitudes or feelings, words highlighted by a number of means such as capitalization, italics, underscore characters, or asterisks, and in the form of dashes or parentheses, which provide suggestions on how the information in a text or sentence may be structured with regard to its informational content. Chapter 2 investigates in detail the genre of written conversation with respect to its place in an emerging continuum between written and spoken language, concentrating on transcriptional devices and their function as indicators of prosody. The implications these symbolic representations bear on the task of reading, by humans as well as machines, are then examined.

In this thesis a study of discourse segmenting is carried out, which investigates both segment boundaries and segment content. The results are related to discourse theory. We study the questions of how the prosody and the text structure influence subjects' annotations of discourse boundaries and discourse prominence. The hypothesis was that the annotations would be influenced by the discourse type. Two studies were carried out. 1) a study of boundary annotation, 2) a study of prominence annotation. All studies were made on four different discourse types, scripted and spontaneous monologue and scripted and spontaneous dialogue. In addition the annotations were carried out under two different conditions 1) based on transcripts alone and 2) based on transcripts together with access to the speech signal. The results indicate that the boundary annotations were less dependent on the speech signal than the prominence annotations. It seems that subjects have segmented on the basis of the text structure, while prominence to a great extent was annotated on the basis of the prosody. In the case of boundary markings the boundary context in terms of parts of speech differs across speaking styles, which is not the case for the prominences. A separate study of segment intentions was also made, and it was found that the interpretation of a specific intention, questions, seems to be arrived at primarily on the basis of the text structure. However, in some cases also the prosody affects the annotations. The picture that emerges indicates a distribution of labour between text structure and prosody, governed by the principle of economy. In cases where the boundaries were less well definied, as in e.g. spontaneous monologue, the pattern of the prominences was clearer. In cases where the boundaries were more clearly indicated, as in read aloud text, the prominences were less clearly communicated. The findings were interpreted within Grosz and Sidner's (1986) discourse theory. It is suggested that differences in the segmenting strategy originating from the interaction of text structure and prosody can be expressed as differences in the contributions from the different components of discourse suggested in the framework of Grosz and Sidner (1986).

För att köpa boken skicka en beställning till exp@ling.su.se/ To order the book send an e-mail to exp@ling.su.se

This thesis explores the variable behavior of bare nouns in Persian. Bare singular nouns realize different grammatical functions, including subject, object and indirect object. They receive different interpretations, including generic, definite and existential readings. However, the task of understanding the reasons for, and limits on, this variation cannot be achieved without understanding a number of pivotal features of Persian sentential architecture, including Information Structure, prosody, word order, and the functions of various morphological markers in Persian. After a brief introduction, chapters 2-3 deal with bare noun objects, firstly comparing them with nominals marked with indefinite morpheme -i suffixed to the noun, and the determiner yek. A bare noun object differs from morphologically marked nominals as it shows properties associated with noun incorporation in the literature (chapter 2). Of particular interest are the discourse properties of these ‘quasi-incorporated’ nominals. With respect to the discourse transparency of Incorporated Nominals, Persian belongs to the class of discourse opaque languages within Mithun’s classification (1984). However, under certain circumstances, Persian bare nouns show discourse transparency. These circumstances are examined in chapter 3, and it is proposed that bare nouns do introduce a number neutral discourse referent. There are no overt anaphoric expressions that could match such number-neutral antecedents in Persian. But covert anaphora lack number features, and hence can serve as means to pick up a number-neutral discourse referent. Also, in case world knowledge tells us that the number-neutral discourse referent is anchored to an atomic entity or to a collection, then an overt singular pronoun or an overt plural pronoun might fit the combined linguistic and conceptual requirements, and may be used to pick up the number-neutral discourse referent. This proposal is phrased within Discourse Representation Theory. In the second half of the dissertation, the interpretation of bare nouns in different positions and with different grammatical functions are discussed. Under the independently supported hypothesis of position>interpretation mapping developed by Diesing (1992), we will see the role of the suffix -ra in indicating that an object has been moved out of VP. Following Diesing, I assume that VP-internal variables are subject to an operation of Existential Closure. In many cases, VP-external –ra-marked objects have a different interpretation to their VP-internal, non-ra-marked, counterparts, because of escaping Existential Closure. For subjects, there is no morphological marking corresponding to –ra on objects, and we have to rely on prosody and word order to determine how a VP is interpreted using theories of the interaction of accent and syntactic structure. We assume that VP-internal subjects exist, under two independent but converging assumptions. The first is prosodic in nature: Subjects can be accented without being narrowly focused; theories of Persian prosody predict then that there is a maximal constituent that contains both the subject and the verb as its head. The second is semantic in nature: Bare nouns require an external existential closure operation to be interpreted existentially, and we have to assume existential closure over the VP for our analysis of the interpretation of objects. So, this existential closure would provide the necessary quantificational force for bare noun subjects as well. It is proposed that both subject and object originate within the VP, and can move out to the VP-external domain. The motivation for these movements are informational-structural in nature, relating in particular to the distinctions between given and new information, and default and non-default information structure.

Thesis (Ph.D.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts & Sciences, 1999.
Includes bibliographical references (p. 209-226).
This thesis links processing in working memory to prosody in speech, and links different working memory capacities to different prosodic styles. It provides a causal account of prosodic differences and an architecture for reproducing them in synthesized speech. The implemented system mediates text-based information through a model of attention and working memory. The main simulation parameter of the memory model quantifies recall. Changing its value changes what counts as given and new information in a text, and therefore determines the intonation with which the text is uttered. Other aspects of search and storage in the memory model are mapped to the remainder of the continuous and categorical features of pitch and timing, producing prosody in three different styles: for small recall values, the exaggerated and sing-song melodies of children's speech; for mid-range values, an adult expressive style; for the largest values, the prosody of a speaker who is familiar with the text, and at times sounds bored or irritated. In addition, because the storage procedure is stochastic, the prosody from simulation to simulation varies, even for identical control parameters. As with with human speech, no two renditions are alike. Informal feedback indicates that the stylistic differences are recognizable and that the prosody is improved over current offerings. A comparison with natural data shows clear and predictable trends although not at significance. However, a comparison within the natural data also did not produce results at significance. One practical contribution of this work is a text mark-up schema consisting of relational annotations to grammatical structures. Another is the product - varied and plausible prosody in synthesized speech. The main theoretical contribution is to show that resource-bound cognitive activity has prosodic correlates, thus providing a rationale for the individual and stylistic differences in melody and rhythm that are ubiquitous in human speech.
by Janet Elizabeth Cahn.
Ph.D.

This study examines nonnative prosody and intelligibility. Past research has suggested that prosody that is unfamiliar or inappropriate in some way can adversely affect the intelligibility of speech (e.g., Hahn, 2004; Tajima, Port & Dalby, 1997; Grover, Jamieson & Dobrovlosky, 1987; Field, 2005). In this study, the effect of overall prosody rather than the effects of particular prosodic features is analyzed. Fifteen native and 15 nonnative speakers were recorded reading identical sets of ambiguous sentences while viewing cartoon drawings. Cartoons viewed by 8 members of each speaker group portrayed one of the two possible interpretations (“Version A”) for each sentence. Cartoons seen by the remaining 7 speakers of each group showed the alternative (“Version B”) interpretations. Recordings were divided and rearranged into new soundtracks containing a different speaker for every sentence. Fifteen native listeners viewed documents showing the Version A and Version B cartoons of each sentence side by side while listening to the new soundtracks, indicating which of the two cartoon versions they believed each speaker had viewed when recording. Listeners identified the cartoon seen by the speaker significantly less often when the speaker was a nonnative, suggesting a relationship between speaker type and intelligibility. Results were further subdivided into 4 categories of structural ambiguity. Of those, compound noun vs. adjective + noun ambiguities (e.g. White House vs. white house) accounted for most of listeners’ errors in disambiguation.

Des projets de science participative (SP) ont stimulé la recherche dans plusieurs disciplines au cours des dernières années. Des citoyens scientifiques contribuent à cette recherche en effectuant des tâches cognitives, favorisant l'apprentissage, l'innovation et l'inclusion. Bien que le crowdsourcing ait servi à recueillir des annotations structurelles en musique, la SP reste sous-utilisée pour étudier l'expressivité musicale. On introduit un nouveau protocole d'annotation pour capturer la prosodie musicale, associée aux variations acoustiques introduites par les interprètes pour rendre la musique expressive. Notre méthode descendante, centrée sur l'humain, donne la priorité à l'auditeur dans la production d'annotations des fonctions prosodiques de la musique. On se concentre sur la segmentation et la proéminence, qui véhiculent la structure et l'affect. Ce protocole fournit un cadre de SP et une approche expérimentale pour réaliser des études systématiques et extensibles. On met en œuvre ce protocole d'annotation dans CosmoNote, un logiciel web personnalisable, conçu pour faciliter l'annotation de structures musicales expressives. CosmoNote permet aux utilisateurs d'interagir avec des couches visuelles, y compris la forme d'onde, les notes enregistrées, les attributs audio extraits et les caractéristiques de la partition. On peut placer des frontières de niveaux différents, des régions, des commentaires et des groupes de notes. On a mené deux études visant à améliorer le protocole et la plateforme. La première, examine l'impact des stimuli auditifs et visuels simultanés sur les frontières de segmentation. On compare les différences dans les distributions de frontières dérivées d'informations intermodales (auditives et visuelles) et unimodales (auditives ou visuelles). Les distances entre les distributions unimodales-visuelles et intermodales sont plus faibles qu'entre les distributions unimodales-auditives et intermodales. On montre que l'ajout de visuels accentue les informations clés et fournit un échafaudage cognitif aidant à marquer clairement les frontières prosodiques, bien qu'ils puissent détourner l'attention de structures spécifiques. À l'inverse, sans audio, la tâche d'annotation devient difficile, masquant des indices subtils. Malgré leur exagération ou inexactitude, les repères visuels sont essentiels pour guider les annotations de frontières en interprétation, ce qui améliore les résultats globaux. La deuxième étude utilise tous les types d'annotations de CosmoNote et analyse comment les participants annotent la prosodie musicale, avec des instructions minimales ou détaillées, dans un cadre d'annotations libres. On compare la qualité des annotations entre musiciens et non-musiciens. On évalue la composante de SP dans un cadre écologique où les participants sont totalement autonomes dans une tâche où le temps, l'attention et la patience sont valorisés. On présente trois méthodes basées sur des étiquettes d'annotation, des catégories et des propriétés communes pour analyser et agréger les données. Les résultats montrent une convergence dans les types d'annotations et les descriptions utilisées pour marquer les éléments musicaux récurrents, pour toute condition expérimentale et aptitude musicale. On propose des stratégies pour améliorer le protocole, l'agrégation des données et l'analyse dans des applications à grande échelle. Cette thèse enrichit la représentation et la compréhension des structures en musique interprétée en introduisant un protocole et une plateforme d'annotation, des expériences adaptables et des méthodes d'agrégation et d'analyse. On montre l'importance du compromis entre l'obtention de données plus simples à analyser et celle d'un contenu plus riche, capturant une pensée musicale complexe. Notre protocole peut être généralisé aux études sur les décisions d'interprétation afin d'améliorer la compréhension des choix expressifs dans l'interprétation musicale
The proliferation of citizen science projects has advanced research and knowledge across disciplines in recent years. Citizen scientists contribute to research through volunteer thinking, often by engaging in cognitive tasks using mobile devices, web interfaces, or personal computers, with the added benefit of fostering learning, innovation, and inclusiveness. In music, crowdsourcing has been applied to gather various structural annotations. However, citizen science remains underutilized in musical expressiveness studies. To bridge this gap, we introduce a novel annotation protocol to capture musical prosody, which refers to the acoustic variations performers introduce to make music expressive. Our top-down, human-centered method prioritizes the listener's role in producing annotations of prosodic functions in music. This protocol provides a citizen science framework and experimental approach to carrying out systematic and scalable studies on the functions of musical prosody. We focus on the segmentation and prominence functions, which convey structure and affect. We implement this annotation protocol in CosmoNote, a web-based, interactive, and customizable software conceived to facilitate the annotation of expressive music structures. CosmoNote gives users access to visualization layers, including the audio waveform, the recorded notes, extracted audio attributes (loudness and tempo), and score features (harmonic tension and other markings). The annotation types comprise boundaries of varying strengths, regions, comments, and note groups. We conducted two studies aimed at improving the protocol and the platform. The first study examines the impact of co-occurring auditory and visual stimuli on segmentation boundaries. We compare differences in boundary distributions derived from cross-modal (auditory and visual) vs. unimodal (auditory or visual) information. Distances between unimodal-visual and cross-modal distributions are smaller than between unimodal-auditory and cross-modal distributions. On the one hand, we show that adding visuals accentuates crucial information and provides cognitive scaffolding for accurately marking boundaries at the starts and ends of prosodic cues. However, they sometimes divert the annotator's attention away from specific structures. On the other hand, removing the audio impedes the annotation task by hiding subtle, relied-upon cues. Although visual cues may sometimes overemphasize or mislead, they are essential in guiding boundary annotations of recorded performances, often improving the aggregate results. The second study uses all CosmoNote's annotation types and analyzes how annotators, receiving either minimal or detailed protocol instructions, approach annotating musical prosody in a free-form exercise. We compare the quality of annotations between participants who are musically trained and those who are not. The citizen science component is evaluated in an ecological setting where participants are fully autonomous in a task where time, attention, and patience are valued. We present three methods based on common annotation labels, categories, and properties to analyze and aggregate the data. Results show convergence in annotation types and descriptions used to mark recurring musical elements across experimental conditions and musical abilities. We propose strategies for improving the protocol, data aggregation, and analysis in large-scale applications. This thesis contributes to representing and understanding performed musical structures by introducing an annotation protocol and platform, tailored experiments, and aggregation/analysis methods. The research shows the importance of balancing the collection of easier-to-analyze datasets and having richer content that captures complex musical thinking. Our protocol can be generalized to studies on performance decisions to improve the comprehension of expressive choices in musical performances