A framework for annotating information structure in discourse

Calhoun, Sasha; Nissim, Malvina; Steedman, Mark; Brenier, Jason

doi:10.3115/1608829.1608836

Cited by 24 publications

(30 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Several possible reasons were marked: correction for when the speaker intends to correct or clarify a previous word or phrase; contrastive when the word is directly differentiated from a previous topical or semantically-related word; subset when it refers to a member of a more general set mentioned in the surrounding context; adverbial when a focus-sensitive adverb such as "only" or "even" is associated with the word being annotated; answer when the word completes a question by the other speaker; background when the word is not intended to be salient and nonapplic for filler phrases such as "in fact", "I mean", etc. A complete description of the annotation guidelines can be found in [8]. ANIMACY Each noun and pronoun is labeled for the animacy of its referent [9].…”

Section: Data and Featuresmentioning

confidence: 99%

The (Non)utility of Linguistic Features for Predicting Prominence in Spontaneous Speech

Brenier

Nenkova

Kothari

et al. 2006

2006 IEEE Spoken Language Technology Workshop

Self Cite

View full text Add to dashboard Cite

Conversational speech is characterized by prosodic variability which makes pitch accent prediction for this genre especially difficult. The linguistic literature points out that complex features such as information status, contrast and animacy help predict pitch accent placement. In this paper, we use a corpus annotated for such features to determine if they improve prominence prediction over traditional shallow features such as frequency and part-of-speech, or over new ones that we introduce. We demonstrate that while correlated with prominence, complex linguistic features do not improve prediction accuracy. Furthermore, the performance of our classifier is quite close to the ceiling defined by variability in human accent placement. An oracle experiment demonstrates, though, that at least some accuracy improvement is still possible.

show abstract

Section: Data and Featuresmentioning

confidence: 99%

The (Non)utility of Linguistic Features for Predicting Prominence in Spontaneous Speech

Brenier

Nenkova

Kothari

et al. 2006

2006 IEEE Spoken Language Technology Workshop

Self Cite

View full text Add to dashboard Cite

show abstract

“…Por lo tanto, más que hablar de una categorización excluyente de tema o rema para cada proposición o segmento de discurso, se trata de una gradación jerárquica y solidaria de subtemas entorno a un tema. De esta forma, para cada proposición temática o subtemática, habrá elementos de foco, background o mediated information (Calhorun, 2012;Calhoun et al, 2005;Steedman, 2000;Vallduví y Vilkuna, 1998).…”

Section: La Estructura Informativa De La Noticiaunclassified

“…En suma, se debe encontrar esa estructura de contextosinterrogaciones temáticos que se articulen jerárquica y solidariamente según criterios sintácticos y semánticos (Ladd, 1980: 85 En segundo lugar, se deben distinguir las "proposiciones tema" de las proposiciones rema" según sea la intención reconocida en el emisor y en base al conocimiento del receptor que la estructura discursiva presupone: Y en tercer lugar, se localiza la entidad representante de la información nueva, vieja y mediada en las partes temáticas o de rema del discurso. Para ello, se deben localizar las palabras-clave novedad, conocimiento y discursividad (Calhoun et al, 2005), esto es, las palabras-clave que por su uso comunicativo en los medios tienen una gran fuerza expresiva. Esa palabra puede ser un sustantivo, verbo, adverbio… o frase hecha, y es capaz por ella misma de comunicar la noticiabilidad de la información:…”

Section: Análisis Discursivounclassified

Caracterización prosódica del foco de tema y rema en las noticias

Manchón

2014

Círculo De Lingüística Aplicada a La Comunicación

View full text Add to dashboard Cite

El estudio de la prosodia con función paralingüística avanza en la actualidad en dos frentes, el análisis discursivo de la estructura informativa y el análisis prosódico de las categorías textuales. Por una parte, el discurso oral informativo en los medios representa un tipo de discurso estable y modélico, por lo que facilita el análisis discursivo. Por otra parte, la locución informativa constituye un reto dadas las especiales condiciones comunicativas del acto. Este estudio analiza un corpus de cinco noticias emitidas en tres canales de televisión y por parte de tres locutores diferentes. Primero, los segmentos discursivos son definidos como unidades de tema y rema. A continuación se localizan los focos de información nueva, vieja y mediada. Y por último, se mide el rango tonal (porcentual), la duración de la sílaba tónica y el pico tonal de un total de 228 acentos. El análisis factorial ANOVA muestra una caracterización del foco de rema mediante alargamiento silábico y una caracterización del foco discursivo mediante pico tonal.mas: prosodia del foco 36Estos rasgos generales se acentúan o matiza en los tres estilos de locución. Mientras el estilo espontáneo usa menos el pico tonal y se limita al alargamiento silábico para marcar la información nueva, el estilo sensacionalista aplica el pico tonal para los tres tipos de focos y el estilo referencial público marca el rema con la duración y el acento discursivo con un pico tonal como inflexión melódica de su cantinela.Palabras clave: estructura informativa, foco, prosodia, locución AbstractProsodic characterization of theme and rheme foci in news. The study of prosody with regard to the paralinguistic function is progressing in two areas, the discursive analysis of the information structure, and the prosodic analysis of textual categories. On one hand, the oral information discourse in the media is an established exemplary type of discourse thus its analysis is patterned-oriented. On the other, the news locution is a still a sort of an announcers' challenge due to the most peculiar communication conditions of the act. This study analyses a corpus of five news announced in three television channels by three different news readers. First, the discourse segments are defined as theme and rheme units. Next, the new, old and meditated information foci are spotted.And lastly, it is measured the pitch range (in percentages), the duration of the stressed syllable and the pitch peak of a corpus of 228 foci. The ANOVA factorial analysis characterizes the rheme focus by the syllable lengthening and characterizes the mediated focus by the pitch peak. These global features are stressed or nuanced by the announcer's personal style. Whereas the spontaneous style does not make a great use of the pitch peak and is centered on the duration structure to mark the news information, the sensationalist style applies the pitch peak to all three types of accents, and the standard public style marks the rheme focus with duration and the discursive focus with the pitch peak as a turning po...

show abstract

“…Our experiment used a subset of the Switchboard corpus [3] that had been annotated with syntactic structure [9] and information structure [10]. We selected sentences containing just one contrastive word pairs.…”

Section: Data Setmentioning

confidence: 99%

Detection and emphatic realization of contrastive word pairs for expressive text-to-speech synthesis

Meng

et al. 2012

2012 8th International Symposium on Chinese Spoken Language Processing

View full text Add to dashboard Cite

This paper addresses the problem of automatic detection of contrastive word pairs and their acoustic realization in emphasis for expressive text-to-speech (TTS) synthesis in English. Support vector machines (SVMs) have been used to automatically detect contrastive word pairs from lexical features, syntactic dependencies and semantic relations. A much better performance is achieved by adding accent ratio and word identity features. Hidden Markov model (HMM) based speech synthesis is then used to generate emphatic speeches by putting emphasis on the detected contrastive word pairs. Subjective experiments show that most of the listeners consider putting emphasis on contrastive word pairs is more acceptable than on non-contrastive word pairs. This indicates the importance of the accurate detection of contrastive word pairs.Index Terms-contrast, expressive text-to-speech (TTS) synthesis, support vector machines (SVMs), hidden Markov model (HMM) based speech synthesis

show abstract

A framework for annotating information structure in discourse

Cited by 24 publications

References 23 publications

The (Non)utility of Linguistic Features for Predicting Prominence in Spontaneous Speech

The (Non)utility of Linguistic Features for Predicting Prominence in Spontaneous Speech

Caracterización prosódica del foco de tema y rema en las noticias

Detection and emphatic realization of contrastive word pairs for expressive text-to-speech synthesis

Contact Info

Product

Resources

About