Prosody of Ukrainian Emotional Speech (Experimental Phonetic Research)

Александрович, Марина Василівна (2011) Prosody of Ukrainian Emotional Speech (Experimental Phonetic Research) Doctoral thesis, Київський університет імені Тараса Шевченка.

[img]
Preview
Text
M_Aleksandrovych_Avtoreferat_IJ.pdf

Download (6MB) | Preview

Abstract

The thesis is devoted to the research of Ukrainian emotional speech prosody based on spontaneous language. The main objects are to investigate prosodic features of emotional utterances, determine regular occurrences of its implementation in Ukrainian spontaneous speech, construct the prosodic models of emotional speech. In order to define specific prosodic features of emotional utterances the research technique which provides the complex analysis of text intonation was developed. It includes: 1) recordings of spontaneous speech and selection appropriate fragments for experimental phonetic analysis; 2) auding experiment for perception and recognition of emotional utterances with two groups of participants (ordinary listeners and expert phoneticians); 3) acoustic analysis of emotional utterances (their tonal, temporal and dynamical parameters) via specific audio software (Praat, Sound Forge). Research data are spontaneous speech of 10 speakers, recorded in direct communication, its recording time is 8 hours. These data make it possible to investigate prosodic parameters of natural emotion in speech as opposed to simulated implementation of emotional speech. After examination of different emotion classification in psychological and linguistic studies, release version of proper classification was found. It includes 19 kinds of emotional utterances, which were formed into classes of positive, negative and contextual emotions. This classification gave basis to construct prosodic models of emotional speech. The set of prosodic parameters, relevant exactly for emotional speech investigation, was summarized from the works of prosody researchers as well. And it became the basis for this research, where data are investigated using 17 items of tonal, temporal and dynamical parameters. The main idea of auding experiment was to determine the influence of context on emotion identification. And this experiment showed that speech context had influence on perception of emotion: emotion changed in 30% of cases for phoneticians and 40% for ordinary listeners because of speech context. The results of another auding experiment proved that such emotions as joy, interest, surprise and disgust recognized the best by listeners. Comparative analysis of 1) spontaneous emotional speech and spontaneous neutral speech and 2) spontaneous emotional speech and delivered emotional speech were held to reveal different prosodic characteristics of each type of speech. It was discovered that emotional speech in contrast to neutral speech is notable for its more frequent usage of fast or slow tempo, higher intensity and extended pitch range. Results received from the experimental phonetic analysis of emotional speech prosody formed the principles of its prosodic models, which combined tonal, temporal and dynamical circuit. Prosodic model of positive emotion gladness consists of: 1) rising prehead, 2) middle pitch level of syntagma’s beginning, 3) falling-rising or falling scale, 4) straight or falling terminal tone, 5) high syntagma pitch level, 6) extended or narrowed tone range, 7) rising or rising-falling nuclear tone, 8) rising interval “prehead – scale”, 9) rising interval “pre-terminal part – nuclear”, 10) high level of pitch maximum and its location on the word with logical stress, 11) fast tempo, 12) narrow intensity range, 13) location of intensity maximum on the word with logical stress. Prosodic model of negative emotion displeasure consists of: 1) rising prehead, 2) low or middle pitch level of syntagma’s beginning, 3) falling scale, 4) falling terminal tone, 5) middle or high syntagma pitch level, 6) narrowed tone range, 7) rising or rising-falling nuclear tone, 8) rising interval “prehead – scale”, 9) falling interval “pre-terminal part – nuclear”, 10) middle level of pitch maximum and its location on the word with logical stress, 11) faster tempo, 12) narrow or middle intensity range, 13) location of intensity maximum on the word with logical stress. Prosodic model of contextual emotion irony consists of: 1) falling prehead, 2) middle pitch level of syntagma’s beginning, 3) falling scale, 4) falling or straight terminal tone, 5) middle or extra-high syntagma pitch level, 6) extended tone range, 7) falling or rising-falling nuclear tone, 8) rising interval “prehead – scale”, 9) rising interval “pre-terminal part – nuclear”, 10) extra-high level of pitch maximum and its location on the word with logical stress, 11) fast tempo, 12) middle intensity range, 13) location of intensity maximum on the word with logical stress. Emotions that revealed to be the best for modelling are: gladness, approval, irony, interest, displeasure, disapproval, contempt.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: prosody; emotional speech; spontaneous speech; emotional utterances; tonal, temporal and dynamical circuit; speech prosodic parameters; prosodic models
Subjects: Abstracts > Спецради у інших ВНЗ
Divisions: Institutes > Інститут журналістики > Кафедра видавничої справи
Depositing User: Марина Василівна Александрович
Date Deposited: 02 Jan 2019 08:33
Last Modified: 02 Jan 2019 08:33
URI: http://elibrary.kubg.edu.ua/id/eprint/26050

Actions (login required)

View Item View Item