TY - GEN

T1 - Can We Detect Irony in Speech Using Phonetic Characteristics Only? – Looking for a Methodology of Analysis

AU - Skrelin, Pavel

AU - Kochetkova, Uliana

AU - Evdokimova, Vera

AU - Novoselova, Daria

N1 - Funding Information: This research was supported by the RFBR grant ? 20-012-00552. Publisher Copyright: © 2020, Springer Nature Switzerland AG. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.

PY - 2020

Y1 - 2020

N2 - The current paper aims to investigate the perception of Russian ironic laboratory speech and to develop a methodological basis for the further research on perceptually relevant acoustic cues of irony. On the first stage of analysis we created short monologues and dialogues including ironic and homonymous non-ironic sentences. Two Russian native male speakers read this material. The recording was accomplished in a sound booth. Then, randomized fragments of the recordings were suggested to native listeners in an auditory perception experiment. The mean intensity, duration of the stressed vowel and F0 range in the reliably identified target fragments were analyzed, as well as functional intonation models in terms of the system suggested by E. Bryzgunova. The results showed that the listeners are able to detect irony without any contextual or lexical marker of it. The acoustic analysis of the reliably identified stimuli demonstrated the difference in intensity level and stressed vowel duration between ironic and non-ironic utterances. The obtained results allow creating a methodology of data collection and validation for the construction of ironic speech corpus and its analysis. They may also have applications in teaching Russian as a second language, human-machine communication and improving natural language processing systems.

AB - The current paper aims to investigate the perception of Russian ironic laboratory speech and to develop a methodological basis for the further research on perceptually relevant acoustic cues of irony. On the first stage of analysis we created short monologues and dialogues including ironic and homonymous non-ironic sentences. Two Russian native male speakers read this material. The recording was accomplished in a sound booth. Then, randomized fragments of the recordings were suggested to native listeners in an auditory perception experiment. The mean intensity, duration of the stressed vowel and F0 range in the reliably identified target fragments were analyzed, as well as functional intonation models in terms of the system suggested by E. Bryzgunova. The results showed that the listeners are able to detect irony without any contextual or lexical marker of it. The acoustic analysis of the reliably identified stimuli demonstrated the difference in intensity level and stressed vowel duration between ironic and non-ironic utterances. The obtained results allow creating a methodology of data collection and validation for the construction of ironic speech corpus and its analysis. They may also have applications in teaching Russian as a second language, human-machine communication and improving natural language processing systems.

KW - Acoustic analysis

KW - Auditory perception experiment

KW - Intensity

KW - Intonation models

KW - Irony

KW - Phonetics

KW - Pitch

UR - http://www.scopus.com/inward/record.url?scp=85092889586&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/aaa9eee9-7e52-3412-885f-b002301c62df/

U2 - 10.1007/978-3-030-60276-5_52

DO - 10.1007/978-3-030-60276-5_52

M3 - Conference contribution

AN - SCOPUS:85092889586

SN - 9783030602758

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 544

EP - 553

BT - Speech and Computer - 22nd International Conference, SPECOM 2020, Proceedings

A2 - Karpov, Alexey

A2 - Potapova, Rodmonga

PB - Springer Nature

T2 - 22nd International Conference on Speech and Computer

Y2 - 7 October 2020 through 9 October 2020

ER -