Acoustic Features of Filled Pauses in Polish Task-Oriented Dialogues

Abstract

Filled pauses (FPs) have proved to be more than valuable cues to speech production processes and important units in discourse analysis. Some aspects of their form and occurrence patterns have been shown to be speaker-and language-specific. In the present study, basic acoustic properties of FPs in Polish task-oriented dialogues are explored. A set of FPs was extracted from a corpus of twenty taskoriented dialogues on the basis of available annotations. After initial scrutiny and selection, a subset of the signals underwent a series of pitch, formant frequency and voice quality analyses. A significant amount of variation found in the realisations of FPs justifies their potential application in speaker recognition systems. Regular monosegmental FPs were confirmed to show relatively stable basic acoustic parameters, which allows for their easy identification and measurements but it may result in less significant differences among the speakers.Keywords: filled pauses, paralinguistics, dialogue, acoustic properties, Polish. Filled pauses as paralinguistic components of spoken utterancesParalinguistic features of utterances (henceforth PLFs) were extensively explored by linguists early in the 1960s (Trager, 1960(Trager, , 1961(Trager, , 1964 Crystal, 1963 Crystal, , 1966 Crystal, , 1974 Crystal, , 1975. Since then, they have been acknowledged as an indispensible component of spoken communication. They may reveal important facts on the speaker him/herself, including his/her age, gender, origin, social background or education, as well as his/her present emotional state, attitude towards the topic of conversation or towards the conversational partner (Gobl, Ni Chasaide, 2003;Wallbott, Scherer, 1979;Pakosz, 1982;Bortfeld, 2001). Some of them may also provide cues in the analysis of the process of speech production (e.g., disfluencies (Fromkin, 1971(Fromkin, , 1973). PLFs often contribute to an individual, idiosyncratic speaking style but most people are also able to control many of them consciously.As PLFs form an extraordinarily heterogeneous group, it is difficult to cover all of them with a single definition. Moreover, they can be defined only as precisely as precise the boundaries of language can be. As a result, researchers tend to define them by enumeration and the inventories proposed in literature are often selective. Among the most frequently mentioned examples of PLFs, there are prosodic features and voice quality parameters (Crystal, 1966). Some of them can be relatively easily measured as local acoustic parameters of speech signal (energy, pitch frequency). Others can be determined only when longer stretches of speech are analysed. Finally, some of them have more "structural" nature. For example, in an emotional utterance, words may be sequenced in an atypical way, accentuation may be also uncommon. But this can be only noticed once the entire utterance is taken into account and compared to some other typical or common structures.In recent decades, many studies of paralinguistic features have been focused o...

Speaking style variation in laboratory speech: A perception study

“…The selected sections of recordings were manually segmented into intonational micro-phrases (MiP, e.g. [25,26]) and transcribed orthographically on another annotation layer. On the basis of the orthographic transcript, speech signals were automatically segmented into phones and transcribed in broad phonetic transcription using Polphone [27] and Salian [28] integrated into Annotation Pro.…”

How Long is Too Long? How Pause Features After Requests Affect the Perceived Willingness of Affirmative Answers

“…Established measures of speaking rate simply either include or exclude pauses instead of trying to model their complex role in tempo perception; and those studies that explicitly investigate pauses typically focus on their formal and deliminative aspects. These aspects include, for example, how strongly pause durations vary across native and non-native speakers, speaking styles, and languages [2,3,4,5,6,46], if pause durations can be organized into different classes [4,7], which kinds of phonetic sounds occur in filled pauses [8,9,10,11,12], and to what extend such pause variables correlate with boundary types of phrases and turns in discourse structure [5,6,9,13,14,15,16,17,18].…”

Special Issue: Slavic Perspectives on Prosody

“…Similarly, research on the functions of prosody in discourse needs more investigation (Karpiński, 2012).…”