ПРОСОДИЧЕСКАЯ ВАРИАТИВНОСТЬ И КОНСТРУИРОВАНИЕ ПРОСОДИЧЕСКОГО ПОРТРЕТА

Introduction

The predominant anthropocentric paradigm sees the focus of linguistic research shift towards the study of man. The tendency is observed in various linguistic fields, and it has contributed to the emergence of a relatively new branch of linguistics – linguistic personality studies [2].

Our research adopts a classic socio-cultural approach, which arguably constitutes the key aspect of anthropocentric paradigm, and applies it to the study of a linguistic persona developed by Yu. N. Karaulov, G. I. Bogin, V. I. Karasik for the purpose of studying prosodic portrait of a linguistic persona. In the study we aim to give an overview of prosodic variation literature and then construct prosodic portraits of ten different book characters voiced by David Crystal, a prominent British linguist, by using previously found data on prosodic aspects of pitch, rate and voice quality and their socio-cultural and emotive cues. Thus, the aim of the study is to provide an overview on prosodic variability; to search for the variability of prosodic features of a linguistic persona; find out prosodic characteristics that are stereotypically associated with a particular social/cultural group and/or an emotional state of the person; find the combination of means of conjuring up the desired image of a character in the mind of the listener.

It seems logical to start with the notion of linguistic stereotypes as it is they that contribute to building the character. Language stereotypes are recognized as a key character building tool due to the capacity of stereotypical linguistic features to trigger social stereotypes [17]. It is maintained that listeners use speech cues to infer speakers’ social group membership(s).  Thus it is agreed that language is “a quick way to build character and reaffirm the stereotype” and we “associate specific characters and life styles with specific social groups by means of language variation” [14, P. 85]; language varieties evoke a stereotypical image and “actors can exploit this” [17, P. 33]. We argue that the idea is relevant for the suprasegmental language level with prosodic cues and prosodic variation acting as a character building tool. We now proceed to look at suprasegmental features that are associated with various socio-cultural and emotive cues.

Prosody variation has been widely explored by linguistic scientific community from different perspectives. The correspondence between emotional state and prosody has been on the scientific agenda since the middle of the previous century and several various ways to classify emotions have been suggested. There exist an approach distinguishing emotional states with high (e.g. anger, fear, joy) and low (e.g. sadness) levels of physiological arousal whereby high physiological arousal is associated with increases in mean F0, F0 variability and vocal intensity [8]. Much research has been done on the so-called basic emotions, such as happiness and sadness, and five broad emotional categories (anger, fear, happiness, sadness, and love-tenderness) [15], [16]. According to these studies, anger and happiness are associated with increases in mean F0, F0 variability, speech rate, vocal intensity and high-frequency energy (voice quality). Sadness and tenderness, on the contrary, are marked by decreases in mean F0, F0 variability, speech rate, vocal intensity and high-frequency energy (voice quality) [12].

An alternative system for studying and classifying emotions is to organize them in a dimensional manner. There exist a circumplex model advanced by James A. Russell (the USA). The model represents emotions through their placement along the two dimensions of arousal/activation and valence/evaluation/pleasure. The emotions and their prosodic cues have been explored according to the model and it has been established that arousal is positively correlated with increases in mean F0, F0 standard deviation, F0 range, mean loudness, shimmer and speech rate. There is less evidence to suggest a clear correlation between valence and prosodic parameters but based on various studies it is possible to conclude that  high valence is positively correlated with lower F0, larger F0 variability, larger F0 range, decreased loudness, less high-frequency energy, “warm” timbral voice quality, increases in speaking rate and shorter pauses [16].

Prosody has also been wildly discussed in terms of its variability due to an array of social factors. The role of gender in prosodic variability and the social status of men and women have been studied by T. G. Medvedeva and T. I. Shevchenko in British English [6]. American female speech has been found to be manifested through a broader pitch range compared to male speech. In contrast with it, British female speech is characterized by a narrower pitch range compared to British male speech. Furthermore, prosodic variability has been studied from the point of view of territorial differentiation (E. A. Babushkina, T. I. Shevchenko [1], [6]). It is known that American Standard speech (General American) is associated with a narrow pitch range while British Standard speech (Received Pronunciation), which reflects high social standing, is associated with a wide pitch range [6, P. 168]. Professional voices have been scrutinized by D. Crystal [10]. O. P. Kryukova explored the said variability within the confines of public speech [4]. L. V. Postnikova brought into focus the prosodic features typical of American presidents’ speech [5]. M. V. Yakutina studied the prosodic features typical of the speeches delivered at U.S. Supreme Court hearing [7]. Interestingly, the author found correlation between the aggressive Prosecution stance and such prosodic cues as high mean F0 and F0 variability while the Defence speech is characterized by lower mean F0 and F0 variability. Concerning the temporal aspect it was found that the Prosecution’s aggression is manifested through a slow speech rate and short pauses while the Defence’s speech is characterized by fast speech rate and long reflection pauses [7].

Thus, the overall conclusion based on the studies conducted is that the main ways to achieve prosodic variability are through the increase in F0 variability, vocal intensity and speech rate though voice quality is becoming increasingly more frequently studied in relation to interpersonal prosodic variability.    

Materials and methods

The material for the research is David Crystal’s autobiography ‘Just a Phrase I’m Going Through’ read by the author. The overall corpus presents an hour long recording (t = 60 min) from where by way of perceptive analysis we have selected the sections where the author-narrator voices ten different characters resorting to variation in his prosody. The selected material was then organized into three main gender and age groups: the ten characters’ speech was described as children’s, adult male or adult female. The main method of the current research is acoustic analysis. For the purpose we extracted the following parameters: minimum, maximum and range of the fundamental frequency (Fo) in Hz (pitch M, pitch min, pitch max and pitch range, respectively); mean loudness in dB (loudness M); Average Syllable Duration (ASD); and the means of jitter in %, shimmer in %. The extracted parameters were measured across the entire sections. (word order) The analysis was conducted with the help of computer software program PRAAT [9].   

Results

The conducted acoustic analysis allowed us to reveal the interdependence of acoustic speech parameters and specific gender and age groups.

We established the narrator’s voice to be the speaker’s modal voice and compared other voices to it. The narrator’s voice is characterized by: F0 min (74 Hz), F0 max (267 Hz), pitch range (22 st), ASD (193 ms), Shimmer (13%), Jitter (5%).

By way of perceptive analysis the following roles (apart from the voice of the narrator) were identified:

1. Adult male voices;

Table 1. Prosodic characteristics of adult male voices (n=4)

2.  Adult female voices;

Table 2. Prosodic characteristics of adult female voices (n=2)

3. Children’s voices. 

Table 3. rosodic characteristics of children’s voices (n=3)

Discussion

Linguistic community seems to agree that F0 parameters are most informative in phonetics experiments. It is known that pitch range of an adult male varies between 75 and 200 Hz while pitch range of an adult female varies between 130-300 Hz. Hence, it is insightful for the purpose of our study to scrutinize the pitch range of the modal voice – the Narrator. David Crystal demonstrates an extremely wide range surpassing the mean male F0 variability by 33%. Furthermore, David Crystal’s modal voice is characterized by a high speech rate and high level of shimmer parameters which indicates his elderly age. These characteristics taken as a mode, we can now compare other characters’ voice characteristics against the Narrator’s to determine prosodic variability of the linguistic persona selected for analysis.