The Pitch of Formal and Informal Russian Discourse
Olga T. Yokoyama
UCLA
Project description
The proposed study examines the similarities and differences between Russian male and female use of pitch in different communicational modes and different speech genres. It focuses on a single intonational parameter, i.e., the fundamental frequencies (Fo). To the best of my knowledge, this is the first statistical study of Russian speaking fundamental frequencies. Outside the Slavic languages, attention has been paid to the average level of pitch (mean fundamental frequency) produced by men and women and/or perceived as pertaining to one or the other sex, as well as to the range of fundamental frequency variability, as associated with male or female speech both in production and perception (cf. Graddol and Swann 1983 for evidence of a wider pitch range in women as compared to men in Southern England, among others).. Gender studies in which gender stereotypes and their social impact are examined have focused most extensively precisely on the perceptual aspect of pitch range (cf. McConnell-Ginet 1983). These studies unequivocally indicate that in the perception of the speakers of different languages a wide pitch range is associated with female (or effeminate male) gender stereotypes, as well as with emotional volatility, a feature that is considered to be typically female. Narrow pitch range, on the contrary, is perceived as masculine.
Experimental evidence on the acoustic side has been statistically on a small scale, typically under 20 informants. They also tend to use elicited data produced in easily controllable no/low noise environments (cf., e.g., the analysis of reading samples by American English male and female voices in Takefuta, Jancosek, and Brunt 1972, one of the earlier studies that found a greater pitch range in female voices). I will examine generic differences in the intonation produced by educated professionals in the informal mode and in the formal mode. All samples were produced without elicitation. Eight native speakers of Standard Literary Russian (four men and four women) were recorded in June 2002, in Moscow, in different communicational settings. The speech genres included: conference presentations, a university lecture, chairing a conference panel, engaging in a discussion after a presentation, informal conversation at home, and informal conversation at a small gathering following a conference. All but the last two genres belong to the formal mode.
All the recordings, totaling over 8 hours, were made in a natural environment where no control for noise or, in dialogic/polylogic situations, for overlapping speech was possible. The result was very "messy" sound files. Given a wish to avoid a narrow band filter which would simultaneously filter out some of the speech frequencies, the extraneous sounds had to be removed manually, upon repeated auditions. Spots where two or more speakers overlapped, as well as laughter, were also excluded from the analysis.
The data used for the final analysis consists of approximately 15 minutes (905,370 msec) of digitized sound files. The statistical analysis is based on the number of readouts (dots comprising the Fo curves). The Fo values were obtained with the help of PCquirer, software developed by Scicon Research and Development. After the calculation of the means () and the standard deviation (Sx) for the Fo values for each genre for each speaker, these values were then subjected to the T Interval Test and the T Test for the significance of , and to the F Test for the equality of variance of Sx. In the formal mode, both the individual formal genres and the pooled formal genres were considered. The paper will focus on the analysis of these statistical data.
Two results emerge as clearly statistically significant.
(1) For both men and women, Fo is lower in the informal genres;
(2) The pitch variance for women
These results agree with those found for other languages; they can now be considered statistically established for Russian
Lacking statistic significance is an otherwise clearly observable interesting convergence with respect to pitch variance of the most monologic and formal of the genres examined (lecture/talk), and the informal two genres (at home/party). For both genders, both generic groups show higher Fo variance than the formal dialogic genres. While a greater variance is to be expected in the informal genres, high variance in the lecture/talk genres requires an explanation. At the moment, an explanation appears to lie in the ÒreadingÓ intonational pattern of public lectures, which forces certain intonational configurations on the speakers regardless of their gender.