Download Acoustic markers of political power

Acoustic markers of political power
Danielle Duez
Laboratoire Parole et Langage, CNRS UPRES-A, 6057
Université de Provence
29 avenue Robert Schuman
13621 Aix en Provence
FRANCE
The effect of political power on acoustic variables was examined in three political speeches
given by François Mitterrand at different periods of his career. F0 data were not found to
reflect the extent of Mitterrand's power. On the other hand, temporal organization appeared
to mirror the politician's distance from power. When Mitterrand was both a challenger and
an opponent, articulation rate was more rapid, while in the two presidential speeches, he
showed a slower articulation rate, which seems to be a feature of power. The pause structure
appeared to reflect the place of the politician within the social hierarchy in such a way that
the higher the status, the longer and the more frequent the pauses. There seems to be a
balance between what is being said and what is left unsaid: silence becomes a power symbol.
2
INTRODUCTION
Speech conveys linguistic information as well as information on the personal characteristics of
the individual. Abercrombie (1967) referred to the latter as indexical information and
distinguished three classes of indices: 1) those which indicate membership in a group, 2) those
which characterize an individual, and 3) those which reveal the changing state of the speaker.
Subsequently, Laver and Trugdill (1979) adopted Abercrombie’s typology but
referring to the three classes of indices as social markers, physical markers, and psychological
markers. The first class refers to group characteristics such as regional or dialectal affiliation,
social status, educational status, occupation, and social role. The second class refers to
individual characteristics such as sex, age, and state of health, The third class describes the
speaker's personality and affective state. Laver and Trugdill (1979) further refined their
typology by distinguishing markers which are not consciously manipulated by the speaker and
have an informative function, from those which respond to the speaker’s intention of
signalling an affect or social role and have a communicative function. These authors called the
former extralinguistic features, and the latter paralinguistic features.
Experimental research conducted in the area of paralinguistic information has tended
to focus less on the communication of affect and more on markers of personality, particularly
as conveyed by dynamic features. Markers of anxiousness and extroversion have been
investigated more than the other features. However, the data collected for these two
personality traits have yielded partly conflicting results. While Murray (1971) and Rochester
(1973) observed a lower proportion of silences in the speech of highly anxious speakers than
in that of slightly anxious speakers, Siegman (1978) reported more frequent longer pauses for
anxious speakers. Similar discrepancies were also observed in the data on extroversion.
Siegman and Pope (1965) found extroversion to be associated with shorter latencies, fewer
"filled" brief pauses, and fewer silent pauses, while Ramsay (1968) did not obtain a significant
extroversion effect for the silent-pause ratio, and suggested that the length of a silence may be
the critical variable that distinguishes introverts from extroverts. Siegman’s explanations
(1978) based on individual differences in cognitive activity were not supported by Scherer’s
findings (1979a) on personality trait markers in German and American speakers. Scherer
obtained different results for the two speech communities. He suggested an effect of
intercultural differences and pointed out the necessity of taking psychosocial factors into
account.
Although speech is a social phenomenon, social speech markers have attracted little
attention in the experimental research on paralinguistic information. Social-role markers in
speech have mainly been investigated in studies on speech styles. For example, Crystal and
Davy (1969) described intonation, pitch range, loudness, rhythmicity, and rate in
conversations, sport commentaries, news, conversations, cricket and funeral state
commentaries, sermons, and lectures. They mentioned the following recurrent features for
conversations: hesitations, short and often incomplete tonal units, frequent F0 falls, and high
speech rate variability. For read speech samples, they reported a tendency towards short units
and isochronous rhythm, narrow pitch range, and a lack of disruptions. A comparative
approach has also been adopted in more recent studies on speech styles. Lucci (1983)
analyzed prosodic features (rhythm, pauses, rate, intonation, emphatic stress) in lectures,
conversations, news readings, and lecture readings by four French speakers. Although he
excluded speech spontaneity cues, the role of acoustic variables in speech style identification
was stressed. Fónagy and Fónagy (1976) studied rhythmic and intonational patterns in French
3
conversations, news, and read fairy tales, and reported stereotypes for the different speech
styles. For example, news reports are characterized by irregular rhythmic patterns, frequent
emphatic stress, and sudden pitch falls. The authors also had subjects identify some of these
patterns in and out of context (in excerpts). They found these patterns to be interpreted
differently, and suggested that their meaning may depend on the situational context.
Other studies have systematically investigated a given parameter across different
speech styles. Graddol (1986) explored the contrasting pitch characteristics of males and
females reading two kinds of text (neutral technical prose and dramatic dialogue) and
examined female and male speech as a function of social conditioning. He demonstrated that
each speech style has a different set of pitch characteristics. Duez (1982) analyzed and
compared the frequency, duration, and distribution of pauses in French political speeches,
political interviews, and casual interviews, and interpreted the results obtained as a function of
spontaneity and the relationship between the speaker and his audience. Non-silent pauses such
as filled pauses, lengthened syllables, repeats, and false starts were found to be less frequent
and shorter in political interviews than in casual interviews, which are less constraining. The
total duration of silent pauses was found to be 50% greater in political speeches than in
interviews, even though the content of the speech in the former case is known to the speaker,
and despite the fact that there were practically no hesitations. It was suggested that some
pauses, particularly the unexpected ones and the longest ones, have an illocutionary function
and serve to increase the strength of the argumentation. Significant pause-time and speechrate differences across politicians were also observed. One of the reasons for these differences
might be a high degree of inter-speaker variability (Goldman-Eisler, 1968). Another reason
might be the heterogeneity of political status. The politicians came from various political
horizons, played various positions in the political "chess-game", and thus had a different
relationship to power. These differences may induce different speech strategies, which may be
reflected at the acoustic level.
The purpose of the present study was to further refine the phonetic description of
political speeches and to test the above assumption that political power has an effect on
speech style. Three political speeches given by François Mitterrand at three different periods
of his career were analyzed. The first speech was produced in 1974 when Mitterrand was both
a challenger and an opponent, the second was a presidential speech (1984), and the third took
place in a particular and interesting situation in which Mitterrand was both the incumbent
president and a candidate to a new presidential mandate (1988). He had the same social status
as in 1984, but the role he played may have depended on a strategic choice. The comparison
of speeches produced by the same politician is one way of neutralizing individual variations
and focusing on relevant situational markers. Several acoustic parameters such as articulation
rate, and the duration, frequency, and distribution of pauses, were studied and compared in the
three speeches. The acoustic analysis was extended to the fundamental frequency range of the
voice since this parameter has been shown to be relevant to the characterization of speech
style (Graddol, 1986; Léon, 1971-93). It was assumed that the above variables are significant
indicators of power, and their comparison should allow us to characterize both the
"presidential" style and the "challenger" style.
METHOD
Corpus
The corpus consisted of three speeches produced by François Mitterrand in 1974, 1984, and
1988. The 1974 and 1988 speeches were recorded from high quality FM broadcasts on a
Revox A77 tape recorder at 19 cm/sec. The 1984 speech was obtained directly from the INA
4
(International Audiovisual Institute). Table 2 gives their duration. The situation in which each
political speech occurred was defined according to Hymes's model (1972).
Situational variables
Hymes (1972) proposed that all speech events take place in a specific situation which can be
defined in terms of basic concepts such as ’setting’, ’purpose’ and ’participants’. The place
and the physical position of the participants vis-à-vis each other are features of setting, as well
as the moment, or period when the speech events occur. Speech purposes are the
communicative intents of the speaker(s). Social markers, physical markers, and psychological
markers (as defined by Laver and Trugdill, 1979) are features of the participants, as are the
relationships between the participants at an interpersonal or social-institutional level.
Setting. Because each of the speeches studied here was televized and broadcast, there
was no direct contact between the speaker and his listeners or viewers. François Mitterrand
was in front of a microphone (and/or a TV camera), in a non-public locale such as the Elysee
Palace or a TV studio, whereas each elector was (potentially at least) watching a TV screen, at
home or in a public place, e.g. a bistro, a pub, or a department store.
Purpose. The three speeches were of the same activity type: public argumentation. They
had the same purpose: persuading the voter of the merits of a program or a decision. The
speeches produced in 1974 and 1988 were part of an electoral campaign. The 1984 speech
differed partially since it was a presidential communication event. However, this difference is
more apparent than actual, since the president had to unremittingly campaign to sustain his
popularity (Gerstlé, 1992).
Participants. In 1974, François Mitterrand was both a challenger and an opponent. He
denounced the errors of the current administration and called for change. In 1984, he was
president and at the top of the power ladder. His legitimacy was not questioned, but he had to
pacify social relationships and convince the French of the necessity of a referendum. In 1988,
he was the current president and a candidate to a new mandate. According to Gerstlé (1992),
his campaign was predicated on his being the incumbent. He did not attack his opponents, but
simply reminded the electorate about the program he already had underway.
Acoustic variables
The long-term modulation of intensity (of the present signal) is partly influenced by the
emotional states and the personality of the speaker. Therefore, it should be interesting and
worthwhile investigating these variations as a function of the situation of communication.
Unfortunately, these values are at best difficult to interpret, due to the large number of factors
that affect the amplitude of a signal such as ambient noise and varying distance and angle with
respect to the microphone. Moreovere, amplitudes often undergo automatic volume control in
broadcast or televized signals. Consequently, the analysis of acoustic variables was only
concerned with duration and fundamental-frequency measurements.
Fundamental frequency (Fo) measurements. F0 values were computed on the voiced
part of the signal. The method of analysis was based on the cross-correlation between the
power spectrum and a comb function (Espesser, 1981-82). The raw data were checked for
octave shifts resulting from second harmonic readings, and then reinterpreted when
5
necessary. The following F0 data were computed for each speech: the minimum + 5%, the
mean, the maximum - 5%, the standard deviation, and the range (5% to 95%).
Temporal measurements and temporal variables. After phonetic transcription from the
tape recordings alone, the oscillograms were segmented while listening to the tapes. The total
speech time was subdivided into total articulation time and total pause time. Non-silent pauses
and silent pauses were classified according to the following definitions.
I. Silent pause: any interval on the oscillographic trace where the amplitude is
indistinguishable from that of background noise. A silent pause is never shorter than the
"minimum" duration taken to be equal to the average duration of an intervocalic stop
increased by four standard deviations.
II. Filled pause: any occurrence of a French hesitation interjection such as euh, eh,
hein.
III. False start: any sequence of segments that is intended to start the next utterance but
is interrupted and replaced by another that will get completed.
IV. Repeat: any unintended repetition of a sequence of phonetic segments.
V. Lengthened syllable: any syllable in which the vowel is abnormally prolonged.
In addition, the following definition was used.
VI. Articulated sequence: any sequence of phonetic segments (excluding filled pauses,
false starts, and repeats) delimited by two silent pauses.
The silent-pause ratio was obtained by dividing the total duration of silent pauses by
the total speech time. The mean number of syllables of an articulated sequence
was
calculated by dividing the total number of syllables by the number of articulated sequences.
The mean duration of a silent pause was obtained by dividing the total pause time by the
number of silent pauses. The articulation rate was obtained by dividing the total number of
syllables by total articulation time (total speech time minus total pause time).
Pause location. The location of the silent pauses was studied by syntactic type. Four
types were considered: pauses located between sentences, with a falling contour; those located
between clauses, with a continuation rise, as defined by Dubois and Dubois-Charlier (1970);
pauses occurring between phrases, which result from the segmentation of sentences and
clauses, again as defined by Dubois and Dubois-Charlier (1970); and within phrase pauses,
i.e. unexpected pauses within a sense group (Grammont, 1914).
6
RESULTS
Fundamental frequency
Table I. F0 data for Mitterrand's three political speeches.
_______________________________________________________________
Political Speech
F0 measurements
1974
1984
1988
Minimum F0 +5% (Hz)
65
65
65
117
136
128
Mean F0 (Hz)
Maximum F0 -5% (Hz)
238
260
242
Standard Deviation (Hz)
34
34
34
F0 Range 5%-95% (Hz)
173
195
177
________________________________________________________________
Table 1 gives the means in Hz of the various F0 measures made for each speech. We can see
that the 1984 speech differs from the 1974 and 1988 speeches: the minimum values are
identical but the mean, range, and maximum values are all markedly larger for the 1984
speech than for the other two speeches, which are comparable. Since the 1984 speech used
an extended range of F0 without altering the baseline, one would indeed expect a higher
maximum and a higher mean, as can be seen in the data.
Temporal variables
Table II. Temporal data for three Mitterrand's three political speeches. Standard
deviations are in parentheses.
Political Speech
Time measure
1974
1984
1988
Total speech time
2’04"
6’00"
3’ 23"
Total pause-time ratio (%)
29.1
39.0
33.5
Mean articulated-sequence duration (syll) 9
7.4
7
(0.58)
(0.42)
(0.53)
Mean pause duration (in ms)
605
972
792
(63)
(45)
(57)
Articulation rate (syll./s)
5.5
4.7
4.7
(1.3)
(1.2)
(1)
__________________________________________________________________
Articulation rate. The mean articulation rates for each speech were 5.3 syll./s (1974),
and 4.7 syll./s (1984 and 1988). The difference between the articulation rates of the 1974
speech and the 1988 speech is statistically significant (p=0.05) contrary to the difference
between the articulation rates of the 1984 speech and the 1988 speech (p=0.89). Changes in
articulation rate seem to mark the different steps in the argumentation. The 1974 speech
began with a denunciation of the opponents’ failings. In the second part, Mitterrand tried to
convince listeners by talking about a world of justice. The articulation rate in the first part was
5.9 syll./sec, which is slightly higher than the articulation ratein the second part (5.5 syll). The
7
difference between the two articulation rates is statistically significant (t=79, p=0.0001). The
first part of the 1984 speech was a reminder of the President’s responsabilities: the articulation
rate was 5.25 syll./sec. In the second and third parts where Mitterrand presented the need for a
referendum, which implied a change in the constitution, he spoke with a solemn tone and his
articulation rate was slower (4.73 and 4.76 syll./sec, respectively). Mitterrand opened the
1988 speech by addressing a question about the President’s role; his mean articulation rate
was 5.35 syll./sec. He continued by drawing a portrait of the ideal president, and then
described his home and foreign policy program. Mean articulation rate in these three parts
was slow 4.7 syll./sec, 4.6 syll./sec, and 4.8 syll./sec, respectively. Finally, a fast "à bientôt"
(see you soon) (6 syll./sec) created a certain proximity between himself and the electorate.
The cross-speech comparison of the total pause-time ratios indicated a higher
percentage of pause time in the 1984 speech (39%) than in the 1974 speech (29%); that of the
1988 speech was intermediate (33.4%). These differences may be related to differences in
mean silent-pause durations. Mean silent-pause duration was shorter in the 1974 speech (581
ms) than in the 1988 speech (792 ms), which in turn was shorter than in the 1984 speech (972
ms). The difference between the three speeches is statistically significant (ANOVA:
F(2,295)=39, p=0.001). The results obtained for mean articulated-sequence durations
expressed in syllables revealed that the pauses were less frequent in the 1974 speech (every 9
syllables) than in the 1984 speech (every 7.4 syllables) and in the 1988 speech (every 7
syllables). This difference is statistically different (ANOVA, F(2,300)=3.5, p=0.03).
Table III. Mean duration (MD) expressed in ms, standard deviation (SD), and number
of silent pauses (N), by location in Mitterrand's political speeches.
Political Speech
1974
1984
1988
Location
MD SD N
MD SD N
MD SD N
Between sentences 827 96 13
2102 67 22
1521 80 15
Between clauses
644 65 21
1112 50 35
907 63 21
Between phrases
637 40 30
663 40 77
534 50 48
Within phrases
357 7
384 6
405
___________________________________________________________________
1
Silent pauses. Silent pause durations by location in the sentence and the power situation
are shown in Table 3. In the three speeches studied here, the pause distributions had a similar
overall pattern: silent pauses between sentences and clauses were more frequent and longer
than pauses between phrases, the latter being longer and more frequent than pauses within
phrases. This is quite apparent in Table 3. Differences in the three speeches can be readily
noticed, however. Between-sentence pauses were longer in the 1984 speech (2102 ms) than in
the 1988 speech (1521 ms), the latter being longer than in the 1974 speech (827 ms). A
similar tendency was observed for pauses located between clauses, whose corresponding
values were 1112 ms (1984), 907 ms (1988), and 644 ms (1974). These differences are
statistically significant (ANOVA between-sentence pauses: F (2, 47)= 71, p = 0.0001;
between-phrase pauses: F (2, 74) = 15, p = 0.0001). The duration of pauses located betweenand within-phrases was very similar in the three speeches. The average durations obtained for
pauses located between phrases were 637 ms (1974), 663 ms (1988) and 534 ms (1984); they
are not statistically different (F (2, 154) = 0.98, p = 0.37). The corresponding values for
pauses within phrases were 357 ms (1974), 384 ms (1984), and 405 ms (1988). Note that the
8
use of within-phrase pauses was more marked in the 1974 speech (7 out of the 71 pauses)
than in the 1984 and 1988 speeches (6 out of 140 and 1 out of 85, respectively).
Non-silent pauses.Only the 1974 speech exhibited non-silent pauses, and only a limited
number of them. One false start was identified. There were also six filled pauses located at
phrase boundaries. Two of them were associated with a silent pause; they were 330 ms and
350 ms long. Non-silent pauses are a characteristic of spontaneous speech. Located before an
infrequent word, they leave time for word searching (Beattie and Butterworth, 1979); located
at the head of a phrase, they leave time for programming the next phrase (Boomer, 1965;
Cook, 1971). In political speeches, the text is known and learned in advance, and this helps
the speaker not to hesitate. Non-silent pauses may be due to anxiety and emotion. One may
also assume that non-silent pauses are voluntary, rhetorical hesitations used to create a certain
proximity between the politician and his audience.
GENERAL DISCUSSION
A major objective of the present study was to examine to what extent acoustic F0 and duration
variables reflect the speaker's social relationship to power. In general, measures based on the
F0 deviations have proven that the realization of pitch is not an exclusively linguistic
phenomenon, but is also influenced by paralinguistic and extralinguistic factors. Mean-F0
values and F0 ranges were shown to depend on the speaker's personality (see Scherer, 1979b
and 1981), sex (Helfrich, 1979), and emotions (Scherer, 1992). For example, F0 was found to
increase with increases in stress or emotional tension (Scherer, 1979b, 1981). The interaction
of these various factors makes the interpretation of F0 variations quite difficult. However,
some similar trends were observed for F0 ranges in the three speeches. They all have a wide
range, with values close to those obtained for lectures and acting, and larger than those
obtained for interviews and conversations (Lucci, 1983; Johns-Lewis, 1986). A wide F0 range
may be partly due to high F0 values, which in turn may reflect the frequent occurrence of
emphatic stress. Graddol (1986) stated that the F0 range is a speech-style marker. The
present results suggest that a wide F0 range is a characteristic of public modes, e.g. of public
address for a multiple audience. The mean-F0 values were found to be higher in the two
presidential speeches (more particularly in the 1984 one) than in the opponent speech. If one
assumes that Mitterrand is more dominant and self confident in the former speeches than in
the latter, the tendencies observed mean that F0 values disagree with the findings on pitch
symbolism in which low pitch is associated with aggressivity, dominance while the meaning
of high pitch is non threatening, submissive, in need of the receiver cooperation (Ohala, 1983
and 1984). One of the reasons for this difference may be Mitterrand’s age (66 years old in
1983) since the average F0 of men appears to rise from about 65 years on (Helfrich, 1979). A
second reason may be a tendency to stay within a relatively larger range of frequencies in the
presidential speeches, giving them a more solemn tone: mean-F0 values increase as there is
less intimacy and greater distance between the interactants (Johns-Lewis, 1986). A third
reason may be the experimental and measurement procedures adopted here. Mean F0 and F0
range are global measures which do not give account of the F0 contours, that change over
time. In a perceptual investigation of spontaneous-speech samples in which F0 was shifted
downward or upward, Ohala (1984) has shown that a steep terminal fall was the one feature
which contributed most to making a voice dominant. The present analysis constitutes but a
first attempt to explore the role of F0 in the process of persuasion and political power.
Further research is needed to investigate the way in which F0 contours are involved in
communicating power situations and used to induce the appropriate emotions and attitudes in
the listener.
9
Of greater interest are the results obtained for the temporal variables. An increase in
articulation rate has been found to be linked to a speaker's degree of practice (Goldman-Eisler,
1968) and to the formalness of the style (Duez, 1982-91). The present results suggest that
being in a situation of power has an effect on articulation rate. As a challenger, Mitterrand
wanted to accumulate arguments, and increased articulation rate in order to do so. As the
incumbent president, Mitterrand was master of the situation; he controlled his speech time
and kept a relatively slow articulation rate. Interestingly, articulation rate in the two
presidential speeches (4.5 syll./sec) is close to that obtained for Georges Pompidou’s
presidential speech (Duez, 1982-91).
The results obtained for pause time in the three speeches confirm previous findings on
political speeches, and in a more general way, on speech styles with an illocutionary function.
In this type of speech, pause distribution correlates with the syntactic structure of the text,
making decoding easier: speakers approach "ideal communication" as defined by GoldmanEisler (1968). However, speakers do not seem to be solely concerned with marking syntactic
boundaries. Sometimes, they neglect to mark sentence boundaries while at other times
making strikingly long pauses or unexpected pauses. In the present case, the longest pauses
lasted more than 2000 ms, which is close to the rhetorical pauses reported by Clemmer,
O’Connell, and Loui (1979) and Kowal (1982). Unexpected pauses were most often
associated with an emphatic accent. Their occurrence strikes the listener and helps
emphasize the next word or the next argument. The existence of such pauses has also been
noted in studies on President de Gaulle’s political speeches (Léon, 1971), on American and
German speeches (Kowal, 1982), and in American sermons (Gumperz, 1982). They may be
an enhancement technique favored by public speakers. A comparison of the silent pause time
of the three speeches allowed us to characterize each one. In the speech produced in 1974,
pauses tended to be less frequent and their duration was close to that of pauses in Mitterrand’s
interviews (Duez, 1982-91). As an opponent and a challenger, Mitterrand had to use time as
efficiently as possible. In the presidential speech of 1984, pauses were strikingly long and
frequent, especially at sentence and phrase boundaries. Mitterrand did not have to argue to
justify a program; he had power and explained a decision calmly and gravely. Pause time
allowed him to maintain a balance between what was being said and what was left unsaid.
Interestingly, a similar use of pause duration was found in Georges Pompidou’s speech
mentioned above (Duez, 1982-91). In the 1988 speech, pause time was intermediate between
that of the 1974 and 1984 speeches. It reflects Mitterrand’s specific situations: both president
and candidate. As a candidate, Mitterrand tried to sway as many voters as he could to his side.
One way to do so was to fill time with ideas and arguments. As the incumbent President, he
could be satisfied with striking the balance. According to Gerstlé (1992), Mitterrand’s late
declaration and intervention intended to reduce the distance between the president and a
"president-candidate" as much as possible. The results obtained for articulation rate and
speech pause time are clearly in line with this interpretation. A slow articulation rate and
silence may be speech markers of power.
Temporal variables may be regarded as power-situation markers in political speeches. In
a more general way, they may reflect the social distance which exists between politicians and
their listeners. The relationship between the social distance, the interpersonal distance and the
speaker's voice was first investigated by Hall (1959 and 1966). Hall divided distance between
participants into four categories (intimate, informal, social-consultative and public) and
related each social distance to a specific interpersonal distance and speakers' voice
characteristics. For example, the public distance (the one of interest here) is characterized by
the so-called "icy style" i.e. the style typical of individuals meant to remain strangers to each
other (as defined by Joos, 1962). A careful articulation, a loud voice and a slow speech rate
10
are features of this style. The specific temporal organisation (slow articulation rate and pause
time) in Mitterrand's presidential speeches perfectly examplifies some aspects of the icy style.
It may be used to symbolize to the extreme the public distance which separates the president
from his fellow citizens.
However, the above assumption raises a certain number of questions, one of them
being its representativeness. Although Mitterrand’s political speeches reflect symbolic power
situations, our data concern a single French speaker in very specific roles. This investigation
must be extended to other speakers, other activity types, and other cultures. Another question
is how listeners use temporal variables to evaluate the speaker’s relationship with power. For
example, studies on identification of a speaker's profession have shown that even though this
is a difficult task, subjects'responses are never at random. Prosodic cues allow listeners to
identify professions and styles (Fónagy and Fonagy, 1976; and Fonagy, 1978). More recently,
Tielen and Koopmans-van Beinum (1991) found that pronunciation and voice allowed
listeners to distinguish nurses, managers, and information agents, and that the scores varied
with the speaker's sex. Some cultures have coded social hierarchy markers. Crystal (1975)
reported that in some Bolivian languages, men use nasality when they speak to a superior, as
do women when they speak to their husband. Speech time use is not coded as a hierarchy
marker in everyday French, but has a communicative function. Therefore, further perceptual
studies of political speeches are planned. They should allow us to specify the role of temporal
variables in the identification of political power.
REFERENCES
Abercrombie, D. (1967). Elements of General Phonetics. Edinburgh: Edinburgh University
Press.
Beattie, G. W. and Butterworth, B. L. (1979). Contextual Probability and Word Frequency as
Determinants of Pauses and Errors in Spontaneous Speech. Language and Speech,
22(3), 201-211.
Boomer, D. (1975). Hesitation and Grammatical Encoding. Language and Speech, 8, 148158..
Clemmer E. J. O’Connell D. C. and Loui W. (1979). Rhetorical Pauses in Oral Reading.
Language and Speech, 22, 397-405.
Cook, M. (1971). The Incidence of Filled Pauses in Relation to part of Speech. Brit. Journal
of Soc. Clin. Psychol. 8, 13-21.
Crystal, D. (1975). The English Tone of Voice. London: Arnold.
Crystal D. and Davy D. (1969). Investigating English style. London: Longman.
Dubois, J. and Dubois-Charlier, F. (1970). Eléments de linguistique française : Syntaxe. Paris:
Larousse.
Duez D. (1982). Silent Pauses and Non-Silent Pauses in three Speech Styles. Language and
Speech, 25, 11-28.
11
Duez D. (1991). La Pause dans la Parole de l’Homme Politique. Editions du CNRS.
Collection les sons et la parole.
Espesser, R. (1981-82). Un Système de Détection du Voisement et de F0. Travaux de l'Institut
de Phonétique d'Aix en Provence, 8, 240-261.
Fónagy I. (1978). A New Method of Investigating the Perception of Prosodic Features.
Language and Speech, 21, 34-49.
Fónagy, I. and Fónagy, J. (1976). Prosodie Professionnelle et Changements Prosodiques. Le
Français Moderne, 44/3, 193-227.
Gerstlé, J. ( 1992). La Communication Politique. Paris : Collection Que Sais-je.
Goldman-Eisler, F. (1968). Psycholinguistics: Experiments in Spontaneous Speech. London,
New York: The academic Press.
Graddol, D. (1986). Discourse Specific Pitch Behavior. In C. Johns-Lewis (Ed.), Intonation
in discourse, pp. 221-237. New York: Springer-Verlag.
Grammont, M. (1914). Traité Pratique de Prononciation Française. Paris: Delagrave.
Gumperz J.J. (1982) . Discourse Strategies. Cambridge : Cambridge University Press.
Hall, E.T. (1959). The silent language. New York: Doubleday and Cie.
Hall, E. T. (1966). The hidden dimension. New York: Doubleday and Cie.
Helfrich, H. (1979). Age markers in speech. In In K.R. Scherer and H. Giles (Ed.), Social
markers in Speech., (pp. 63-107). Cambridge and Paris: Cambridge University Press and
Editions des Sciences de l'homme.
Hymes, D. (1972). Models of the Interaction of Language and Social Life. In J.J. Gumperz
and D. Hymes (Eds.),
Directions and Sociolinguistics: the ethnography of
communication. New York.
Johns-Lewis, C. (1986). Prosodic differentiation of discourse modes. In C. Johns-Lewis (Ed.),
Intonation in discourse, (pp. 199-219). New York: Springer-Verlag.
Joos, M. (1962). The five clocks, International Journal of American Linguistics.
Kowal, S. (1982). Pauses to Persuade: Politicians Reading their Speeches. Proceedings of the
International Conference on reading. Padova, 116-121.
Laver, J.D.M. and Trugdill, P. (1979). Phonetic and Linguistic Markers in Speech. In K.R.
Scherer and H. Giles (Ed.), Social markers in Speech., (pp. 1-32). Cambridge and Paris:
Cambridge University Press and Editions des Sciences de l'homme.
12
Léon P. (1971). Essais de Phonostylistique. Montreal, Bruxelles, Paris: Studia Phonetica,
Didier.
Léon, P. (1993). Précis de Phonostylistique. Parole et Expressivité. Paris: Nathan.
Lucci, V. (1983). La Variation Situationnelle en Français Standard. Grenoble: Presses de
l’Université.
Maclay H. and Osgood C.E. (1959). Hesitation Phenomena in Spontaneous English Speech,
Word, 15-19.
Murray, D.C. (1971). Talk, Silence and Anxiety. Psychological Bulletin, 75, 244-60.
Ohala, J. J. (1983). Cross-language use of pitch : An ethological view, Phonetica, 40(1), 1-18.
Ohala, J. J. (1984). An ethological perspective on common cross language utilisation of Fo in
voice, Phonetica, 41, 1-16.
Ramsay, R.W. (1966). Personality and Speech. Journal of Personality and Social Psychology,
4, 116-18.
Rochester, S.R. (1973). The Significance of Pauses in Spontaneous Speech. Journal of
Psycholinguistic Research, 2, 51-81.
Scherer, K. R. (1979a). Personality Markers in Speech. In K.R. Scherer and H. Giles (Eds.),
Social Markers in Speech, (pp. 147-209). Cambridge and Paris: Cambridge University
Press and Les Editions des Sciences de l’Homme.
Scherer, K. R. (1979b). Non linguistic indicators of emotion and psychopathology. In C. E.
Izard (Ed.), Emotions in personality and psychopathology (pp. 493-529). New York :
Plenum Press.
Scherer, K. R. (1981). Vocal indicators of stress. In J. Darby (Ed.), Speech evalutation in
psychiatry (pp 189-220). New York : Grune and Stratton.
Scherer, K.R. (1992). Vocal affect expression as symptom, symbol and appeal. In H.
Papousel., U. Jürgens, and M. Papousek, (Eds), Nonverbal vocal communication.
Comparative and developemental approaches (pp 43-60). Cambridge: Cambridge
University Press, and Paris: Editions de la Maison des Sciences de l'Homme.
Siegman, A.W. (1978). The Tell-Tale Voice: Non Verbal Messages of Verbal
Communication. In Siegman and S. Feldstein (Eds.), Non Verbal Behavior and
Communication. N.J.: Hillsdale.
Siegman, A.W. and Pope, B. (1965). Effects of Question Specificity and Anxiety Producing
Messages of Verbal Fluency in the Initial Interview. Journal of Personality and Social
Psychology, 2, 522-30.
13
Tielen, M.J.J. and Koopmans-von Beinum, F.J. (1991). Evaluation of Voice and
Pronunciation Characteristics of Men and Women. Actes du XII Congrès International
des Sciences Phonétiques d’Aix-en-Provence, Vol 2/5, 178-79.
14