Document 15931720

advertisement

Comparing American and Palestinian Perceptions of Charisma Using Acoustic-Prosodic and Lexical Analysis

Fadi Biadsy, Julia Hirschberg, Andrew Rosenberg, and Wisam Dakka

The Department of Computer Science, Columbia University, New York, USA

Why study Charismatic Speech?

Inter-Subject Agreement Acoustic/Prosodic Analysis Abstract

Charisma, the ability to lead by virtue of personality alone, is difficult to define but relatively easy to identify. However, cultural factors clearly affect perceptions of charisma. In this paper we compare results from parallel perception studies investigating charismatic speech in Palestinian Arabic and American

English. We examine acoustic/prosodic and lexical correlates of charisma ratings to determine how the two cultures differ with respect to their views of charismatic speech.

What is Charisma?

• Ability to attract and retain followers by virtue of personal characteristics – not traditional or political office (Weber ‘47)

• What makes an individual charismatic?

• Their message?

• Their personality?

• Their speaking style?

• Charisma arising from the faith of a leader’s listener-followers

(Marcus, 1967)

• It as a combination of a ‘gift of grace’, an inspiring message and an important crisis (Boss, 1976)

• Tannen (1984) identifies a number of pragmatic dimensions that vary cross-culturally including when to talk, formulacity, and degree of indirectness, cohesion and coherence.

What is Charismatic Speech?

• It is an intriguing phenomenon

• To identify potential charismatic leaders

• To provide a feedback system for individuals who want to improve their speaking style – politicians, professors, students …

• To create a charismatic Text-to-Speech system, when compelling speech is needed (e.g.,

Intelligent tutoring system)

Our Approach

• Collect tokens of charismatic and noncharismatic speech from a small set of speakers on a small set of topics

• Ask listeners to rate the ‘ The speaker is charismatic ’ plus statements about other 25 attributes (e.g., The speaker is

boring, charming, persuasive,…

)

• Correlate listener ratings with lexico-syntactic and acoustic-prosodic features of the tokens to identify potential cues to perception of charisma

American and Palestinian Perception Studies

• Speech that leads listeners to perceive the speaker as charismatic

• What aspects of speech might contribute to the perception of a speaker as charismatic?

• Content of message?

• Lexico-syntactic features?

• Acoustic-prosodic features?

• Subjects for English experiment: 12 native

American English speakers (6 f, 6 m)

• They were presented with 45 speech segments of 2 –28 seconds duration, 5 each from 9 candidates for Democratic nomination for U.S. president in 2004

• Topics:

greeting, reasons for running, tax cuts, postwar Iraq, healthcare.

Is Charisma a Culture-Dependent

Phenomenon?

• Do people of different languages and cultures perceive charisma differently?

• Do they perceive charismatic speech differently?

• Do Arabic listeners respond to American politicians the same way Americans do?

• Do Americans hear Swedish professors the same way Swedish students do?

• Subjects for Arabic experiment: 12 native

Palestinian Arabic speakers (6 f, 6 m)

• Presented with 44 speech tokens of 3–28 seconds, 2 each from Palestinian politicians and authors

• Tokens extracted from Aljazeera talk shows, topics:

the assassination, of the Hamas leader, the debate among the Palestinian, groups, The

Intifada and resistance, the Israeli separation wall, the Palestinian Authority and calls for reforms

‘Functional’ Definition of Charisma

H-H%

Influence of Speaker and Topic on Charisma Ratings

• Identity of speaker of a segment: significantly influences subjects ’ ratings of charisma in

both

studies.

• Subject ’s Recognition of speaker: positively influences perceptions of charisma in

English

.

• Topic (in English): approaching statistical significance on subjects ’ ratings of charisma. Healthcare (mean rating 3.31), postwar Iraq (3.29), reasons for running (3.28), content-neutral

(3.07), and taxes (2.97).

• Topic (in Arabic): influences charisma ratings. The Israeli separation wall (3.96), the assassination of the Hamas leader

(3.37), the debate among the Palestinian groups (3.23), the

Palestinian Authority and calls for reforms (3.21), and the

Intifada and resistance (3.17)

Recent results: Prediction of Charisma (actual vs. predicted charisma score)

Arabic English

Lexical Analysis

Token duration: charisma in

Pause: in the token gives a positive correlate only in

Arabic both

Duration

positive correlate with studies number of pauses to number of words

• Sdev of length of pauses: negative correlate in English but positive in Arabic

• Number of words: positively correlated with charisma in both studies

• Disfluency: rate of disfluencies (repetitions, repairs, and filled pauses) negatively correlated with charisma in both

• The filler “yaEony” (ينعي): negatively correlate with charisma in Arabic

• Ratio of repeated words: positive correlate in both

• The use of Arabic regional dialect: negative correlation with charisma

Speaking Rate

• Speaking Rate: positive correlate in English and approaching negative correlate in Arabic

Speaking rate of fastest intonational phrase: positive correlation in both

Pitch

• Mean F0: positive correlate in both studies

• Min F0: positive correlate in English but negative in Arabic

• Max and Sdev F0: positive correlate in Arabic only

HiF0 max and Sdev:

Arabic only

Mean HiF0: positive correlate in positive correlate in both

Prominalization

• First person plural pronoun: positive correlate in English only

• Third person singular pronouns: negative correlate in Arabic but positive in English

• Third person plural pronouns: negative correlate in English but positive in Arabic

Part of Speech

• Ratio of adverbs, prepositions, and nouns: negative correlate in Arabic

• Ratio of adverbs and adjectives: negative correlate in English

• Max F0:

Normalized Pitch

approaching significance with negative correlation in

• Mean HiF0: positive correlate in both

• Max and Sdev HiF0: positive correlate in

Arabic only

English

Mean and Sdev of rms of IPs: correlate in both positive

• Sdev Intensity: positive correlate in Arabic only

Intensity

ToBI Labels

• Ratio of H* pitch accents: negative correlate in English only

• Ratio of !H* and L+H* pitch accents: positive correlate in both

• Ratio of L* pitch accent: negatively correlates with charisma in both languages

• Ratio of H- phrase accent: phrase accent approaches positive significance in English only

• L% boundary tone: positively correlate with charisma in Arabic only

• H% boundary tone: negatively correlate with charisma in Arabic only

Conclusions and Future Work

• While American and Palestinian speakers share some notion of a ‘functional’ definite of charisma — both find speakers who are persuasive, charming, enthusiastic and not boring to be charismatic — American speakers also find passionate and convincing speakers to be charismatic, while Palestinians associate the qualities of toughness and powerfulness with charisma

• Arabic subjects tended to be more homogenous in their judgments than were

English subjects

• When we examine the acoustic/prosodic and lexical correlates of charisma in each study, we again find some broad similarities and some major differences

Future Work

• Machine learning experiments to predict degree of charisma for a given speech token based on our features

• Perception experiments:

• Palestinian and Swedish subjects rate

English

• American subjects rate Arabic

Download