What tools are needed to account for prosody in its relation with discourse

1
What tools are needed to account for prosody in
its relation with discourse ?

• A. Di Cristo, C. Portes, C. Auran, R. Bertrand
• Laboratoire Parole et Langage
• Aix en Provence

8th International Pragmatic Conference, Toronto,
13-18 july 2003
2
Introduction

• A session devoted to  Prosody and text units  
  !
• Our talk
• A preliminary issue
• What kind of prosodic information is needed to
  deal with Discourse analysis?
• Our object
• A complex interface network of heterogeneous
  forms and functions

3
Some precisions

• Form
• in a general meaning, any verbal, prosodic and/or
  gestural configuration used in order to convey
  meaning
• Function
• these pragmatic meanings
• Prosodic information
• a real question and a complicated problem
• What kind of prosodic cues?
• How to represent them?
• Working on French

4
Hypotheses about discourse

• Discourse
• A contextualization activity
• The product of this activity
• Discourse interpretation based on integrated
  representations
• Interpretation of the combination of
  propositional structures (Roulet et al., 2001)
• Interpretation of embodied experiences (Gibbs,
  2003)

5
What about prosody?

• Prosody contributes to the construction of these
  integrated representations through
• Its multilevel parsing function
• Its relational and inferential functions
• Its identification function
• Types of prosodic markers
• Local, global, iterative
• Categorical, scalar, gradual
• Perceptually relevant (interpreter strategy,
  Reboul Moeschler, 1998)

6
What tool(s) ? A multilinear grid

• Two patterns of interaction for discourse cues
  (verbal, prosodic and mimo-gestural cues)
• Convergence
• Dispersion
• A multilinear grid for analysis
• Direct interpretation of vertical alignment or
  syntagmatic combination of cues
• Emergence of original interactions (heuristic
  aspect)
• Emergence of relevant cues masked by their
  superposition

7
What tool(s) ? A multilinear grid
8
INTSINT tier and  list effect 

• Repetition of intonational cadences (LUT)
  frequent in enumerations

9
Downtrend tier and syntactic cohesion

• downtrend on successive accented syllables (A)
• Syntactic cohesion of the first four accentual
  groups
• Broken by the final commentary group
  (rising-falling contour)
• Role of pauses

10
Pitch Range and parenthesis

• Raised level and span expansion emphatic
  parenthesis

11
Pitch Range and polyphony

• Normal ReSp/ReLe speaker proper voice

• Raised/expanded episod reported speech

• Normal/compressed segment speaker self quotation

12
Rythmical tiers and syntactic iteration

• Repetition of rythmical pattern 2 1 3
• Repetition of syntactic structure
• Metrical iterationsyllable lengtheningpausespit
  ch variations, etc.

13
 arc accentuel  (accentual arc) and textual
grouping

•  arc accentuel  secondary and primary accent
  framing a text segment
• Associated with the compression of the register
  span
• Iconic of  entre guillemets 

14
Conclusion

• Those different samples show the diversity of
  prosodic means to materialize pragmatic meanings
• They have allowed us to show how we conceive
  prosodic cues and represent them in order for the
  description to better fit discourse variability
• This versatility is allowed by the collaboration
  of different prosodic cues associated with cues
  from other levels of language analysis
• This approach is work in progress, the aim of
  which is to identify sets of forms and functions
  ,and their relations, relevant to discourse
  analysis

15
Thank you for your attention
This presentation is available online at the
following address http//www.lpl.univ-aix.fr/p
rodige/
16
The textual part of the multilinear grid