Prosodic phrasing in oral narration: Intentional control and operational constraints - PowerPoint PPT Presentation

Loading...

PPT – Prosodic phrasing in oral narration: Intentional control and operational constraints PowerPoint presentation | free to view - id: d4fe9-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Prosodic phrasing in oral narration: Intentional control and operational constraints

Description:

An intonation phrase may consist of a single syllabe up to syntactic phrases, ... involved in the melody, 'the music', the intonation of. speech. ... – PowerPoint PPT presentation

Number of Views:76
Avg rating:3.0/5.0
Slides: 30
Provided by: annie71
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Prosodic phrasing in oral narration: Intentional control and operational constraints


1
Prosodic phrasing in oral narration Intentional
control and operational constraints
  • Monique Vion Annie Colas
  • Laboratoire Parole et Langage
  • UMR 6057 CNRS-Université de Provence
  • 29 av. R. Schuman
  • F-13621 Aix-en-Provence cedex1
  • E-mail mvion_at_up.univ-mrs.fr

2
K. Bock (1995, p. 185)
3
 The intonation phrase (also referred to as
Intonation unit, prosodic sentence, breath group,
etc.) refers to coherent intonation structures
with no major prosodic break. An intonation
phrase may consist of a single syllabe up to
syntactic phrases, clauses and sentences and a
larger utterance may thus be pronounced with one
or severai intonation phrases which may not have
any predictive one-to-one correspondance with
syntactic semantic unit.  Botinis, Granström,
Möbius (2001, p. 269)
4
When the speaker does take a break he
factually completes an intentional phrase (IP).
1. IP are, to some extent pragmatic devices under
the speakers intentional control.
2. speakers prefer to make IP with a particular
range of duration (tendency to equal durations).
3. speakers usualy break when they reach sentence
boundary.
4. a speaker may be inclined to break shortly
after a very prominent speech accent.
5. If at a potential break point no further
surface mate- riels have become available for
phonological encoding, the speaker must take the
option in order to gain proces- sing time.
from Levelt (1989, p. 385-86)
5
 The global function of prosody mainly consists
in assisting the coding and decoding of speech
which essentially operates on the levels of the
semantic and pragmatic interpretation of
utterances.  Di Cristo (2000)
Intentional Control
6
 The lack of new amnunition may have different
causes, where may be planning trouble at the
message level of grammatical encoding. Lev
elt (1989)
Operational Constraints
7
Hypotheses
H1 Cues of the cognitive activities occurring
during planning are provided by the way words
are grouped by prosody.
H2 Prosodic boundaries are exploited while the
listener is interpreting the produced utterance.
8
H1 Prosodic structure varies as a function of
the speakers intentionally exerted control.
operational constraints related on mnemonic
activity and inferential resolution.
H2 prosodic structure can help the listeners in
interpreting the utterance.
9
Narratives
Cross-sectional evolutive method 7-, 9-, and
11-year-old children
10
Arbitrary Sequence
11
Ordered Sequence
12
(No Transcript)
13
Perceptual Evaluation
84 recordings
naïve French listeners
14
Speech band-pass filtered Lehiste Wang (1977)
Lehiste (1979) Kreiman (1982) Swerts
Geluykens (1994) Text vs. TextSpeech Grosz
Hirschberg (1992) Grosz, Hirschberg Nakatani
(1994) Grosz Hirschberg (1995) Nakatani,
Hirschberg Grosz (1995) Hirschberg Nakatani
(1996) Swerts (1997) Auran, Portes, Rami Rigaud
(2001) Speech Lehiste (1975a) Sluijter Terken
(1993) TextSpeech Swerts (1998) Portes (2000,
2002) Swerts Geluykens, (1994) Swerts
(1997, 1998) Portes (2000, 2003)
15
Instructions (extracts)
We are interested in the cues which listeners
usually use in order to structure and understand
utterances.
We are particularly interested in sound phenomena
involved in the melody, "the music", the
intonation of speech.
We hypothese that these phenomena take part in
the structuring of what is told.
Your task will consist in listening to the
recordings and to indicate on the written text
the breaks that you perceive on the basis of the
sound phenomenon which accom- panies the words.
16
Procedure
1. Listen.
2. Listen and label (/) versus (//).
3. Listen and scale.
4. Listen and check.
17
(No Transcript)
18
7 judges by recording
One judge listens to 6 recordings
112 judges
19
Results from eleven-year-old childrens
corpora
Predictions
Display mode x Sequence-type interaction
20
C.S. fisherman/ B.D. pêcheur
  •  the boy settles down to fish. 
  • -  Le garçon sinstalle pour pêcher.
  •  the boy goes near the river,
  • sits down with his net beside him,
  • diving his fishing rod into the water and waits.
  •   le garçon sapproche de la rivière,
  • sasseoit avec son épuisette à côté de lui,
  • Trempe sa canne à pêche dans leau et attend. 

21
From group segmentation data one can derive a
hierarchical discourse structure . It is
logical to assume there will be less
disagreement about stronger breaks since they
present clearer transitions in the flow of
informations. Swerts, 1997, p.515
22
Criteria
  • Break if 4 judges out of 7 at least
  • Category (continuative/final) majority choice
  • Strength average of the weights corresponding
    to the
  • majority choice.

23
Coding Patterns 5 slots
24
Tools
CHILDES  Child Data Exchange System, MacWhinney,
1991. CHAT Codes for the Human Analysis of
Transcripts). CLAN  Computerized Language
Analysis. Available at http//childes.psy.cmu.edu
/
PRAAT, a system for doing phonetics by computer,
Boersma, P., Weenink, D., 1992-2000. Available
at http//fonsg3.let.uva.nl/praat/
25
Example _at_begin _at_participants CHI Benoît
child _at_ID 209.09.Ex.g _at_group of
DEP _at_G filper _at_bg V1m CHI alors y'a y'a c
quelqu'un et i va pêcher avec son père
c. cod c7VIIc157C. c7IV225C. _at_bg V2m
CHI i prépare ses affaires. _at_bg V3m CHI et i
jette sai jette son fil c. cod 7cVII214C. _at_b
g V4m CHI et puis après y'a quelque chose au
bout duau bout d'laau bout du fil
c. cod c5III233C. _at_bg V5m CHI oui mais le
quelque chose, c'est un chaussure
c. cod t7IV35T. _at_bg V6m CHI et puis
dedans la chaussure, y'a un poisson
c. cod c7IV125T. _at_bg V7m CHI il l'attrape
avec son épuisettec. cod c7VII114C. _at_bg V8m
CHI puis est bien content d'avoir eu un poisson
c. cod t7VII386T. _at_end
26
Two Results
Number of unanimously identified breaks
Sequence type x Display interaction
F(1,24) 4,087 plt.0545.
27
(No Transcript)
28
Display-mode effect F(1,24) 4,292 plt.0545
The judges have perceived more final boundaries
in the recordings collected in  consecutive
display  (3,07 units on average) than in the
recordings collected in  simulta- neous
display  (2,07 units on average).
29
to follow
7- and 9-year-old children
toward a stratificational approach to the
relations of prosody to discourse (ProDiGE)
About PowerShow.com