On the phonological status of pitch falls in English and Dutch: Evidence from semantic judgements Experimental Studies on Intonation: Phonetic, Phonological and Psycholinguistic Aspects of Sentence Prosody, 5-7 January 2009, Potsdam University Carlos Gussenhoven Queen Mary, University of London
52
Embed
Experimental Studies on Intonation: Phonetic, Phonological and Psycholinguistic
Experimental Studies on Intonation: Phonetic, Phonological and Psycholinguistic Aspects of Sentence Prosody, 5-7 January 2009, Potsdam University Carlos Gussenhoven. On the phonological status of pitch falls in English and Dutch: Evidence from semantic judgements. - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
On the phonological status of pitch falls in English and Dutch:
Evidence from semantic judgements
Experimental Studies on Intonation:
Phonetic, Phonological and Psycholinguistic
Aspects of Sentence Prosody,
5-7 January 2009, Potsdam University
Carlos Gussenhoven
Queen Mary, University of London
Outline
Two analyses of pre-nuclear steep falls in English:
• and we then kept ALL the bottles that had a dePOSit on them
%L H*L H*L L%
DISPLACEMENTCONTINUATION
CONTINUATION
ToDI
• and we then kept ALL the bottles that had a dePOSit on them
%L H*LH H*L L%
DISPLACEMENTCONTINUATION
CONTINUATION
ToDI
• and did you keep ALL the bottles that had a dePOSit on them
%L H* L*H H%
CONTINUATION
CONTINUATION
ToDI
CONTINUATION
• and did you keep ALL the bottles that had a dePOSit on them
%L H*L L*H H%
ToDI
• and we then kept all the bottles that had a dePOSit on them
%L H* %
CONTINUATION
ToDI
CONTINUATION
maar kep NIET gezegd dat we niet toe moeten naar herverdeling van ARbeid‘But I haven’t said we shouldn’t consider redistribution of labour’(Neelie Smit-Kroes)
%L H*L H*L L%
ToDI’s tritonal prenuclear H*LH
%L H*LH% %L H*L L%
%L H*LH H*L L%
Research question
Does a steep F0 fall from an accented syllable signal an IP-boundary?
MethodologySemantic judgements about F0 contours on two-accent source utterances that are disambiguated by an IP-boundary
Experiment I: English
Verbal adverb vs. sentence adverb
She TREATED the poor man(,) HONESTLYI THOUGHT she responded(,) ODDLYHe NEVER acted(,) STRANGELYHe DEALT with the woman(,) HONESTLY
Experiment I: English
f0 manipulation on durationally hybridized source utterances by female AmE speaker
- Three pitch accents H*L, H*LH and H* (all before H*L L%)- Boundary vs No boundary- %L and %H- Female speaker read eight sentences (two sets of four)- Durational hybrids were created by judicious splicing and cutting of sections in the speech wave form, per segment. - Two sentences used the ‘comma’ source utterances, two the ‘no comma’ source utterances- 12 F0 contours on each hybridized speech files (i.e. only f0 varied)
Experiment I: English
2 x 6 contours
%H%L H*L (L% %L) H*L L%
%H%L H*LH (% %L) H*L L%
%H%L *H ( % %L) H*L L%
Experiment I: English
%L %H
%H%L H*L L% %L H*L L%
%H%L H*L H% %L H*L L%
%H%L *H H*L L%
%H%L H*L H*L L%
%H%L H*LH H*L L%
%H%L *H % %L H*L L%
Experiment I: English
Task
- Pairs of contours: which is more likely to have the comma? (‘1, 2 or neither’); ditto ‘lack’- %L and %H sets of 5 x 6 contour pairs- Sentences and contours Latin-squared- 15 American English subjects (12 vs 3 per order)
Experiment I: English
Your task in this 10-minute experiment is to listen to a number of pairs of sentences and to decide which of the two pronunciations expresses better that there is coherence between the verb and adverb. You do that by checking either the first or the third box. In the following example, the listener judged that the second pronunciation of the sentence more clearly expresses that there is coherence between the verb and the adverb, i.e. that the treatment of his case was honest:
She dealt with him honestly
In the next example, the listener thought that her treatment was honest: 1
She dealt with him honestly
You are always encouraged to make a decision. You will find that ...
She dealt with him honestly
1 0 2
1 0 2
1 0 2
Experiment I: English
Your task in this 10-minute test experiment is to listen to a number of pairs of sentences and to decide which of the two pronunciations expresses the ‘comma intonation’ better. You do that by checking either the first or the third box. In the following example, the listener judged that the second pronunciation of the sentence more clearly expressed the comma than the first:
She dealt with him, honestly
In the next example, the listener thought that the first better expressed the comma:
She dealt with him, honestly
You are always encouraged to make a decision. You will find that often ...
She dealt with him, honestly
1 0 2
1 0 2
1 0 2
Experiment I: English
Experiment I: English
Experiment I: English
Conclusions
1. Steep fall followed by rise (the ‘Jackknife’ of O’Connor & Arnold) does not define a boundary;
2. Off-ramp analyses (British, ToDI) are supported if H*LH is accepted. A rival analysis is L+H*+L followed by H+H*. Neither of these are available in MAE-ToBI.
Experiment I: English
And something needs to be done such that H+H* and H+L* always appear after L+H*+L .
Tone Concord
Experiment II: Dutch
(e.g. Wells 2006, Intonation)
- Identical melodic structure - separated by an IP-boundary
Pitch accents
Tone Concord
Experiment II: Dutch
- Apposition Mr. Evans, the butcher (cf. John the Baptist)
- Reformulation They had about ten, about a dozen
- Optional adverbial Act normally, like William
- Non-restrictive relative clausesHe chose to keep the watch, which was never repaired - ...
Tone Concord
Experiment II: Dutch
- Restrictive apposition Mr. Evans, the butcher (cf. John the Baptist) - Reformulation They had about a dozen, more than ten
- Optional adverbial vs. modal adverb ‘Act normally, like William’ - Non-restrictive relative clauses He chose to keep the watch, which was never repaired
Doe gewoon, zoals Willem
‘Act normally, like William’
vs
Doe gewoon zoals Willem
‘Just act like William’
Two hypotheses
1 No boundary after prenuclear H*LH (like English experiment)
2 The off ramp (=the fall from the peak) represents the pitch accent (and not the on-ramp=the rise to the peak)
Experiment II: Dutch
Two hypotheses
1 No boundary after prenuclear H*LH (like English experiment)
2 The off ramp (=the fall from the peak) represents the pitch accent (and not the on-ramp=the rise to the peak)
Experiment II: Dutch
Semantic difference due to ±bounday (identical melodies)
Semantic difference due to variation in shape of pre-nuclear pitch accent (+boundary)
Identification experiment
Semantic task: meaning modal adverb or otherwise? Female speaker3 ambiguous words2 two sentence lengths6 contours (Praat, f0 manipulation)5-point scale Two counterbalanced orders, reversed scales 5 filler contours20 listeners
Experiment II: Dutch
zonder anderen in ’t café [ ] [ ] [ ] [ ] [ ] slechts in ’t café
repeated stimulus presentation
doe normaal, zoals Willem [ ] [ ] [ ] [ ] [ ] doe maar net als Willem
in een file op de snelweg [ ] [ ] [ ] [ ] [ ] zeker op de snelweg
Experiment II: Dutch
%L H*L L% %L H*L L%
%L H*L H% %L H*L H%
%L L*H % %L L*H H%
Experiment II: Dutch
210180
290
160
200
240
270
250
180 160
165 160
240240
280
%L H*L L% %L H*L L%
%L H*L H% %L H*L H%
%L L*H % %L L*H H%
210180
290
175
270
250
180 160
165 160
240
240
280
190
%L H*L H*L L%
%L H*LH H*L H%
%L L*H L*H H%
No boundary -> Modal meaningBoundary -> Concord
Experiment II: Dutch
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Short Long
Mod
al m
eani
ng Boundary
NoBoundary
Hypothesis
L*H Long
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
gBoundary
No Boundary
H*L Short
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
Boundary
No Boundary
H*LH Short
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
Boundary
No boundary
L*H Short
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
Boundary
No Boundary
H*L Long
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
Boundary
No boundary
H*LH Long
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
Boundary
No Boundary
‘She’s surely there?’ is semantically incongruous
Identity of pitch accent
Experiment II: Dutch
%L H*L % %L H*L L% % L+H* L- % L+H* L- L%
On-ramp (ToBI) vs Off-ramp (ToDI)
%L H*L % %L H*L L%
% L+H* L- L% % L+H* L- L%
Experiment II: Dutch
%L H* % %L H*L L%
% L+H* L- L% % L+H* L- L%
%H H*L % %L H*L L%
%H H* L- L% % L+H* L- L%
hij zit alleen in het café
hij zit alleen met die man in het café
%H H*L % %L H*L L%
%L H* % %L H*L L%
Time (s)0 2.95009
100
400
Time (s)0 2.95009
100
400
%L H*L % %L H*L L%
Time (s)0 2.464
100
400
Freq
uenc
y (H
z)
Non-identity -> Modal meaningIdentity -> Concord
Experiment II: Dutch
Hypothesis
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Short Long
Mo
dal
mea
nin
g
%H H*L
%L H*
%L H*L
Experiment II: Dutch
Prenuclear Rise vs %H-Fall vs %L-Fall
Long
0
0.51
1.5
22.5
3
3.54
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
%H H*L % H*LL%
%L H* % H*L L%
%L H*L % H*L L%
??He is aLONE with that man in the PUB
‘alone’ has shifted to ‘only’
Prenuclear Rise vs %H-Fall vs %L-Fall
Short
0
0.5
11.5
2
2.5
33.5
4
4.5
alleen gewoon vast
Mo
dal
mea
nin
g
%H H*L % H*LL%
%L H* % H*L L%
%L H*L % H*L L%
Experiment II: Dutch
Conclusions
• Pre-nuclear H*LH exists (Jackknife has no boundary after first peak)
• An accentual peak is a fall, not a rise.• Listeners respond sensibly to meaning identification task.
•
Experiment II: Dutch
1. Two meaning identification tasks, one for English and one for Dutch, show that pre-nuclear steep falls exist, analysed as H*LH
2. Tone Concord for a pitch peak is perceived when a fall precedes, not when a rise precedes. This suggests a peak is (L) H*L, not L+H* (L).
3. The morphological and phonological structure of the intonation of West Germanic is an under-researched field.