Diapositive 1

1
Perceptual explanations of articulatory
variability in the realisation of the nasal
feature for the consonants.
J. Vaissière
2
Perceptual explanations of articulatory
variability in the realisation of the nasal
feature for the consonants.

• Jacqueline Vaissière
• Laboratoire de Phonétique et de phonologie -UMR
  7018, Paris

3
1) The nasal feature?

• Most of the languages use the feature nasal for
  contrasting consonants
• Only 20 of the languages use it also for
  contrasting the vowels (but often no real pairs,
  such as in French)
• Most of the nasal vowels emerge from contextual
  anticipatory nasalisation due to N in coda,
  followed and N-loss
• Easy articulatory, difficult acoustically,
  complex perceptually

4
The death of nasal consonant in coda and the
birth of nasal vowels?

• Regressive vowel nasalisation
• Final nasal consonants are more likely to
  nasalize the preceding vowels
• Ex  pan  in English
• But also  cama  first /a/ nasalized
• Final nasal consonants are more likeky to
  dissappear (N-loss) than initial consonants
  (nasal vowels)
• Ex  panum  in Latin gt pE in French
• Intervocalic nasal consonants may dissappear too
• Ex portuguese
• Luna Lua Solo Só
• Some cases of spontaneous nasalisation.
• .

5
2) Goal of this communication?

• the observed articulatory and aerodynamic
  inter-speakers and inter-languages differences
• between initial nasal consonants (/n/), and
  unreleased final consonants (/N/)?
• Perceptual ?
• Other known types of explanations for velum
  behavior in general
• articulatory (for anticipation)
• aerodynamic (for stops and fricatives)

2) goal of this communication?
6
2) Goal of this communication?

• Does not deal directly with the vowels
• or with the place of articulation of the nasal
  consonants
• -----
• This paper does deal with the nasalisty feature,
  the perception of nasality
• and the differences between initial and final
  nasals
• Data from in a number of languages
• French and English (X-ray), Japanese, and others
• Based on the available litterature, work in my
  lab, my own work.

2) goal of this communication?
7
3) the effect of the position of a consonant in a
syllable in general
8
the effect of the position of a consonant in a
syllable in general The syllable as a unit?

• Unit of physiological organization? (Stetson,
  1951)
• no
• Coextensive with the temporal domain of
  coarticulation (Kozhenikov and Chistovichn 1965)
  ?
• no
• Thoughs in the mvt pattern of one articulator
  (Gay, 1978)?
• no
• Characteristics patterns of articulatory
  organisation (Krakow, 1989)?
• yes, in careful speech at least
• Syllable organisation of phonological patterns
  (Ohala et Kawasaki, 1984)
• yes

3) the effect of the position of a consonant in a
syllable in general
9
the effect of the position of a consonant in a
syllable in general
Straka and Durand
1) stress
2) Position in syllable, word, syntagma
10
Sonorant in coda

• Sonorants  fused  with the preceding vowel
• Becoming more vowel-like
• /l/ gt u (chevals gt chevaux)
• /r/
• Nasal consonant gt nasal vowels or nasal glide or
  glide
• Backing also for nasal (velar nasal)

11
4) what is well known about the nasality feature?

12
A lot of very nice studies

• Japanese (Fujimura, Sawashima, Honda, etc.)
• English (Cohn, Krakow, Ohala, Bell-Berti,
• French (Benguerel, Amelot Rossato, Delveaux,
  Basset al Cohn, Badin
• Spanish (Solé )

13
Which have shown the many different factors
influencing the velum behavior
The many different factors influencing the velum
behavior
14
Many factors
dialects
South French
Canadien French
Nasal features first !
/a/ lower thet /i/ /p,t,k/ gt sonorants
Intrinsic velar height
Stress and effort
Impedance extremely important
/a/ less thet /i/ Around stops
Position in sentence
Language influences production and perception
Syllable boundaries
Speakers strateegy
Coarticulatory nasalized phonologized in English
style
Spontaneous/carefully
15
Anticipation attendue avant N
style
Basset al, 2001
Less than expected
More than expected
Nasalized /k/
No anticipation
Nasalized /v/
Basset al, 2001
16
Position in sentence
sentence

• Open velopharyngeal port is the unmarked case
• Natural  coda nasalisation 

17
/atu/ sequence
Final position favors nasalisation Pause behave
as a nasal consonant in French cVpause final
nasalisation

• From the university of Strasbourg (France)

Less well known, wrongly ignored in current
litterature
18
/atu/
19
/atu/
20
/atu/
Maximum opening (before /a/!!!)
21
/atu/
Jaw starts to rise Closing gesture starts At the
vowel beginning !!!
22
/atu/
Vocal tract already half Closed (in the middle of
the  acoustic  vowel !!!
23
/atu/
Protrusion of the lip starts
24
/atu/
25
/atu/
Lip continue to round (in anticipation of /u/)
26
/atu/
Lip continue to protrude
27
/atu/
28
/atu/
Lip continue to protruded, they are As protruded
as for /u/
29
/atu/
30
/atu/
31
/atu/
Deprotrusion starts Velum lowers
32
/atu/
33
/atu/
34
Lips still very protruded Carry-over velum low)
35
Natural final nasalisation

• You dont hear it.
• You barely see it on spectrograms
• But it is there.
• in French
• mV pause the velum does not rise again
• Pause act as a nasal phoneme
• This may explain presence of N as a coda

36
But conflicting conclusions

• Transitional because unspecified in English
• Cohn airflow

• phonologized anticipatory nasalisation in English
• Malécot, Ohala, Vaissière, and others

37
Initial /n/
Final /N/
Release
Onset
n
V
V
N
V
38
Partly due to different instrumentations

• In all cases, the nasal feature is realized by
  lowering the velum
• Connection oral and nasal cavities
• If enough acoustic coupling, the phoneme is
  perceived as nasalized or as a nasal vowel or
  consonant

39
Principal results

• A phoneme is perceived as nasal when there is
  enough coupling between the main VT and the nasal
  cavity
• Coupling necessary depends on the phoneme
  identity and on the speaker native language.

40
1) Velar height

• velum only down for the realisation of the nasal
  phonemes?
• Yes, but sometimes down for oral low vowel
• Same height for all oral consonants?
• No, lower for stops than sonorants, lower for low
  vowels than high vowels
• Same height for all nasal consonants?
• No, ower for final consonants, than initial,
• Yes, for /m/ and /n/
• Few data because invasive?
• X-ray, yes, but MRI, no

41
2) Nasal airflow

• 1. Nasal airflow is always a by-product of an
  open velopharyngeal port?
• No, positive airflow may be the result of a
  pumping effect
• No, Negative airflow is possible when velum goes
  down long the velo-phayrngeal wall
• No, glottal consonants
• No cold
• 2) Nasal airflow proportionnal to VP opening
• No, impedance (/i/ gt /a/)

42
3) Velopharyngeal port

• The best!
• but, lateral opening of theVP port is possible
  (IRM)

43
4) Articulatory synthesis

• Rather easy
• Maedas model
• Vowels and consonants

44
5) acoustics

• Not too bad, but dangerous to infer the velum
  behavior from only acoustic data
• Consonants /l/ has also zeros
• Vowels
• Zeros due to context

45
5) Perception

• If easy to create the sensation of nasality
• More difficult to do it in a well-motivated way
• Nasal tract very complicated
• Aerodynamics constraints difficult to model
• Nasalisation of sonorants , fricatives and vowels
  may go unnoticed because no nasal counterparts
  (non-native contrasts)
• For stops, corresponding nasal may be more easily
  heard
• The listeners would be sensitive to the total
  nasalisation in VN sequence (Beddor, 2007)

46
Is differences between initial and final
consonants strickly physiologically necessary
(production) ?

• No !

47
Initial and final Consonants Could behave the
same way
Anticipatory and carry-over are not strictly
necessary
48
what differences between initial and final
consonants generally observed ?

• Yes !
• Why?

49
a) Differences in velar height and VP opening

• Lower velum for final than initial nasal

50
Differences in velar height and VP opening
Also X-ray, X-ray microbeam system, IRM, French,
Japanese, etc.
51
Krakow, English
52
b) Differences in coarticulatory patterns
53

• More anticipation than carry-over
• More coarticulation when tautosyllabic but no
  blocking
• More anticipation when tautosyllabic (cvN)

54
what differences between initial and final
consonants generally observed ?

• Some conflicts in the conclusion
• But I will concentrate on similarities

55
Variability observedsame instrumentationsame
language (French)

• Variability beginning and middle of murmur in nV
• Maximum at /n/ release
• the only instant where nasality is necessary

Benguerel
Cohn
Ouvaroff
56
Maximum nasal airflow Minimal velum height
A point of rendez-vous The release of the nasal
consonant If missed no nasality perceived
But masking
Perceptual requirement
Close for aerodynamic reason
57
demo

• Most of the nasal murmur
• in nV, or VnV or VnV
• is not necessary for /n/ to be perceived as nasal
• Jadmets gt Jeanne met
• Et demi gt ennemi
• Maintenant gt mainnant
• Nombre, number gt nomme
• What counts perceptually is the release only the
  release

58

• So if the release perceptually counts for initial
  /n/
• What happens when a non released final consonant
  to be perceived as a nasal?
• Observations
• Articulatory observation minimum VH at VN
  boundary, that is at the N onset.
• A perceptual complication If short and no
  anticipatory nasalisation of the vowel, masking
  and murmur not perceived

105
59
Maximum nasal airflow Minimal velum height
But masking
Perceptual requirement
Close for aerodynamic reason
60

• So if the release perceptually counts for initial
  /n/
• What happens when a non released final consonant
  to be perceived as a nasal?
• Observations
• Articulatory observation minimum VH at VN
  boundary, that is at the N onset.
• A perceptual complication If short and no
  anticipatory nasalisation of the vowel, masking
  and murmur not perceived

105
61
What solution(s)?

• What to do the nasalisity of the coda to be
  perceived
• The French way
• Released it !
• Liaison
• Enchaînement
• Make it long !
• Vowel anticipatory nasalisation ! (Beddor, 2007
  integration)
• Anticipatory nasalisation even in French

62
language
Anticipatory nasalisation In languages With Even
in languages With nasal vowels
Cohn
63
(No Transcript)
64
VN/ C

• Different behavior depending on the surrounding
  consonants
• Nstop nasal release of the stop, maximum velar
  height
• If unvoiced stop VP close at the beginning
• If fricative maximum velar height at the middle

65
Maximum nasal airflow Minimal velum height
But masking
Perceptual requirement
Close for aerodynamic reason
66
Conclusions?
67
1) Perceptual consideration

• nV, VnV,
• A long murmur is not necessary
• Not audible for most of its lenght
• Around the consonant release
• It is essential importance

• VN, VNC,
• Long murmur necessary
• If short, masked
• Anticipatory nasalisation as a compensation
• Integration of nasality over V and N (Benguerel,
  Beddor)

68
Recall

• The same perceptual constraints for
• /k,g/
• Perceptual necessity and not articulatory ease!
• F2

69
2) alignement

• Consonants
• Onset of unvoiced stop is - nasal (of better VP
  close), aerodynamics
• Release of stops has to be -nasal , perception,
  nasal counterparts
• Fricative middle, variability at the edge
• Nasal vowels middle and last part, perception

70
3) From phonetic to phonology

• Much advantage to phonetically divide the
   phoneme  into three or four parts
• Onset, steady state, just before release,
  release, transition

71
4) multi-instrumentation

• Necessary for the nasality feature

72
5) Integration of multiple considerations

• Release
• And duration
• And anticipatory phenomena
• Should be faced in once
• Because of perception

73
Merci!