Post on 30-Dec-2015
description
transcript
Nasal venting and the early mastery of Greek voiced stops.
Eun Jong Kong*, Mary E. Beckman† and Jan Edwards*
* University of Wisconsin-Madison, † Ohio State University
ICGL 2009 (OCT 29-31: 10:30 -10:55 a.m. 31th)
1. Introduction• Stop voicing/aspiration categories and Voice Onset Time (VOT) as an acoustic measure:
VOT = Burst – Voicing Onset
Burst Voicing OnsetTime (s)
121.2 121.30
5000
Fre
quen
cy (
Hz)
Short lag VOT
Long lag VOT
Time (s)0 0.2006
0
9000
Freq
uenc
y (H
z)Time (s)
15.13 15.380
5000
Fre
quen
cy (
Hz)
Lead VOT
BurstVoicing Onset Burst Voicing Onset
Word-initial stops with a voicing/aspiration contrast (e.g., /d/ vs. /t/ vs. /th/) are differentiated by VOT (Lisker andAbramson, 1964).
[do'mata] tomato toast[tixos] wall donut
Lead Long lagShort lag
Language Lead Short lag Long lag
Cantonese unaspirated aspirated
Spanish voiced voiceless
English voiced voiced voiceless
Thai voiced unaspirated aspirated
• Laryngeal categories and VOT
Language Lead Short lag Long lag
Cantonese unaspirated aspirated
Spanish voiced voiceless
English voiced voiced voiceless
Thai voiced unaspirated aspirated
Short lagLong lagLead
• Phonological development and VOT
- Universal order of mastery:
[t] [th] [d]- Kewley-Port & Preston (1974):the relative difficulty of producing the different VOT.- Late mastery of voiced stops: maintaining the supra-glottal air pressure lower than the sub-glottal pressure. : Spanish (spirantization error) French (nasal venting)
• A puzzle in the mastery of Greek stops
Language Lead Short lag Long lag
English voiced voiced voiceless
Spanish voiced voiceless
Thai voiced unaspirated
aspirated
Greek voiced voiceless
Greek adult VOT patterns predicts:
Lead Short lag
HOWEVER…The studies found adult-like lead VOTs in 2 year-olds’ stops. (Okalidou, Petinou,Theodorou & Karasimou et al 2002, Kong& Beckman, 2006)
[t] then, [d]
: voiced vs. voiceless
& adult like lead VOT at 4 or 5.
• Possible explanation of this puzzle can be …the fact that Greek voiced stops are not always ‘truly voiced’ but can be prenasalized: they are historically developed from a nasal + voiceless stop cluster (Arvaniti & Joseph 1999).
-- the nasal venting facilitates voicing.
• VOT cannot capture this prenasality in Greek voiced stops. In Burton, Blumstein and Stevens (1992), time-course of amplitude change of the first resonance peak was a successful measure but not a duration in differentiating voiced stops, prenasalized stops and nasals in Moru.
Figure. The schematized version of phonetic analysis of prenasalized stops in from Burton et al. (1992).
Hypothesis:
• If prenasalized variants are accepted in children’s productions, this could explain the earlier acquisition of the voicing contrast by Greek-speaking children relative to French- or Thai-speaking children.
Goals of the study
• This study provides acoustic evidence that Greek voiced stops are at least partially prenasalized in adults’ and children’s productions.a. direct comparison between voiced stops and nasals
in word-initial positionb. examination of their spectral quality because a
duration measure of VOT is not sufficient.
• Materialsword-initial voiced stops (/b/,/d/) and nasals (/m/, /n/).: cross-sectional data from the Paidologos project http://www.ling.ohio-state.edu/~edwards
• Task picture naming task for children e.g., ['bala] ball, [mixa'ni] motorbike, [du'lapa] wardrobe, [ne'ro] water
word-reading task for adults.e.g., [bu'kali] bottle, [mu’ɡos] mute, [do'mata] tomato, [no'mizo] I think
• Subjects93 children (2;0 – 5;11) and 6 adults (18;0 – 30;0) monolingual Greek speakers. Recordings were made in Thessaloniki, Greece.
2. Method
• Two acoustic measures
a. Duration of voicing lead (i.e., VOT) in the voiced stops and nasal murmur in nasal consonants.
b. Amplitude changes during the voicing lead and nasal murmur:
Burst
Time (s)33.78 33.98
glottal pulse
[berðevo]
25ms
- the first peak amplitude in the FFT spectrum made from a 6 ms Hamming window centered at each glottal pulse starting at the burst.- normalized by the amplitude of the following vowel.
6ms
Frequency (Hz)
Am
plitude (dB)
peak amplitude
1. VOT distribution of voiced stops and nasals by children.
3. Results (duration)n
um
ber
of
token
s
VOT in seconds
• The percentage of prevoiced stops:
girls : 84.7% (72/85) boys: 82.2% (74/90)
2 yos ~ 5 yos: 85.7%, 79.1%, 87.5% and 81.8%
Lead
2. Amplitude trajectories: 6 Adults
3. Results (amplitude)
vd. stops
nasals
burstvoicing onset
voiced stop (gtf07)
begins with energy as high as her nasals followed by an energy decrease over time toward the burst.
voiced stop (gtm08)
begins with lower energy than his nasals and maintains this same amplitude difference.
Smoothing spline ANOVA (Gu 2002, Davidson 2006)
nasal:
gradual increase over time
2. Amplitude trajectories: 6 Greek adults
Figure) The schematized version of phonetic analysis of prenasalized stops in Moru (pp. 137) from Burton et al. (1992).
3. Results (amplitude)1. Amplitude trajectories: by children’s age groups.
nasal:
gradual energy increase over time
voiced stop: begins with an amplitude as high as the nasal murmur amplitude followed by a gradual separation toward the burst.vd. stops
nasals
Df
2. Amplitude trajectories: Greek children
Figure) The schematized version of phonetic analysis of prenasalized stops in Moru (pp. 137) from Burton et al. (1992).
adults
3. Results: summary 1. Duration:
Young Greek-speaking children could produce adult-like prevoicing lead (unlike Spanish, French and Thai).
2. Amplitude trajectories: • The amplitude trajectory of the prevoicing lead was a
successful acoustic measure to capture the prenasalized quality in Greek voiced stops.
• Prenasalization in Greek adults’ voiced stops; -- an amplitude as high as in the nasals followed by an amplitude drop.-- speaker variability
• Prenasalization in Greek children’s voiced stops-- a high degree of nasality that was extended over the entire duration of the prevoicing lead.
4. Discussion and conclusionGreek voiced stops …
• are in contrast with voiceless stops• have lead VOT values• can be prenasalized.
Mastery of Greek voiced stops• adult-like lead VOT in children’s voiced stops • acoustic evidence of prenasalization in Greek voiced stops
: children were taking advantage of nasal venting to lower the supraglottal pressure in order to maintain vocal fold vibration.
4. Discussion and conclusion
Acoustic properties of stop voicing categories: • There exist various language-specific differences in the
acoustic realization of stop voicing categories• VOT is not always sufficient to describe stop voicing
categories, given language-specific acoustic properties.
Children’s mastery of stop voicing categories …• Cannot be fully understood based on transcription… • Or even based on VOT measurements (at least for
Greek)• Understanding the language-specific fine phonetic
details allows us to explain seemingly exceptional patterns.
• Ultimately, this will allow us to capture the universal patterns in the mastery of stop voicing categories.