Development of coarticulatory patterns in spontaneous speech Melinda Fricke Keith Johnson University of California, Berkeley.

Slides:



Advertisements
Similar presentations
Acoustic and Physiological Phonetics
Advertisements

Sounds that “move” Diphthongs, glides and liquids.
SPPA 403 Speech Science1 Unit 3 outline The Vocal Tract (VT) Source-Filter Theory of Speech Production Capturing Speech Dynamics The Vowels The Diphthongs.
Some speech disorders in school-aged children originate as normal behavior.
Basic Spectrogram & Clinical Application: Consonants
Acoustic Characteristics of Consonants
Philip Harrison J P French Associates & Department of Language & Linguistic Science, York University IAFPA 2006 Annual Conference Göteborg, Sweden Variability.
Glides (/w/, /j/) & Liquids (/l/, /r/) Degree of Constriction Greater than vowels – P oral slightly greater than P atmos Less than fricatives – P oral.
Voice quality variation with fundamental frequency in English and Mandarin.
“Connecting the dots” How do articulatory processes “map” onto acoustic processes?
Coarticulation Analysis of Dysarthric Speech Xiaochuan Niu, advised by Jan van Santen.
Spoken Language Analysis Dept. of General & Comparative Linguistics Christian-Albrechts-Universität zu Kiel Oliver Niebuhr 1 At the Segment-Prosody.
“Speech and the Hearing-Impaired Child: Theory and Practice” Ch. 13 Vowels and Diphthongs –Vowels are formed when sound produced at the glottal source.
Digital Systems: Hardware Organization and Design
Statistics for Linguistics Students Michaelmas 2004 Week 6 Bettina Braun
ACOUSTICAL THEORY OF SPEECH PRODUCTION
The Human Voice Chapters 15 and 17. Main Vocal Organs Lungs Reservoir and energy source Larynx Vocal folds Cavities: pharynx, nasal, oral Air exits through.
Structure of Human Speech Chris Darwin Vocal Tract.
TEMPLATE DESIGN © Listener’s variation in phoneme category boundary as a source of sound change: a case of /u/-fronting.
Stop Place Contrasts before Liquids Edward Flemming MIT.
Yao LSA Separating speaker- and listener- oriented forces in speech – Evidence from phonological neighborhood density.
On the Correlation between Energy and Pitch Accent in Read English Speech Andrew Rosenberg, Julia Hirschberg Columbia University Interspeech /14/06.
On the Correlation between Energy and Pitch Accent in Read English Speech Andrew Rosenberg Weekly Speech Lab Talk 6/27/06.
ARTICULATORY AVOIDANCE Sylvia Moosmüller Acoustics Research Institute Austrian Academy of Sciences, Vienna IAFPA 2006, Gothenburg Acoustics Research Institute.
Stop Place Contrasts before Liquids Edward Flemming MIT.
Stop Place Contrasts before Liquids Edward Flemming MIT.
-- A corpus study using logistic regression Yao 1 Vowel alternation in the pronunciation of THE in American English.
Two-Way Analysis of Variance STAT E-150 Statistical Methods.
ElectroScience Lab IGARSS 2011 Vancouver Jul 26th, 2011 Chun-Sik Chae and Joel T. Johnson ElectroScience Laboratory Department of Electrical and Computer.
Inference for regression - Simple linear regression
Present Experiment Introduction Coarticulatory Timing and Lexical Effects on Vowel Nasalization in English: an Aerodynamic Study Jason Bishop University.
Segmental factors in language proficiency: Velarization degree as a signature of pronunciation talent Henrike Baumotte and Grzegorz Dogil {henrike.baumotte,
Speech Production1 Articulation and Resonance Vocal tract as resonating body and sound source. Acoustic theory of vowel production.
Some thoughts on modelling phonetic effects in corpora.
Confidence Intervals for the Regression Slope 12.1b Target Goal: I can perform a significance test about the slope β of a population (true) regression.
Acoustic Phonetics 3/9/00. Acoustic Theory of Speech Production Modeling the vocal tract –Modeling= the construction of some replica of the actual physical.
Do Socio-Religious Characteristics Account for Later Alcohol Onset? Paul T. Korte, B.A. Jon Randolph Haber, Ph.D.
Speech Science Fall 2009 Nov 2, Outline Suprasegmental features of speech Stress Intonation Duration and Juncture Role of feedback in speech production.
VOT trumps other measures in predicting Korean children’s early mastery of tense stops Eun Jong Kong Mary E. Beckman Jan Edwards LSA2010 January 7 th.
Speech Science Fall 2009 Oct 26, Consonants Resonant Consonants They are produced in a similar way as vowels i.e., filtering the complex wave produced.
Speech Science Fall 2009 Oct 28, Outline Acoustical characteristics of Nasal Speech Sounds Stop Consonants Fricatives Affricates.
Connected speech processes Coarticulation Suprasegmentals.
♥♥♥♥ 1. Intro. 2. VTS Var.. 3. Method 4. Results 5. Concl. ♠♠ ◄◄ ►► 1/181. Intro.2. VTS Var..3. Method4. Results5. Concl ♠♠◄◄►► IIT Bombay NCC 2011 : 17.
Sh s Children with CIs produce ‘s’ with a lower spectral peak than their peers with NH, but both groups of children produce ‘sh’ similarly [1]. This effect.
Speech Science IX How is articulation organized? Version WS
The vowel detection algorithm provides an estimation of the actual number of vowel present in the waveform. It thus provides an estimate of SR(u) : François.
Objectives 2.1Scatterplots  Scatterplots  Explanatory and response variables  Interpreting scatterplots  Outliers Adapted from authors’ slides © 2012.
Electrophysiological Processing of Single Words in Toddlers and School-Age Children with Autism Spectrum Disorder Sharon Coffey-Corina 1, Denise Padden.
Stops Stops include / p, b, t, d, k, g/ (and glottal stop)
Takeshi SAITOU 1, Masataka GOTO 1, Masashi UNOKI 2 and Masato AKAGI 2 1 National Institute of Advanced Industrial Science and Technology (AIST) 2 Japan.
Lecture 10 Chapter 23. Inference for regression. Objectives (PSLS Chapter 23) Inference for regression (NHST Regression Inference Award)[B level award]
28. Multiple regression The Practice of Statistics in the Life Sciences Second Edition.
Tongue movement kinematics in speech: Task specific control of movement speed Anders Löfqvist Haskins Laboratories New Haven, CT.
4.1.4 The four groups’ average performances of / ʃ /, /t ʃ / and /d ʒ / 3176Hz English native speakers place their tips of tongues in a further back location.
Bayesian Speech Synthesis Framework Integrating Training and Synthesis Processes Kei Hashimoto, Yoshihiko Nankaku, and Keiichi Tokuda Nagoya Institute.
Speech Production “Problems” Key problems that science must address How is speech coded? How is speech coded? What is the size of the “basic units” of.
IIT Bombay 17 th National Conference on Communications, Jan. 2011, Bangalore, India Sp Pr. 1, P3 1/21 Detection of Burst Onset Landmarks in Speech.
A. R. Jayan, P. C. Pandey, EE Dept., IIT Bombay 1 Abstract Perception of speech under adverse listening conditions may be improved by processing it to.
1 Acoustic Phonetics 3/28/00. 2 Nasal Consonants Produced with nasal radiation of acoustic energy Sound energy is transmitted through the nasal cavity.
Acoustic Phonetics 3/14/00.
Corresponding author: Ruth Raymaekers, Ghent University, Department of Experimental-Clinical and Health Psychology, Research Group Developmental Disorders;
Stop/Plosives.
A STUDY ON PERCEPTUAL COMPENSATION FOR / /- FRONTING IN A MERICAN E NGLISH Reiko Kataoka February 14, 2009 BLS 35.
An Articulatory Analysis of Phonological Transfer Using Real-Time MRI Joseph Tepperman, Erik Bresch, Yoon-Chul Kim, Sungbok Lee, Louis Goldstein, and Shrikanth.
1 Probing the Big Bang with ultrasound: Retraction of /s/ in English Adam Baker, Jeff Mielke, Diana Archangeli University of Arizona Supported by James.
Elaine R. Hitchcocka, Ph.D., Laura L. Koenigb,c, Ph.D.
Understanding Variation of VOT in spontaneous speech
Month 2002 doc.: IEEE /xxxr0 Nov 2003
/r/ Place: palatal Articulatory phonetics Acoustics
A Japanese trilogy: Segment duration, articulatory kinematics, and interarticulator programming Anders Löfqvist Haskins Laboratories New Haven, CT.
Presentation transcript:

Development of coarticulatory patterns in spontaneous speech Melinda Fricke Keith Johnson University of California, Berkeley

Why study spontaneous speech? Laboratory speech is not always natural… – articulatory differences – psycholinguistic/plannin g differences (Zharkova, Hewlett, & Hardcastle, 2011)

Why study coarticulation? Coarticulation reveals speech planning. – articulatory planning, motor control – psycholinguistic planning, higher level processes (Zharkova, Hewlett, & Hardcastle, 2011)

What is coarticulation? Coarticulation: when an articulatory target affects adjacent targets. Anticipatory: [s] in “seat” vs. [s] in “suit” Perseverative: [s] in “geese” vs. [s] in “goose”

Research questions Can we use acoustic measures to detect fricative- vowel coarticulation in a corpus of spontaneous speech? If so… which ones? differences between anticipatory and perseverative coarticulation? differences between adult and child articulatory patterns?

The Corpora Buckeye Corpus of Conversational Speech (Pitt et al., 2007) – 40 adults (20 men/women), ~ 1 hour each – sociolinguistic interviews Davis Corpus, CHILDES Database (Davis et al., 2002; MacWhinney, 2000) – 21 children, ~ 1 hour/week – spontaneous interactions with caregivers

Adult data total # of tokens = 3794 front [i, ɪ, e, ɛ]round[u, ʊ, o] anticipatory 1362[si]1535 [su] perseverative 618[is]279[us] TOTAL

Child data 11 children (5 boys) produced tokens of [s] in identifiable words age range: 1;1 – 3;1 total # tokens = 3035 total # unique words = 425 – “this” (630), “yes” (179) – “juice” (139), “house” (53), “nose” (33) front [i, ɪ, e, ɛ]round[u, ʊ, o] anticipatory 615 [si] 103 [su] perseverative 1801 [is] 516 [us] TOTAL

Child data Token Contributions by Child age in months each color = 1 child

Fricative measurements adult spectra 0 – 8 kHz child spectra 0 – 11 kHz all fricatives hand labeled measurements taken at 4 locations – 40 ms Hamming window centered at 20%, 50%, 80% duration of fricative 20 ms into vowel today: high frequency centroid, amplitude ratio, kurtosis

High frequency centroid Inversely correlated with length of front cavity. low value = longer front cavity = rounding + PoA Weighted mean frequency above… – 2125 Hz (men) – 2500 Hz (women) – 3500 Hz (children) (McGowan & Nittrouer, 1988; Li, Edwards, & Beckman, 2007)

Statistical modeling Linear mixed effects regression – random effects speaker, word (for child data only) – fixed effects measurement location (20% vs. 80%) context (round vs. non-round vowel) interaction term Separate models for adults vs. children, and for perseverative vs. anticipatory coarticulation

High frequency centroid Results: Adults, anticipatory intercept3737 Hz main effect round vowel-31 Hz No effect of measurement location. Adults begin anticipating an upcoming round vowel at fricative onset. Adult High Frequency Centroids, Anticipatory measurement location 20% 80% non-round round i u

High frequency centroid Results: Adults, perseverative intercept3703 Hz main effect round vowel-45 Hz interaction72 Hz location:round Adults correct for perseverative lip rounding by the end of the fricative. Adult High Frequency Centroids, Perseverative measurement location 20% 80% non-round round i u

Amplitude ratio Related to tongue posture high ratio = palatal articulation Find peak above F2 region… mean amplitude in 1000 Hz band around high frequency peak – mean amplitude in 1000 Hz F2 region find peak here

Amplitude ratio Results: Adults, anticipatory intercept15.1 main effects location-1.1 round vowel-1.0 More palatal articulation at beginning of fricative, and in non-round context. Adult Amplitude Ratios, Anticipatory measurement location 20% 80% non-round round i u

Amplitude ratio Results: Adults, perseverative intercept13.4 no significant main effects interaction2.4 location:round Adult /s/ more palatal by the end of /us/ sequence. Adult Adult Amplitude Ratios, Perseverative measurement location 20% 80% non-round round i u

Kurtosis Correlated with lip rounding high kurtosis = more peaked distribution = more lip rounding Calculated following Forrest et al. (1988)

Kurtosis Results: Adults, anticipatory intercept3.1 main effect round vowel0.01 Adults show early anticipatory lip rounding, lasting throughout the fricative. Adult Kurtosis, Anticipatory measurement location 20% 80% non-round round i u

Kurtosis Results: Adults, perseverative intercept3.1 main effect round vowel0.05 interaction-.07 location:round Lip rounding disappears by the end of the fricative. Adult Kurtosis, Perseverative measurement location 20% 80% non-round round i u

Summary: Adult results Adults begin anticipating upcoming round vowel at fricative onset. Perseverative coarticulation lasts into the beginning of the following fricative, but is greatly reduced by the end.

High frequency centroid Results: Children, anticipatory intercept5697 Hz main effect round vowel-20 Hz Children also begin anticipating round vowel at fricative onset, but to a lesser degree. Preview: lack of effect on kurtosis may indicate this difference is due to PoA, not lip rounding. Child High Frequency Centroids, Anticipatory measurement location 20% 80% non-round round i u

High frequency centroid Results: Children, perseverative intercept5693 Hz main effect location- 10 Hz round vowel- 20 Hz Perseverative effet lasts throughout fricative. Main effect of location < utterance final position of most fricatives? Child High Frequency Centroids, Perseverative measurement location 20% 80% non-round round i u

measurement location Amplitude ratio Results: Children, anticipatory intercept3.63 no significant predictors Child Amplitude Ratios, Anticipatory non-round round % 80% i u

Amplitude ratio Results: Children, perseverative intercept3.31 main effect location- 1 Main effect of measurement location < most fricatives being utterance final? Child Amplitude Ratios, Perseverative measurement location 20% 80% non-round round i u

Kurtosis Results: Children, anticipatory intercept1.81 no significant predictors Much lower values than adults (intercept = 3.1) Suggests lack of lip rounding: coarticulation observed in centroid data may have been related to PoA. Child Kurtosis, Anticipatory measurement location 20% 80% non-round round i u

Kurtosis Results: Children, perseverative intercept1.81 main effects location-.003 Again, no significant difference due to round vowel context. Child Kurtosis, Perseverative measurement location 20% 80% non-round round i u

Child vowel spectra

Summary: Child results Evidence for only gross motor control: – overall flatter spectrum < lack of tongue groove – lack of change/compensation in perseverative data – little evidence for lip rounding: differences in centroid may have come from PoA BUT children anticipate upcoming round vowels as early as fricative onset (even though the gestures used to produce both fricative and vowel are different from adults’) Suggests children’s planning is similar to adults’, but they lack the motor control needed to produce adult-like articulation

Comparison with previous findings Zharkova et al. (2011b) concluded children don’t have differential control of tongue tip vs. dorsum by 7;7 – Our data are consistent with this conclusion – Also consistent with Nittrouer’s (1995) conclusion that different types of gestures develop along different timescales Most previous studies have not looked at the interaction between age group and direction of coarticulatory influence.

Conclusion We identified several acoustic measures that reveal fricative-vowel coarticulation in spontaneous speech. Similarities between adults and children: – planning – constraints on articulation Difference: – Motor control necessary to compensate for constraints

Thank you! (especially to Barbara Davis and Brian MacWhinney for making the child data available, and to Vanessa Chew for help segmenting fricatives)

Future work Investigating additional variables: – random effect for word, for adults – age effects, for children – individual variation – lexical predictors (word frequency, neighborhood density) – control for neighboring segments, speech rate – compare within- vs. across-word coarticulation

Adult fricative spectra

Adult vowel spectra

Child fricative spectra

Child vowel spectra

Examples of child speech Cameron, age 22 months Rebecca, age 17 months

Child data