An important point…• When discussing source-filter theory, the sound

source was the glottal spectrum• When discussing stops (and fricatives and

affricates), we introduce a new sound source, noise produced within the oral cavity

• However, source-filter theory still holds even though the sound source is different…the vocal tract still filters the sound source, whether it is the complex periodic signal from vocal fold vibration, or a transient aperiodic signal produced during a stop release

Unit 4The Articulatory System II

I. The DiphthongsII. The GlidesIII. The LiquidsIV. The StopsV. The FricativesVI. The AffricatesVII. The Nasals

Fricatives

• Place– Labiodental /f/ /v/– Interdental // //– Alveolar /s/ /z/– Palatal // //– Glottal /h/

• Voicing– Voiced /v/ // /z/ // – Voiceless /f/ // /s/ // /h/

Fricatives

• Manner of production– severe vocal tract constriction– Air pressure behind constriction builds up– Air flow through the constricted path is very

high– At a critical point, the airflow becomes

turbulent– Turbulent flow is heard as noise – frication

Fricatives

• Aerodynamics• Airflow for vowels is laminar – molecules are

moving along in an orderly fashion (like the flow of water in a river or cars on the freeway)

• Airflow for fricatives is turbulent – molecules are moving is a disorderly way – (like the eddies of water when a large rock impedes the river’s flow)

Fricatives

• The physics of turbulence• For a given constriction/obstruction, there

is a critical flow velocity above which turbulence occurs -Reynolds number

Fricatives

Equation for turbulence

Re= V*h/– Re: Reynolds number– V: flow velocity : kinematic coefficient of viscosity (.15 cm/sec for air)– h: characteristic dimension (size of constriction)– Critical Re for speech ~1800

Fricatives

• In theory, the spectral characteristics of “white” noise, which has all frequencies in equal amplitude

• sound source characteristics should be the same regardless of place of articulation

Fricative source spectrum

Frequency

Am

plitu

de

Fricatives

• Question…• How do we distinguish different fricatives if

the sound source is the same?

Fricatives

• Answer…• The transfer function of the vocal tract will

shape the otherwise flat spectrum

Fricative: Vocal tract features

Fricative: Vocal tract features

• Vocal tract has – a back cavity (behind the constriction)– a front cavity (in front of the constriction)– Front cavity plays a more important role in

shaping the fricative spectrum– Longer the front cavity the lower the resonant

frequencies

Fricatives:

• Labiodental/interdental• very short front cavities = very high

resonant frequencies• Practically, there is little effect on shaping

the noise energy• Low energy diffuse spectrum

Labiodental /f/

Fricatives:

• Alveolar• front cavity length ~ 2.5 cm• F1=34000/4*2.5 = 3400 Hz• Intense energy at/above 3400 Hz

Alveolar /s/

Fricatives:

• Palatal• front cavity length longer than for /s/• Intense energy around 2000 Hz• Lip rounding increases front cavity length

and helps to reduce the frequency of the prominent energy

Palatal //

// vs. /s/

// /s/

Fricatives:

• Glottal• Spectrum shaped by whole vocal tract• Low energy diffuse noise with apparent

vowel-like formant values

Glottal /h/

Voiced/voiceless distinction

• Voiced fricative have two simultaneous sound sources

• Glottal sound source (voicing)• Frication (noise)• Both sound sources are shaped by the

vocal tract shape• Voiced fricatives will have low frequency

energy in the spectrograph (voice bar)

/z/ /s/

/z/

/s/

/s/ /z/ // // - The stridents

• These fricatives have much greater energy when compared to others

• Teeth serve as an obstacle to the airflow, which increases the turbulence and amplitude of the noise energy

Transitions

• Formant transitions also play a role in fricative identity

• More prominent cue for “weak” fricatives such as /f/ and // since energy for these is typically low and diffuse

Unit 4The Articulatory System II

I. The DiphthongsII. The GlidesIII. The LiquidsIV. The StopsV. The FricativesVI. The AffricatesVII. The Nasals

Affricates

• Place: – palatal (/t/, /d/)

• Voicing:– Voiceless (/t/)– Voiced (/d/)

Affricates

• Manner of production– Features of both stop and fricative– Vocal tract occlusion– Release from occlusion into a severe

constriction– Spectral features of //– “Rise-time” of burst differs for stop and

affricates

Affricate

Silent gap frication

/t/

Rise-time: stops vs. affricates

/t/

/t/

Nasal

• Place– Bilabial /m/– Alveolar /n/– Velar //

• Manner of production– Velopharyngeal port is open– Oral cavity is closed– Sound source: glottal spectrum

Nasals

• Distinct vocal tract configuration

Pharyngeal cavity

Oral cavity (closed)

Nasal cavity (open)

Nasal

• Acoustically, nasals are characterized by – Antiformants– Nasal formant

Nasal

• Closed oral cavity produces antiformants in the transfer function

• Antiformants are regions where energy is damped

• Location of antiformants is related to place of articulation

• As place of articulation moves back, the frequency of the anti-formant increases

Nasals

• /m/:antiformants 750-1200 Hz• /n/: antiformants 1450-2200 Hz• //: antiformants > 3000 Hz

Nasals

• Nasal formant• Strong low frequency band 250-500 Hz• Most prominent acoustic feature of nasals

Nasals

• Have formant transitions similar to oral stops

Initial position• Bilabial-rising F1 and F2• Alveolar-rising F1 and dropping F2• Velar-F2 and F3 “C” shaped

Nasals

bilabial alveolar velar

XI. THE NASALS

A. Define an antiformant and how their values change with place of articulation.

B. Draw the vocal tract configuration for a nasal.

C. What is a nasal formant?

Outline: ArticulationI. THE VOCAL TRACTII. SOURCE FILTER THEORY OF SPEECH

PRODUCTIONIII.CAPTURING SPEECH DYNAMICSIV. THE VOWELSV. THE DIPHTHONGSVI. THE GLIDESVII. THE LIQUIDSVIII. THE STOPSIX. THE FRICATIVESX. THE AFFRICATESXI. THE NASALSXII. PUTTING IT ALL TOGETHER: STUDYING

CONNECTED SPEECH PROCESSES

Name those acoustic events!

XII. PUTTING IT ALL TOGETHER: STUDYING CONNECTED SPEECH

PROCESSESA.Identify how coarticulatory processes may

be revealed in speech-related signals.B.Distinguish between the phonetic properties of

speech and suprasegmental features of speech.C.Identify and describe suprasegmental properties

of speech.D.Identify some key “problems” features that

speech production theories must address. E.Describe how speech disorders may be revealed

in articulatory processes.

What is coarticulation?

What is coarticulation?

• “An event in speech production in which adjustments of the speech production system are made simultaneously for two or more speech sounds” (Kent)

What is coarticulation?

• In other words, the features of speech elements will vary depending upon the context in which they are produced

Terms used that refer to this general concept

• Coarticulation• Coproduction• Contextual variation

Kinds of coarticulation

• A speech event can be influenced by a previous event

OR

• A speech event can be influenced by an upcoming event

Coarticulation

• Anticipatory (right-to-left) coarticulation– A segment’s features are influenced by

upcoming segment

S1 S2

Coarticulation

• Carryover (left-to-right) coarticulation– A segment’s features are influenced by a

previous segment

S1 S2

Examples of anticipatory coarticulation

• Lip protrusion has been observed three (or more) consonants in advance of a rounded vowel

• e.g. /stu/ (“stew”) will exhibit lip rounding through the /s/ and /t/

An (in)famous coarticulation study

Anticipatory coarticulation: Some implications from study of lip rounding

byFredericka Bell-Berti & Katherine Harris

Haskins LaboratoryNew Haven CT

Published in JASA Vol 65(3) 1979

Examples of anticipatory coarticulation

• Velopharyngeal opening can occur two vowels in advance of a nasal consonant

• e.g. /an/ will exhibit V-P opening during the /a/

Examples of anticipatory coarticulation

• Jaw opening for an open vowel may be observed two consonants in advance of the vowel

Examples of carryover coarticulation

• Velopharyngeal opening can continue into a vowel following a nasal consonant

• e.g. /nat/ will exhibit V-P opening during the /a/

Why is there coarticulation?

• Articulators cannot make quantum leaps from one static position to another

Carryover coarticulation

A possible reason?• Articulator are ‘sluggish’ and it takes time

to move on to the next sound

Anticipatory coarticulation

A possible reason?• Articulator are ‘sluggish’ and it takes time

to move on to the next sound

Suprasegmentals

• Intonation• Stress• Duration

Intonation

• Manner in which Fo is varied to mark linguistic aspects of speech

• Fo/pitch contour• Rise-fall pattern typical for declarative

sentences• Start-frequency - variable• End-frequency - stable

Sound pressure waveform

Fo contour

start Foend Fo

Intonation

• End-frequency may be related to the physiology of phonation

• Completing a speech breath – Psg is lower ~ lower Fo

But, this doesn’t have to be the case…

Intonation

• Questions are marked by a rising Fo contour

• Can override patterns for linguistic purposes

Stress

• Not the kind of stress you get around exam time

• Stress is applied to parts of speech• For example,

– Lexical stress– Emphatic stress

Stress

• Stress typically marked by– Higher Fo– Higher intensity– Longer duration– Vowels will be more clearly articulated than

unstressed– Perception of stress will result from some

combination of these acoustic features

Duration

• The length of speech sounds• Why are sounds as long as they are?

– Physical requirements of their production– Phonetic distinction (i.e. vowel length)– Context in which they are produced– Overall rate of speech