Csd 353 Speech Perception

Front 1

What is speech perception?

Back 1

Assigning meaning to sound

Front 2

We perceive ________ events; we hear __________ events

Back 2

We perceive LINGUISTIC events; we hear ACOUSTIC events

Front 3

What are some of the challenges in discriminating between and identifying the acoustic-phonetic features of speech sounds?

Back 3

- lack of segmentation
- lack of phonemic invariance
- lack of linearity

Front 4

What perception cues do we rely on in addition to the acoustic information?

Back 4

body language
facial expression
content-topic
familiarity w/speaker & speech patterns
movement of articulators as speech cues
knowledge of language -- what do you expect in terms of linguistic rules

Front 5

What are the acoustic cues for MANNER of articulation?

Back 5

I. Harmonics - no noise; relatively low frequency energy

A) high intensity formants:
i. formant glides
a) semivowels - faster transitions
b) diphthongs - slower transitions
ii. vowels -- relatively steady state formants

B) Low-intensity formants:
i. nasals -- relatively steady state formants; low frequency resonance

---

II. Non-periodic component; relatively high frequency energy

A) Duration of noise
i. Stops: transient burst
ii. Affricates: sharp onset of noise
iii. Fricatives: longer noise

Front 6

What are the cues for PLACE of articulation?

Back 6

I. Formant Spacing

A) Vowels
i. F1 by F2
ii. F2 high - front V; F2 low - back vowel
iii. F1 high - low V; F1 low - high V

B) Semivowels
i. F2: high - /j/; mid - /r,l/; low - /w/
ii. F3: low - /r/; flat - /l/

II. F2 Transition:

A) Stops & Nasals:
i. Low - labial
ii. High - alveolar
iii. Varied - palatal-velar

B) Fricatives:
i. Especially /f/ and theta: lowm - /f/; high - theta

C) Frequency of Noise:
i. Friction: Fricatives: High - /s/; Lower - /sh/; Wide band - /f, theta, h/
ii. Burst: Stops & Affricates: High - alveolar; Low - labial; Varied - palatal-velar

Front 7

What are the cues for VOICING distinctions?

Back 7

I. Voice bar:
A) stops - during closure
B) affricates - during closure
C) fricatives - with friction

---

II. Timing:

A) Noise duration
i. longer aspiration - voiceless stops
ii. longer friction - voiceless affricates; voicelss fricatives

B) VOT
i. stops
a) longer, voiceless;
b) shorter, vocied

C) F1
i. stops
a) rises from base - voiced
b) cutback - voiceless

D) Closure duration:
i. Stops & Affricates
a) longer - voiceless
b) shorter - voiced

E) Duration of preceding vowel
i. longer before voiced
ii. shorter before voiceless

Front 8

What are the perceptual cues for VOWELS?

Back 8

long duration
higher intensity
strong formant patterns

In running speech:
transitions to/away from vowels; changing information
- phonetic context (esp. because vowels are neutralized)

Front 9

What are the perceptual cues for DIPHTHONGS?

Back 9

gradual transition from one vowel to another -- how fast formants change is very important for perception

intense formants

Front 10

What are the general perceptual cues for CONSONANTS, to differentiate from vowels?

Back 10

shorter in duration
less intense
construction
voiced/voiceless

Front 11

In consonant perception, there is lots of ___________ to help us perceive

Back 11

REDUNDANCY -- many speech cues to help us perceive

Front 12

What is categorical perception?

Back 12

the ability to discriminate only as well as one can identify

Front 13

Categorical perception:

Sounds differing in some small way are perceived as the same ________ until a _______ is reached

Back 13

Sounds difering in some small way are perceived as the same PHONEME until a BOUNDARY is reached

i.e., there is a whole category of what /k/ would be -- lots of different phones, with different physical data, are all perceived as /k/ and cannot be discriminated

Front 14

T/F: Categorical perception exists only for place cues.

Back 14

FALSE: CP exists for place, manner, and voicing cues.

Front 15

Describe the ba/da/ga place study and discrimination task.

Back 15

The vowel /a/ was synthesized; the formants for /a/ in each presentation were kept the same except for the only thing changing was the F2 transition (which gives info about which stop-plosive the /a/ is coming from). Frequency of F2 transition varied in equal increments. When presented to listeners, distinct boundaries were found to be present. Some presentations were clearly ID'd as ba, some as da, some as ga; even though no two sounds were the same.
Evidence for categorical perception -- our brains like to put info into categories.

Discrimination task: Two-step pattern (present 1&3; 2&4; difference between 1&3 and 2&4 is equal). Presented A, B, and X (which could be either A or B). Task is to decide if X is more like A or more like B. If you can't discriminate A vs. B, X will be chance (you're guessing). If you can, your X will be close to 100%. I.e., when A and B were within the same perceptual category (both /ba/), chance of getting X correct was 50%. When A and B were from different perceptual categories (e.g. one /ba/ and one /da/), people were close to 100% at getting it right.

Front 16

What are some of the considerations for categorical perception?

Back 16

- Do people perceive speech differetly than they perceive nonspeech? (do we have CP for other sounds)

- Does learning of a language alter speech perception?

- Is CP innate or learned? (nature vs nurture)

- Auditory vs. linguistic contributions

Front 17

What did Within-language and Cross-language Studies of CP in Adults find?

Back 17

- categorical perception for place: b,d; r,l
- CP for manner (transition duration: b/g to w/j/ to ua/ia) and voicing (VOT from -150 to +150)
- perception is more categorical for consonants than vowels (dynamics/speed of consonant is perceived more readily than for vowels)
- Influence of linguistic knowledge (learning a 2nd language; age; shifting boundaries depend on expectation of what lg to be heard - e.g. Puerto Rican)

Front 18

T/F: There are multiple acoustic cues in consonant perception.

Back 18

TRUE. Example of VOT in voicing contrasts for plosives.

Front 19

What are the 7 acoustical parameters that are important cues for IDing stop consonants in VCV environments:

Back 19

1) Rate, degree, extent of vocalic transition from V to C

2) Duration of closed phase (stop phase)

3) Presence or absence of voicing during closed phase

4) Voice onset time VOT

5) Duration of noise burst

6) Spectral characteristics of noise burst

7) Rate, degree and extent of vocalic transition from C to V


---- all of this is evidence that there are multiple acoustic cues in consonant perception

Front 20

What does it mean when we say that "trading relations exist among perceptual cues"?

Back 20

Many cues contribute to our perception -- but in certain conditions, certain cues will be more salient. There are trade-offs in terms of which cues will prevail.

Front 21

Coarticulation/Rate:

How do coartic/rate affect speech perception?

Back 21

Perceptually we integrate information that is spread out over several phonetic segments to make a decision about phoneme identity

Front 22

Coartic/rate:

What is an example of how context influences perception?

Back 22

"sh" vs "s" -- (when fricative noise in midrange of bandwidth; slushy "sl") -- percept varies according to vowel context. Will perceive as /s/ if preceding a back vowel, b/c relatively big difference. Will perceive as /sh/ if preceding a front vowel b/c relatively smaller difference

Front 23

Coartic/rate:

What is an example of how rate influences consonant ID?

Back 23

an intermediate VOT will be perceived as voiceless in fast speech; voiced in slow speech

Front 24

What are the acoustic cues for liquids?

Back 24

formant transitions; F3 lower for /r/ than for /l/

Front 25

What are the acoustic cues for glides?

Back 25

shorter transition duration (40-60ms to </110-150 ms) than for diphthongs (>100ms)

Front 26

What are the acoustic cues for nasals?

Back 26

internal formant structure; vowel formant transitions.
Manner cue -- weak formants; nasal formant;
Place cue -- F2 transition frequency and duration

Front 27

STOPS -- (cues are intertwined with cues for surrounding vowels & consonants)

What are the manner cues for stops?

Back 27

preceding silence or sound attenuation; transient noise burst; rapid formant transitions

Front 28

STOPS --

What are the place cues for stops?

Back 28

noise burst frequency; formant transition to vowels; VOT (shortest for bilabials; longest for velars)

Front 29

STOPS --

What are the voicing cues for stops?

Back 29

duration of silent interval; initial position stops -- VOT, f1 cutback, F0 of vowel (higher for voiceless stops); final position stops -- vowel duration (longer before voiced stops)

Front 30

FRICS --

What are the manner cues for fricatives?

Back 30

friction noise; longer in duration than for stops (130ms or greater)

Front 31

FRICS --

What are the place cues for fricatives?

Back 31

sibilants more intense than non-sibilants and high-frequency spectral peaks; nonsibilants flat spectra
/s,z/ prominent peak @ 4000 Hz;
sh, zh @ 2500 Hz;

F2 transitions

Front 32

FRICS --

What are the voicing cues for fricatives?

Back 32

friction noise (longer duration and more intense for voiceless fricatives)

Front 33

What are the three roles of context in speech production?

Back 33

ACOUSTIC context -- info regarding several phonemes converge on a specific sound

LINGUISTIC context -- language specific sound sequences (e.g. ng and nd at end of syllable; never at beginning); other language based expectations re: syntax and semantics (we are biased by what we expect to hear)

TOPIC/situational context -- familiarity with subject matter

Front 34

bring

Back 34

trazer