Psycholinguistics : Speech perception Flashcards

1
Q

Speech perception ?

A

Mapping of auditory input speech to stored linguistic representations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

2 major properties ?

A

High acoustic variability + multimodal cues

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Variability ? (3)

A

Btw talkers : Regional accent, gender, age, social status

Within talkers : co-articulation, dialog Noise in speech

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Multimodal cues

A

lip, head, face, jaw mouvements

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

BOLD ?

A

Blood Oxygen Level Dependent

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Basal ganglia

A

Motor control + motivation + prediction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Cerebellum

A

Time perception - Timing

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Which network for prediction ?

A

Subcortical - cortical for temporal processing and predictions in speech perception

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Neural dissociation in processing noise and accent in spoken language comprehension.

A

Adank P, Davis MH, Hagoort P. 2012
The brain doesn’t do the same thing when dealing with noise or unfamiliar accent distortion
Ventral + Dorsal : noise
Ventral only : unfamiliar accent

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Why only ventral for unfamiliar accent processing ?

A

Only areas relative to phonological processes, coz the system tries to adapt, learn new representations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Non native phonemic contrast

A

Ability to perceive speech sounds greatly depends on the inventory of phonemes of the native language acquired during first months of life (Japanese don’t make a difference btw “r” and “l”)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Adank 2012
Méthodo

A

fMRI STUDY (BOLD)

Participants
26 monolingual dutch speakers
right-handed
homogenous sample

Stimuli
204 sent. in dutch
204 sent. in unfamilliar accent
204 sent. sent with noise

Task
Sentence true or false (button box)

Behavioral measures
From the offset of the sound file

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Adank 2012
Results

A

Behavioral results
Response time : more difficult for nosie & accent (no signf diff)
The two distortions were matched for the difficulty of processing

BOLD signal results
Increased activation in bilateral inferior frontal area (IFG and FO) for NOISE vs CLEAR
Increased activation in left temporal cortex (STG and STS) for ACCENT vs CLEAR

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Adank 2012
Interpretation

A

The dual streams model assumes that left frontal areas are more engaged inadverse speech conditions
→ Data not totally in favor with the dual stream model

External distortion (noise)
Recruitment of cortical areas involved in higher order processing (semantics, syntax) of spoken language

Speaker-related distortion
(phonetic and phonological variation)

Recruitment of cortical areas involved in auditory and phonological processing

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q
A
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Perceptual distance

A

Goslin et al 2012

ERP : home, regional, foreign accent
(time locked final word)

  • P : native monolingual English, right-handed
  • S : sentences recorded in home accent, regional accent and foreign accent
  • T : press button with their dominant handwhen the final word was the name of an animal.

Activation (200ms)
hyp : foreign>regional>home
res : regional > home > foreign

hypothese de distance perceptuelle non confirmée

Regional accent
greater responses only at early stage suggests a quick adaptation at prelexical level

Foreign accent
lower responses at early and late stages suggest a reduction of prelexical and lexical activation

17
Q

Hanulikova 2012

Brain correlates of on-line in speech perception

(text - method)

A

Hanulikova et al 2012
ERP
“Expectation is different when a non-native accent is perveived”

Participants
17 Dutch speakers

Stimuli
grammatical incorrect sentences with either gender disagreement btw the definite determiner and the noun, or inflective adjectives

Task
Answer comprehension questions

Semantic manipulation is a good control to check if the effect is related to the knowledge of the world and not a specific gramatical process (same level of comprehenson to check)

18
Q

Hanulikova - Results

A
  • Speaker adaptation can lead to be not sensitive to grammatical errors
  • These adjustments differed as a function of the accent of the speaker
  • Listeners adjust their probability model following the speaker’s characteristics during spoken language comprehension
  • Probability modal means brain predictions about language

We can totally adapt when the mistake is predictable, and focus on comprehension

19
Q

Prediction

A

Ability to generate predictions of forthcoming events based on prior knowledge ad experience about the world is a result of evolutionnary pressure for survival

KEY ROLE OF PREDICTION

Pickering & Garrod 2007
Tip of the tongue
Online lexical prediction driven by sentence context (written & spoken language)
Syntactic & semantic predictions
Prediction of the end of speaker’s turn
→ Subcortical-cortical network (involving basal ganglia-cerebellum)

Imitiation (repeat word, grammar, speech rate…)

20
Q

Lesage

A

2012 - rTMS study
Role of cerebellum in language predictions
from speech inputs

Finish the sentence by looking at a drawing on the screen

Result
more time to choose the object when cerebellum stimulated
= No prediction

→ Cerebellum involved but not alone

21
Q

Multimodal cues of speech

A

Relevance of natural head motion in speech perception

Impact of visual prosody
Munhall 2004