05 speech perception

8/12/2019 05 Speech Perception

1/36

Language !Speech

! Speech is the preferred modality for language.


2/36

Outer ear

Collects sound waves. Theconfiguration of the outer

ear serves to amplifysound, particularly at

2000-5000 Hz, afrequency range that is

important for speech.

Middle ear

Transforms the energy of a

sound wave into the internalvibrations of the bone

structure of the middle ear

and transforms thesevibrations into a

compressional wave in the

inner ear.

Inner ear

Transform the energy of acompressional wave within

the inner ear fluid intonerve impulses which can

be transmitted to the brain.


3/36

http://www.sis.pitt.edu/~is1042/html/audtrans.gif

Auditory

input reaches

primary auditorycortex about

1015 msec after

stimulus onset

(Liegeois-Chauvel,

Musolino,

& Chauvel, 1991;Celesia, 1976).


4/36

http://ist-socrates.berkeley.edu/ ~jmp/LO2/6.html

Primary auditory cortex (A1): Brodmann areas 41 and 42.Auditory-association cortex (A2): Brodmann area 22.


5/36

Questions

! Is speech special?" Is there brain activity that is specifically sensitive

to the properties of speech (as opposed to those ofother sounds)?

" Can brain damage lead to a selective loss of speech

perception?

" Are the mechanisms of speech perception different

from the mechanisms of auditory perception ingeneral?

! Role of audio-visual integration


6/36

An auditory evoked neuromagnetic field

Roberts, Ferrari, Stufflebeam & Poeppel, 2000, J Clin Neurophysiol, Vol 17, No 2, 2000

M100 field pattern


7/36

M100 (N1 in ERP terms)

! Intensity

" higher intensity "shorter latency, larger amplitude

! Frequency" higher frequency "shorter latency

! Phonetic identity

" e.g., shorter latencies for athan for u(explainable inspectral terms)

! Function of M100 activity in speech processing

unclear.


8/36


(Roberts & Poeppel, Neuroreport, 1996)


9/36


(Poeppel et al., Neuroscience Letters, 1996)


10/36

Questions


to the properties of speech (as opposed to those ofother sounds)? Unclear.


perception?





11/36

Questions




perception?





12/36

Disorders of auditory processing

! Rare.

! Signal from each ear

is processed in bothhemispheres (contravisual system).

! Bilateral damage oftennecessary." Generally requires

two separate

neurological events.


13/36

! Inability to hear sounds without apparent damageto the hearing apparatus or brain stem

abnormality.! Extensive bilateral damage to auditory cortex

(BAs 41 & 42).

Disorders of auditory processing:

CORTICAL DEAFNESS


14/36

! Inability to recognise auditorily presented sounds

(e.g., coughing, crying)independent of any deficit in

processing spoken language.

! Damage in auditory association cortex (BAs 22 &

37)


AUDITORY AGNOSIA


15/36

! Impaired in tasks requiring pattern recognition in

music.

! Relative sparing of speech and (other) non-speech

perception.

! Damage generally in righttemporal areas.


AMUSIA (a subvariaty of AUDITORY AGNOSIA)


16/36

! Inability to understand spoken words while

" auditory perception is otherwise intact &

" other linguistic skills are intact (reading, speech

production)


PURE WORD DEAFNESS


17/36

Either

! Bilateral damage to auditory

cortex

Or

! A subcortical lesion in the lefthemisphere that severs bothipsilateral and contralateralprojections to Wernickes area(Ziegler,1952; Geschwind, 1965; Coslet et al.,

1984; Jones and Dinolt, 1952).


PURE WORD DEAFNESS


18/36

! Impairment in the ability to recognise familiar

voices." Speech comprehension is intact.

" Intact ability to identify nonverbal sounds.


PHONAGNOSIA


19/36

! Van Lancker et al., 1988: Double dissociation between

memory for familiar voices and the ability to discriminate

between unfamiliar voices.


PHONAGNOSIA

! Performance on a

discrimination task was

impaired by a lesion to either

temporal lobe.

! Performance on the famous

voices task was impaired by

lesions to the right parietal

lobe.


20/36

Questions




perception? YES, but no one-to-one correspondence

between lesion site and type of disorder. E.g.,

bilateral temporal damage can lead either to

cortical deafness or pure word deafness.


from the mechanisms of auditory perception in

general?



21/36

Poeppel (2001), Poeppel & Hickok (2000):

! Speech is fundamentally bilateral.

! But LH and RH have different temporal integrationwindows.

" LH processes fast phenomena, 25-50ms

! Acoustic changes in frequency that allow phoneme perception.

" RH processes slow aspects of speech, 150-200ms.

! Syllable level phenomena. Intonation. Prosody.

Bilaterality of speech processing


22/36

Questions





between lesion site and type of disorder. E.g.,

bilateral temporal damage can lead either to

cortical deafness or pure word deafness.



general?



23/36

What makes speech sound like speech?


24/36

Sine-wave speechA spectrogram, or anacoustic "picture" of aspeech utterance. Time is

represented on the

horizontal axis, frequency onthe vertical axis. Amplitudecorresponds to the darkness.

A sinewave replica of thenatural above. All fine-

grain acoustic propertiesof speech are discardedand only the coarse-grainchanges in the spectraover time are retained.


25/36

Sine-wave speech! Remez et al. (1981):

"Nave subjects fail to perceive sine-wave

stimuli as speech.

"However, if subjects are instructed aboutthe speech-like nature of sinewave speech,

they can easily understand it.

"Argument for a special speech mode of

processing.


26/36

Sine-wave speech

OriginalSWSSentence 6







27/36

Role of articulatory gestures on speechperception! Recognition of sine-wave speech or speech that

has been degraded by noise is improved if thestimulus is accompanied by congruent

articulatory gestures.


28/36

McGurk effect


29/36

Role of articulatory gestures on speechperception! Incongruent articulatory gestures can changethe

auditory percept even when the signal is clear.


30/36

Is audio-visual integration specific to speech?! Saldana and Rosenblum (1993): No.

" Audiovisual integration of the plucks and bows

of cello playing.

#Not only speech, but also other ecologically validcombinations of auditory and visual stimuli canintegrate in a complex manner.


31/36


32/36

Question! Is speech is perceived as all other sounds or is there a

specialized mechanism responsible for coding theacoustic signal into phonetic segments?


33/36

Method experiment 11. Training in non-speech mode: Subjects were first trained to

distinguish between Stimulus 1 and 2, which were the sinewavereplicas of the Finnish nonwords /omso/ and /onso/.

2. Non-speech test:SWS tokens were presented alone or audio-visually with a congruent or incongruent visual articulation.

Subjects were instructed to indicate by a button press whether theyheard stimulus 1 or 2.

3. Natural speech test:As in 2, but now the auditory stimuliconsisted of natural tokens of /onso/ and /omso/. Subjects were toldto indicate whether the consonant they heard was /n/, /m/ or

something else.4. Training in speech mode:Subjects were taught to categorize the

SWS stimuli as /omso/ and /onso/.

5. Test in speech mode:Indicate whether the consonant they heardwas /n/, /m/ or something else.


34/36

Results


35/36

Experiment 2: Speech mode training andtest occur first


36/36

Questions


to the properties of speech (as opposed to those of

other sounds)? Unclear.



between lesion site and type of disorder.



general?

! Role of audio-visual integration may be special forspeech.

05 speech perception

Documents