Perceptual abilities of infants (1-4 months)

Infants exhibit adult-like categorical perception of speech stimuli

Evidence suggests that these organized behaviors are part of the initial conditions of the infant:
• They are already organized at birth
• They do not require specific experience with speech.

Categorical Perception

Adults' abilities to discriminate members of a stimulus continuum differing in a consonant-related "cue" tend to be no better than their ability to categorize the consonants:
Within category differences are poorly discriminated.

 

One-month old infants show the same tendency

e.g., Voicing

Voicing contrast:
presence (/pa/) vs. absence (/ba/) of laryngeal spreading gesture

This difference causes systematic difference in voice onset time (VOT)

Time between closure release and onset of voicing

• /pa/ interval is long.
• /ba/ interval is short.

Possible to construct a VOT stimulus continuum

VOT = 10 ms	VOT = 55 ms

 

Boundary for English-speaking adults is 30 ms

Eimas et al (1971) present three types of stimulus changes to infants in HAS technique:

• 0 ms-20ms (within-category voiced)
• 60ms-80ms (within-category voiceless)
• 20ms-40ms (across-category)

Sucking rate increases significantly (compared to controls) when stimulus change is across adult categories, by not when a change is within adult categories.

Acoustic change is the same in both cases (20 ms).

 

Almost every consonant category tested is perceived categorically by young infants.

Experience vs. initial conditions

Early experience appears not to be a necessary condition for observing this organization to emerge.

Infants' performance does not depend on the language spoken in the infants' environment.

Thus, results cannot be due to any learning infant has accomplished by 1 month of age.

Kikuyu (Streeter, 1976)

has contrast between fully voiced and voiceless unaspirated stops (like some examples of "b,d,g" in English).

VOT of 20ms, 40ms are both in the same category for adult speakers of Kikuyu.

Infants being raised in a Kikuyu-speaking community (in Kenya) shows the same pattern of responses to these 20ms--40ms differences as do infants raised in English-speaking communities.

Within-organ constriction location contrasts

Infants can discriminate them when they are not contrastive in the adult's language, and when adults have difficulty discriminating them.

Nthlakampx

velar vs. uvular

Hindi

dental vs. retroflex

Development of sensitivity to gestural structure of the ambient language (7-12 mos.)

Decreased sensitivity to nonnative within-organ contrasts

At some point, infants keen ability to discriminate must decrease, given adults difficulties.

Experiments by Werker show that this decrease occurs during the first year of life.

Why do the behaviors of Hindi and English infants diverge at 10-12 months?

Presumably the phonologies of the languages have not yet been acquired by infants

Role of distributional information in percpetual discrimination (Maye et al, 2002)

Distribution of tokens in short (~100 item) familization session affects perceptual discriminability

This would provide a basis for the different behavior of e.g., Hindi and English-raised infants with respect to discrimination.

But is there evidence for bimodal distributions in the enviroment?

 Distribution of Tongue Tip Horizontal Position in Coronal stops (in running speech)

Hindi	English
Front <--------------------------------------------------->Back	Front <--------------------------------------------------->Back