Speech perception

Question 1

Pena et al. (2003) brain region while neonates listened to forward/backward speech and silence

Answer 1

Left hemisphere specialisation. Neonate brain responds specifically to forward speech shortly after birth.

Question 2

DeCasper and Fifer (1980) sucking rate mother’s and father’s voices

Answer 2

Adjusted sucking rate to listen to mother’s voice rather than that of another female. No preference for father’s voice.

Prenatal learning. Discriminate familiar vs. unfamiliar voice.

Question 3

DeCasper and Spence (1986) listening to a poem during pregancy

Answer 3

Neonates prefer to listen to this poem over a new one read by mother. Discriminate old vs. new poem, even though both are read by mother.

Question 4

2 potential roles for language input

Answer 4

1. To trigger innate knowledge and populate it with language-specific info.
2. To construct the language system.

Question 5

5 features of infant-directed speech (IDS)

Answer 5

1. Higher in pitch.
2. More exaggerated intonation contours.
3. Shorter utterances and longer pauses.
4. Simplified sentence structure.
5. Vowel-space expansion.

Question 6

Fernald (1985) 4-5-month-olds listening preference for IDS

Answer 6

Prefere IDS over ADS.

Question 7

IDS preferences in other languages

Answer 7

Even newborns show this preference. IDS preferred even if it is German/Italian being spoken/low-pass filtered speech to an English infant. Not content, but intonation or tone that is being attended to.

Question 8

3 reasons for IDS

Answer 8

1. Perceptual-attention factors (increases salience and helps attract infant’s attention/heighten arousal).
2. Linguistic effects (helps perceptual discrimination/segment words/map meaning).
3. Social connectivity.

Question 9

2 problems in speech perception

Answer 9

1. Same words sound very different when produced by different speakers/same speaker in different contexts.
2. Functional equivalence (different acoustic signals treated as identical in the brain).

How to learn when to differentiate and when to ignore differences between different sounds?

Question 10

3 possible solutions to speech perception

Answer 10

1. Motor theory.
2. Universal theory.
3. Attunement theory.

Question 11

Motor theory

Answer 11

Nativist view. Innate motor speech system that produces and detects speech sounds. Infants are born sensitive to all possible sounds in all possible languages, and then just need to figure out what’s relevant for own language. Categorical perception of all possible speech sounds as an innate capacity (we don’t have categorical perception of non-speech sounds).

Question 12

Categorical perception

Answer 12

Discriminate well between categories but poorly within categories.

(eg. [p] vs. [b] VOT but we do not differentiate different sounds if they are on the same side of a category boundary).

Question 13

Eimas et al. (1971) high-amplitude sucking habituation procedure for within-category vs. cross-category switch

Answer 13

1-4 months. Bin size of change in VOT always 20msec. By 1-month, infants show increase in sucking for cross-category switch but not within-category switch.

Categorical perception of phonemes

Question 14

Does categorical perception evidence support motor theory?

Answer 14

No. Other species (macaques, quail) show categorical perception + categorical perception for non-speech sounds. So categorical perception not speech-specific or species-specific, as argued by motor theory.

Question 15

Universal theory

Answer 15

Innate but not language-specific. Universal set of contrasts, categorical in nature, and then learning is about becoming tuned to those that are relevant to native language.

Question 16

Werker and Tees (1984) English infants perceiving non-native speech contrasts

Answer 16

6-12 months. Hindi (/d/ - /t/ continuum) and Salish (/g/ - /k/ continuum). Hard for adult native English speakers to distinguish.

6-8 months: perceive a distinction for both.
10-12 months: stop perceiving a distinction for both contrasts.

Sensitive period for discriminating non-native contrasts.

Question 17

2 critiques of universal theory

Answer 17

1. Contrasts differ in difficulty. Don’t fully lose ability to discriminate all non-native contrasts.
2. Some contrasts aren’t perceived well early but develop later if they are relevant to the language (Narayan et al., 2010).

Question 18

Attunement theory

Answer 18

Infants begin life with ability to partition sounds into categories. Experience with language tunes basic perceptual abilities (not maintenance/loss, but rather ‘perceptual reorganisation’ in which innate perceptual abilities are tuned, modified, and re-organised by exposure to language).

Question 19

Kuhl et al. (1992) prototypes and magnets

Answer 19

Infant brain generalises across examples and represents prototypes. Prototypical sounds function as perceptual magnets (sounds that are perceptually similar to prototype are grouped with it, so different variants of a phoneme start to sound the same). Sounds that are experienced as similar varies across languages.

Question 20

Maye et al. (2002)

Answer 20

Infants heard exemplars along ta-da continuum (either unimodal or bimodal). Bimodal group listened longer to non-alternating test samples. Unimodal group listened longer to alternating test samples.

Learnt different patterns of categorical perception, depending on distributions of input. Use distributional info in input to detect phonetic category structure.

Question 21

2 challenges for attunement theory

Answer 21

1. Not all infants show the effect of categorical perception within a particular experiment.
2. Language isn’t just about 2 contrasts presented repeatedly. Normal speech contains much noise and variability.

Question 22

Kuhl et al. (2003) American infants live vs. televised interaction with Chinese speakers

Answer 22

9 months. Worse Chinese phonetic discrimination for television/audio-only exposure.

Social interaction influences language learning. Attention/joint attention?