Found. of SLD Exam 2: Lectures 9-12 (Social Learning, Word Learning, IDS)

Lecture 9 big question: how do children learn language?

(L9: Is language learning socially mediated?)

• To what extent is language learning mediated by social factors?

• The environment plays a critical role in language development

Kuhl et al. (2003): what was the starting observation?

(L9: Is language learning socially mediated?)

• Infants are born able to discriminate phonetic contrasts from many languages

• By 10-12 months, this declines for non-native sounds and improves for native sounds

Kuhl et al. (2003), study 1: research question

(L9: Is language learning socially mediated?)

Can short-term live exposure to a foreign language preserve sensitivity to non-native phonetic contrasts in infants?

Kuhl et al. (2003), study 1: hypothesis vs. alternative hypothesis

(L9: Is language learning socially mediated?)

• Hypothesis: exposure to a new language (Mandarin) during this period lets infants retain or regain sensitivity to its contrasts

• Alternative: short-term exposure in the lab won't be enough; infants need 10-12 months of listening

Kuhl et al. (2003), study 1: participants and groups

(L9: Is language learning socially mediated?)

• 9-month-old English-learning infants

• Live Mandarin group: 12 sessions of live interaction with native Mandarin speakers

• Control group: came to the lab and heard English (no Mandarin)

Kuhl et al. (2003): what was the intervention dose?

(L9: Is language learning socially mediated?)

• 12 sessions, 25 minutes each

• 4 different talkers

• About 33,000 syllables

Kuhl et al. (2003): how were infants tested after exposure?

(L9: Is language learning socially mediated?)

• Head-turn preference procedure

• Detecting a Mandarin contrast (/tɕʰi/ vs. /tɕi/)

Kuhl et al. (2003), study 1: results

(L9: Is language learning socially mediated?)

• Live Mandarin group discriminated the Mandarin sounds; control group did not

• Mandarin group was statistically comparable to infants who had been listening to Mandarin for 10-12 months

Kuhl et al. (2003), study 1: conclusion

(L9: Is language learning socially mediated?)

• There is a critical or 'sensitive period' for learning language sounds

• Babies absorb information like a sponge at a particular point in time

• Short-term exposure is enough to induce phonetic learning at 9 months

Kuhl et al. (2003), study 2: research question

(L9: Is language learning socially mediated?)

Is the presence of a live person critical to phonetic learning in infancy?

Kuhl et al. (2003), study 2: hypothesis vs. alternative

(L9: Is language learning socially mediated?)

• Hypothesis: if learning is triggered by linguistic input alone, DVD exposure (audiovisual or audio only) should produce learning

• Alternative: if learning is enhanced by social interaction, exposure without a human may not be enough

Kuhl et al. (2003), study 2: groups

(L9: Is language learning socially mediated?)

• Live exposure group

• TV exposure group (audiovisual or audio only)

Kuhl et al. (2003), study 2: results

(L9: Is language learning socially mediated?)

• Live exposure group discriminated Mandarin sounds

• TV/audio groups showed no learning

• Babies can't learn language sounds from screens; language is socially mediated

Kuhl et al. (2003), study 2: conclusions

(L9: Is language learning socially mediated?)

• Children learn language from humans significantly better than from machines

• Infants can learn from simultaneous exposure to two languages (implications for bilingualism)

Kuhl et al. (2003): why does it matter for products?

(L9: Is language learning socially mediated?)

Commercial products (DVDs, flashcards, apps) that claim to teach infants foreign languages lack solid empirical support

Kuhl et al. (2003): clinical science implication

(L9: Is language learning socially mediated?)

Children with social deficits (e.g., ASD) also tend to show language deficits

Lytle et al. (2017): research question and hypothesis

(L9: Is language learning socially mediated?)

• What could improve learning from a disembodied source (a video screen)?

• Hypothesis: phonetic learning from a screen is enhanced when a peer is present vs. learning alone

Lytle et al. (2017): alternative (not mutually exclusive) hypothesis

(L9: Is language learning socially mediated?)

Learning from a screen is enhanced if infants control the video clips themselves with a touch screen

Define mismatch negativity (MMN)

(L9: Is language learning socially mediated?)

• An event-related potential (ERP) brain response recorded from the scalp

• Shows the brain detecting a change in a repeated sound (ba ba ba pa ba)

• Measures speech sound discrimination

MMN: what does a more negative response mean?

(L9: Is language learning socially mediated?)

Greater neural discrimination of the contrast

Lytle et al. (2017): results and conclusion

(L9: Is language learning socially mediated?)

• Social partners enhance learning, even from screens

• Infants paired with a partner showed more negative MMN (better discrimination); more partners, better learning

• Social interaction matters especially for learning from screen media

Conboy et al. (2015): research question

(L9: Is language learning socially mediated?)

Does infants' social engagement during foreign-language exposure predict how much they learn (ERP non-native discrimination)?

Conboy et al. (2015): how was social engagement measured?

(L9: Is language learning socially mediated?)

Joint attention: video analysis of infants' eye-gaze after the tutor introduced new toys

Conboy et al. (2015): results

(L9: Is language learning socially mediated?)

• Infants who shifted gaze between the tutor's eyes and the new toys showed more negative MMN (better discrimination)

• Infants who just stared at the tutor or the toy had fewer gaze shifts and less negative MMN

Why is learning better in person than on a screen?

(L9: Is language learning socially mediated?)

• More information is available in person

• Screens aren't interactive, so there is less joint attention

Kuhl (2024) zebra finches: what do they show?

(L9: Is language learning socially mediated?)

• Zebra finches must learn to produce their species' vocalizations in the presence of another bird

• Vocal learning is socially mediated in animals too

Meltzoff et al. (2009): research question

(L9: Is language learning socially mediated?)

Can young children learn language from interactive social robots?

Goldstein et al. (2003): research question

(L9: Is language learning socially mediated?)

• Are infants' early speech-like vocalizations shaped by social feedback?

• Do caregivers' reactions to infant sounds affect vocal development?

Goldstein et al. (2003): hypothesis

(L9: Is language learning socially mediated?)

If infant vocal behavior is influenced by social partners, vocal frequency and/or quality should differ when maternal behavior is contingent vs. non-contingent

Goldstein et al. (2003): design

(L9: Is language learning socially mediated?)

• Contingent condition: mother responded right after the infant vocalized (moved closer, smiled, touched)

• Yoked control: mother's responses were timed to another infant, not her own baby's sounds

• Phases: baseline, social response, extinction

Goldstein et al. (2003): results

(L9: Is language learning socially mediated?)

• Contingent, but not non-contingent, maternal behavior led to more complex, mature vocal behavior

• Infants vocalized more and produced more canonical babbling

• The social feedback was non-auditory

Define contingent feedback

(L9: Is language learning socially mediated?)

A response that comes right after, and depends on, the infant's behavior

Goldstein et al. (2003): clinical questions it raises

(L9: Is language learning socially mediated?)

• Word gap and SES factors

• What counts as high-quality exposure

• How SLPs can coach parents (especially of preemies, chronic OME, ASD)

Kuhl (2003) vs. Goldstein (2003): receptive vs. expressive

(L9: Is language learning socially mediated?)

• Kuhl: receptive; infants learn a new language's sounds from live social interaction

• Goldstein: expressive; contingent social feedback makes infants vocalize more and babble more maturely

Language learning is social, imitation: key ideas

(L9: Is language learning socially mediated?)

• Children don't learn language indiscriminately; social cues highlight what and when to learn

• Infants are predisposed to attend to people and motivated to imitate

Meltzoff & Moore (1977): what did newborns do?

(L9: Is language learning socially mediated?)

As young as 42 minutes old, they matched gestures like tongue protrusion and mouth opening

Meltzoff & Moore (1977): why is newborn imitation remarkable?

(L9: Is language learning socially mediated?)

• Infants can't see their own faces and have never seen a mirror

• Suggests shared representations for the acts of self and others

Imada et al. (2006): infant MEG finding

(L9: Is language learning socially mediated?)

• When passively listening to speech, infants activate the same brain areas that will later produce speech

• Social and language learning use brain systems linking self and other; imitation requires sensorimotor integration

Kuhl & Meltzoff (1996): what did infants imitate?

(L9: Is language learning socially mediated?)

The acoustic properties of vowels they saw and heard (non-interactive presentation)

What aspect of infant learning can't be explained by imitation?

(L9: Is language learning socially mediated?)

Goldstein's result: non-auditory feedback enhanced infant vocal behavior

Shared attention: earliest component

(L9: Is language learning socially mediated?)

Gaze following: in the first half-year, infants look in the direction of an adult's head turn

Shared attention: what changes by 12 months?

(L9: Is language learning socially mediated?)

• Infants are sensitive to the direction and state of the eyes, not just the head turn

• They follow the gaze to a target only if the adult's eyes are open

Yu & Smith (2008): what tool studied shared attention?

(L9: Is language learning socially mediated?)

Head-mounted eye trackers on the child and the parent

Define the adult 'social ensemble'

(L9: Is language learning socially mediated?)

• Infant-directed speech

• Contingent responses to infants' actions and vocalizations

• Positive reactions through mutual eye gaze and smiling

Bosseler et al. (2024): hypothesis

(L9: Is language learning socially mediated?)

Brain activation in attention and sensorimotor areas in response to the social ensemble predicts language skills more than 2 years later

Bosseler et al. (2024): social vs. nonsocial conditions

(L9: Is language learning socially mediated?)

• Social: caregiver looks at and talks with the baby (live face-to-face)

• Nonsocial: caregiver looks at a phone

Bosseler et al. (2024): brain results

(L9: Is language learning socially mediated?)

Live, responsive face-to-face interaction significantly increased activity in attention brain regions vs. the nonsocial control

Bosseler et al. (2024): longitudinal results

(L9: Is language learning socially mediated?)

• Neural activation in attention and sensorimotor areas during the social condition (not nonsocial) positively predicted language skills at 18, 21, 24, 27 and 30 months

• More brain activity, higher vocabulary scores

Lecture 9 key point: social interaction

(L9: Is language learning socially mediated?)

• Infants learn language best through live, socially contingent interaction (back-and-forth with caregivers)

• Passive exposure (TV, audio) is not sufficient, especially for new or foreign languages

Lecture 9 key point: joint attention

(L9: Is language learning socially mediated?)

• Caregiver and child focusing on the same object or event helps the child learn speech and language

• A foundational social-cognitive skill

Lecture 9 key point: social feedback

(L9: Is language learning socially mediated?)

• Contingent, not non-contingent, feedback facilitates more complex infant vocal behavior

• Feedback does not have to be auditory

Linking language and cognition: what happens early in life?

(L10: How do infants link language and cognition?)

• Children tune into the signals of language, laying the foundation for speech motor learning

• Shaped significantly by social interaction and engagement

Define perceptual narrowing

(L10: How do infants link language and cognition?)

Tuning in to the native language's sounds (and faces and voices) while losing sensitivity to non-native ones

Acquiring language involves more than attuning to signals. What else?

(L10: How do infants link language and cognition?)

• The signals must be linked to meaning

• Core of word learning: forming concepts and linking them to the right linguistic signals

What does the precision of the cognition-language link depend on?

(L10: How do infants link language and cognition?)

• How precisely children perceive speech signals

• How abstractly they represent objects in the world

Ferry et al. (2013): evidence that prelinguistic infants think

(L10: How do infants link language and cognition?)

• Infants look more at novel things

• They perceive and think in visual object categories even without a label for the category

Izard et al. (2009): evidence that prelinguistic infants think

(L10: How do infants link language and cognition?)

• Infants look longer when what they see matches what they hear

• They look at displays whose number of shapes matches the number of syllables

How do infants establish the cognition-language link? (2 tasks)

(L10: How do infants link language and cognition?)

• Cull words and phrases from the melody of language (the 'segmentation problem')

• Learn word meanings and how words combine, linking them to core cognitive capacities (objects, relations, number)

Define the segmentation problem

(L10: How do infants link language and cognition?)

Pulling individual words and phrases out of the continuous melody of speech

Hummingbird/bee example: what problem does it show?

(L10: How do infants link language and cognition?)

When an adult names something, the child must figure out the scope: the whole object, a surface quality (yellow), or an action (flying)

Medina et al. (2011): point about learning environments

(L10: How do infants link language and cognition?)

Real-world learning environments are complex (cluttered)

Bergelson & Swingley (2012): main finding

(L10: How do infants link language and cognition?)

• Infants know the meaning of common nouns at 6-9 months

• They look longer at the pictured object that matches the word (e.g., 'Where is the bottle?')

Bergelson & Swingley (2012): why it matters

(L10: How do infants link language and cognition?)

• The words weren't trained in the lab, so infants learn ordinary words through daily experience

• By 12 months infants robustly looked at the named object even in never-before-seen pictures

Fulkerson & Waxman (2007), 12 mo: familiarization and test

(L10: How do infants link language and cognition?)

• Familiarized with different exemplars of one category (dinosaurs differing in color and shape)

• Test: a novel member of the category vs. a novel object outside it (fish)

Fulkerson & Waxman (2007): logic of the novelty test

(L10: How do infants link language and cognition?)

• Infants prefer novelty

• If they formed the category, they should look longer at the out-of-category object

Fulkerson & Waxman (2007), 12 mo: research question 1

(L10: How do infants link language and cognition?)

Does naming the objects help infants form the object category?

Fulkerson & Waxman (2007), 12 mo: no word vs. novel word results

(L10: How do infants link language and cognition?)

• No word ('Look here!'): no consistent preference

• Novel word ('Look at the toma!'): looked longer at the novel object; formed the category

Fulkerson & Waxman (2007), 12 mo: conclusion 1

(L10: How do infants link language and cognition?)

Words have a powerful effect on visual object categorization

Roger Brown quote about words

(L10: How do infants link language and cognition?)

'Words are invitations to form object categories'

Fulkerson & Waxman (2007): variable word condition

(L10: How do infants link language and cognition?)

• Each object got a different name (toma, dax, blick, modi)

• Tests whether the label must be applied consistently

Fulkerson & Waxman (2007): conclusion about variable words

(L10: How do infants link language and cognition?)

• Infants expect caregivers to name objects consistently to highlight category relations

• Consistent labeling induces category formation

Fulkerson & Waxman (2007): non-speech tones condition and conclusion

(L10: How do infants link language and cognition?)

• Consistent tones paired with objects did not produce categorization

• The effect is specific to language; other consistent signals can't induce it

By 12 months, infants' language-cognition link is (3 properties)

(L10: How do infants link language and cognition?)

• Sufficiently constrained to pick out linguistic signals, not tone sequences

• Sufficiently powerful to promote categorization

• Sufficiently early to guide word learning

Fulkerson & Waxman (2007), 3-6 months: results

(L10: How do infants link language and cognition?)

• Younger infants never formed the category with non-speech tones

• They always formed it with speech

Fulkerson & Waxman (2007), 3-6 months: conclusion

(L10: How do infants link language and cognition?)

By as early as 3 months, listening to speech boosts cognition

Why might speech boost cognition? Hypotheses considered

(L10: How do infants link language and cognition?)

• Arousal

• Familiarity (more familiar with speech than tones)

• Complexity

Newborns to 3 months: listening preferences

(L10: How do infants link language and cognition?)

Same preference for human speech and non-human primate vocalizations

Ferry et al. (2013): conditions tested at 3-6 months

(L10: How do infants link language and cognition?)

• Lemur (non-human primate) vocalizations

• Backwards speech

Ferry et al. (2013): results at 3-6 months

(L10: How do infants link language and cognition?)

• 3-4 months: lemur calls also supported categorization; backwards speech did not

• By 6 months: only human speech worked

Infants' cognition-perception link: summary of its development

(L10: How do infants link language and cognition?)

• Precocious link by 3 months

• Starts from a broad template including human and non-human primate vocalizations

• Tunes specifically to human vocalizations by 6 months

• Adaptive: tuned to the signals that will form word meaning

Vouloumanos et al. (2010): newborns

(L10: How do infants link language and cognition?)

Newborns prefer speech over non-speech, but experience fine-tunes this preference

Ferry et al. (2013), 6 months with lemur exposure: results

(L10: How do infants link language and cognition?)

With systematic home exposure to lemur calls, 6-month-olds succeeded at categorization with lemur vocalizations

Ferry et al. (2013): conclusion about exposure

(L10: How do infants link language and cognition?)

• Mere exposure can sustain the effect for lemur vocalizations

• But it can't create the link from nothing (de novo): exposure to backwards speech didn't work

Vouloumanos et al. (2009): what can infants infer?

(L10: How do infants link language and cognition?)

The sources of human speech and rhesus monkey calls (they match the vocalization to the right face)

Ferguson & Waxman (2016): research question

(L10: How do infants link language and cognition?)

Can 6-month-olds infer the communicative function of arbitrary non-speech tones and link them to meaning?

Ferguson & Waxman (2016): result and conclusion

(L10: How do infants link language and cognition?)

• Tones embedded in a rich social, communicative exchange let 6-month-olds use them for categorization

• A non-speech signal works if it appears to have communicative intent

• More evidence that social-communicative cues enhance language learning

How many words does a typical child know by age 2?

(L11: How do children learn object names in a cluttered real world?)

• About 1,000 words

• About half are object names (Fenson et al., 1994)

How do children learn object names?

(L11: How do children learn object names in a cluttered real world?)

Through observation: connecting heard words with seen things

Why are object names learned before verbs?

(L11: How do children learn object names in a cluttered real world?)

It's easier to link a word to an object than to figure out which part of an event a verb describes

Bergelson & Swingley (2012): what were 6-9 month olds visually attentive to?

(L11: How do children learn object names in a cluttered real world?)

Objects in their home environments (foods, body parts)

Smith et al. (2003): what can toddlers do by 24 months?

(L11: How do children learn object names in a cluttered real world?)

After seeing one instance and hearing its name, generalize the name to the whole adult-like category

Define the shape bias

(L11: How do children learn object names in a cluttered real world?)

Categorizing (and generalizing names to) objects by their global shape

Smith et al. (2003): caricature finding

(L11: How do children learn object names in a cluttered real world?)

• At 18 months, toddlers recognize richly detailed objects better

• By 24 months, recognition is equal for rich objects and simple caricatures

• This ability comes with object naming

Why do lab eye-tracking studies underrepresent the challenge of word learning?

(L11: How do children learn object names in a cluttered real world?)

• Lab objects aren't in cluttered scenes

• Babies' natural environments are visually cluttered

Define egocentric vision

(L11: How do children learn object names in a cluttered real world?)

• Studying the visual world from the child's momentary perspective (head camera or eye tracker on the baby)

• More ecologically valid

Jayaraman, Fausey & Smith (2015): what do young babies see at home?

(L11: How do children learn object names in a cluttered real world?)

• Faces, lots of faces

• The youngest babies see a face about every 15 minutes of every waking hour

Jayaraman et al.: how do faces in view change with age?

(L11: How do children learn object names in a cluttered real world?)

• Faces (and faces within 2 feet) decline with age

• People are still around, but babies aren't looking just at faces

Kuhl & Meltzoff (1982); Yeung & Werker (2013): faces and voices

(L11: How do children learn object names in a cluttered real world?)

• Young babies perceive faces and voices as unitary events

• They recognize the correspondence between facial movements and speech sounds

By 6 months, infants know names for what?

(L11: How do children learn object names in a cluttered real world?)

Common body parts (Bergelson & Swingley, 2012)

What is the dominant early input to the visual system?

(L11: How do children learn object names in a cluttered real world?)

Faces; early visual experience is largely shaped by faces