Infant Speech Perception

• Child typically learns their first words around 1 year old
• Common early utterances: "Byebye,” “Nana,” “Mama,” “Uh oh”

High Amplitude Sucking Procedure

• Method to test infant perception by measuring changes in sucking rate in response to auditory stimuli
• Infant given pacifier with sensor
• Each time the baby sucks, a sound is played

Sucking Rate Changes

• Habituation: Sucking rate decreases as stimulus becomes familiar
• Dishabituation: Introduction of a new sound leads to increased sucking rate if the baby detects change
• Increased sucking rate indicates the baby noticed a difference in the stimuli

Categorical Perception of Phonemes

• Categorical perception of phonemes tested using high amplitude sucking procedure
• Phonemes like /pa/ and /ba/ are acoustically similar but sound different
• Heightened sensitivity to the boundary between similar sounds results from knowing a language such as English
• Ability to measure how quickly babies suck on pacifier to identify if they know difference between sounds
• Babies have heightened discrimination at the boundary of /pa/ vs /ba/
• The study used voice onset time (VOT) differences between sounds like /pa/ and /ba/

Conditions for the Study

• Between-category: "pa" vs. "ba" (different phonemic categories) lead to babies increased sucking rate as they dishabituate and babies perceive a categorical difference between the sounds
• Within-category: "pa1" vs. "pa2" resulted in babies not changing sucking rate as no dishabituation occurred and babies did not perceive meaningful difference
• Infants experience categorical perception and treat certain acoustic differences as meaningful (phonemic) and others as irrelevant

Acquiring Phonemes as a Matter of Loss

• Infants born with the ability to distinguish all phonemes used across world languages
• Over time, they lose ability to hear contrasts not used in their native language
• Werker & Tees (1984) used conditioned head-turn technique to test English-learning vs. Hindi-learning infants on Hindi dental vs. retroflex stop contrast

Results from Werker & Tees

• English-learning infants (6–8 months) could detect Hindi contrast; at 10–12 months, they could no longer detect the contrast
• Hindi-learning infants continued to detect the difference
• Phonemic development is a loss of sensitivity to contrasts not relevant in the child's linguistic environment and is a “selectionist model” of development

Animal Innateness of Categorical Perception

• Chinchillas trained to respond to a phoneme change (pa to ba) by jumping to avoid a shock, showed categorical perception, like human infants
• Categorical perception is not uniquely human or linguistic, it reflects an innate, domain-general auditory ability that language acquisition builds upon
• Categorical perception is innate, but more related to general auditory perception than to language-specific mechanisms

Babies' Understanding of Physical World

• Young babies are smarter than originally thought in understanding physical world (objects, events, properties, etc.)
• Old view: Babies were passive learners, “blank slates” needing extensive exposure to form associations
• New understanding: Babies are active learners

Capabilities of Babies

• Tracking statistical patterns in speech (e.g., syllable pairings)
• Discriminating between real words and non-words based on transitional probabilities

Saffran et al. (1999)

• Exposed infants to a continuous stream of nonsense words with no pauses
• Babies used statistical learning to identify "words” based on which syllables reliably followed each other (e.g., ta-bi vs. do-go)
• After just 2.5 minutes, infants could discriminate real word-like units from random syllable sequences

Implication

• Word learning is more than imitation as it involves sophisticated computation and pattern recognition, even before full word production begins
• Piaget discovered that kids knew a lot, they knew the difference between birds, cats and dogs
• Part of brain that represents rich mental life. Children have a rich, conceptual understanding of objects

Object Permanence

• Object permanence is an understanding that objects continue to exist even when they are no longer visible
• Piaget: Infants do not have object permanence in early months
• According to Piaget: 0-4 months: Babies only understand objects as part of current action schemes (e.g., grabbing, sucking)
• According to Piaget: 4-9 months: Babies begin integrating more schemes but still don't understand object permanence
• According to Piaget: 9-12 months: They begin to search for hidden objects but do so through action patterns, not abstract understanding
• Piaget concluded that object permanence emerges around 9–12 months, and fully develops by 18–24 months, later in the sensorimotor stage
• The A-not-B error is when an after an object is repeatedly hidden under Location A, and the baby successfully finds it there, the object is then hidden in Location B, even after watching it moved, the infant still searches under Location A

Piaget's Interpretation

• Lack of object permanence and reliance on motor schemes

Modern Reinterpretation

• Infants look more at Location B, and are surprised when object is revealed at A
• Cummins & Bjork, 1983; Ahmed & Ruffman, 1998 suggest infants know the object's true location, but make the error due to:
• Motor memory bias: Inability to inhibit a practiced motor response, likely due to delayed frontal lobe development, not lack of conceptual knowledge

Baillargeon's Rotating Screen Study

• Infants were habituated to a screen rotating back and forth (like a drawbridge). Then a solid box was placed behind the screen
• Possible and impossible event results in 4-month-old infants looking longer at the impossible event, indicating surprise, which suggests that infants expected the box to stop the screen and knew the box still existed, even though it was occluded
• Object permanence is present far earlier (as young as 4 months) than Piaget believed

Quinn, Eimas, and Rosenkrantz (1993)

• Infants were familiarized with a category of images (e.g., cats), then shown a novel item from the same category vs. an item from a different category (e.g., bird or dog)
• Infants looked significantly longer at the new category, indicating discrimination between categories at ~63-65% for novel categories, showing preference for novelty
• Infants have conceptual categories for object types, like cats, dogs, and birds, even before they have words for them
• Word learning must be like a mapping problem where words are mapped onto pre-existing concepts
• Symbolic representation of the world (thought by Piaget) is in place very early, perhaps from birth

Referential and Semantic Uncertainty

• Referential Uncertainty: When a word is spoken in a scene with multiple objects, the child doesn't know which object it refers to (e.g., "vash")
• Example: Parent says "doggy" while dog, leash, and bowl are present-how does the child know "doggy" refers to the dog?
• Semantic Uncertainty: Even if the child identifies the correct referent, it's unclear which aspect of the object the word refers to
• Example: "Vash" could mean shoe, sneaker, sportswear, laces, or even “things encountered on Tuesdays" (Quine, 1969)

Dare Baldwin study

• 18-month olds tried to learn the meaning of a novel word while a mother was or wasn't looking inside a bucket
• 18-month-olds inferred meaning of "dax" based on what mother was attending to with "dax" referring to what was in the bucket
• Children are intention readers as they use gaze and body posture to infer reference. Word learning is not passive association; it's socially guided
• Younger infants (~16 months) only learned if they and the mother were jointly attending to the object and it suggests that the ability to use disjoint attention for word learning develops between 16 and 18 months

Word Learning Bias

• Whole Object Bias: Children assume a new word refers to the whole object, not just a part or property of it Ex: Shown a cup and told “This is a mido,” the child assumes “mido” = entire cup, not the handle or color
• Helps with semantic uncertainty by narrowing down which part of the object the word likely refers to-the object as a whole
• Shape Bias: Children generalize a new word to other objects with the same shape, rather than based on color or texture
• Shown a novel object called a “dax,” and then presented with a shape match and color match, they choose the shape match as another "dax”
• Shape Bias guides generalization to category-level concepts, especially useful when learning object nouns
• Mutual Exclusivity: Children assume each object has only one label and if they know one object's name, a new word likely refers to the unnamed object
• Example: If shown a dog (known) and tongs (unknown), and asked “Where's the mipen?”, child picks tongs. helps with referential uncertainty and semantic uncertainty of category of something new

Human Simulation Paradigm

• Research method where adults view muted video clips of real parent-child interactions
• At the exact moment the parent utters a word, adults guess which word was said
• Purpose: To estimate how clear the referential context is by using adults as “ideal learners” of context clues
• Provide an upper-bound estimate because adults can guess the word and it's unlikely a baby could

Referential Context for Children Learning

• Most occurrences are NOT referentially clear and over 60% were classified as “junk” as adults guessed the target word less than 20% of the time.
• Only ~15% of contexts were “gems” as clearly timed cues and successful adult guesses accounted for 50-100% with 25% falling into an intermediate category
• "Gems” are rare but powerful: Gems are defined by precisely timed cues that align with the moment the word is spoken and these cues include:
• Parent gaze or attention directed at the referent
• Child looking at the object being named
• Referent being manipulated or gestured toward by the parent Object being clearly visible in the child's view

Clear Referential Intent

• Occurs when all of cues are in tight temporal alignment with the spoken word
• Timing is critical. Cues must be time-locked to moment that word is said and experiment showed having beep turn gems into junk
• Verbs are especially hard to learn from observation, 100% of verb occurrences tested in HSP were classified as junk
• Verbs help children struggle to learn abstract relational terms that explain why children tend to learn nouns before verbs
• More "gems" in a child's input predict better language outcomes and families varied in the proportion to that

Most Occurrences of Words

• Most occurrences of words in the home are not referentially clear
• Successful word learning often depends on rare but well-timed, high-quality learning moments (gems)
• These gems are characterized by joint attention, gesture, referent visibility, and temporal alignment of these cues with speech
• Children aren't learning from every word instance they hear, they must filter the noisy input and focus on the most informative moments
• Findings challenge the idea that children passively absorb language from constant exposure but are selective, targeted learners who rely heavily on contextual clarity

Global Learner vs Local Learner

• Both try to infer the meaning of a word (e.g., "dax") over multiple exposures.
• Both learn from context, but differ in how much they store and recall
• Global Learner Stores all possible referents from every situation and compares and intersects this to find a common meaning
• Global learner treats learning as a cumulative, statistical process across situations. If "dax" is heard five times, the takes what object was present in all five instances-that's "dax"
• Local Learner Guesses the meaning in each individual instance and guessed meaning is reinforced when confirmed later, or new hypothesis can be made
• Doesn't store every past context—focuses only on most recent guess
• Propose but verify strategy avoids memory overload and learning becomes diluted from irrelevant referents

Empirical Approach

• Medina, Snedeker, Trueswell & Thorpe (2011) looked at how much people recal
• Adults watched muted parent-child interaction clips, each time the parent said a known noun, it was replaced with nonsense word
• Findings reveal adults remembered their own prior guess, not other plausible referents and When prior guess was correct, participants guessed correctly again. When prior guess was incorrect, performance dropped to chance levels
• Adults act like local learners and use propose-but-verify guess test cycle to update feedback is wrong
• Includes: Suggests that humans do not aggregate all information globally across word occurrences and incrementally over trial and error
• Different ways to learn
• Why are verbs learned later than nouns during early vocabulary growth? as Two theories help explain: Conceptual change and learning from observation. Conceptual Change
• Conceptual Change Hypothesis: Argues that verbs require more abstract thinking-understanding actions, intentions, and states • This suggests that young children lack the conceptual capacity early on to understand what verbs refer to • Once children's conceptual system matures, they can start learning verbs Ex: verbs involve complex relational ideas, babies get objects and then events

Learning From Observation

Learning-from-Observation Claims that verbs are harder to observe and label than nouns—not because of conceptual limits, but due to informational structure in input.

• Object exists in space for a long period of time so it is more likely that you would have the concept shoe to make that mapping
• Verbs exist in space and time for a short period of time with limited labeling

Studies Show

• 15% of noun uses are “gems” , but 0% of verb uses are gems-hard to infer verbs like “give” or “see” from visual context alone
• Even blind children learn verbs just as well as sighted children, suggesting that language input and structure, not visual access or conceptual growth, are the bottleneck

Research

• Gillette, Gleitman, Gleitman & Lederer tested that word learning particularly stems from certain limitations or from encountering that word. Tested videos on nouns and verbs
• If language learning were limited in "put" adults should be able to to guess the meanings easily as they already know the answer

Results

• Adults performed much better than identifying nouns than verbs
• suggests it is not conceptual but observational: nouns label visible objects and are easier to infer, while verbs describe abstract or less visually salient events that are harder to identify through observation alone
• Information change: explains patterns of word learning and vocabulary growth that favors the learning of easier concrete words
• Children use multiple sources of information beyond the visual learning of senses of events and actions.

Syntax and Nouns

• Knowing the other words in a sentence helps reduce the amount of information needed to learn the meaning of verb

Syntactic Frames

• Grammatical structure (number and order) provides clues to people involved.

Syntactic Bootstrapping (SB)

• How adults use the rules of of how one talks with the same meaning to help with unfamiliar
• The grammar of the sentence provides clues to types of relations between entities
• Children can use what they know can help their verb knowledge.

Research

• 2 Studies showed about 24 months old babies listening to adults while using sb
• (Nagiles) In 1 video with sides by side one had a transitive one not

Kinderpathen

• the children always make similar mistakes.
• is described when children are always going incorrect and fail when giving sentence and inter.

Study

• 3 sentences in 1 to prove and is often missed often
• conclusion : all types of language all show this no language
• key notes : most of it the brain controls with that they help to show.

Key Examples

• Some Adults ( ig363
• studies show other task is used to help
• Support from many groups says language dont count.
• So people change it earlier/ later and if it dont happen than its a reason children dont learn faster-

A theory

A natrual experiment is a situation where researches observe And study children with learning disabilities and they see

• that it creates.
• Children can have both problems and thats the proof.

Home Signs

• is just gestures
• all kids start at 2 words • The idea comes to both and is proven to all.