psych 143

statistical information

they use stats to understand what an item refers to and whether it refers to a specific type of thing( a dalmatian) or the broader category itself ( dog) or the characteristics (fur)

What properties of a word's meaning affect our categorization judgments?

shape, material, texture, color are neither necessary or sufficient either function (something you sit on is not necessarily a chair) But perceptual features can be used as cues to object kind

Fast Mapping (Carey & Bartlett, 1978)

"Give me the chromium tray, not the red one" they then remember what chromium can refer to and can then pick the color our a month later in time. they learn how words relate to others. Fast mapping may not be specific to words

What happens when parents label objects children aren't looking at?

- 30-50% of the time the parent isn't talking about the object that the child is looking at - Does the child attach the label to the object that she sees, or can she infer the adult's focus of attention?

Human Simulation Studies on Verb Learning

- 7 information conditions, which vary whether subjects receive the following information: 1. Scenes: the silent video clips 2. Lists of the nouns that occurred with the verb in each scene 3. Syntactic frame that the verb was used in All combinations of information presented to participants: 4. Scenes + Nouns 5. Scenes + Frames 6. Nouns + Frames 7. Full Information (Scenes + Nouns + Frames) Sample noun set for wug 1.elephant, piano 2. mommy 3. I, it, you 4. it, you 5. drums 6. music, you what does wug mean?

Problems with maintenance-loss model

- Ability to distinguish between phonemes is not completely lost with lack of exposure: - Adults can be trained to make distinctions - Learn new languages without accents even when exposed after 10 or 12 months - What differs between languages isn't that children aren't exposed to some phonemes, but instead the distribution of sounds they are exposed to

Relation between phoneme perception and word learning

- Ability to focus and detect language-relevant contrasts is important for word learning "bear" vs. "pear" "big" vs. "dig" - Ability to discriminate native contrasts at 6 months positively correlates with later vocabulary size at 13, 16, and 24 months - Discrimination of non-native contrasts negatively correlatedwith later language skill

Speech Perception

- About 200 distinct sounds are used across languages, and English uses 45 - Other languages use different sounds (e.g., click consonants in Khoisan languages from southern Africa) The first step in understanding a language: perceiving its sounds Three problems in perceiving sounds of a language: Discrimination: bat vs. pat difference in phones Invariance: /b/ changes over varitation in speakers, intonation, context (bat,bit,but,tub...) Categorization: which sounds count as /b/?

The Human Simulation Paradigm

- Adults "simulate" being a 3-year-old: watch silent videos of actual scenes of mothers playing with children - Common nouns & verbs chosen as "targets" - Can subjects identify a word (that they already know) from the scenes alone? - Subjects are shown 6 scenes for each word Have to identify the mystery word (noun or verb) represented by a beep No context given: Only the scene (world) is given. Nouns (45% correct) can be identified from scenes alone, but verbs(15%correct) cannot

Describing speech sounds: phonetics

- Alphabet letters aren't adequate for describing sounds (e.g., s can sound different depending on where it appears) - Vowels: unimpeded airflow through vibrating vocal cords - Vary by placement of tongue and rounding of lips - Consonants: sound is impeded in some way - Consonants also vary by manner of articulation (how vocal tract is closed) and voicing

Neuroconstructivism

- Brain as a bucket of legos: the brain starts out as bunch of multi-purpose parts - Parts may be relevant to some kinds of inputs but usable for other processing - The more a part is used for one kind of processing, the more domain-specific it becomes - By adulthood these different domainspecific parts can be impaired - Mutation found on gene that helps control the assembly of actin - Actin helps determine rigidity of cells throughout body - Almost no effect anywhere except hair cells in ear - Results in deafness but it's misleading to call this a "gene for hearing" even though there is a selective deficit

Do Developmental Disorders Provide Evidence for Domain-Specificity?

- Case study of DH: had severe cognitive deficits but could nonetheless carry on a fluent and grammatical conversation Lends support to the idea that language (and especially grammar) is dissociable from other non-linguistic cognitive functions - Down syndrome: language more impaired than other cognitive functions - Grammatical development especially affected; what takes most children 30 months can take children 12 years (Fowler, 1994) - Is there a domain-specific deficit to language? - Need to know more about how cognitive deficits affect specific parts of language

Linguistic cues also affect word extension

- Children are more likely to attend to shape for count noun ("a dax") than a mass noun ("some dax") - More likely to pick substance matches when shown mass nouns paired with non-rigid substances

Problems with learning via association

- Children could figure out what a word might refer to by compiling evidence from multiple observations - But can infants hold a whole matrix of word-referent probabilities in memory and update them with observations? - What about cases when children aren't looking at the referent?

Left Hemisphere Function

- Dichotic listening task: subjects report more syllables presented to right ear -Language in LH - Split-brain patients have corpus callosum severed to prevent seizure - Stimuli from RVF go to LH and vice versa - Prevent hemispheres from communicating - LH > RH activation for speech sounds and syllables, RH > LH activation for nonspeech sounds beginning in infancy (Dehaene & Lambertz et al., 2006) - Degree of lateralization increases with age (using LH predominantly by age 2) - Of 65 kids with left or right lesions, aphasia almost always followed damage to LH (Woods & Teuber

Pinker's Proposal

- Gradually evolved since split from chimps, through changes to multiple genes that affected multiple language components - Language faculty shows signs of complex adaptive design; selected for communication of thoughts through speech - Could confer reproductive benefits: useful to know if there are "animals you can eat" or "animals that can eat you" - There has been enough time for this to happen (200-300K generations) - Would have still been useful in earlier forms (as pidgins are)

Williams Syndrome

- Moderate to severe mental disability - Relatively good language and face processing ability - Similar level of cognitive impairment to individuals with Down syndrome, but have rich language Karmiloff-Smith: essential to go beyond the behavior to study the underlying processes in detail - Syntax is not normal in WS children, despite earlier claims (syntax mental age lower than vocabulary mental age) - Abnormal patterns of electrical brain activity in language and face processing tasks (no progressive lateralization) - Model of "normal" brain with one module knocked out incorrect: "normal" behaviors achieved in other way

What happens to the sensitivity shown in human infants?

- Newborns are universal listeners: they are sensitive to every phonological distinction yet tested, including non-native ones - Infants' perceptual abilities must be innate - But what happens to their ability to discriminate among phonemes that are not used in their native language?

What happens to sensitivity shown in human infants?

- Newborns are universal listeners: they are sensitive to every phonological distinction yet tested, including non-native ones - Infants' perceptual abilities must be innate - But what happens to their ability to discriminate among phonemes that are not used in their native language? - How do they focus in on contrasts that are used in their language?

Insights from non-human animal communicationƒ

- Not clear if animals communicate to affect others' mental states, as opposed to behavior (e.g., to fulfill goals) - Don't seem capable of learning complex, productive systems like human language

Are children's meanings broader than those of adults?

- Object words are thought to define units How many things are in this room? How many shoes are in this room? Should be counted as a 'shoe': a shoe should not be counted as a 'shoe': a sock or half a show "Count the shoes" (image of 2 whole shoes and a show with a half) Adults never count pieces as wholes when counting and are less likely to say there are 4 shoes. kids from 4/6 years count 4 56% of the time. children's errors are not limited to counting when one guy has 2 shoes and other guy has 1 shoe broken in 3 parts, kids will say the guy with 3 parts has more shoes children understand difference between whole and broken object they recognize that its broken 76% of the time

Are there developmental changes in word learning?

- On the surface, the meanings of child's first words appear different than those of adults Are there developmental changes in word learning? - Complexive meanings: "duck" for ducks, the pond they swim in, anything with feathers, etc... - Overextensions: "dog" for all four legged animals "dog" - Underextensions: "car" is only cars seen from window - No evidence that children's word meanings are radically different than adults - Overextensions and underextensions similar to adult word meanings, but child doesn't know all features - "map" = flat large paper with picture on it - No evidence for complexive meanings: children may stretch words to fill gaps in their vocabulary - Word learning begins slowly then accelerates - Many claim that there is evidence for step like progression ("a vocabulary burst")

is Specific Language Impairment all that specific?

- Proposal: genetic changes lead to deficit in domain-general ability to process fast auditory sequences in SLI patients - In normal children, this ability is recruited for processing phonology, morphology, syntax (e.g., agreement marking) - SLI kids may thus have domain-specific difficulty caused by deficit in domain-general ability - Evidence: Rapid acoustic training sometimes improves grammatical performance

Phonological rules

- Rules that govern how phonemes are pronounced in different contexts: help to smooth out pronunciation - Rule is implicitly known, and generalizes to novel words (e.g., wugs vs. ricks) - When two consonants occur together, they match in terms of voicing (demo: bugs vs. bikes) - Phonological systems can vary across different regional dialects

What do children's word extension biases reveal about their word meanings?

- Smith: attentional mechanism; linguistic cues and properties of objects automatically bias attention toward certain features (e.g., shape) which guide extension - Although word extension appears "smart" and context-specific, it's based on "dumb" attentional mechanism - An alternative: shape as a cue; children have rich categories and can extend words based on deeper conceptual properties; They use shape only as a cue to category membership

Phonotactic knowledge

- Speakers know how sounds of language can combine to form words - Kpakali and zloty cannot be words in English, but they are in other languages - We have intuitions for what legal words are, even when they don't have meaning:

Other unique and language-specific components:

- Speech perception - Phonology (phonemes and ways they are combined and adjusted) - Word learning (intentionality/reference, perspectives, mechanisms supporting rapid word learning )

A second approach: Social-pragmatic approach to word learning

- The child is not just mapping a word to whatever they see in the world - She is attempting to discover another person's intention to refer - Initially, child may not be able to use linguistic context ("Look at the bunny next to the baseball") - But speaker intention could be inferred through social cues: - eye gaze - gestures (pointing, etc.) - verbal exclamations ("aha!" "oops!") - Adults monitor children's attention - Adult follow child's eyes and gestures - Mothers label the object the child is looking at 50-70% of the time (follow-in labeling) - Follow-in labeling leads to learning - Correlated with vocabulary size WARNING: all work done on middle-class American mom

Tomasello's Proposal

- Unique to humans:evolved ability to recognize others as having mental states, allowing us to share attention - Words as symbols: through which we triadically direct others' mental states - Other animals don't use signals to direct attention in same way

subtypes of SLI that seem very specific

- Van der Lely: With psycholinguistic tests, we can isolate more coherent SLI subpopulations, instead of basing it on exclusion - Grammatical-SLI: take morphosyntax tests and have to produce 20% or more errors - Inflections that produce extra syllable (pu-shes, wan-ted), are more perceptually salient than non-syllabic inflections (cats missed) but are harder for SLI kids - Many SLI children don't have impaired auditory abilities, and many typically developing kids do

Are these constraints specific to word learning or do they emerge from general properties of cognition or social-pragmatic reasoning? (bloom and markson)

- Whole object constraint may reflect salience of objects both linguistically and non-linguistically - Even infants parse the world into objects, and expect something that moves as one unit to also obey the following principles: continuity and solidity Domain-specific: I know this is a cat, and every object has one name, so the other thing must be a dax

Phonological structure

- Words are made up of smaller syllabic units - We can recognize words that rhyme: differentiate between onset (n in nap) and rhyme (ap in nap) - Spoonerisms: different units are moved between different words in sequence "My dear old queen" becomes "My queer old dean"

A shift from concrete to relational words?

- by 12 months (concrete): - mommy, hi, bottle, telephone, cup - by 20 months (concrete relational): - go, sit, hug, hand, big, up - 30+ months (abstract and grammatical): - By, around, think Why an advantage for concrete words? - Cognitive development: Young infants don't have abstract concepts and initially can't learn relational words - Contingent acquisition: It's a logical necessity to learn concrete words first, because relational words relate concrete words

Faculty of Language (broad) (involved, but not specific to language, could be unique or shared)

- memory - theory of mind - statistical learning - motor control -auditory processing

Specific Language Impairment

- their language skills are below those expected for their age - another explanation for language delay has been excluded (e.g., hearing loss, IQ, autism, etc.) - A lot of variability in the severity of the impairment and specifics of the symptom - Vocabulary generally delayed, but greatest delays are in morphology and syntax - When length of utterance is matched, SLI children have deficiencies in morphology (e.g., tense/agreement marking)

Prediction of Eimas Study

-Innate Categorical Perception = ICP -Untuned Sensitivity = US -Insensitive = IS -Within Category PA1 and PA2 =WC -Btwn Category BA and PA1 = BC ICP+WC = remain habit ICP+BC = dishabit US+WC = dishab US+BC = dishab IS+WC = remain habit IS+BC = remain habit

FOXP2 and language evolution

-Mutation on gene that encodes FOXP2 protein leads to language difficulties -Two differences in FOXP2 since separation from common ancestor with chimp; all humans share normal version of gene -When FOXP2 is "knocked out" in mice, songbirds, results in abnormal vocalizations -Probably selected in last 200,000 years, when spoken language is thought to have emerged

Faculty of Language (narrow) unique + domain-specific

-syntaxctic devices -speech perception -speech production -word learning ...etc.

Forming higher-level generalization (smith)

1) children learn names for individual objects 2) they make first-order generalizations (balls are round, cups are cup shaped) 3)they make second-order gen,(if I know that balls are round, and cups are cupped shaped, then _____ is _______) shape bias once children have learned words and start constructing shape categorizes shows you can also take kids that have not developed a shape bias and train children to develop a shape bias (17 months year old) 17 month olds show shaped bias when students are trained for 8 weeks to 9 week

How do children meet the challenges posed by word learning?

1. Have to use many kinds of information and mechanisms to learn words: the scenes, lexical constraints, social cues, context 2. Only scenes, associative mechanisms, lexical constraints, and social cues are available when learning begins (not context) 3. Only concrete words (mostly nouns) can be acquired from these data sources and mechanisms alone 4. Once children learn words, they can use sentence context to acquire more relational and grammatical words

Developmental change in word learning: contingent acquisition proposal

1. Initially, the child may rely on the scene they observe alone, and thus learn only very concrete words like object labels 2. Having learned nouns, the child has a richer understanding of sentence context, and can learn concrete relational words sentence to world mapping are possible ....man.....GORP...apple....you GIVE? ( since it has a predicate with 3 arguments, gorp may be give) ....man.....GORP...apple EAT? HOLD? (only has 2 arguments) 3. Learning relational words allows the child learn the basic grammar and syntax of her language - Syntactic bootstrapping: can use sentence structure to guess meaning - Allows children to learn even more abstract meanings - Range of structures verb can and cannot appear in constrains meaning

Syntax may be required to help children learn verb meanings

1. Lo WIF what the fimmet mibs. 2. Well, bo WIF what your rucky zavvy smegs are verkly kemp, don't bo? 3. Po WIF lo pung mo. 4. Lo WIF what lo can mib with it? 5. Do lo WIF where the kax's lif is? 6. Do lo WIF where a fimmet is in your runk? what does wif mean?

Do children use social cues to find referents?

1. Look in Box X, but don't retrieve object, say "blicket" 2. Retrieve object in Box Y 3. Retrieve object in Box X yes they do, such as eye gaze and motions

What is learned in word learning?

A mapping between a sound sequence and a conceptual representation - English: d-aw-g DOG - Spanish: p-e-rr-o DOG - French: sh-iy-en DOG - Meaning should allow us to categorize objects (call different things "dogs") and comprehend sentences (combine words together like "big white dog")

hildren are sensitive to deeper factors like intended function when naming objects

Accidental Jane was holding a newspaper. Then she dropped it by accident, and it fell under a car. She ran to get it and picked it up. This is what it looked like. Intentional Jane went and got a newspaper. Then she carefully bent it and folded it until it was just right. Then she was done. This is what it looked like. 3-year-olds more likely to say "hat" when given evidence of intended function (gelman and bloom)

Problem of Discrimination

Although speech sounds seem separable and sequential, they lie on a continuum

Where does categorical perception come from?

Alvin Liberman Motor Theory - although acoustic information for speech is continuous and depends on context, the actions that produce speech are categorical explains difference between adults and children phonemes produced by discrete actions

The Mutual Exclusivity Constraint

Assume that every object has just one name - Related idea - the principle of contrast: assume there are no synonyms (allows for multiple labels for an object) - Evident in 15-month-olds For experiement, given microscope and telephone. Experimenter "look heres a dax" mutual exclusivity show the child already know a telephone exist so the phone can not be the dax. That must mean the microscope is a dax (works with the whole object constraint)

Evidence For Specialization in the brain

Broca's aphasia: patients have great difficulty initiating speech, especially with function words and inflections - Damage to RH structure causes visuo-spatial problems

Syntactic bootstrapping: Semanticprojections from structure

By observing structure in which verb can and cannot be used, child can learn about meaning of verb Intransitive ≈ Agent's self-caused acts John sleeps *John sleeps the bed (not correct) Transitive ≈ An agent acting on a thing affected *John hits John hits the ball. - Child can infer that "sleeping" involves self-causation, but not action on a thing, and vice versa for "hitting" Ditransitive ≈ Transfer John puts the ball on the table *John puts *John looks the ball on the table Sentence Complement ≈ Perceptual/Mental John sees/thinks that the ball rolled away *John puts that the ball rolled away - Helps explain how we can learn meanings of verbs like think - Powerful cue to meaning: once you've heard verb in a frame, you can conclude something about it (especially because parents are speaking grammatically)

Native Listener Experiment

Categorical Perception: /l/ vs. /r/ in english and japanesse japanese speakers can less easily disstinguish between the 2 while the english speakers can. It is because /l/ and /r/ sounds are more specific to english, not japanese spanish speakers have a different VOT than english

Taxonomic Constraints

Children assume words label things of the same kind, not things that are thematically associated one experiment is when they label a dog dax and have an item that is a thematic match( bone) and one that is a taxonomic ( a cat) if they are given a label ("look at this dax! find another dax!") they will more likely pick the taxonomic but if they are not given a label, then they will pick the thematic Children are more likely to select other category member when noun is used Also guides extension of words with novel meanings

How do children find the referent of a new word?

Children can compile evidence from multiple exposures to hone in on the correct referent of a new word (associative theory) - Kids use knowledge of others' intentions to make inferences about reference that go beyond what they see (social theory)

The Whole Object Constraint (markman)

Children interpret word for unfamiliar objects as labeling whole object, not its material or parts sometimes leads to errors hot = stove

Proportion of nouns decreases as vocabulary size increases

Chinese adoptees and English learning toddlers have a similar type of vocabulary size even though they do show patterns in the same stages, the adoptees learn more quickly.

Language Complexities

Complexities do not pose problems or speech perception -adult native speakers have no problem perceiving phonemes -solve discrimination (dad vs. bad) -solve invariance problem (same /d/ in dad and dud) solving discrimination problems leads to categorical perception

Children's interpretations may be narrowed by thinking about alternatives

Count the shoes! If she wanted me to countpieces of shoe, she would have said so... I should only count whole shoes

How do children figure out what phonemes are relevant to their language?

Does word learning play a role? Children learn that "bear" is different than "pear" and differentiate /b/ from /p/? - But infants have difficulty discriminating minimally-different words like "bear" and "pear" until at least 17-months - There are also effects of native language experience on phoneme perception by 6-months (before word learning) -babies take statistics of sounds. English has high and low inflections when mothers speak motherese pronounce /l/ and /r/ whereas Japanese have more middle sounds which is their /r/ which is different from ours

Do children speak to communicate?

Experiment where kids were given objects yes, Children communicate to get their point across, not just to get what they want if given a toy and misunderstood they were more likely to repeat "want duck"/"duck please" than when compared to when they were understood and given the duck. if given a sock, they were more likely to repeat that they want the duck

Do 8-month-olds use transitional probabilities to find "words"?

Familiarization: Exposed to artificial language composed of 3 words in continuous speech (e.g., golatu) use statistical learning listen longer to ones that are not words

Study of Categorical Perception

Identification - Person is played a sound (ba,da,ga) they say that there was a steep drop off when categorizing the different between sounds discrimination - in the same places that people were able to identify ba, da, or ga a person had the ability to discriminate there is some points on the spectrum where you cant discriminate but, there comes a point where you can which shows that there are categorizes

Focusing on the right part of the world: Via association? associative learning

If we will observe how children learn languages we shall find that .... people ordinarily show them the thing....and then repeat to them the name John Locke, 1690 You learn the referent of word across multiple observations, because it is repeatedly associated with something in the world Occasion 1"cat" "cat" = BALL or CAT Occasion 2 "bird" "bird" = CUP or BIRDO ccasion 3 "cat" "cat" can't = BALL "cat" = CAT?

Development of Speech Perception

Infants dis and cat some speech sounds before they begin producing those sounds. The motor theory of speech perception is disconfirmed. Experiments since Eimas: US infants discriminate every speech contrast in English What are the origins of this ability? One possibility: learning in utero - babies can hear in utero -newborns recognize / prefer moms voice and native language in utero = in the wound

Starting over learning language

Internationally-adopted children: input situation parallels infant learners - But they possess cognitive abilities to entertain abstract concepts If they go through these vocabularyshifts, it suggests they are due to logical task of learning language ABSTRACT—Language development is characterized by predictable shifts in the words children produce and the complexity of their utterances. Because acquisition typi-cally occurs simultaneously with maturation and cognitive development, it is difficult to determine the causes of these shifts.Weexploredhowacquisitionproceedsintheabsence of possible cognitive or maturational roadblocks, by examining the acquisition of English in internationally adopted preschoolers. Like infants, and unlike other sec-ond-language learners, these children acquire language from child-directed speech, without access to bilingual informants. Parental reports and speech samples were collected from 27 preschoolers, 3 to 18 months after they wereadoptedfrom China.Thesechildren showed thesame (snedeker)

Word Segmentation: How do we find words in a continuous speech stream?

Jenny Saffran Higher transitional probabilities for syllables that are word-parts "Pretty baby" High TP: "Pre"+ "tty" Low TP: "tty"+ "ba" important because it gives infants the knowledge of what words are acceptable in a specific language - Statistical learning also applied toward learning phonotactic regularities (e.g., blicket vs. *zloty) and grammatical rules? - There are other cues to word segmentation, such as stress PEN-cil gui-TAR Strong-Weak: Weak-Strong: - Children can use stress as a cue to word segmentation (but availability of this cue will differ across languages)

Constructivist Approach to Language Evolution, development, and cross-linguistic universals

LE:-Adaptation for theory of mind (not specific to language) LD:Cpmstruct abstract categories using domain-general mechanism; grammar arises over history CRU:-Fewer universals; only universals of thought/communication

Nativist Approach to Language Evolution, development, and cross-linguistic universals

LE:-Complex adaptation resulting in many domain specific components LD:-Guided by domain-specific universal grammar; have abstract linguistic categories CRU:-Many universals; constrained by universal grammar

Nativist Proposal

Language as an evolved domain-specific structure Gradually evolved via natural selection, through changes to genes that target specific language components

Alternative Proposal:

Language as the modification of other cognitive capacities - Language could be a new machine that is built out of old parts - No uniquely linguistic capacity: we acquire language with capacities that also support other cognitive functions - May not be brain areas for specific tasks, or developmental disorders dissociating language from other abilities

Tomasello's Theory

Language has its roots in uniquely human ability to recognize that others have mental states

Autism as a test of the role of theory of mind in language learning?

Language in lower functioning autism is almost absent - Language in "higher functioning" autism: do acquire language, but can be both delayed and deviant - Problems with prosody, pitch, volume, expressing affect - Difficulty using words for mental states (believe, idea) - Deficit in communicative competence (joint attention, indirect requests, etc.) dont know if these result from theory of mind deficit

Non-native Speech Discrimination

Logic: If learned, infants should only be sensitive to distinctions in native language. If innate, then infants should discriminate non-native contrasts.

Difficulties in word-to-world mappings (gleitman)

Many words are difficult to learn by observing the world and might require other sources of information Difficulties in word-to-world mappings 1. An event that the child observes could be interpreted in multiple ways - Verb labeling this event could mean: - Pushing - Moving - Rolling - Sliding ... 2. Some verbs seem to apply to exactly the same kinds of events: "Buy" or "Sell"? "Chase" or "Flee"? Perhaps children eventually hear these verbs in situations that disambiguate them (e.g., when you flee a city it doesn't chase you) 3. But children will also hear words in situations that have no relation to their meaning at all! Where is your mother? (being asked because their mother is not their) Eat your peas! (being told because they are not eating) 4. Many words stand in subset-superset relations: MOVE super set WALK sub set SAUNTER sub-sub set - If children thought "saunter" meant MOVE, seeing more "sauntering" wouldn't help them recover from this misinterpretation - Especially an issue if children don't receive corrections (negative evidence) 5. Many verbs are not clearly linked to the physical world at all Mental verbs: think, know, guess, wonder, know, hope .... "Try, for example, to learn the meaning of the word think by titrating discourse situations into those in which thinking is going on, somewhere when you hear "think" versus those in which no thinking is happening. Remember that there isn't always brow furrowing or a Rodin statue around to help. Keep in mind that you're going to have to distinguish as well among think, guess, wonder, know..."

Does word learning have a critical period?

No

Is transitional probabilities Is this ability specific to language or unique to humans?

No, it is not, it is domain general. Tamarin monkeys also use it.

Sounds of Language

Phonemes: systematically vary, contrastive example : bat vs. pat (b vs. p) Allophones: Systemically vary, not contrastive example: pill vs. spill (p vs. sp)

Pinker and Combinatorial Devices

Pinker: There are other devices that are not necessarily recursive but are specific to language and unique to humans 4 Combinatorial devices of syntax 1) Hierarchical Structure 2) Order of Words or phrases within a phrase "Dog bites man" vs. "Man bites dog" 3) Agreement: inflections corresponding to number, person, gender The boy walks The boys walk I am He is el perro rojo la vaca roja 4) Case marking: noun phrases marked with inflections depending on their grammatical role Nominative: subject of verb ("We went to the party") Accusative: direct object of verb ("They high-fived us")

Sentence context as a cue to word meaning

Sentence context is useless to the beginner - Children initially don't understand any words in the sentence - But associations, social cues, and word learning constraints could help children learn concrete nouns (e.g., object labels) - Children could then use context to learn more abstract and grammatical words, like verbs, prepositions, abstract nouns Across languages, children learn nouns before other words (despite hearing lots of verbs) Concrete verbs before abstract verbs

word extension

Smith: Children don't have the deep understanding of word meanings that adults do based on attentional learning mechanism when first encountered with nouns, children organize things into the shape category table = table-shaped things

Chomsky's theory

The only unique and language-specific ability is recursion

Pinker's Theory

There are many unique and language-specific components; language a complex adaptation for communication that evolved piecemeal

Language acquisition in internationally adopted children

These children lose L1 and become native speakers of L2 • By adulthood no evidence of access to birth language (Pallier, Dehaene, Poline, LeBihan, Argenti, Dupoux & Mehler, 2003) • Adopted toddlers are linguistically delayed relative to age mates but show catch-up growth. (Glennen & Masters, 2002)

Eimas Study Results

They are sucking when sound (pa) is presented and stop sucking when same sound is presented overtime(bored), when new sounds presented, they suck more (ba) they do no discriminate or categorize between pa1 and pa2 categorical perception does not require articulation

Knowledge of Articulation

This is involved in our perception guy with conflicting sound ("ba") and visual place of articulation ("ga") lead to perception of ("da")

Categorical Perception

Voice Onset Time (VOT) - when the voice begin - when we are able to discriminate phonemes by different time indicators.

Testing the shape-as-a-cue theory

What if children are given reason for why 2 objects share same shape but are from different categories? This is a fendle I'm going to put my fendlein here shape match substance match standard - 3-year-olds less likely to choose shape match if they learn it's a container - Should have chosen shape match if they have automatic attentional bias Diesendruck et al. (2003)

What is recursion in language?

When the category being defined is applied within its own definition.

Can babies take statistics on new language?

american babies heard mandarin for the first time and 6/8 months they were equivalent with taiwanesse infants - 2 months later, the taiwanese infants did better at distinguishing sounds. test this experiment again to cancel out human interaction and it with 1) a tv 2) just audio and found that it did not improve performance

Early version of motor theory

as children produce speech, they associate the sounds they make with these different action categories when they har speech, the sounds they hear automatically call up associated articulatory actions explains invariance- same articulation actions in different context (/da/ vs/ di/) explains discrimination - different actions required to produce different phonemes cat percep - articulaltions are categorical even though resulting sounds arent

Chinchilla Discrimination of Human Speech Sounds

avoidance conditioning in chinchilla when ever they hear a Pa, they will get shocked (A) and when they hear Ba (B), they dont get shocked if they hear sound A and run to the other side of the cage, they wont get shocked if they hear B, they wont need to run to the other side.

Is mutual exclusivity specific to word learning?

cat and whisk Domain-specific: I know this is a cat, and every object has one name, so the other thing must be a dax If children's behavior stems from socio-pragmatic reasoning, it should apply beyond word - 3-year-olds treat both words and facts exclusively (Diesendruck & Markson, 2001) - Consistent with idea that mutual exclusivity depends on socio-pragmatic reasoning and not a domain-specific constraint - But reasoning about words and facts may correspond to different processes - 2-year-olds and autistic children are better at words than at facts - Exclusivity with words correlated with vocabulary size; exclusivity with facts correlated with socio-pragmatic skill

Do 14 month olds represent precise phonological forms when learning words?

child habituates to one image with 'Dih' and another with 'Bih', then in the test phase they will see a sound which does not match the image and they will be aroused because they would be different in 14 month kids, they dont represent the pairings they then saw with 1 image and sound, and 8 monthold succeed in distinguishing switch and same trial whereas 14 months can not as well then more specifcally on 14th month olds, have a pairing with the word lif and distinguish it from neem 8 month-olds succeed, but 14 month-olds still do not - Rules out complexity as explanation new experiment with bih and dih with just a checker board - 14 month olds detect phonological changes (/b/ vs. /d/) - But they fail to use this ability when learning words! - One hypothesis: because they initially have small vocabularies, they don't need to represent words in fine detail - Additional functional reorganization brought about by word learning

Piraha Amazonian Language (counter to recursion)

claim is this language lacks recursive was told 1 man killed a jaguar, 1 killed a bird, and the one that killed the jaguar fell. when asked who fell he said the "cat fell down" actually meaning the man that killed the cat fell down

Problem of Invariance

each sound is also affected by elements around it, creating problem of perceiving an invariant phoneme (rather than contrasting ones)

Speech perception specific to humans and language

evo scenario 1) humans evolved to perceive the relevant auditory info. evo scenario 2) primate audition systems are sensitive to rapid transitions when language evolved, it was constrained by pre-exisiting limits on auditory perception.

Children use part of speech to constrain interpretations of new words

expposed to hes sebbing, look a seb, and look some seb What does seb mean? - Preschoolers infer meaning based on sentence structure and grammatical category (brown

Preferential Looking Paradigm and Categorical Perception

face with [a] and [i], when infants ( 4 to 5 months) see the shape and then hear the sound, they match the shape of the face with the sound

What accounts for decline in percetion of non-native contrasts?

functional reorganization - language-relevant categories represented in additional layer maintenance or loss - "use it or lose it"

Brain AS Swiss Army Knife

genes are selected to solve specific problems evolution creates "a toll for each job" - Should find evidence for localization of functions in the brain Deletion, reduplication, or mispositioning of genes should result in selective developmental impairments

Do children speak in order to communicate? (study results)

go back over lecture 8

Predictions for Chinchilla

if they have categorical perception, they should have steep drop of at specific time by VOT if they have graded perception, then it should be a linear decline categorical perception is not specific to humans

Rejection Onject (sock/toy)

if they were given a toy, and where misunderstood ( they would reject it) if they were given a sock and were understood then would still sometimes reject it but, where more likely to reject if misunderstood

Non-native Speech Summary

infants initially are sensitive to the main sound distinctions used by all natual languages they have not heard non-native contrast so their ability to discriminate is not learned by experience with those sounds.

Testing Motor Theory

infants only begin to produce phonemes around 6 months Do younger infants show signatures of phoneme perception? use pacifier with sensor to monitor sucking -infant sucks to produce speech sounds which are found to be rewarding -habituate to familiar sounds and suck less - if infants hear a novel sound and start sucking more, it shows that they are able to discriminate

So how instead might children learn these words? The role of sentence context (gleitman)

jabbawock poem example which shows that adults can understand the meaning or form (noun, verb, adjective) of a word based on the surrounding vocab. (context clues) What does gyre mean? What does slithy mean? What does toves mean?

Neonate preference for speech

neonate = new borns tested newborn infants preference for human speech relative to sign wave transformation of the same speech method - sucking example 1 = human speech example 2 = sine wave infants have no different at initial block the preference is then seen in the second block and prefer human speech over sign waves

Eimas Study

played sounds for infants during sucking task s1= ba, s2 = pa1, s3 = pa1 s1-3 are spread out by VOT - time between consonants adults place phoneme category boundary at 25 ms (difference between ba and pa)

Infants rapidly learn word-referent mappings via cross-situational statistics Linda Smith *, Chen Yu

presents infants with novel objects paired with new words 12 and 14 month olds will be presented 2 objects and the two words then must figure out which one is different. Test trials: Present one word with correct "target" object and "distracter" object in view; which do infants look longer at? both 12 and 14 months correctly observe what are the different word meanings.

preferential looking paradigm with verb learning

pushing and arm moving training phase: when two actions are happening at once "rabbit is gorping the duck" (transitive) and"the rabbit and duck are gorping" (intransitive) test phase: when asked to find gorping, they will start at either the transitive or intrastive depending on their bias.

Conditioned Head-turn paradigm

sucking only works on young infants, while children and adults can be asked to indicate changes conditioned to turn head when sound changes children begin to anticipate reward when different sound is played which show the children notices the differences between sounds. 6-8 month olds and 8-10 month olds do good. when done on 1 year old, they have a harder time discriminating

Perceptual Magnet Effect (kuhl)

the prototype of is in the middle of the sound and when sounds are similar to the perception prototype, they have a harder time distinguishing

Do kids always show shape bias, or does prioir word meaning constrain interpretation of new one?

this stuff is called wug/kiv (material) this thing is called a kiv (with specific shape) are they (object) when asked if new items are a kiv, will they base it on shape or material in study that is unambiguous (where the word for the material is different form the word of the objects kiv and wug) kids share shape bias in polsemy where they have the same name (kiv and kiv), kids are likely to pay attention to the material more than shape - Children initially appear to extend newly-learned words according to perceptual features like shape - But they may do so because they use shape as a cue to object category membership - When they have another way of determining category membership, they base their word extension on that - Children's early word meanings are not so different from those of adults in this respect

The problem of radical indeterminacy (quine) (word learning as a problem of induction)

we associate word with a referent but this doesn't explain how we extend to new instances The "dax" = dog example from the smith reading. level of categorization - only this dog? other dogs? other animals by other properties - furry thing? brown things? small things? not by category - dogs and leashes? dogs and bones?

Figuring out a word's meaning: A problem of induction

when given a new word

Some evidence for early difficulty with relational reasoning?

when given a picture of 2 pigs facing each other, a relational mapping would 2 turtles facing each other in the same manner and an object match would be an image with a pig and a gold fish. Children are more likely to do the object matching

The insufficiency of word-world pairings

when told "go show granny what youre doing. Shell think you push the truck so well." the actions however go push-go-show-think There are systematic mismatches between when we hear words and what we see Could lead to mistaken hypotheses if children are just making word-world associations

word learning as an inductive problem

which allows taxonomic constraint to arise due to a need for category barriers smith reading young children extend words to similar-shaped objects (by age 2) when given dax and asked to categorize whether a different item is dax, they were more likely to say it was dax if it had the same shape even if the texture was different

Does knowledge of prior senses constrain how children extend new senses?

wiping and wipe - kind membership defined in part by shared function glass and a glass - kind membership defined in part by shared material Material names label artifacts made from a particular material (but of varying shape) if you had a cup made of glass, and called it a glass and then see a cup made out of plastic, you dont call it a glass.

Nouns and sentence structure helps with mapping known concepts onto novel verbs

with just nouns and scenes, not a good prediction with verbs but when you combine nouns and scenes with varying other factors, they are able to do better.