Acquiring words + their meaning Flashcards
(21 cards)
Babbling
-Initial attempts of communication
- repetitive use of syllables with no meaning
- 6-15 months
- universal + May reflect attributed of lang
- universal = muscular exercise where the babbles have no meaning
- second stage where there is an intention behind the babbled speech
- goes from non-referential to referential (intention)
- deaf babies babble too, orally + manually
is babbling hard-wired? study
-Petitto + Marentetle (1991)
- tracked proportion of manual babbling (hand gestures) of 2 deaf + 3 hearing children
- Deaf had greater proportion of manual babbling
-Babbling is automotive + modal, hard-wired in dev
Is babbling a precursor of lang- discontinuity hypothesis
- No
-Jakobson
-Babbling is pre-programmed, muscular exercise
-deaf children don’t babble orally as much as hearing but still acquire lang - silent period between babbling + onset of real words
- No strong correlation between babbling + lang onset
Is babbling a precursor of lang- continuity hypothesis
-Yes
-Mowrer + Vihman et al
- commonalities between late babbling + native lang
-Adults hear differences in babbling for diff lang
- acoustic analyses show difference in babbling for diff lang
- There is no silent period
-correlation between babbling + lang onset are detectable
-Babbling has linguistic manifestations
Linguistic manifestations of babbling - Holowka + Petitto (2002) - mouth
- left hemisphere controls muscles on right + is Assoc with lang
- when we speak we open our mouths more on right
- Graves et al -> right asymmetry for 150/ 196 adult speakers
- filmed babies babbling + found greater right opening of mouth
- babbling shows neuro-linguistic characteristics similar to adults, suggesting babbling is related to lang
Phonemes
- Are classes of sounds in a lang
- At 1.5 years, 50-60% phonemes are correctly articulated
-At 3 years 90% of consonants + 100% of vowels are correctly articulated
-earliest produced consonants occur during babbling (P,h, n,b, m)
words - when does it start? studies
-classic method : indirect count + reports (average 9 months
- Tincoff + Jusczyk preference looking paradigm,6 months
-Hear mumy = look longer at pic of mother
- Associate sound pattern of mummy to their mother
word knowledge as a device for word discovery
-Bortfeld et al - head-turn preference
- 6 months
- pre-check = infants knew word mummy but not tommy
- familiarised with mummy/ToMmy + feet passage
- infants who heard mummy passage showed preference for feet over cup
- Infants who heard tommY passage didn’t show any preference
-known words trigger learning of more words
Reference problem with words
-child holding duck in bath says bear
-> Thinks a duck is a bear ?
-> Doesn’t know word for duck?
-> pretends duck is bear?
-> Names duck bear?
- Difficult to infer intention through beh
- At 18-24 months: Juxtaposition of 2 words - 2 sub-stages:
1) random arrangement
2) syntax-like constructions
word segmentation
-less than 10% of infant- directed speech consists of isolated words (Brent + Siskind. 2001)
-segmenting speech into words is a challenge
childrenhearwordsinasequencelikethis
statistical learning - Safran, Aslin + Newport
- 8 month olds
-head turn procedure
-initially appears as a sea Of sounds - Then progressive detection of sequential regularities (transitional probabilities)
- sperate syllables to create regularity
- Infants can discriminate familiar from novel items -longer listening to novel (novelty effect)
- They AssoC high transitional probabilities between syllables as likely words
vocabulary growth
-Receptive Vocabulary (comprehension)
-> 6 months recognise some familiar words
-> 12 months avg = 100 words
- Expressive vocab (production)
-> 12 months mean = 10 words
-> 24 months mean = 300 + words
- production lags behind comprehension
-external factors can hinder production (shyness)
Lexical Explosion
-sudden word learning spurt around 2 years (Bloom)
-initially explained by qualitative changes in learning: naming insight + syntactic dev
- Far from systematic (Goldfield + Reznick) -> happens at different times for different children + some doesn’t happen At all so may be simply vocab acquiring
computational simulations can account for lexical explosion without changes in learning (McMurrary, 2007)
-exponential learning observed it : words acquired in parallel + most take a long time
- Both are reasonable + realistic
-seen if learning of words may lead to cost in learning others
Multiword utterances
-combination of words can occur with vocals of 50-100 words
-As vocab increases children produce sentences
- children differ in age their sentences begin to increase in length it’s independent of vocab
- constituency in trajectory not age
. At 24 Months, long mean length utterana comes with
increase in syntactic complexity
2 4 months = few words, present tense
-36 months = more words, various tenses
semantic development - ease depends on the word
- Nouns are referential (physical rep) + meaning is inferred from env
-verb meaning is more difficult
semantic development - syntactic bootstrapping principles
= meaning learned from syntactic structure
2 core learning principles - Markman+ Wachtel (1988)
1) whole-object bias: words refer to whole object - teaching features is hard
2) Mutual exclusivity: 1 word=1 object - novel words must refer to something else (counter-intuitive if bilingual)
semantic development -role of infant directed speech
-exaggerated prosody, short sentences, slower, repetition, novel word at end
-vital in learning syntax (grouping features) + semantically
semantic dev- role of visual attention
- smith Yu + Pereira -18-24 month olds
-mothers teach new object- word pairing - infants wear camera tracking eye movement
-successful learning associated with field occupied by object during utterances
How does semantic dev scale up? - 2 unknown words
- Smith + Yu
- 1 year olds
-event 1: present bat+ ball
-event 2: present dog + ball - Through cross-situational activation can infer ball is item that appears twice + bat + dog are the other thing
- known as Hebbian learning : associative learning that co-occurs
Hebbian learning
-semantic learning doesn’t require specialised process
- associative mechanisms are active
- Involve camp between options + suppressing unsupported associations
-Analogy to neural pruning
