how do children learn words lec 4 Flashcards
step 1 = detect words in speech: speech preferences: TD:
infants are born with an innate preference for speech over non-speech and particularly infant directed speech (IDS).
infant directed speech (IDS)
unconsciously when in front of a baby - higher in pitch and tone variation, smile more, widen eyes, raise eyebrows, language simpler - helps infant learn quicker. infants prefer IDS to ADS - more neurons firing when listening to IDS
IDS evolutionary?
IDS may be evolutionary as the higher pitch may be to sooth babies when the babies are near to them but not holding them. other animals also do it.
how does IDS support lang learning?
IDS supports lang learning by facilitating word segmentation and providing redundant cues to grammatical structure.
how does the infant’s auditory perception diminish?
listening to speech narrows infant’s auditory perception so they become attuned to their native language - when they are born they can discriminate between different sounds from different languages but as they hear more sounds from their own language they lose their ability to discriminate between the different languages. infant’s ability to discriminate non-native sounds diminishes by 10 months, while discriminatory perception for native sounds is retained
step 2 = identify meaning
identification of a novel word’s meaning involves isolating the intended referent and forming a stable word-coherent pairing
fast mapping: social cues
upon hearing a novel word, TD infants spontaneously refer to a speaker’s gaze, pointing or affect to determine their attentional state/focus. consequently they tend to map labels to the object of the speaker’s attention
fast mapping: lexical heuristics
word-referent mapping in TD infants is also guided by linguistic assumptions . TD infants apply the ‘mutual exclusivity’ principle to assign novel words to unfamiliar objects when viewed alongside objects with known names.
cross situational mapping
some word learning situations are highly ambiguous, making accurate fast mapping impossible. children disambiguate meaning over multiple exposures by tracking cross situational stats
step 3: store meaning: mapping = learning??
can get everything right in a task while mapping but they don’t remember it and it isn’t learning. two-year-olds often forget new words just 5 minutes after identifying their meanings. working out what a word means ‘in the moment’ is not sufficient for long term learning
store meaning
long-term word learning occurs through building and strengthening associations between words and objects over time. children’s retention can be increased by providing feedback when meanings are first being identified - redirect their attention towards it after mapping - highlights the target and decreasing attention to other things
dynamic association model: referent selection
…explains the relationship between referent selection and retention in terms of two interacting processes occurring on different timescales - draws on attentional mechanisms to increase activation for correct referents while decreasing activation for incorrect competitors and operate in ‘situation time’. referent selection doesn’t necessitate learning - as demonstrated in experiments with children
dynamic association model: retention
retention draws on associative learning mechanisms and occurs over ‘developmental time’. retention involves strengthening correct word-referent connections and pruning incorrect connections. when a word-referent pairing becomes sufficiently strong, hearing the word activates the correct concept without external support. retention doesn’t require a solution to referential ambiguity - a word can initially link to multiple referents
step 4 = extend meaning
flexible use of language also involves generalising words across contexts to previously unseen and unlabeled referents - need to find out the correct rules. basic nouns refer to categories, rather than specific exemplars
step 4 = extend meaning - shape bias
learning object names selectively tunes children’s attention to shape. mapping a word to multiple category members leads children to recognise their similarity in structure. once children have aprox 50 nouns in their expressive vocab, they infer the general rule that noun-referent relations are constrained by shape - apply this general rule when learning novel words. by 24 months, TD children demonstrate a shape-bias when generalising new words. the shape bias is driven by TD children’s sensitivity to word-shape co-occurrences during infancy - driven by non-linguistic abilities - categorise, abstraction of prototypes.
