Lecture 12 Flashcards
why we didn’t know a whole lot about individual differences in language development
individual differences got swept under the rug to a degree
they focused on finding linguistic universals: overall patterns that held across cultures and SES
we didn’t know that much about language acquisition: had to characterized usual/typical patterns before looking at variability around those patterns
• If all kids acquire language similarly
(assumption), then studying one kid in depth
was a good method
• So they would pick a kid with clear articulation so you can study what they’re saying (‘jargon’ kids hard to transcribe)
• Advanced/imitative things kid said dismissed
as anomalous
then, child language become interested in more than just syntax
people starting looking at: learning meaning, sound patterns of words
researchers looked at more “jargon-y” kids
deviations from some “perfect universal pattern” became more evident
researchers looked at kids from varied languages and socioeconomic backgrounds: do we see patterns or differences?
referential vs. expressive style
clearly expressed (good single word, highly articulate) language vs. fluent language but not as intelligible utterances (more jargon-y)
segmentation
how they break up words
how they learn things as chunks: is “mrrogers” a single word? or are there other mrs?
early sentences
two word stage: fix noun and then another noun
formulas: pronoun, verb, noun, “me eat”, “Sarah eat.”
kids use slightly diff formulas but using formals is a common pattern across kids
referential vs. expressive
diary study from Nelson
17 kids in a diary study (where you give the parents a diary and have them write down all the things kids say over a period of time)
looked at the child’s first 50 words
divided into two general classes: referential and expressive
referential kids (who were origionally mostly focused on) who are more referential tend to talk about things using words to categorize the environment and expressive kids using language more as a tool for socialization (communicating with other people): the two types have diff hypothesis for what language is for
referential kids
n=10
mostly words for object
went from 1 to 2 word stage
spurt of words near 50 word level
spurt preceded syntactic combos
talked a lot about objects
language used to categorize the world!
problems with Nelson’s distinction conclusion drawn from the referential vs. expressive diary study
1) she was using parental report: parents are overly generous in the things kids are saying, they tend overreport nouns
2) composition of lexicon (mental dictionary) vs. word frequency (there could be words that they know but that they don’t use very much = difficult to detect those words that they know but don’t use often)
- likely to miss low frequency words (didn’t produce words within earshot)
- two kids could look the same, but use that word differently depending on who the kid is
3) criteria to decide type of word kid’s using
- Nelson’s criteria for ‘noun’ mixed form (‘dog’ = noun) and function (saying ‘dog’ to get a doggie-ride = verb)
- just because our understanding of what a noun is what it is, doesn’t mean that that’s the kid’s understanding of that word’s function
expressive kids
n=7
fewer object words, more pronouns and function words
social phrases (go away, stop it)
less clear transition to syntactic combos
no word spurt
using language for socializing and communicating with other people
different ways people characterize variation referential vs. expressive dimension
referential words…….expressive words
code-oriented (how language is used to mean particular things)…….message-oriented
more nouns……more phrases
analytic strategy (break things down into component parts)……..holistic strategy (try to produce overall forms)
noun combos (early word combos noun+noun) or…….pronominal combos (pronoun + noun)
segmentation
frozen phrases?
- what does the kid use as independent words?
- may say “little red wagon,” but to the kid it might be “littleredwagon”(expressive style) is one, solid object (frozen phrases): are they naming all the component parts or is it just one object?
• possibly related to ‘cautious’ vs. ‘risk-taking’ phonology
- cautious: use rule system religiously or avoid words that they can’t say (referential)
• you’re more accurate if you produce shorter words
- risk-taking: ‘sloppy’ phonology, less clear rule system, more likely to use imitated forms (less worried about getting it right) (expressive)
= neither strategy is faster for acquiring vocabulary
early sentences
bloom, lightbown, & hood
observed 4 kids doing 2 word combinations
found that all produced negation (“no socks” because they’re not wearing socks, or “no more beans” i don’t want any more beans) / nonexistence, recurrence (“more beans” because they want beans)
- no + X, no more + X - more + X -pivot word strategy (two word formula = function word (no, no more, more) +content (noun or verb))
different in expressing action and location
function words
fewer of them, they’re closed class
pronoun, preposition, article, negatition
early sentences
eric and peter pivot word strategy for location and possession
action, location, possession w/pivot word strategy
- articule I + X
e. g. “I eat” - location: X + here
e. g. sit here, eat here, sleep here - possession: My + X
e. g. “my Gijoe” - pronominal strategy: have a formula with one fixed value and one that you fill in
early sentences
Kathryn and Gia
combined content words that were not fixed
- “Gia push”, “Kathryn sock”
- nominal strategy: combining two content words (noun+noun)
early sentences
around MLU of 2.5….
…. the patterns converged (nominal and pronominal strategies became one)
maybe these early differences end up coming back to the same place later in language development
how stable are these individual differences?
or do they happen randomly and everyone ends up in the same place eventually
pronominal strategy-using kids end up also using nouns and vice versa
Stability
Nelson experiment
once kids hit 2-word speech:
- the referential kids started with lots of nouns (nominal strategy users)
- expressive kids mixed nouns and pronouns
(pronominal strategy users)
Stability
Starr - 1975
‘object’ kids do object-attribute relations more in 2-word speech
e.g. “Yummy cake”
Stability
Bates 1988
27 kids from 10-28 months: longitudinal study
- kids that were analytic at 13 mos –> advanced grammar @ 28 mos
- holistic style @ 13 months, their level of language development was uninformative as to their performance at 28 months
Pine and Lieven disagree with Bates
11 to 20 months: kids who start out with more frozen phrases at 11 months–> produce more productive word combos at 20 months
more consistent with words breaking down frozen phrases
Peters uniting: different strategies are useful
fission (if you start out doing frozen phrases you can break those down into individual words)
and fusion (if you start out with individual words you can put them together)
both useful!
in adult speech we do a little bit of both: we use chunks of over learned words/phrases and we also use word-word combos
Sources of variation
- the child itself (the nature part: individual differences that would show up reliably regardless of the environment)
- child’s input (the nurture part: if they get more complex input vs. less input that makes a difference)
- type of language being learned (structure of the particular language)
- variety of other people, social situations (most studied is maternal speech, though)
- mom-to-kid: direct, converse, instruct (stable differences across moms, related to child speech)
- directing: fewer nouns, more verbs, and social interaction (vs. routinized naming games)
- refer/describe, request objects: more nouns
sources of variation
variability in learning rate
- first word around 1st birthday to 18 months
- not related to referential/expressive dimension
sources of variation
shyness
scared of talking, of making mistakes
- talk less and w/less complexity k-1st
- receptive abilities (compression) may be fine they may just be shy about production
- may be slower at second language learning
- hard to tell if slow acquirers turn out to be shy: how do you quantify shyness?
sources of variation
cognitive differences
- referential kids more attentive to toys, so they’re better at object categorization
- where expressive kids do more social (puppets) and symbolic (banana phone) play
sources of variation
routinized naming games
Nelson: for referential kids, 28% of first 50 words were body parts (probably from routinized naming)
- influence whether a child will be referential or expressive (word doping)
- no body parts within the first 50 words for expressive kids
sources of variation
expressive-speech kids’ moms
conventional social expressive:
please, bye, oh dear, etc.
sources of variation
different kid style may make….
…different use of input
mom uses a lot of conventional social phrases but you don’t pay attention and you just really like trucks or dolls so you just pay attention when someone talks about trucks or dolls
inclined toward being referential (trucks) or inclined toward expressive (listen more to the conventional social phrases)
sources of variation
Socioeconomic status
BIG SOURCE OF VARIATION
children in families with higher SES have larger vocabularies
why? word doping? the amount of child directed language that the kid gets (not getting as much language input)
not just amount of input but quality of input (variety of the input)
low quality input for 4: all they ever say is “hello baby” but with the same 4 hours they read books or do naming games
most of the kids being studies are above average because they’re in higher SES
even in lower SES more language input and better language input leads to better language outcomes
sources of variation
input languages differ greatly…..
…may lead to different patterns of acquisition
- more vs. less nouns (noun bias in English but not in Mandarin)
- synthetic (e.g. Polish: hard to learn all the case endings) vs. agglutinated (e.g. Turkish: dummy syllables for unlearned) morphology
sources of variation
interaction factors
e.g. child’s preferences and parent’s response
depends on whether the parent indulges the child’s preference or not: so names a lot of objects cause kid likes it or use a lot of social phrases cause they feel they need more work there
maybe child attends to objects a lot and does parent notice and engage in joint attention labeling a lot?
implications to keep in mind when we’re talking about the study of language development
- not just English, which as dull (impoverished) morphology
- expressive kids underrepresented in language studies
- kids and adults use formulas (not just rules)
- don’t talk back to Darth Vader – he’ll get ya!
- Donttalkbackto _____ hellgetya! (frozen form with one variable element)
• contexts tested (toy play) usually elicit object labels
- may misrepresent (underestimate) some kids: not tapping the language resources that they have
• diff cultures place diff values on language behaviors (English has a strong preference to label things)
Theoretical approaches
structuralism vs. functionalism
competence vs. performance
nativism vs. empiricism
Is there an answer to the problem of language acquisition?
we don’t have a perfect, grand unifying theory of language development that physics, biology, etc..
we have a lot of complex data and people tend to focus on one part of the data = if you’re only working on one small area it’s hard to see commonalties across domains
there are different theoretical approaches that drive experimentation
different people think different things constitute evidence: “they can’t produce subject-verb sentence but they can comprehend those sentences: what does that mean?”
structuralism vs. functionalism
- linguistic structure as explanation: what you’re studying is how do children come to form syntactic utterances that match their language community
- behavior (language in –> behavior out)
• don’t necessarily care about linguistic form
competence vs. performance (Chomsky’s distinction)
• competence = knowledge of language rules
- mostly structuralist
• performance = things (memory, attention) that make you mess up
- mostly functionalists - these things are relevant and suggest constraints
nativism vs. empiricism
- language is mostly governed by biology
- language is mostly governed by environment, like any other behavior
interactionist or emergentists
language is a constant recombination of nature and nurture (they interact in complex, dynamic ways)
really old: behaviorist
associate language stimulus with behavioral response
but kids don’t get reinforced/punished by parents
Skinner
research approaches
mid 20th century: linguistic (nativist)
- language acquisition device (module)
- poverty of the stimulus: you couldn’t get the grammar back from the language input kids get, so it must be built into your brain
- mostly about grammar, but some speech is special arguments too: note that reading is also really automatic, yet we don’t seem to have innate reading areas
- recursion (embedding of rule within itself): “ the singer Natalie likes died.” (but you can only get like 3 until meaning is lost - nativists account for it by saying it’s just a performance problem)
- serial processing: processing bit by bit
Interactionist (emergentist)
- Piaget, Bates, McWhinney
- Associated with UCSD
- approach to development generally, not just language
- language has complex structure, but many factors affect it
- social interaction is very important
- performance limitations are relevant
- parallel, not serial, processing
- PDP models: can learn things that look like rules (will over-regularize and can even do embedding)
pivot word strategy
two word formula = function word (no, no more, more) +content (noun or verb)
- no + X,
- no more + X
- more + X
synthetic languages
Polish
some of the morphemes in the language have multiple pieces of meaning attached to them
e.g. one morpheme might indidicate that I’m singular, feminine, and in the accusitve cases (3 pieces of meaning in the same morpheme)
hard for a kid to learn all the case endings in Polish cause it’s computationally more difficult
so Polish kids may look less advanced
agglutinated languages
have several morphemes that each carry a single meaning attached to it
somewhat more simple because there’s a one to one mapping between meaning and morpheme
kids may produce dummy syllables for unlearned words: fill in nonsense syllables
