Study Guides (238,599)
Canada (115,242)
LING 355 (8)
k (1)

LANGUAGE ACQUISITION Midterm 1 Study Prep.docx

23 Pages
Unlock Document

McGill University
LING 355

INTRODUCTION TO FIRST LANGUAGE ACQUISITION Developmental Sequence: -Common to all cultures -Can be viewed as a series of ‘evolving’ grammars, i.e. G G1..2G n -G 1 Initial state grammar -G n End/final state grammar -G T Target grammar -G 2 G n-1= Interlanguage grammar 5-7 months: babbling 12-18 months: one-word stage 18 months: two-word stage; 50 word vocab 2-3 years: adult-like sentences 3-5 years: most difficult aspects of G acquired T 7 years: Ability to learn language begins to decline Learnability Issue: -How do children learn language? -Different theories, but which ones are consistent w/ acquisition facts? Empiricism/Behaviourism -Dominated study of child language until 1960s -Language treated as a behaviour, not a mental phenomenon -L believed to be learned from the environment by a process of habit formation -stimulus-response theory; children are mimicking input, then are reinforced or corrected. Once forms mastered, children generalize them, extending what they know by analogy PROBLEMS: -Assumes positive reinforcement w/ grammatical sentences only -BUT parents seem to treat grammatical & ungrammatical sentences alike (understand and express approval equally for both) -Parents actually care more about TV of children’s utterances, eg “Momma isn’t a boy, he’s a girl” “That’s right” -BUT children ignore corrections -BUT children’s linguistic experience differs: -some parents ask clarification questions for ill-formed Ss, some for well-formed, so how can kids figure out what correlates w/ grammaticality in their situation? -Assumes language is taught 1 -BUT parents themselves aren’t aware of the majority of grammatical rules, so how can they teach them? -Children usually ignore instructions -Children don’t just repeat things, they use language creatively, eg. “who deaded my kitty cat?”; “Don’t giggle me” Constructivism -Children learn language from linguistic input ALONE, by INFERENCE, using DOMAIN-GENERAL learning mechanisms -Children extract relevant generalizations from input, THEN entertain various logically possible hypotheses to learn language PROBLEMS: -Input often degenerate (contains pauses, stuttering, slips of tongue, errors, background noise, etc.) -SOME evidence suggests their input isn’t degenerate: **STUDY** examining speech of 15 mothers with their kids show that “motherese” is remarkably well-formed (99.44%) -Motherese is potentially helpful, esp. for non-syntactic components of grammar, but NOT NECESSARY: some cultures don’t even speak to children until they can produce multiword speech -Motherese is also syntactically oversimplified. Sentences are much shorter than IRL -Embedding and conjunctions extremely rare -Plato/Chomsky’s poverty of stimulus argument: certain properties of language are too abstract, subtle, and complex to be acquired from input alone if input is degenerate/syntactically oversimplified (output is more complex than input) → must be some form of innate linguistic capacity (UG & LAD - both needed) -Whenever environmental factors are insufficient to predict a developmental fact, biologists hypothesize a genetic factor -Two inborn components allow humans to acquire language: UG & LAD (genetically predisposed to learn a human language) -What to children need to acquire? Grammar + Lexicon -Based on input, children “choose” relevant components of the target grammar from UG, using LAD -UG constrains the kinds of hypotheses children can form about their language & prevents them from acquiring a “rogue grammar” -This is why interlanguage is still always consistent with some other natural language(s), i.e. children’s ‘errors’ are sanctioned by UG -Children may also make performance errors - mistakes that result from their limited 2 processing capacities TYPES OF EVIDENCE AVAILABLE TO CHILDREN: -POSITIVE EVIDENCE (crucial!) - they hear a string, so they know it’s acceptable -DIRECT NEGATIVE EVIDENCE (not available/effective)- explicit evidence that a certain string isn’t acceptable “You can’t say “chase mice cats!” -Children aren’t systematically corrected (parents care more about TV), and when they are, the kids ignore the parents -Adults understand and approve of ~same amount of grammatical vs ungrammatical sentences -**STUDY** 40 mothers and their kids - no significant correlation between grammaticality of child’s utterance & mother’s approval; in fact mothers of 2-year olds tended to repeat ungrammatical sentences MORE often than grammatical. These researchers thought repetition might help kids establish which were ungrammatical BUT this doesn’t work b/c there’s no consistency in parental responses! -Conclusion: Repetitions = noisy feedback (wouldn’t allow children to properly separate ungrammatical from grammatical sentences b/c parents don’t react consistently) -INDIRECT NEGATIVE EVIDENCE (may play a role) - if a string isn’t encountered, it must not be grammatical -BUT fact that a certain string has never been encountered may be purely coincidental; we would need an elaborate mechanism that would tell us at what point we can conclude that the string we haven’t encountered is unacceptable -Theories that incorporate some notions of probabilistic/statistical learning may provide a solution (later we’ll look at the variational learning model, which is one theory) CONSTRUCTIVISM VS. NATIVISM: THE ROLE OF THE INPUT -Constructivists think L1A is done on basis of input alone -Nativists believe input + UG -Lately constructivists have developed various statistical models of LA, where it’s claimed children learn language based on statistical regularities using domain-general mechanisms BUT statistical learning can explain acquisition of a very small set of data Why Nativism? -Constructivists haven’t shown how exactly abstract linguistic properties that aren’t easily extractible from the input can be acquired based on input alone -Eg. THAT-TRANCE EFFECT 3 -Can’t move subject out of a CP whose head is occupied by THAT eg.Who did Harry say [CP the stranger mistrusted _t_?] *Who did Harry say [CP that mistrusted the stranger] -Even young children know this constraint, but they’re never told sentences like *are ungrammatical -Nativists assume this knowledge is innate, but constructivists have no explanation -Constructivists assume language learning is based on linguistic input alone, so it predicts that children should produce ONLY utterances that reflect the input -HOWEVER, children’s interlanguage is often distinct from the target language, although compatible with other human languages -Constructivists predict that stages of acquisition as well as type of learned constructions should vary substantially depending on target language -HOWEVER, all children of all cultures go through same developmental stages, and the types of constructions that they build have limited variability -MODULAR VIEW: language is independent of general intelligence. It constitutes a separate module within human cognition -Distinguishes between language acquisition vs. learning -Acquisition: subconscious, automatic; involves language-specific cognitive module like UG & LAD; linked to procedural memory -Learning: conscious and more effortful; involves a general problem-solving mechanism; linked to declarative memory -Constructivists think L1A is based on input using inference alone, so children should entertain all logically possible hypotheses -BUT they don’t entertain linguistically implausible hypotheses, even if they’re logically impossible -e.g. Subject-Aux inversion: For yes-no questions, two logically possible hypotheses: front first auxiliary, or front main auxiliary -Children don’t hear enough examples to witness a case where the two hypotheses yield diff. results, and based on simple sentences with one aux, can’t decide between the two -If hierarchical rule weren’t innately known, we’d expect some children to acquire the 4 incorrect rule, but they never do, so it must be innate. -Nativists think L1A is done based on input using innate knowledge, so children entertain only UG-compatible hypotheses -**STUDY** 7 year old children can’t learn a rule like “drop the first four words of a sentence” b/c children can’t learn structure-independent rules, although they’re logically sound -Conclusion: to the extent that mechanisms postulated by constructivists & proponents of statistical learning allow for learning of structure-independent rules, they’re very diff from rules humans use SPECIES-SPECIFIC DEBATE -Nativists claim both UG and LAD are species-specific; only humans are predisposed to learn language -Doesn’t mean animals don’t have their own systems of communication, but these differ qualitatively from human language; not creative in ways human language is -Lack limitless expressive power & usually consist of a limited range of non- syntactically structured vocab tokens -Evidence for innate knowledge of certain basic language features: -All languages consist of: Cs and Vs; Ns and Vbs; Subj & Obj -Biologist Eric Lenneger defined a list of characterists typical of innate behaviours: -Maturationally controlled, emerging when critically needed (2-3 years = coherent use of language; this is when kids start playing together) -Don’t appear as a result of conscious decision -Don’t appear due to an external trigger -Are relatively unaffected by direct teaching/intesive practice (kids ignore parents’ corrections) -Follow a regular sequence of “milestones” in development -Generally observe a critical period for their acquisition (Critical Period Hypothesis: There’s a small window of time for L1A to be natively acquired - first 5 years) -Genie (not exposed to speech til 13): grammar all messed up, though can make sense, e.g. “Man motorcycle have” -Chelsea (not exposed to speech until 31): No sense at all “The small a the hat” -Conclusion: UG necessary but NOT SUFFICIENT RESEARCH METHODS -The majority of research on language acquisition focuses on: -children’s early utterances -order in which they emerge -the kinds of errors they contain 5 -Two complementary methods of collecting production data: -Naturalistic observation -Experimental testing -NATURALISTIC APPROACH -typically longitudinal (examine language development over an extended period of time) -Two types: Diary studies and Spontaneous speech recordings Diary Studies: Investigator (us. parent) observes and records child’s linguistic development during some (long) period of time Spontaneous speech recordings: child’s spontaneous speech is videotaped/recorded, to give samples of him interacting with his caregivers. Sessions usually about an hour at a time, spread over a long period of time. Then detailed transcripts are made & analyzed -Many transcripts available on CHILDES (Child Language Data Exchange System); contains thousands of hours of recordings from more than 20 different languages Advantages of Naturalistic Approach: -Data collected in a natural (for child) situation -Provides detailed reports of language development with lg number of examples of a wide range of linguistic phenomena -Allow to build up a clear picture of the child’s actual linguistic development Disadvantages of Naturalistic Approach: -Takes a long time -Difficulties in transcribing/interpreting data -Only a small sample of utterances at any given time -Spontaneous production doesn’t necessarily reflect the child’s linguistic competence since we can’t target particular, non-frequent structures -EXPERIMENTAL TESTING -Researchers create specially designed tasks to elicit linguistic activity relevant to the phenomenon they wish to study -These results are compared with adults’ performances on the same task -The child’s performance is then used to formulate hypotheses about the type of grammatical system acquired at that point in time (note: can’t prove a hypothesis, can only disprove) -Typically cross-sectional - across groups of kids at diff. ages (typically same #/group) -If enough subjects are selected for each group, presumably typical behaviour is observed 6 -We can then make inferences about the change of behaviour over time without actually observing any one child changing its behaviour -ADVANTAGES: -Allow researchers to target a specific structure/property of UG, and to work with many children -DISADVANTAGES: -Can’t directly observe progression of acq process over time; -Hard to design a good experiment -Children’s performance may be influenced by external factors like shyness, inability to understand procedure, inattention, etc. -Three types: -Imitation -Production -Comprehension -ELICITED IMITATION -Children’s ability to repeat a certain structure is a good indication of how well they’ve mastered it; children can’t repeat things that their grammar doesn’t allow for -Embedded sentences: learn to acquire easiest ones first. S, DO, OP, IO, GEN (this order) -ELICITED PRODUCTION -NOTE: Children’s comprehension is often more advanced than production, so prod. studies by themselves offer an overtly conservative picture of linguistic development -Experimenter presents the child with a situation that calls for a particular type of words or utterance -**STUDY** wug test (This is a wug. Now there are two of them. There are two ______) -Past-tense formation (This is a man who can RICK. He is RICKING. Yesterday he ____) -Question formation (Ask the puppet if the dog is smiling) -COMPREHENSION STUDIES -Picture matching task, TV judgement task, act out task -Picture matching task: children are asked to choose (point to) picture that matches the sentences they hear (the dog pushes the cow VS the cow pushes the dog) -TV judgement task: children are asked to judge the truth of statements made about particular pictures, video or act-out situations presented to them (is every ball on the 7 box?) -Act-out task: children are supplied with toys and asked to act out the meaning of a sentence (the truck was bumped by the car) EXPERIMENTAL PARADIGMS FOR TESTING INFANTS (VERY YOUNG) -HAS (High Amplitude Sucking) Procedure -Head Turn Preference Procedure -Preferential Looking Paradigm HAS: -Under 3 months of age -Infant’s heart rate and sucking rate change when he/she is exposed to new stimulus -Special equipment to measure these things -Three phases: 1) Baseline Phase: measure sucking rate w/out any stimulation to get each infant’s baseline 2) Habituation Phase: stimuli presented via loudspeaker. Infants eventually become habituated 3) Experimental Phase: infants presented w/ new stimuli, while control group continues to receive same stimuli as in habituation HEAD TURN PREFERENCE PROCEDURE: -6-12 months -Blue light blinks at front until child looks at it -THEN red light on R or L starts to blink -When child looks at red light, hears speech sample -Sound stops if child looks away for >2 seconds -When sound stops, blue light blinks again. Repeat. -Allows researchers to determine if a child has a preference for some speech samples over others (native vs non-native; grammatical vs ungrammatical; frequently vs infrequently-occurring words; words with stress on first syllable vs words with stress on second syllable PREFERENTIAL LOOKING PROCEDURE -6 months to 3 years -Children look at visual stimuli that match the audio they hear -Dependent measure: looking time to the left vs. to the right 8 PHONOLOGICAL DEVELOPMENT Perceptual System: -Children are born with a perceptual system that is specially designed for listening to speech -Newborns respond differently to sounds than adults -Prefer mother’s voice -Can discriminate their native language from a foreign language -**STUDY** using HAS -Habituation Phase: French infants heard Russian utterances -Experimental Phase: control group kept hearing Russian while experimental group was exposed to French -Results: Experimental group detected change (sucked more vigourously) HOW? -Infants rely on utterances’ prosodic and suprasegmental information (intonation, rhythm) to discriminate between languages -When this info is removed (L is played backwards), infants can’t tell between French and Russian -Can tell even before born: -**STUDY** -Familiar rhyme was read by mothers for 4 weeks; caused a decrease in heart rate in 37-week-old foetus -Can also distinguish between 2 diff foreign languages -**STUDY** -Four-day-old infants born into a French-speaking enviro can distinguish between English and Italian HOW? -Infants classify languages on the basis of their global rhythmic patterns -Helps them discriminate languages from different rhythmic classes RHYTHMIC SYSTEMS: 1)Stress-timed languages: Dutch, English, Russian, Swedish -interval between Vs is long & irreg, b/c of greater variability in syllable structure 2)Syllable-timed languages: Italian, French, Greek, Spanish -interval between Vs is shorter & more regular b/c fewer types of syllables 3)Mora-timed languages: Japanese, Tamil -distance between Vs is even shorter than in syllable-timed languages Phonemic contrast: children select sounds instantiated in the input from the entire set of 9 sounds that are encoded in the UG. Build phonemic inventory of target language -Set of universal sounds in finite -Each language chooses the different number of phonemes from that set -Language can have 20-60 phonemes -English has ~40 Italian 33 Perception of phonemes: -Perception of contrast between consonants is categorical; adults presented with synthetic speech can easily discriminate between sounds belonging to two distinct phonemes. However, they have trouble discriminating between two instances of [b] etc. -Adults: very good at contrasts in their language; very bad at contrasts in other languages -One month old infants are amazing at it -**STUDY** -Habituation phase: Infants were presented with serious of identical [ba] sounds -Experimental phase: Infants in experimental group were presented with [pa] -Results: infants could perceive diff between the two -Infant’s Perception: can discriminate Cs in terms of voicing, place, and manner of articulation; can also discriminate between phonemes in unfamiliar languages -EVIDENCE: **STUDY** English infants (6-8 months) can discriminate between Hindi retroflex and dental stops and between voiced and voiceless aspirated stops -Also can tell between glottalized velar and uvular voiceless stops -6-8 month old Japanese infants can discriminate between English [ra] and [la] Losing Sensitivity: -With experience, ability to discriminate unfamiliar phonemic contrasts rapidly declines -Only sensitivity to contrast valid Cs in the native language is maintained -6-8 months - can discriminate non-native contrasts; 8-10 less successful; 10-12 can’t -By age of 12 months, children behave like adults Vowel Perception: -perceived more ‘continuously’ (less categorically) -Children set on repertoire of native Vs before Cs (prob. acquire Vs first) Decline in Sensitivity - good or bad? -Good! Children attend only to sounds w/ phonemic value in their language = distinguish meanings -Preps children to learn the lexicon -Restricts speech space and minimises changes of making errors 10 The Functional Reorganisation Hypothesis: -Loss of sensitivity may not be permanent -System for detecting foreign contrasts may be reactivated in young children (4-6ish) Perceptual Assimilation Model -Adult L2ers tend to filter unfamiliar speech sounds through their native phonemic system -4 scenarios for assimilating non-native sounds, depending on phonetic and acoustic properties 1) Non-native phonemes resemble two diff native phonemes - prob. assimilated into these phonemes 2)Non-native phonemes resemble a single phoneme - will struggle with discrimination 3)Non-native sounds resemble a single native category, but with quite distinct acoustic properties - harder than 1, easier than 2 4)Non-native sounds don’t resemble any native category - discrimination fails -Some studies indicate that infants perceive and organise speech in terms of syllable- like units -EVIDENCE: Infants can detect an increase in # of syllables but not in number of segments -Newborns detect a new syllable when the change concerns the vowel but not consonant (build phonemic representations in terms of nucleus of syllable - carries acoustic info of length, intensity and height) -Suggests it may be nucleus that fxns as universal unit of representation in 1st stages of LA (could be that infants use syllable structure to organize speech) Discriminating words: -Harder as word length increases -Russian linguist N. Shvachkin developed a technique for studying chlidren’s ability to distinguish among words that are minimal pairs -Picture pointing task (coat vs goat) - can also use made up words, real words, or mixture -Results: children can make distinctions between words long before they can produce them in speech -By age 2 or earlier, children can disti
More Less

Related notes for LING 355

Log In


Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.