LING 2120 Chapter 2- The logical problem

recap: Grammar is a system of

principles, rules, and constraints that govern…

- The production, perception, and patterning of speech sounds

- The formation and interpretation of words and sentences

Some constraints are _____ and others are _______

language specific (parameters)

universal (principles)

General observation

in spite of the complexity exhibited in every language, the vast majority of kids acquire grammar of their native language with seemingly little effort

note

The induction problem

On the basis of linguistic input, learned must come up with a grammar for the lang being acquired

- However, the input doesn’t give learners enough information to work out all the properties of a lang that they eventually come to know

Our knowledge is more complex, abstract, and subtle than would be expected if

a language was acquired from input alone

therefore.. INPUT ALONE IS NOT SUFFICIENT

Every set of input data allows for multiple generalizations

-> induction problem

If there are so many different ways to generalize, there’s no way all people who speak the same

lang induced the same rules. There has to be some other underlying reason we fundamentally know the same rules

generalization

you need to be able to generalize from stimuli you see

it is also at the heart of human lang ability

notes on generalization

- You need to be able to notice the RIGHT properties of the input, otherwise you aren’t able to figure out the principle

- Every set of input allows for multiple generalizations >> the problem of induction

- Generalization is at the heart of human language ability > you can comprehend and produce smth new, and do so in a way that’s consistent w your existing lang system

Every set of input allows for multiple generalizations, hense

the problem of induction

Where do generalizations come from?

- Most agree that learners attain a set of generalizations abt their lang

- The bigger questions are: what is the basis for these generalizations?

What are the cognitive mechanisms involved in the making of these generalizations

Poverty of the stimulus

Input is necessary for acquisition, but is seemingly not sufficient

- Kids’ errors go beyond the input to which they’re exposed, but the errors are principled

(they respect UG constraints)

Ex: yes-no questions

Single auxiliary questions:

The man is in trouble.

Is the man __ in trouble?

What theories can explain this?

Linear order hypothesis- move the first aux verb to the front of the sentence.

linear order hypothesis is important when we have

2 auxiliary verbs

based on this example that b is wrong, why is the linear hypth wrong?

a) (the man who is in custody) is in trouble

b) (the man who is in custody) is in trouble

children never make mistakes like this

Structure based hypothesis

move the aux verb in the MAIN clause to the front of the sentence

This fixes the problem from b)

Syntactic movement relies on ___ while linear order relies on _____

Phrase structure

Counting

Legate & Yang (2002)

in the input to two children, yes-no questions where the second auxiliary had been fronted over the first occurred 0 times in a sample of 29,540 questions (both kids combined).

Poverty of the stimulus:

the input is insufficient for kids to learn all they need to know abt the

grammar (rules and structures) of their lang.

Input cannot explain the things we come to know about language

Innate learning biases guide kids in their analysis of the …

linguistic input

- These biases prevent kids from considering analyses of the input that are incompatible with the grammatical systems of all langs

- Movement rules are sensitive to structure, not linear order

what are Movement rules sensitive to?

sentence structure, not linear order

recap: The induction problem

every set of input data allows for multiple generalizations

recap: Poverty of the stimulus

the input is lacking

negative evidence

kids dont hear words like goed from their parents

Do caregivers tend to correct grammatical errors?

No.

When kids do receive corrections, they often don’t listen to them (in order to apply them, they need to understand which part of the utterance is ill formed and in what way it’s ill formed). Kids often can ‘t repeat what they can’t say spontaneously

NOTE: Children can not repeat something that they have yet acquired.

Caregivers tend to not correct grammatical errors, and instead correct (2)

Semantic overextension- the extension of the meaning of a word based on similar properties> calling any animal a dog

Morphological overregularization- overapplying of a rule to forms which it should not apply to (irregulars) > brought vs bringed

For children to benefit from correction, minimally, they need to understand

which part of an utterance is ill-formed and in what way it is ill-formed.

Many adults talk to kids in a special way (motherese/ child directed speech) characterized by….

long vowels, dramatic intonation, slow speech rate, etc

Do kids all over the world learn to speak on a very similar timeframe?

Yes. However, any cultures do not use a special way of speaking with babies and some cultures hardly speak to babies at all

Most speech is not fluent but halted

summary

Types of evidence that kids use/don’t use:

Kids build their grammar based on POSITIVE EVIDENCE: discovering that a structure is well formed from it’s presence in available data

direct negative evidence- being explicitly told that a structure is wrong isnt effective

Behaviourism- skinner

According to behaviourism.. Children learn language through… (4)

Imitation

Reinforcement

Analogy

Structured input

All of these ways are insufficient. But, we just reviewed evidence that children do not imitate their parents; reinforcement is non-effective; there are many analogies children do not entertain; and the input is not structured

Nativism (Chomsky)

Learning lang isn't something a kid does, its something that happens when the kid is in the appropriate environment

Kids don't only learn their language through meticulous care from adults, reinforcement, etc. > children of immigrants often learn a second language fluently and quickly just by being exposed to it in the streets

UG is a PART of the answer

Kids are born with certain biases about what a language looks like

these biases limit the kid’s hypotheses about what is possible in a lang

Biases kids may be born with (3)

1. Knowledge of types of categories (N, Det, Adj, etc) and operations found in grammars of all langs

2. Knowledge that categories combine in a principled way, resulting in hierarchical representations

3. Movement is sensitive to constituency

T/F: biases that are innate and purely linguistic are assumed to form a part of the UG

true

Factors suggesting a biological basis of lang acquisition (9)

1. The behavior emerges before it’s necessary

2. It’s not the result of a conscious decision

3. The critical period for lang acquisition

4. teaching/intensive practice have little effect

5. Milestone of lang development (kids go through the same stages in the same order, regardless of their native lang. *but not necessarily at the same rate/age)

6. Lang has a regular onset (milestones)

7. Lang occurs regardless of upbringing (*a few exceptions)

8. Similarities in lang acquisition across diff langs

9. Areas of the human brain are specifically devoted to lang (broca’s in the left frontal lobe of most people for production and wernicke’s in temporal lobe for comprehension)

for right handed people lang is found in the…

left hem

does input largely vary across lang learners?

yes!!!

Stages of prelinguistic vocalization (11)

12 wks: cooing (vowel sounds- ooooo)

16 wks: cooing in response to human sounds

20 wks: consonants added to cooing

6 mos: babbling, 1 syllable

8 mos: canonical babbling (repetitive syll sounds ex ba ba, reduplication of syllables (mama, baba), recognizable intonation patterns of native lang

10 mos: attempts to imitate speech sound in native/ surrounding env

Some stages of linguistic development

12 mos (1 year): 1st words, understanding of words and simple commands

18 mos: vocab of ~50 words, single word utterances

24 mos (2 year): 2 word utterances (correct word order)

30 mos: large vocab, simple sentences (telegraphic speech> 2-3 words, mainly content

words), understand most things said to them

3-4 yrs: vocab of at least 1000 words, segmental inventory that’s almost target-like, certain complex stces

In the 2 word stage

word proto-sentence have the correct word order

Summary

- Theres a biological basis to lang and lang acquisition, but the amnt is debated

- There are milestones in lang acquisition

- “Speech perception” considers the earliest stages: perception and production by infants who cant speak (aka prelinguistic development)

Biases that are_______ and _____are assumed

to form part of Universal Grammar (UG

(i) innate and (ii) purely linguistic