23 - Grammar Flashcards
How are words related to one another?
Both semantically and co-occurring in speech
Why does “priming” work?
Because words are related to each other semantically and co-occur in speech, so hearing one word makes it easier to think of other similar words.
What 3 things predict word learning?
1. Word length and frequency
This predicts the age of acquisition of a word
2. Contextual diversity
- The amount of different contexts they hear a certain word in*
- If a word coexists with a lot of other known words it predicts word learning (cross-situational learning)*
3. Knowing words associated with a word
Predicts the learning of that word (distributional theories)
What is a “network graph”
A visual representation of how works are connected to one another
What does this network graph show us?
Left is a typical talker, right is a late talker (less words than is expected for their age)
Late talkers connections are more sparse, it shows that words are less predictive of other words and there are less opportunities for priming
- Important to build word connections for language*
- Association between words helps us build our vocabulary*
What impact does age have on the acquisition of a word when the person reaches adulthood?
Words learned earlier are:
Rated as more familiar
Recognised faster (lexical decision)
Named faster
What changes happen to the effects of learning a word earlier in life change when you control for frequency, length, familiarity and imaginability?
Nothing, words are still rated as more familiar, recognised faster and named faster.
Shows the importance of learning words as a young child going into adulthood
What is “grammar”?
How we put words together
What are “open class” words?
Adjective, adverb, verbs and nouns
They carry most of the content and it is easy to add new words
They are also the first words infants learn and the easiest for second language learners
What are “closed class” words?
Consists of: pronouns, auxillary verbs, conjunction, prepositions and determiners
These carry most of the grammar and it is difficult to add new words
These are hardest for second language learners
Is it easier for second language learners to learn “open class” or “closed class” words?
Open class words are easier for second language learners
Do infants learn closed class or open class words first?
Open class words
What do “most sentences” consist of? (3 things)
A verb (an action doing word)
A subject (agent of the action)
The object not involved in the action (the patience)
The most common sentence structures around the world are
Subject, object verb
And Subject, verb, object
Which two sentence structures does english follow?
Subject, verb and object
Object verb and subject
Name the 3 parts of the sentence shown here
What is “passive language” in english?
Object, verb, subject
- Not used in practical reports*
- E.g. “mistakes were made”*
Which part of the sentence is referred to as the “kings of the sentence”
The verb
Why are the verbs important?
They determine what the arguments are and where they go
- E.g. “like” tells us we need a subject and an object*
- Someone to do the liking of something*
What are the 5 theories that attempt to explain how we learn grammar?
Mimicry
Negative evidence
Subtle negative evidence
Bootstrapping
Statistical learning
Is language limited?
No, it is generative and infinite
We are constantly coming up with new words and sentences that we have never heard or said before
What is meant by ‘mimicry’?
Children copy what they hear and learn language this way
What is meant by “children overgeneralise in language”?
Children utter false utterances (they generalise what they hear too far and state false statements)
They make up new phrases or use grammar that isn’t used
E.g. “do you want to disappear my head”
What is the problem with ‘mimicry’?
Children need to generalise to utterances outside of what they have heard
- There are problems with meaning, novel sentences, generalisation*
- Not all verbs work the same way and therefore sometimes it wont make sense*
- E.g.*
- Jess is mooping jack the ball -> he is mooping the frisbee to her (works)*
- She is pilking the data to the screen -> she is pilking the screen the data (does not work)*
What is “negative evidence” when learning grammar?
Adults provide some sort of negative evidence when the childs sentences are not grammatical
How prolific is “negative evidence” when children are learning grammar?
Not at all, observational studies have shown that it is rare for a parent to fix a child’s grammar
Even rarer to correct syntax
What is the no-negative-evidence problem?
Children don’t seem to understand when you try to correct them using negative evidence
Parents rarely correct their child’s grammar
Describe “subtle negative evidence” in regards to learning grammar?
Rather than fully correct a child, parents subtly repeat the sentence a child makes back in a more grammatical way
E.g. yesterday I go to the playground - (parent repeats) “oh you went to the playground yesterday? Was it fun?”
Is the “subtle negative evidence argument” accurate?
More intuitively correct and is done more, but children rarely notice and fix their sentences
It also fails to explain novel words
Parents rephrase more when children make errors (first line is when sentences have been rephrased back when the sentence was correct, second column is when the sentence was rephrased back when a grammatical mistake was made)
Do parents rephrase children when children don’t make grammatical errors
Yes, but not as much as when children do make grammatical errors
Is the “subtle-negative-evidence” argument deemed accurate?
Why?
Not really,
it still succumbs to the no-negative-evidence problem
Also fails to account for why children tend to overgeneralise in such common ways
What is the “semantic bootstrapping” grammar learning theory?
There are many subtle differences in meaning in verbs (and sentences) that are used in different ways
- (An extension of mimicry, we learn something in the meaning of a sentence and then transfer this to many sentences)*
- E.g. if a verb refers to the subject of a possession, it can only go a certain way (e.g. subject, verb, object)*
What is the issue with the “semantic bootstrapping” theory?
No linguist has come up with any explanation of what the subtle differences between verbs are
- E.g. “he carried the girl the ball” doesn’t sound right*
- However “he kicked the girl the ball” does*
What is the “statistical learning” theory for how we learn grammar?
People track the frequency of what you have heard (the more you hear the more you build up a general idea of what works and what does not)
The more examples of a verb they see but not the other way around means they begin to understand that the way the sentence should be phrased
This is called negative implicit evidence
- E.g. “he brought his son a bike - (his son (indirect object) and bike (direct object)*
- They sent him a postcard (him (indirect object) and postcard (direct object)*
- Very rarely do we see DO - IO and so children pick up on that*
Define “negative implicit evidence”
Where you track the frequency of the way a verb is used and then imply what the proper way of using the verb is
What is the evidence for the statistical learning theory?
Children overgenerate verbs a lot, adults do not as much potentially but have heard the common verbs more
For less common verbs, it is more likely that adults and children get them incorrect
- Implies that we learn from the number of examples we hear of the word*
- Theory can solve why children use “overgeneralisation”*
Does “implicit negative evidence” in statistical learning solve our questions regarding how we learn grammar?
No
- It doesn’t explain how children know what arguments are in the first place or how they know which out of an infinite amount of semantic features are plausible*
- There is evidence that ‘bootstrapping’ can help, but statistical learning is playing a major role*
What are “regular grammars” in grammar rules?
Regular grammars state that there is a chain of rules where one word leads to another
- Words within a sentence are produced based on the previous word*
- We can apply rules based on the previous words to generate sentences*
- Can be thought of as a ‘word chain grammar’*
Explain the grammar chain shown here
S = subject
Then you can select any “the A” (happy, boy. dog)
Happy calls on itself (called a recursive)
This then leads to B, which leads to C etc.
The rules follow along the chain and are linked one after the other
What is the problem with “regular grammars”?
They cannot explain “long-distance dependencies”
This is the “argument from simplicity”
- In order to learn long-distance dependencies, they would have to be incredibly complicated*
- Language seems to instead have a hierarchical phase structure*
- The mistakes children make cannot be explained by regular grammars*
What are “long-distance dependencies”?
Where a word a long way down the sentence could be dependent on another word far away from it
Explain “context-free grammars”
Trees of hierarchical phase structures called “pastries”
Basic idea: there are a bunch of phrases nested within each other (rather than words)
Phrases are clusters of words that map onto larger units of meaning
Regular grammars cannot grasp this structure
This explains the types of mistakes children make
Context-free grammar argues that…
Grammar is written with production rules based on “phrases” (or chunks of information)
- E.g. subject is always followed by a “noun-phrase” or a “verb-phrase”*
- This allows the clusters of phrases rather than chains*
How does context-free grammar explain long-distance dependencies?
You move up and across the second part of the phrase rather than having to memorise everything along the way
What is “substitution” in grammar?
When we substitute whole phrases rather than individual words
What evidence supports “context-free grammar” and phrase learning?
Children make most mistakes at the boundaries of phrases
They seem to perceive phrases as a whole rather than individual words
We use “substitution” - We can substitute whole phrases rather than individual words
How does “context-free grammar” link back to the “poverty of the stimulus” argument? (or why do some people believe that grammatical rules are innate - poverty of the stimulus argument)
Children show behaviour and understanding which is not possible without certain knowledge, and understanding of this knowledge is not possible without first knowing it
Children would not be able to learn a specific behaviour without learning a strict grammatical rule
However, it is impossible to have learnt “G” just because of “D” in the example above
Therefore, many believe that knowledge about grammatical rules is innate
Do we perceive grammar in a hierarchical structure or a chain-like manner?
Hierarchical structure
This accounts for long-distance dependencies, grammatical errors that are made and the substitution of whole phrases rather than words
Also supported by “poverty of stimulus” learning.
