Ling 203 1st edition Exam 2 Study Guide Syntactic Development 1 Early Structures and Semantic Bootstrapping 1 Synthesizing a Grammar a Find meaningful units in language i Phonemes ii Phonotactic and syllable constraints iii Word and morpheme boundaries iv Phrases and clauses b Find out what can be a word how words are formed morphology c Find out how words can be put together word order function words how ordering of words relates to meaning d Recall the basics of syntactic structure X bar structure for all phrases Sentence phrase is IP Inflectional phrase 2 What needs to be learned a Constituent structure How do you know which words group together John saw the dog but not John saw the dog b Functional structure negation tense marking c Displacement movement How do constituents move in your language i Your tongue is numb is your tongue numb ii Does movement work the same in all languages No d Reference pronouns and missing elements i Pronouns have variable reference sometimes are missing completely We see a bee Now we see three null bees 3 Why is it hard potentially to learn the right structure a Even after word boundaries can be detected how do learners parse the underlying structure of a given string i a John ate a sandwich b John ate Sometimes objects can be omitted ii a John likes tomatoes b John likes Sometimes objects can t be omitted iii a John grew tomatoes b John grew Sometimes they can be omitted but the meaning is changed b Each of these pairs is associated with a different structure General principles at work help learner distinguish what is correct 4 Early Multi word utterances a Simple Semantic Relations b Early 2 word productions involve very simple semantic relations Relation Agent action Action object Agent object Action location Entity location Possessor possession Entity attribute Demonstrative entity Example mommy push pull car Mommy cookie walk street Baby table Mommy sock Big doggie that doggie Notice word order they use the correct word order for their language c Semantic bootstrapping i How children come to have a knowledge of the structure of their language 1 Pinker I t is plausible that when speaking to infants parents refer to people and physical objects using nouns that they refer to physical actions and changes of state using verbs that they communicate definiteness using determiners and so on ii Structure dependent distributional learning Later non prototypical members of a category e g inanimate subjects can be interpreted because they will share other properties of that category But knowing syntax is more than just knowing categories of words it also entails knowing how words group together to form constituents How can Semantic Bootstrapping help children identify the subject as separate from the predicate of a sentence and learn where the subject goes in the structure iii Semantic Bootstrapping rests on the following assumptions 1 children can perceptually distinguish objects and events and attributes etc and children expect them to be labelled differently by language 2 children possess the formal categories to sort objects vs events i e noun verb as well as formal categories that relate to grammatical functions e g subject predicate 3 once children have identified certain things as subjects or nouns or verbs etc they assume that those things will have other properties of subjects nouns verbs found crosslinguistically A prototypical subject has these properties 1 Agent of action 2 Receives Nominative case 3 Highest NP directly under S IP 4 Animate sentient 4 Concepts are primary they precede language 5 A certain amount of lexical semantic information can be deduced by observing the world 5 Syntactic Bootstrapping a Landau and Gleitman have some objections to the Semantic Bootstrapping approach b Their solution Assuming the following 1 nouns are acquired first 2 knowledge of predicate argument structure N argument V predicate 3 Ability to contrast a surface parse of simple sentences Then the number and syntactic position of nouns can be used to learn verb meanings Subject Verb sleep hit give Subject Verb Object sleep hit give Subject Verb Object Indirect Object sleep hit give What is the evidence in favor of a syntactic bootstrapping type of approach 6 Experiments on verb learning adults Human Simulation paradigm show adults a scene of mother child play with audio turned off make them guess what word was uttered when they hear a beep Gillette et al 1999 experimenters pick some word that occurs 6 times in the mother child conversation children s MLU 2 Each time the word occurs subjects hear a beep At each beep subjects make a guess about what the word is Then give final guess With no other info on average 44 9 correct on nouns as high as 100 15 3 correct on verbs highest is 85 Gillette et al series of different conditions adults again asked to guess mystery verb alphabetical list of nouns cooccurring w verb no video e g Why don t you call Grandma Grandma you avg correct ID of verb 16 5 video clips plus list of nouns 29 syntactic frame nouns and verbs changed to nonsense word order function words left intact e g Why don t you call Grandma Why don t ver gorp telfa 51 7 no video but all linguistic info except the verb e g Why don t you gorp grandma 75 4 video plus all linguistic info except the verb 90 4 Kako 1998 gave adults sentences w nonsense words e g George daxed the ball into the hole and asked them to rate how likely the verb is to mean something about causation perception motion mental state transfer etc found correlation between syntactic frame number and type of arguments and possible likely meanings of the nonsense verb 7 Experiments on verb learning children Fisher et al 1994 children 3 5 years shown videos of 2 characters interacting In each case the action could be described with an intransitive or a transitive verb e g scene rabbit is feeding elephant with a spoon The elephant is gorping eating The rabbit is gorping the elephant feeding child asked what does gorping mean result children tend to give an intransitive V response when given intrans sentence transitive V response when given trans sentence 54 68 of the time give other response 12 23 In no sentence condition children give each type of response equally no bias Naigles et al 1989 given a sentence with a real English verb in a wrong syntactic frame children alterthe meaning of the verb 2 5 year olds The zebra goes the lion children act out the zebra making the lion go