Você está na página 1de 26

Using Developmental Science to Design a Computerized Preschool Language Assessment Grant from IES

Jill de Villiers Athulya Aravind With: coPIs: Roberta Golinkoff, U. Delaware Kathy Hirsh-Pasek, Temple Aquiles Iglesias, Temple Mary Wilson, Laureate learning, VT.

Why?
What is needed is a quick and easy to administer language assessment that can be used
a) as an early warning signal for children with language issues; b) as a way to compare curricular and language interventions; and c) as a research tool.

The assessment will yield profiles in two areas of competency:


(a) vocabulary and word learning strategies and (b) grammar and the use of syntax in comprehension.

These competencies comprise a broad profile of verbal abilities that reflect both
a) the products of learning, or milestones, and b) learning processes, or the use of learning strategies.

Test comparison and the advantages of what we propose

Why do this with a touch screen and computer?

The decisions about what to include based on evidence, age range.

Product Vocabulary Known Nouns Known Verbs Spatial Relational Prepositions Clausal Connectives Grammar Simple and Complex WhQuestions Two-Clause Sentence Complements Prepositional Phrases Past Auxiliary

Process Fast Mapping: Nouns Fast Mapping: Adjectives

Syntactic Bootstrapping Generalizing Novel Active Verb to Passive Voice

Known Nouns
Category: Parts Target Phonological Foil Leg Conceptual Foil Branch Category: Superordinate Target Phonological Foil Tray Conceptual Foil Beverage (Milk)

Leaf Thematic Foil Tree (with no leaves)

Toy Thematic Foil Children

Difficulty: this was a compromise with reviewers who thought it was important. We think its a ridiculous goal to find 4 items that have the right properties.

Known Verbs
Cleaning/Drawing Singing/Waving Clapping/Drumming

NB: maybe equally impossible, but verbs are a bit more culturally neutral, and we have The huge advantage of animation. But can children take in two things at once?

Spatial Relational Prepositions

We are testing a small number of contrasts in later developing prepositions: in front, behind, between, above, below. See later preposition phrases test too.

Clausal Connectives
Temporal Connectives: Target:
Grandma cleaned while the baby slept

Causal Connectives:
Target: The mother held the babys hand in case he fell down. Foil: The mother held the babys hand because he fell down.

Foil:
Grandma cleaned until the baby slept
These connectives are important in early reading, and very little is known about their development: CHILDES searches reveal interesting errors.

Fast Mapping Nouns


Type 1 2 Fast Mapping Basic Level Instructions "Where is the blick" Target Green snake-like bubble blower Foils Bubbles, Cheese, Keys Shoes, Bird, PVC Pipe (novel distractor) Doll, Toy truck, Blocks

Extension Basic "Can you find another Blue snake-like Level blick" bubble blower
Fast Mapping Superordinate Category Extension Superordinate Category "Where is the dax" Purple squishy, puffy object

Toy airplane, "Can you find another Red squishy, puffy Rubber duck, dax" object Shaking barbell (novel distractor

Fast Mapping Adjectives


Trial Type Ostension

Types of Trials Fast Mapping


1. Within-basic (dog another dog) 2. Across-basic (but within superordinate) (dog cat) 3. Across-superordinate (dog truck)

Look, this is a blishish cup!

Target Object with target property (e.g.,blue leopard print cup)

Foil 1

Foil 2

Extension
Across-basic (but within superordinate) (dog cat) Across-superordinate (dog truck) Novel object (dog unknown object)
Extension Can you show me another blickish one? Different object with target property (e.g., red leopard print plate) Object with same unique property as foil 1 (e.g.,red striped plate) Same object as object used in ostension demonstration (e.g., red cup) Fast Mapping Can you show me a blickish one? Object with target property (e.g., green leopard print cup) Object with unique property (e.g., green striped cup) Green novel object

Based on work by e.g. Waxman, we are testing when children can generalize a property word to new items: fast mapping and generalization.

Wh-questions
Tom and I have been working for several years to develop a product with Laureate Learning called Question Quest. It trains all the different levels of simple (one clause) wh-questions, and we tested a big group of children this summer on the 80 items to see what seemed promising to distinguish these ages. The items were e.g.: Who/what is on the bed? (lexical diffs between who/what) Who pushed the baker/who did the baker push? (subject/object) What did the baby eat/what did the baby eat with? (Preposition stranding) How/why did the boy wash the dog? (lexical contrast of harder wh) Who ate what? (paired exhaustive)

Wh-Questions

What is the baby eating with? (What is the baby eating?)

Who is pushing the baker? (Who is the baker pushing?)

Wh-Questions

Who is playing what?

Why is the boy filling the washtub? (How is the boy filling the washtub?)

Pilot Data for Wh-Question Modules


0.9
0.8 0.7

A 0.6 v R g 0.5 e s C 0.4 p o o r 0.3 n r s e 0.2 e c t 0.1


0 3 4 5

Who/What What/With What Subject/Object Exhaustive Sets How/Why

Age Group

We are now doing two things: finding the BEST items and testing a wider sample of SES

Complements
We know that one of the most sensitive changes from 3-5 is in mastery of complementation. In the grant, we posited that the best test would be a contrast between: Whinfinitival complement a) What did the man say to bring? (where what was brought did not match what was said: saying was unfulfilled, not false) Wh tensed complement b) What did the man say he brought? (where what was brought did not match what was said: saying was false). In our BU paper, we will show that with the same verb say, and stories, children aged 3 to 5 treat a) and b) quite differently: i) ii) They more often answer b) with reality i.e. what was brought. They answer a yes/no version c) with yes, but variably on d):

c) Did the man say what to bring? d) Did the man say what he brought? iii) They answer the medial question of c) wrt the top verb. But with d), the bottom verb

Issue
Together with Tom and Peter and Helen Tager-Flusberg. Laureate also made a new training module on Theory of Mind containing these very contrasts.The advantage was that the stories were all enacted in animation, with direct speech events, reducing the need for so much narration. The answer is just pointed at, not spoken. BUT, we found pilot children doing equally well on infinitivals and tensed! Is it perhaps because the disparities between what was said and what was done were not encoded in speech, OR, because both kinds of speech acts occurred in the tensed event, perhaps children were not really answering the final question? e.g. they might have been answering
What did the mother say to Amy to do?

Instead of
What did the mother say Amy was doing?

OR, is it that direct speech is much easier than indirect?

Two-Clause Complements
Pilot Testing Scheme
Group 1 Infinitival Complement Clauses Direct Speech No verbal encoding of alternate action Direct Speech No verbal encoding of alternate action Group 2 Direct Speech Alternate Action encoded Direct Speech Alternate Action encoded Group 3 Group 4

Indirect Speech

Finite Complement Clauses

Indirect Speech
Instructions, Alternate action, Report

Indirect Speech
Instructions, Report, Alternate action

Two Clause Complements

Group 2: Direct Speech Infinitival

Group 2: Direct Speech Finite

Two Clause Complements

Group 3: Indirect Speech Infinitival

Group 3: Indirect Speech Finite

Two Clause Complements

Group 4: Indirect Speech Infinitival

Group 4: Indirect Speech Finite

Prepositional phrases
Most studies have been of the semantic contrasts in prepositions. Some work has been done on parsing and attachment (e.g. Trueswell: put the frog on the blanket in the cup) We wanted to see if children had the syntax mastered and what errors might occur with first single then double prep phrases.

Prepositional Phrases
Single-PP Recursive PPs

Adjunct PPs

Past Auxiliary

Who was crying? We chose this because it is a dialect neutral piece of morphology about tense, found to be useful in production on the DELV. But can we make it work in comprehension? Opinions!

Syntactic Bootstrapping
Design
Actions are presented with audio (e.g., Look! The boy is wezzling!) (4s) Actions are presented with prompt (e.g., Touch where the boy is wezzling!) (4s)

Action presented without audio to allow the child enough time to respond (10 seconds)

Generalizing Novel Verb to Passive


Verb
Glorping

Action
One actor bends another actor sideways from behind The girl is glorping the boy

Passive Question
Which one got glorped?

Type
Reversible action

Choices: boy/girl Target answer: boy Which one got chazzed?


Choices: Boy/ball Target answer: ball

Chazzing

A boy is moving a basketball between his hands in a back and forth motion The boy is chazzing the ball

Nonreversible action

Você também pode gostar