2013.05.21 studentenplannen spoorzone presentaties rick & alexandra
Automated Assessment of 4-Skills in English (and Other ... presentaties/CIN… · Automated...
Transcript of Automated Assessment of 4-Skills in English (and Other ... presentaties/CIN… · Automated...
Automated Assessment of 4-Skills
in English (and Other Languages)
for Placement and Vocational Purposes
Ryan Downey
Pearson
Katrin Windsor
Pearson
CINOP workshop
24 May 2012
The problem• Large volume of students requiring English skills
– HBO (from hbo-raad.nl; estimated 2011)
• 102.525 intake first year students
• 423.776 total enrollment
– MBO (from mboraad.nl; estimated 2009)
• 58.000 graduating students
• 230.000 total enrollment• 230.000 total enrollment
• Assessing language skills:
– time-consuming
– resource intensive
– managing subjectivity / training raters
2
A solution
• Solution: Automated test delivery and scoring– objective, standardized, time-efficient testing with immediate results
– used now for “passive” language skills such as listening and reading
• Ideal solution: Automated assessment of all 4 major language skills: – Reading, Writing, Listening, Speaking
• Technology allows innovative task design– Integrative tasks = more data in less time
– Tasks relevant to target language use domain• Listen to a story and retell it
• Read and summarize an article for a colleague
• Read and respond to an email
• Testing efficiency returns time to the teacher
3
What is meant by “completely” automated?
Testing Type Responses
collected by…
Scoring
Direct Human Human judges
4
- Speech recognition technology
- Natural language processing
- Latent Semantic Analysis
Semi-directComputer
(recording device)Human judges
Automated Computer Computer
What it is /How it works
Student Student Responses are
Scoring system scores the test Student
accesses the Versant Testing
System by computer
Student responds to
the test questions or
prompts
Responses are sent to the automated
scoring system
scores the test and returns
scores to teachers via
web reporting tool
50-110 minutes for 4-skills tests 1-2 minutes for most tests
Outline
1. Introduction
2. The Versant English Placement Test
3. The Versant Professional Test3. The Versant Professional Test
4. Automated language assessment technology
5. Discussion
6
Scoring structurePart Group Name # of Items
PresentedScores Produced
A Passage Reading 2 Pronunciation, Fluency
B Repeats 16 Pronunciation, Fluency, Grammar
C Sentence Builds 10 Pronunciation, Fluency, Grammar
D Conversations 12 Listening Comprehension
8
E Typing 1 N/A, but…
F Sentence Completion 20 Reading Comprehension
G Dictation 16 Listening Comprehension
H Passage Reconstruction 3 Reading Comprehension, Writing
I Summary & Opinion 1 Reading, Writing (Opinion, Content)
Automated Scoring Technology:Speaking
Content Manner
Vocabulary Sentence Mastery Pronunciation Fluency
w1 w2 w3 w4 w5 w6 75-90 Words/Min
p p pppp p p p p p pp ppp pp p p p p p 5.8 Phones/Sec
waveform
Automated Scoring Technology:Speaking
10
Phoneme & Word Alignment
spectrum
segmentation
words
AccuracyFluencyPronunciation
Automated Scoring Technology:Speaking
11
Performance Comparison
5.502 seconds
Learner
3.026 seconds
Native speaker
Automated Scoring Technology:Writing
• “Latent Semantic Analysis” knows that…
“Surgery is often performed by a team of doctors.”
“On many occasions, several physicians are involved in
an operation.”an operation.”
= mean basically the same thing, even though they share no words.
• Enables evaluating the content of what is written rather than just matching keywords
12
Automated Scoring Technology:Writing
• Content
– How well did the writer address the topic?
• Voice & Tone
– Word choice – like that in high scoring essays or low
Latent Semantic Analysis and other scoring technologies compare each written
response to a large (n = hundreds) set of reference responses with human-
assigned scores
– Word choice – like that in high scoring essays or low
scoring essays?
• Organization
– Word and sentence flow – do the words and
sentences naturally follow each other?
– Coherence – does each sentence logically follow the
next? Does each sentence contribute to the essay as
a whole?
• Grammatical range & accuracy
– Grammar, word usage, punctuation, spelling, …
Speaking Listening Reading Writing
Overall
RepeatsRead Alouds
Sentence Builds
Conversations DictationSentence CompletionTyping PassageReconstruction
Summary and Opinion
Outline
1. Introduction
2. The Versant English Placement Test
3. The Versant Professional Test3. The Versant Professional Test
4. Automated language assessment technology
5. Discussion
16
You have successfully entered your TIN and are now ready to take Versant Professional English Test.
MENU
You are now ready to take the Versant Professional English Test.
The Versant Professional English Test has 13 parts. The test takes approximately 100 minutes. Click ‘Start’ to start the test.
Task Items Presented Scores Produced
Passage Reading 2
Repeats 16
Questions 20
Sentence Builds 10
Story Retells 3 SpeakingStory Retells 3 Speaking
Response Selection 16 Listening Comprehension
Conversations 12
Passage Comprehension 9 Listening Comprehension
Sentence Completion 20
Dictation 16
Passage Reconstruction 4
Email Writing 2 Grammar, Vocabulary, Voice &
Tone, Organization, Reading
Comprehension
Versant Professional
5-page score report (each module)
• Scores & subscores
• Explanation of subskills and
capabilities
• Detailed can-do statements
(linked to CEFR)
• Recommendations for • Recommendations for
improvement
• Relationship to other
tests/scales (CEFR, TOEFL, etc.)
21
Outline
1. Introduction
2. The Versant English Placement Test
3. The Versant Professional Test3. The Versant Professional Test
4. Automated language assessment technology
5. Discussion
22
Technology
• Patented, researched, and validated over last 15 years
• Publications describing technology and validation process are publicly
available (most on website)
• Technology certified, recognized, or approved by external (objective) • Technology certified, recognized, or approved by external (objective)
agencies
– 2012 SIIA CODiE Award finalist
– 2011 Tech and Learning Award of Excellence
– 2009 ICC Certificate of Quality Assurance
– 2009 CRM Excellence Award From "Customer Interaction Solutions“
23
Tests
ENGLISH
SPANISH
DUTCH
C2
C1
B2
CEFR
Testing
System
24
ARABIC
DUTCH
CHINESE
B1
A2
A1
FRENCH
Test validation process is rigorous and includes aligning to CEFR
Technology in use•Teacher/TA certification
•ELL placement
•Oral reading fluency
•Student assessment
•Study abroad
•Recruiting selection
•Training placement
•Leadership programs
•Promotion
CorporateSchools and Universities
•ESL, EFL language skills
•Oral reading fluency
•Assessment and scoring service for publishers
•Training
•Employment screening
•Language certification
•Immigrants screening
Private Language
Schools and Publishers
Government
25
Technology in useSpoken Tests Sample Users
SpanishUS Government: Department of Homeland Security and US Dept of
Defense; UC, Davis
Dutch Dutch Government: civic integration and naturalization exams
Arabic US Defense Language Institute: Arabic training program
English
Singapore Ministry of Education, Texas school districts, Stanford,
UConn, U Washington, Rhode Island Dept. of Education, Navitas,
Laureate, Hong Kong University of Science & Technology, FIFA
AT&T, Dell, IBM, P&G, Accenture, CitiBank, LG, Convergys, Amazon,
Deloitte, 3M, Bell Deloitte, 3M, Bell
Academic EnglishStudents for college and university entrance; recognized by >1,300
institutions
Junior EnglishGovernments of Singapore, S Korea, Chile
Yoons English School
Aviation EnglishFAA, Boeing, Emirates Airlines, Belgian Government, Indian
Government, Air Asia
French Canadian Government: Department of Education, Bell Canada
(Chinese) Peking University and Consortium
Outline
1. Introduction
2. The Versant English Placement Test
3. The Versant Professional Test3. The Versant Professional Test
4. Automated language assessment technology
5. Discussion
27
DiscussionTest Duration Skills required Target educational level
Placement Test 50 mins Basic English skills Entrance for higher
professional education
(HBO)
Professional Test Speaking module = 25 mins
Writing module = 45 mins
Combined = 110 mins
Basic English skills,
with work-related
content
Final exam, secondary
vocational education
(MBO)
28
CEFR Score
A1 20-29
A2 30-43
B1 44-53
B2 54-66
C1 67-76
C2 77-80
CEFR Score
A1 26-35
A2 36-46
B1 47-57
B2 58-68
C1 69-78
C2 79-80
Wri
tin
g
Sp
ea
kin
g
Discussion
Benefits Caveats
Accurate, consistent, objective School administrators and ICT prepare for
initial set-up
Time-efficient and cost-effective May be most appropriate for candidates
A2/B1/B2
Scalable and practical (any time, any place)
Automated solution for 4-skills testing
29
Scores returned immediately, available for
batch download by teacher
Administrators/teachers can view scores
online
Can be delivered on any supported
computer with Internet connection
(including WiFi)
Bedankt, en… Take a Test!
• Check the Pearson table to take your own test
– Sampler
– Demo
• Contact [email protected] for more • Contact [email protected] for more
information
• Check website @ www.kt.pearsonassessments.com
30