Transcript ppt
Introduction
“a technique that enables the computer to
encode complex grammatical knowledge
such as humans use to assemble sentences,
recognize errors and make corrections”
CALL
Hardcoded instructions
Pre-configured assessment items
Pre-specified mapping between learner response
and error category
ICALL
Adaptive instructions
Dynamic assessment item generation
Automated mapping using NLP techniques
CALL
Teacher centric rather than learner centric
Explosion in learner responses Explicit learner response
to error mapping not feasible
Highly constrained learner responses
Not sufficient for self-learning
ICALL
Abstract away from specific string entered by learner to
more general classes of properties
Generation of feedback, learner modeling, instructional
sequencing can be based on small number of abstract
properties
NLP systems are not robust
Feedback
Student
Response
Instruction
Feedback Designer
NLP
Learner
Modeling
LM
Instructional
Sequencing
Tutoring System
In form-focused ICALL, the interaction workflow
proceeds as follows:
In response to some prompt or question by the tutor,
the student enters a sentence
The sentence is forwarded to the parser for analysis
of syntax validity
The sentence passes the syntax validity check or
The parser will fail in case the learner response is illformed
The error is classified into generic error classes
The error handler generate appropriate feedback to
be presented to the learner.
Tutoring subsystem moderation of parser
output
withholding information
alerting the student that something is wrong
highlighting the location of errors
classifying the errors
correction or hint on errors
showing the structural analysis of the sentence
assign score against the learner response
revisiting instruction sequence dynamically
update student model
Vocabulary learning
Diagnosis of learner error
Correcting learner errors
Language learning exercise generation
Lexical Hypothesis: Speaking is essentially
lexicon driven (Levelt, 1989)
Grammatical form can only be activated once a
lexical item has been chosen
Lexical items need to have a rich internal structure
▪ Meaning, syntactic, morphological and phonological
properties
Learning types
Intentional
Incidental
University of Groningen
“An intelligent word-based language learning assistant” – Nerbonne and Dokter
The task
Given a learner provided response, mark the
errors
Need to parse learner response
Erroneous sentences are ill-formed
Parsers expects the input sentence to be well-
formed.
Parsers should show tolerance to error
Overgenerate and rank
Imposing ranking constraints on grammatical
rule violation
Mal-rules to allow parsing with specific errors
Parse fitting
Generate fragmented parse trees and try to fit
them together
Do not allow analysis of completely arbitrary
ungrammatical input
Issues with English Language Learners (ELL)
Concentration of errors are much higher than
native learners.
Using proofreading tools (e.g. MS Word)?
▪ Designed for native users
▪ Not very robust against foreign learner errors
▪ Targeted errors are small subset of learner errors
Error correction in machine translation output
Data driven approach
Classification approach
▪ Whether an article will be followed by a noun?
▪ Whether an article appearing before noun is correct?
▪ What would be the correct article?
Language modelling approach
▪ Errors will most likely be located in the area with low LM
score
A hybrid system?
Influence of L1
No equivalent for a feature.
▪ Japanese and Russians face difficulty in learning articles
Languages sharing features
▪ German and French learners find it easy to learn English
article systems
Transfer problem
▪ Positive transfer
▪ Negative transfer
Spelling errors
Article usage
Preposition usage
Collocation errors
Negative transfer
Correspondence between prepositions of any two
languages is many-to-many
घर पर at home, सड़क पर on the road
Prepositions imposes semantic variation
in the summer vs. during the summer
Argument of predicates
Nomilalization (removal of hazard vs remove the hazard)
Type of argument (book in the box vs book on the table)
Verb alteration (They loaded hay on the wagon vs They
loaded the wagon with hay)
Phrasal Verbs (verb+particle)
Non-compositional
give vs give up
Particles can move (put the switch off)
Phrasal verbs often used with prepositions (give in
to their demands)
Idioms (in the house vs on the house)
This is Kalyani in the house with all your
favourite tunes
“in the house” venue
All the drinks were on the house
“On the house” free
Indefinite article depends on countability of
nouns
Countable vs uncountable
▪ The price of a Spring Fest hoody is Rs. 700.
▪ The price of freedom is constant vigilance.
Syntactic
Some uncountable nouns can take indefinite article
when attached with a preposition phrase (a
knowledge of English)
Discourse
World knowledge
Learner Error Corpora
Grammatical Error Detection
Grammatical Error Correction
Evaluation of Error Detection/Correction
System