Main heading for entire slide show

1

Future Trends in Language Learning Technology
2

• Background books
• The Future of English?
• The Internet and ELT
• The Language Machine

British Council English 2000
3
Published for online use

• Books available paperback or PDF
• http//www.leeds.ac.uk/library/secure/books/eastme
  nt.pdf
• http//www.leeds.ac.uk/library/secure/books/atwell
  .pdf

4
English 2000

• Aimed to forecast where ELT is going
• Survey of practitioners and contributors
• Surveys of relevant IT Internet, Speech And
  Language Technology (SALT)
• Now a Forecasting/Intelligence service

5

• Overview of www technical developments
• LOTS of URLs
• ELT perspective

6

• The Future of English?
• Trends in world languages
• Influences on development of English
• Futurology techniques

British Council English 2000
7
Conclusions

• ELT should be guided by futurology
• ELT industry can be leading-edge provider of
  cultural and knowledge-based products if we
  recognise the potential of English!

8
How to predict the future

• Rely on Visions and Dreams?
• Consult the Oracle of Delphi?
• More scientific / rational approaches

9
Formal futurology / future studies

• British Futurology Society
• Leeds Univ MSc module Future Directions in
  Distributed Multimedia Systems
• Professional futurologists BT, banks,
• Govt bodies British Council, UK Foresight
• Research papers, theses have future plans

10
Strategy 1 extrapolate from past trends

• Plot numeric data over past, extrapolate
• Eg Demographic population-based
• Changes in language use English 2000
• BUT models cant include ALL factors,
• Training data too sparse / insufficient
• wildcards eg Asimov Foundation trilogy

11
How many speak world languages?
12
Decline in international (European) languages
13
English as a second/foreign language L1 lt L2 ltlt
EFL
750 million EFL speakers
370 million L2 speakers
370 million L1 speakers
14
Trends for prediction

• Demography
• World economy
• Technology

• Globalisation
• Cultural flows
• Global inequality

15
Example why is English a global language?

• Trade, political military power, colonialism
• Nation states
• Rise of USA

• Global financial institutions
• Science technology
• IT, the Internet

16
Strategy 2 Plan the future

• Futurology planning
• LeedsMU Future Studies Town Planning
• Decide what SHOULD happen
• and stay in control
• BUT this should still be guided by predictions

17
Examples

• EPSRC, EU research strategy plans used to decide
  what research is funded
• Microsoft XP with speech input/output
• Government initiatives in language learning and
  technology (?)

18
Strategy 3 consult experts

• British Telecom technology calendar
• Choose several X future technology
• Experts consulted on when X should happen
• Scientific Oracle method
• Research papers may give Authors expert
  predictions

19
BT Predictions by Futurologist Ian Pearson

• ? real-time MT for print and voice
• ? voice synthesis to human quality
• ? portable MT device
• ? AI imitating human thought
• ? thought recognition for PC I/O
• ? full brain link to computer

20
BT Predictions by Futurologist Ian Pearson

• 2004 real-time MT for print and voice
• 2005 voice synthesis to human quality
• 2007 portable MT device
• 2018 AI imitating human thought
• 2025 thought recognition for PC I/O
• 2030 full brain link to computer

21
More Predictions

• ? Natural Language for home PCs
• ? TV computer personalities
• ? IT literacy essential for employment
• ? domestic robots get attractive, cuddly
• ? more robots/computers than people in
• developed countries

22
More Predictions

• 1998 NLP for home PCs
• 2000 TV computer personalities
• 2003 IT literacy essential for employment
• 2005 domestic robots get attractive, cuddly
• 2025 more robots/computers than people in
• developed countries

23
BTTJ Millennium Edition

• Timelines of technology past
• and futurology predictions
• BT senior managers give their views
• Outlines of current BT RD projects

24
The Language Machine
Natural Language Processing
Speech Recognition
language generation
Speech and language Technology (SALT)
Speech Synthesis
25
Other SALT components

• OCR, written input with scanners
• Machine Translation, document processing
• Dialogue modelling Call Centres
• Multimodality 500-channel TV presenter
• Maths models Hidden Markov Models
• Language resources Corpora, lexical resources,
  software
• ICAME, ELSNET, ELRA, LDC,

26
What is a Language Machine?

• Includes (some of) above components
• Includes LANGUAGE MODELS (AI)
• Includes expertise from linguistics, computer
  science, (ELT?)
• Contributes to goals of Language Engineering
  research

27
Goals of Language Engineering research

• Computer models of language
• Computerised language resources
• Communication people and computers
• Communication person to person
• (?computer-to-computer communication?)
• Wealth creation

28
SALT uses and users

• Machine Translation for Peace-keeping
• Telephone Call centres
• World Wide Web, Email translation
• Information Retrieval
• Working while Driving
• Search for Extra-Terrestrial Intelligence

29
Interactive Spoken Language Education

• European Union RD project
• Pronunciation tutoring, feedback on target
  phonemes
• Speech recognition and error diagnosis
• Users involved throughout

30
amalgam-tagger_at_comp.leeds.ac.uk

• Part-of-Speech tagger for English email text
  messages
• Auto-replies with your email PoS-tagged,
  according to one of 8 standard PoS-tagging
  schemes.

31
8 rival PoS-tagging schemes
Brown ICE LLC LOB parts POW SEC
UPenn select VB V(montr,imp) VA0 VB verb
M VB VB the AT ART(def) TA ATI
art DD ATI DT text NN N(com,sing) NC
NN noun H NN NN you PPSS PRON(pers)
RC PP2 pron HP PP2 PRP want VB
V(montr,pres) VA0 VB verb M VB VBP to
TO PRTCL(to) PD TO verb I TO
TO protect VB V(montr,infin) VA0 VB verb M
VB VB . . PUNC(per) . . .
. . .  

32
Users of amalgam-tagger

• English language teachers (spot the errors!)
• Artificial Intelligence student exercise
• English linguistic research
• English corpus tagging
• Text compression research
• SETI Search for Extra-Terrestrial Intelligence
• Comparing linguistic analysis models

33
Natural Language Learning and the Search for
Extra-Terrestrial Intelligence

• SETI search for ET signals
• When we find them, what next?
• We have developed AI algorithms for natural
  language learning
• - to tokenise unknown data into characters,
  words, phrases
• http//www.comp.leeds.ac.uk/jre

34
Are you using SALT on your computer? Why not?

• Mini user survey
• 1) Do you use SALT
• All the time / occasionally / never ?
• 2) give me 2 reasons why not
• Lets see if prospective users can guide me!
• (delegates please email me your expert opinions
  on Language Machine)

35
Why ask? user-guided system development

• Atwell et al, User-guided system development in
  Interactive Spoken Language Education in
• Language Engineering journal, Special Issue on
  Best Practice in Spoken Language Dialogue
  Systems.

36
SALT for Language Learning why not?

• Expensive
• Not user-friendly
• Prone to mistakes
• Just not appropriate
• Needs a new approach to working
• Needs training
• ENGLISH TEACHERS DONT KNOW ABOUT SALT - need
  advice they can TRUST

37
Evaluationof Language Machines

• EAGLES Guidelines Functionality, Reliability,
  Usability, Efficiency, Maintainability,
  Portability
• CALICO Journal Technological features,
  Activities, Teacher fit, Learner fit.
• (Shakir 2000) EAGLES for developers, CALICO for
  users

38
Language Machines will be...

• All-pervasive
• Customised to individual users
• As error-prone as humans (!)

• Helpful tools, not rivals - customers will come
  to realise the added value of human language
  professionals

39
A way forward

• More debate about the Language Machine
• Build better forecasting models
• Brand management
• English language industry as a leading-edge
  provider of knowledge-based products
• English linguists needed in IT research!

40
Language Machine for Learning?
What do YOU think?
http//www.britcoun.org/english/ eric_at_comp.leeds.
ac.uk amalgam-tagger_at_comp.leeds.ac.uk