Computational Linguistics for Low-Resource Languages October 26, - - PowerPoint PPT Presentation

▶
computational linguistics for low resource languages
SMART_READER_LITE
LIVE PREVIEW

Computational Linguistics for Low-Resource Languages October 26, - - PowerPoint PPT Presentation

Dec 18, 2023 251 likes •461 views

Computational Linguistics for Low-Resource Languages October 26, 2011 Alexis Palmer Wednesday, October 26, 2011 CL for LRL Questions of interest What is a low-resource language? (aka less-studied language, resource-poor language,

slide-1
SLIDE 1

Computational Linguistics for Low-Resource Languages

October 26, 2011 Alexis Palmer

Wednesday, October 26, 2011

slide-2
SLIDE 2

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

CL for LRL

2

Questions of interest

  • What is a low-resource language? (aka less-studied

language, resource-poor language, minority language, less-privileged language, ...)

  • What are the challenges posed by LRL, and what are

the major approaches to addressing these challenges?

Wednesday, October 26, 2011

slide-3
SLIDE 3

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

CL for LRL

3

Questions of interest

  • What is a low-resource language? (aka less-studied

language, resource-poor language, minority language, less-privileged language, ...)

  • What are the challenges posed by LRL, and what are

the major approaches to addressing these challenges?

Some major themes

  • Role of labeled/annotated data
  • Role of expert/linguistic knowledge (anno & beyond)
  • Single language vs. “universal” solutions
  • Resource creation: does it always make sense? how

can it be done most efficiently?

Wednesday, October 26, 2011

slide-4
SLIDE 4

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

And another question... Why do we care? ✦ practical reasons ✦ theoretical reasons

4 Wednesday, October 26, 2011

slide-5
SLIDE 5

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Course requirements & organization ✦ reading & participation: read papers prior to relevant meeting, discuss ✦ presentation: 30-45 minute presentation of selected paper(s), discussion after ✦ additional: 1 lg. resource case study, 2 critical reviews (1-2 pages each) ✦ term paper: original research or in-depth survey and analysis (15-20 pages) ✦ optional: guest post(s) on Cyberling blog

5 Wednesday, October 26, 2011

slide-6
SLIDE 6

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Language endangerment

6

Language loss

  • Current estimated rate of language death: one every 2

weeks (Crystal 2000)

  • Half of world’s languages extinct by end this century
  • UNESCO Endangered Languages Programme (under

auspices of Section on Intangible Cultural Heritage)

  • UN General Assembly: 2008 was International Year of

Languages

UNESCO endangerment status

  • six levels: safe, unsafe (or vulnerable), definitively

endangered, severely endangered, critically endangered

  • criteria go beyond number of speakers

Wednesday, October 26, 2011

slide-7
SLIDE 7

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Evaluating language endangerment

7

Criteria to consider (UNESCO 2003)

  • Intergenerational language transmission
  • Absolute number of speakers
  • Proportion of speakers within the total population
  • Trends in existing language domains
  • Response to new domains and media
  • Materials for language education and literacy
  • Governmental and institutional attitudes and policies,

including official status and use

  • Community members’ attitudes toward their own

language

  • Amount and quality of documentation

Wednesday, October 26, 2011

slide-8
SLIDE 8

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Globally, 2488 languages in danger

8

source: UNESCO Interactive Atlas of the World’s Languages in Danger, 2009 edition

Wednesday, October 26, 2011

slide-9
SLIDE 9

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

528 ʻseverely endangeredʼ languages

9

source: UNESCO Interactive Atlas of the World’s Languages in Danger, 2009 edition

Wednesday, October 26, 2011

slide-10
SLIDE 10

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Germany: 13 endangered languages

10

source: UNESCO Interactive Atlas of the World’s Languages in Danger, 2009 edition

Wednesday, October 26, 2011

slide-11
SLIDE 11

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Documenting endangered languages

11

The realities

  • Most projects are individual or small-group endeavors

with very small budgets

  • Each project seems to find its own workflow
  • Basic workflow: collection, transcription, translation,

detailed linguistic annotation (NOT a pipeline)

  • Tangible end products: orthographies, grammars,

dictionaries, language teaching and learning materials, collections of stories, websites, etc.

  • Such materials support survival of the language
  • Do they support CL/NLP???

Wednesday, October 26, 2011

slide-12
SLIDE 12

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Uspanteko : 1320 speakers, ʻunsafeʼ status

Uspantán, Quiché Department, Guatemala

12 Wednesday, October 26, 2011

slide-13
SLIDE 13

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Scenario: IGT for Uspanteko

Corpus of texts in the Mayan language Uspanteko

Produced by OKMA (Oxlajuuj Keej Maya' Ajtz'iib') 66 texts, mostly oral history, personal experience, and stories Total 284K words of transcribed text, 74K words glossed

IGT-XML: representational format specifically for IGT

13

# texts

# morphemes

train 21 38802 dev 5 16792 test 6 18704

Wednesday, October 26, 2011

slide-14
SLIDE 14

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Types of resources

14

Data

  • primary: audio, video, texts (archiving)
  • machine-readable corpora
  • data with annotations
  • parallel corpora, comparable corpora

Linguistic resources

  • traditional: grammars, dictionaries, word lists
  • WordNet, other ontological resources
  • treebanks, etc.

Tools

  • user-oriented: spell checkers, input systems, etc.
  • for NLP: tokenization, POS tagging, parsing, etc.

Wednesday, October 26, 2011

slide-15
SLIDE 15

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Challenges and approaches

15

Having to do with insufficiency of data

  • create more data?
  • leverage resource-rich languages
  • use semi- or unsupervised methods
  • use rule-based methods
  • ...

Having to do with the nature of the data

  • use linguistic knowledge to seed unsupervised models
  • use linguistic knowledge to adapt models/approaches
  • change the data to look more like familiar languages
  • ...

Wednesday, October 26, 2011

slide-16
SLIDE 16

Topics and scheduling

Wednesday, October 26, 2011

slide-17
SLIDE 17

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Topics

17

Data/resource creation

  • annotation; crowd sourcing; active learning
  • lexicon building
  • “low-level” issues: orthography, character sets/

encoding, spell checkers

POS tagging and morphological analysis

  • unsupervised POS tag induction
  • unsupervised morphological analysis (e.g. Morfessor)
  • morph. by alignment and projection
  • universal POS tag set, universal linguistic ontologies

Wednesday, October 26, 2011

slide-18
SLIDE 18

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Topics

18

Syntactic analysis

  • grammar engineering [guest lecture]
  • grammar induction
  • parse projection; evaluation; treebanking

Other topics

  • machine translation; crisis MT
  • cross-lingual approaches to information retrieval,

word-sense disambiguation, etc.

  • leveraging resource-rich languages

Linguistic universals and typology

  • inducing language classifications; linguistic universals
  • empirically-driven linguistic typology

Wednesday, October 26, 2011

slide-19
SLIDE 19

Palmer, CoLi, UdS CL4LRL, 26 Oct 2011

Scheduling

19

  • 2 Nov: resource case studies; Bird/Simons [me]
  • 9 Nov: no meeting
  • 16 Nov: guest lecture, Antske Fokkens [grammar

engineering, Grammar Matrix]

For next week:

  • Bird and Abney on building a Universal Corpus
  • Bird and Simons on requirements for good data
  • Language resource case study (1-2 pages)
  • Meet with me to finalize topic and schedule

Wednesday, October 26, 2011