The BLARK Matrix and its relation to the language resources situation for the Celtic languages Delyth Prys Language Technologies Unit, Canolfan Bedwyr.

The BLARK Matrix and its relation to the language

resources situation for the Celtic languages

Delyth PrysLanguage Technologies Unit,

Canolfan BedwyrUniversity of Wales, Bangor

The target audience of the BLARK is:

• researchers (both in academia and in industry), and educators

• policy makers and organisations

Welsh Language Technologies1992 Welsh Language Act1993 Terminology Spelling and

Centre Grammar Checker1998 Welsh Assembly Government2000 NAACLT Conference, Limerick2001 e-Welsh Unit founded2003 + Speech Technology

+ Place-name Centre2006 Welsh Language Board IT Strategy2006 e-Welsh>Language Technologies Unit

LR Activities• Terminology Standardization

– 23 projects

• Lexical projects– English/Welsh, Irish/Welsh, digitization

• Language tools– spelling, grammar checkers, hyphenators

• Place-name archives– historical, contemporary

• Speech Technology– speech processing resources for Welsh and Irish

• Computer-assisted language learning– CDs and web-based e.g. BBCs’ LearnWelsh web-site

[Please imagine knight in shining armour galloping across screen carrying BLARK banner]

Celtic Community

• Common linguistic heritage• Common socio-political situation• Some common pan-Celtic institutions• Grant possibilities• Celtic is sexy!

Celtic overview

Irish Scots Gaelic

Manx Welsh Breton Cornish

Official Status

xxx xx xx xx x x

No. of users

xx xx x xxx xx x

Prognosis xx xx xx xxx xx xx

Academic community

xxx xx x xxx xx x

States & Languages

Irish Scots Gaelic

Manx Welsh Breton

Cornish

Member State

Ireland /UK

UK Non EU

UK Fra-nce

UK

Bilingual with

E E E E F E

Lang-uage packs

C + V V V C + V

V V

BLARK Recipe

• Applications (e.g. CALL, access control)

• Modules (e.g. morphological analysis, speech synthesis)

• Language data (e.g. data sets, descriptors)

Available Written Data

Irish Scots Gaelic

Manx Welsh Breton Cornish

Religious xxx xxx xxx xxx xxx xx

Legal xxx xx xx xx x x

Bilingual xxx xxx xxx xxx xxx xxxNewspapers

x x x x x x

Available Spoken Data

Irish Scots Gaelic

Manx Welsh Breton Cornish

Religious xx xxx x xxx x x

Broadcast xxx xx x xxx x x

Other formal

xxx xx xx xxx xx xx

Free dialogue

xxx xx x xxx xx x

Conclusions• BLARK not only good but desirable• Develop Pre-Blark in order to be

realistic• Also look for other solutions in data-

poor environments• Archive and store all possible data• Educate and train• Mobilise wider academic community• Lobby policy makers