[Corpora-List] POS-tagging learner corpora

Rayson, Paul rayson at exchange.lancs.ac.uk
Fri Jan 27 12:23:01 CET 2006


There were a couple of messages last July on corpora about POS tagging
for learner data (and spoken language). Please see my message at:


You should also look at the work of John Milton:



Resource-Rich Web-Based Feedback: helping learners become independent

To appear in Hyland, K & Hyland, F. (Eds) Feedback on ESL Writing.
Cambridge University Press.



Dr. Paul Rayson

Director of UCREL

Computing Department, Infolab21, South Drive, Lancaster University,
Lancaster, LA1 4WA, UK.

Web: http://www.comp.lancs.ac.uk/computing/users/paul/

Tel: +44 1524 510357 Fax: +44 1524 510492


From: owner-corpora at lists.uib.no [mailto:owner-corpora at lists.uib.no] On
Behalf Of Xiaotian Guo
Sent: 26 January 2006 23:32
To: corpora at uib.no
Subject: [Corpora-List] POS-tagging learner corpora

Dear All

Has anybody reviewed or assessed the rate of accuracy of POS-tagging
(not error tagging) to learner corpora, or simply reported the use of
POS-tagging in their learner corpora studies? Due to the unexpected
non-native features of learner English, POS tagging systems designed for
NSs will need some modification or adjustment when used on learner
corpora. I wonder whether somebody has already worked out some better
solutions than simply using POS tagging software for general purposes. I
am aware that this thorny issue of learner corpora study was addressed
somewhat before (for example, Granger in Granger 2002) and there were
few studies based on POS tagging to learner corpora, but in this fast
growing area there might be some progress which has already largely
outdated the problem (POS-tagging to learner corpora) (fingers crossed).

All the best

Xiaotian Guo

PhD candidate

The Department of English

The University of Birmingham

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.uib.no/public/corpora-archive/attachments/20060127/52c96f36/attachment.html

More information about the Corpora-archive mailing list