Existing commercially available spellcheckers for Afrikaans (i.e. PUK/Microsoft Speltoetser; Pharos Speller (and hyphenator); Ispell vir Afrikaans) still don’t comply with the general desiderata for spellcheckers (i.e. user-friendly, technically elegant, and linguistically justified, that gives the user high recall, high precision, and adequate suggestions), especially with regard to their lack in efficiency to cope with the structure and nature of Afrikaans as a semiagglutinative language that allows for productive compound formation. This project under discussion is aimed at the development of an enhanced spellchecker for Afrikaans, which will not only improve in functionality and performance on existing Afrikaans spellcheckers, but also reach the benchmarks set by other comparable projects (e.g. the SCARRIE project; see also Van Huyssteen, 2002). The three main areas of improvement are coverage/recall (i.e. the recognition of valid words as valid, and of errors as errors), flagging/precision (i.e. the correct indication of (potential) errors), and suggestion adequacy (i.e. to make useful suggestions for correcting errors) (Paggio & Underwood, 1995: 6-8). The project commenced in February 2002, and a first version of the new spellchecker will be released in July 2003. However, the project will continue after that, until the benchmark results are reached.
[1]
A. G. Jenkinson.
DIE PROBLEEM VAN FLEKSIE EN AFLEIDING IN AFRIKAANS
,
1993
.
[2]
李幼升,et al.
Ph
,
1989
.
[3]
Wessel Kraaij,et al.
Porter's stemming algorithm for Dutch
,
1994
.
[4]
Chris D. Paice,et al.
Another stemmer
,
1990,
SIGF.
[5]
Martin F. Porter,et al.
An algorithm for suffix stripping
,
1997,
Program.
[6]
T. G. Vosse.
The Word Connection. Grammar-based Spelling Error Correction in Dutch
,
1994
.
[7]
Patrizia Paggio,et al.
Validating the TEMAA LE evaluation methodology: a case study on Danish spelling checkers
,
1998,
Nat. Lang. Eng..