Spelling and Grammar Correction for Danish in SCARRIE

Spelling and Grammar Correction for Danish in SCARRIE Patrizia Paggio Center for Sprogteknologi Copenhagen (DK) patrizia@cst, ku. dk Abstract This paper reports on work carried out to de- velop a spelling and grammar corrector for Dan- ish, addressing in particular the issue of how a form of shallow parsing is combined with er- ror detection and correction for the treatment of context-dependent spelling errors. The syn- tactic grammar for Danish used by the system has been developed with the aim of dealing with the most frequent error types found in a parallel corpus of unedited and proofread texts specif- ically collected by the project's end users. By focussing on certain grammatical constructions and certain error types, it has been possible to exploit the linguistic 'intelligence' provided by syntactic parsing and yet keep the system robust and efficient. The system described is thus superior to other existing spelling checkers for Danish in its ability to deal with context- dependent errors. 1 Introduction In her much-quoted and still relevant review of technologies for automatic word correction (Ku- kich, 1992), Kukich observes that "research in context-dependent spelling correction is in its infancy" (p. 429), and that the task of treating context-dependent errors is still an elusive one due to the complexity of the linguistic knowl- edge often necessary to analyse the context in sufficient depth to find and correct such er- rors. But progress in parsing technology and the growing speed of computers seem to have made the task less of a chimera. The '90s have in fact seen a renewed interest in gram- mar checking, and proposals have been made for systems covering English (Bernth, 1997) and other languages such as Italian (Bolioli et al., 1992), Spanish and Greek (Bustamante and Ldon, 1996), Czech (Holan et al., 1997) and Swedish (Hein, 1998). This paper describes the prototype of a spelling and grammar corrector for Danish which combines traditional spelling checking functionalities with the ability to carry out com- pound analysis and to detect and correct cer- tain types of context-dependent spelling errors (hereafter simply "grammar errors"). Gram- mar correction is carried out by parsing the text, making use of feature overriding and error weights to accommodate the errors. Although a full parse of each sentence is attempted, the grammar has been developed with the aim of dealing only with the most frequent error types found in a parallel corpus of unedited and proof- read texts specifically collected by the project's end users. By focussing on certain grammati- cal constructions and certain error types, it has been possible to exploit the linguistic 'intelli- gence' provided by syntactic parsing and yet keep the system robust and efficient. The sys- tem described is thus superior to other existing spelling checkers for Danish in its ability to deal with certain types of grammar errors. We begin by giving an overview of the sys- tem's components in Section 2. In Section 3 we describe the error types we want to deal with: Section 4 gives an overview …

Content truncated to 500 words