C-WEP―Rich Annotated Collection of Writing Errors by Professionals

C-WEP—Rich Annotated Collection of Writing Errors by Professionals Cerstin Mahlow Institut für Deutsche Sprache mahlow@ids-mannheim.de Abstract This paper presents C-WEP, the Collection of Writing Errors by Professionals Writers of German. It currently consists of 245 sentences with grammatical errors. All sentences are taken from published texts. All authors are professional writers with high skill levels with respect to German, the genres, and the topics. The purpose of this collection is to provide seeds for more sophisticated writing support tools as only a very small proportion of those errors can be detected by state-of-the-art checkers. C-WEP is annotated on various levels and freely available. Keywords:error collection, multi-layer annotation, grammar, German 1. Introduction Current state-of-the-art checkers for grammar and spelling are aimed to support experienced authors writing in their native language. For German, however, only a very small proportion of errors in texts produced by professional authors can be detected. Most often, these authors introduce errors by revising and editing texts, we almost find no spelling errors or typos. With this paper, we close the gap on error collections by pro- fessional writers of German. We focus on grammar issues. C-WEP, the Collection of Writing Errors by Professionals Writers of German provides seed information for the devel- opment of more sophisticated writing support tools beyond spell checking. It currently consists of 245 sentences with grammatical errors.1 All sentences are taken from published texts. Published text is understood in a broad sense and includes advertisements, formal e-mail messages, late drafts of conference papers, etc. Most of the errors were found in well-respected German and Swiss newspapers, which have a copy-editing stage. The texts are written by various authors and cover various topics; however, all authors are professional writers with high skill levels with respect to German, the genres, and the topics. Almost none of the errors in C-WEP are detected by the state- of-the-art grammar checkers in standard word processors (we used the Microsoft grammar checker (Heidorn, 2000) as used in MS Word and Hunspell2 as used in Apple TextEdit or OpenOffice). The sentences in C-WEP only contain a small number of typos, most errors are “grammatical errors.” However, both checkers mark a high number of “spelling issues,” which are listed in table 1. Hunspell and the Microsoft grammar checker correctly find 5 spelling issues each (one is a name). None of the grammar issues are detected by Hunspell. The Microsoft checker correctly identifies 7 instances of repeated words and also provides the correct suggestion (deleting one instance). However, this is only a very small portion of the errors. We first present the state of the art in error annotation in various fields in section 2. and then report on C-WEP in detail in section 3. In section 4. we look at use case for C-WEP. 1 All examples used in this paper come from C-WEP. 2 http://hunspell.sourceforge.net/ 2. Related Work on Annotating Errors 2.1. What to Annotate Most of the available error corpora focus on texts either produced by learners writing in …

Content truncated to 500 words