Godkänd i svenska? : Bedömning och analys av gymnasieelevers texter

• Main Author: Östlund-Stjärnegårdh, Eva
• Format: Doctoral Thesis
• Language: Swedish
• Published: Uppsala universitet, Institutionen för nordiska språk 2002
• Subjects: Scandinavian languages - general; Upper-secondary school; school texts; grades; assessment; Swedish; national tests; writing; manipulated texts; sentence structure; holistic scoring; Nordiska språk - allmänt; Gymnasieskola; elevtext; betyg; bedömning; svenska; nationella prov; skrivande; manipulerade texter; meningsbyggnad; helhetsbedömning; Scandinavian languages; Nordiska språk
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-1967; http://nbn-resolving.de/urn:isbn:91-506-1555-6

This thesis deals with the assessment of school texts by students in Swedish upper-secondary school or in the corresponding adult education and concentrates on what differs between the grades Pass and Fail. The 60 texts used in the survey come from the 1997 archives of the national test construction group. A questionnaire to teachers asks what criteria are most important when distinguishing between Pass and Fail. The five criteria pointed out are Holistic scoring, Relevant content, The connecting thought, Sentence structure and How the student has followed the instructions. The most salient result regarding grades is the difference between grades from the students’ own teachers and from the independent assessors. Ten texts have received a Fail from the student’s own teacher, but as many as 35 get an average Fail from three assessors. There is variation in assessment, but 18 Fail texts and 18 Pass texts get a unanimous grade. Quantitative analyses of the 60 texts show a definite correlation between grade and number of words. However, the Fail-groups among narrative, expository and argumentative texts contain both the shortest and the longest texts. School texts are longer now than thirty years ago, especially texts with low grades. Sentences and words have become shorter. Coherence is investigated by a method of reference cohesion. No clear difference between Pass and Fail texts can be found, but between types of essay topics. Also important is the organization of the text and its paragraphs, a factor which separates Fail and Pass texts. Various aspects of sentence structure show better results in the Pass texts. The thesis is concluded with a commentary on the demands of the last compulsory course in Swedish. The needed level is argued to be the ability to write for an unknown reader.