|dc.description.abstract||This thesis looks at how Norwegian learners of English acquire the English laryngeal feature distribution in two types of suffixes; s-endings and the past tense. The past tense forms are distributed almost identically in both Norwegian and English, only differing after sonorants. The s-endings in Norwegian are all voiceless, whereas in English these behave in a parallel way to the past tense.
A cross-sectional study of Northern Norwegian learners in three age-groups, year 9, 11 and 13, were conducted, and the results from this study make up the foundation for the analyses. Two learning algorithms; the Gradual Learning Algorithm (GLA) and the Error Driven Constraint Demotion principle (EDCD) were applied to the data. I argue that the GLA cannot account for the findings, whereas the EDCD can. The GLA’s promotion principle works against its purpose as it triggers constraints to move in a fashion which leads to inter language grammars that are not to be found. The EDCD is found to be the algorithm that best describes the Norwegian learners’ patterns as it makes use of universal constraints that are hidden in the native language of the learners, and so can explain the difference in acquisition rate between certain patterns. At the same time it rejects the principle of constraint promotion and therefore does not run into the same problems as the GLA.||en