An Error Analysis of Subject-Verb Agreement by Swedish Learners of English : Corpus-Based Study of the Difficulties Surrounding Subject-Verb Agreement for Swedish Students

Sammanfattning: This study investigates subject-verb agreement errors produced by students in upper secondary school in Sweden. The research was conducted in an attempt to understand why students acquiring English as their L2 struggle with learning subject-verb agreement, and to categorize the grammatical environments which cause students to produce errors related to subject-verb agreement. There are two extralinguistic variables at the center of the study: (i) gender and (ii) education level. Error analysis was the method of choice for this study, and the study used essays produced by students from the Uppsala Learner of English Corpus to identify and classify subject-verb agreement errors according to linguistic and extralinguistic variables. The linguistic variables were separated into four clusters: (i) subject type, (ii) subject number, (iii) distance and (iv) verb. The variables in each cluster were: (i) common noun, pronoun, existential clause; (ii) singular, plural; (iii) subject-postmodifier, other distance, no distance; (iv) auxiliary verb, regular, irregular. In total, 117 texts were included in the analysis with 57 texts being written by female students, and 60 from male students, to ensure that gender could be used as a variable. Previous research suggests that female students are slightly better academically than male students and suggests that subject-verb agreement is an issue across all levels of education. These findings were supported by this study, as the normalized frequencies for gender show that not only do female students perform better than male students, but they also improve across level while male students get worse. The grammatical environments that proved to be most difficult for the participants to produce correctly were common nouns 57.7% (n=75), plural subjects 54.6% (n=71), irregular verbs 59.2% (n=77) and distance 30.8% (n=40).