Subordination, subject-verb agreement and the CEFR: An analysis of linguistic complexity and accuracy in written L2 English and L3 French in relation to the Common European Framework of Reference for Languages

Abstract: This investigation concerns written output produced by Swedish learners of English as a second language (L2) and French as a third language (L3). The analysed texts were rated according to the Common European Framework of Reference for Languages (CEFR), which is a widespread and language-independent document commonly used to assess communicative competence based on six main proficiency levels (A1, A2, B1, B2, C1 and C2). More specifically, the present study aims at comparing texts placed on different CEFR levels within the same language as well as samples from the same CEFR levels across different languages, in relation to linguistic complexity (i.e. the ability to use elaborate structures) and accuracy (i.e. correctness) (Housen & Kuiken, 2009). The targeted linguistic phenomena are subject-verb agreement and subordination, as it is hypothesised that the former is an indicator of accuracy, whereas the latter can be used to measure complexity (Norris & Ortega, 2009). The data set consisted of email messages and narrative texts in English (N=24) and French (N=24) written by learners who were between 15 and 19 years old. All the data were manually coded for T-units (i.e. a measure referring to a main clause and all its associated dependent clauses) and clause-types by means of the CLAN toolbox (MacWhinney, 2000). This enabled a measuring of the complexity of the samples, as well as an examination of the kinds of subordinate clauses produced by the informants. The current investigation also contains an accuracy analysis based on the proportion of correct present tense verb forms in the texts. According to the present findings, the analysed samples are relatively homogenous with regard to linguistic complexity, as the selected complexity measures only yielded statistically significant differences between the English B1 and the French A2 samples. However, the accuracy analysis showed that the percentage of correctly conjugated verbs in the texts generally increased with higher CEFR levels.