Valutare le competenze orali in italiano L2: variazione longitudinale e situazionale in apprendenti a livello avanzato

This dissertation deals with the problem of assessing oral proficiency among immigrant learners of Italian L2 at an advanced level. This is an area of research rarely investigated in general and still neglected in Italy, where today assessment is a delicate issue for school operators, who are required by recent school reforms to evaluate linguistic competences and progress for immigrant students. The dissertation aims at contributing to the solution of this problem, by discussing the nature, development and measurement of linguistic proficiency. Referring to recent discussions at international level, it tries to combine two perspectives: second language acquisition and testing. Together with theoretical results, this combination can offer suggestions for practical implications of both the teaching and assessing of language proficiency of migrant students enrolled at Italian schools. From an acquisitional perspective this project analyses the longitudinal and situational variation of advanced learners involved in a three year time period in four communicative tasks. From a testing perspective, it aims at verifying the relationship between two types of measurements: an analytic one which quantifies complexity, accuracy and fluency (CAF), used in SLA research by the so called developmental index studies, and a more subjective one based on the descriptors of the Common European Framework of References (CEFR), more closely linked to testing. The corpus comprises the oral performance of four migrant students from different nationalities and mother tongues, as well as that of two native speakers of Italian with similar characteristics, who constitute the benchmark for the comparison with the learners. The language analysed has been elicited with two monologic tasks (film retelling and picture story telling) and two interactive ones (interview and telephone call openings). The study is explorative in nature, in so far as it aims at answering research questions and generating hypotheses, rather than verifying pre-formulated hypotheses or convalidate specific theories. In particular it aims at answering three sets of questions: (1) How do complexity, accuracy and fluency develop longitudinally? How do they vary in different communicative activities and in comparison with native speakers? What is the individual path of development of each learner? What are the possible relationships between these dimensions? (2) Are the selected CEFR descriptors effective in assessing oral proficiency? In particular are they sensitive to longitudinal and situational variation? (3) What is the relationship between CAF measures and CEFR ratings? The analysis shows that interlanguage varies systematically in relation to the task and learner. First, in the longitudinal perspective, all learners show progress in almost all the measures considered. Secondly, like the Italian students, all the learners show variation since the first data collection. However, whereas for native speakers the monologic tasks are syntactically more complex, and less accurate and fluent than the interactive ones, learners are more complex and fluent in the monologic tasks than in the interactive ones. Over the years learners achieve a progressive alignment with the native speakers’ linguistic behaviour, even if the manner and rate of progress change from task to task and from learner to learner. Thirdly, the higher variation amongst learners is due to the initial level of competence. The two learners that were weaker at the fist data collection, particularly in complexity and fluency, register changes that are not only more significant, but also more linear than those achieved by the two more advanced learners. An important result of this study is the observation that progress does not always correspond to a linear and generalised increase in complexity, accuracy and fluency. In an interactive task, for instance, a high index of syntactic complexity can indicate incompetency at the interactional level. This means that learners can achieve progress through a reduction in complexity. Furthermore complexity, accuracy and fluency in some cases can progress together, in other cases they develop one at the expense of the other. This seems to be connected to a series of factors: the initial level of competence, the type of task, the speech style chosen, and finally the interactions amongst the three dimensions, which can result in trade-off effects. Therefore CAF measures cannot be considered absolute indexes of proficiency, as an increase or decrease does not correspond automatically to a positive or negative change. As a consequence it is important to combine the different measures and compare them with native speaker variation, if the objective is the elaboration of more effective proficiency profile. The CEFR seems generally an effective tool, even if it is not always sensitive to task variation. This is the case with interactive tasks, in particular telephone calls. The raters used in this thesis perceive the descriptors cohesion-coeherence as unsuitable for the peculiarity of this interactive task, as they do not provide enough detailed information about the quality of the expected performance. However, in general, the study underlines the usefulness of both CAF measures and CEFR ratings, showing how they can both reflect longitudinal and situational variation, revealing in some cases the same tendencies, and in others complementing each other’s interpretation. The study shows the usefulness of a broader interaction between acquisition and testing. Those who are involved in testing can find various suggestions for a better definition of the linguistic elements related to different levels of proficiency in acquisition research. Those who are involved in acquisition can find instruments for a more rigorous validation of the elicitation and measurement instruments in the testing literature. Finally, the study has useful implications both for teaching and testing. In implementing a communicative syllabus it is important to consider the details of different types of interaction and their related routines, whereas in testing it is useful not to separate rigidly linguistic competence from interactional competence.