Abstract
The project Early Language Learning in Europe (ELLiE) has studied the longitudinal effects of an early foreign language (FL) start in seven European contexts. This article presents a sub-study of ELLiE that investigates the impact of out-of-school factors on learners' listening and reading skills in year four of formal FL instruction. More specifically, we include parents' educational level, parents' use of the FL professionally, exposure, interaction and cognate linguistic distance. Data were collected by means of listening and reading tests and a parents' questionnaire. Results of the statistical analyses show that cognate linguistic distance was the strongest predictor of both listening and reading scores, followed closely by exposure, and parents' FL use at work and international interaction at some distance. Parents' educational levels only impacted on reading scores, and domestic interaction did not have any effect on listening or reading. Furthermore, the results confirm previous research on young learners' incidental FL acquisition through watching subtitled films, as watching films was the most powerful exposure type for both listening and reading. Parents' use of FL at work correlated significantly with exposure, indicating that the influence of parents would have an effect on the opportunities for their children's FL exposure.
Acknowledgements
This research was supported by a European Commission grant under the Lifelong Learning Programme (Project number 135632-LLP-2007-UK-KA1SCR) and by the British Council. We gratefully acknowledge our colleagues in the ELLiE team who have been part of instrument development and data collection: Janet Enever, Evelien Krikhaar, Lucilla Lopriore, Gun Lundberg, Jelena Mihaljević Djigunović, Magdalena Szpotowicz and Elsa Tragant. We thank Ivana Režek and Agata Borecka for their illustrations used in the listening task. We also acknowledge the children, parents and teachers who generously helped us with their participation. Finally we warmly thank the two reviewers who provided thoughtful comments and suggestions on an earlier version of the article.
Notes
1. An examination of the dependent variables suggests that the assumption of normal distribution could be violated, with significant Kolmogorov–Smirnof values of 0.153 for listening and 0.143 for reading. Further investigation shows that the skewness lies in the upper half of the result range for both listening and reading (−0.930 and −0.154, respectively), which is also supported by the relatively high mean values (25.13/32 for listening and 4.09/7 for reading). Kurtois values further indicate that the results on the reading test have a flat distribution (−0.940). However, lack of normal distribution is not unusual in large samples and it has been suggested that with large sample sizes skewness and Kurtois values as presented previously are still within an acceptable range for analysis (Tabachnick & Fidell, Citation2001).
2. A timely move in this direction has been recently taken in Spain, one of the countries with a stronger dubbing tradition (initiated as a way of political censoring under Franco's dictatorship). See the document ‘Fomento de la versión original en la exhibición de obras audiovisuales: conclusiones, propuestas y recomendaciones’ (Ministerio de Cultura, ICAAV), http://www.mcu.es/cine/docs/Novedades/COMISION_FOMENTO_VO.pdf
3. A correlation analysis showed no significant relation between test scores and number of instruction hours.