Page 62 - Second language development of newly arrived migrant kindergarteners - Frederike Groothoff

62 Chapter 3 3.3.2 Research design and data collection Ortega and Iberri-Shea (2005) remark that research on second language development should include “time” and “learning,” this could be done via data collection that is longitudinal or cross-sectional. Data collection in longitudinal studies used to be carried out with one single subject or a small group of participants, over a prolonged period of time with regular intervals. Longitudinal data usually consist of several aspects of language development and other cognitive development. Although a lot of data is collected in longitudinal studies, it is sometimes difficult to generalize with such a small group of participants. An alternative for that is to do cross-sectional research. In cross-sectional research data is usually collected by means of large groups of participants with different ages at a single point in time. In these studies the focus is mostly on one specific aspect of language development. The idea is that with such a large group one is “able to see a slice of development, which is used to piece together actual development” (Gass & Selinker, 2008, p. 56). However, it is always problematic to make sure that the different age groups are comparable. Both methodologies have advantages as well as disadvantages; fortunately, the distinction is not as rigid as it seems. There is flexibility in categorizing research as cross- sectional or longitudinal (Gass & Selinker, 2008) as is, for example, the case with the Pre- Cool data collection (e.g., Veen, van der Veen, Heurter, & Paas, 2012a) and with our data collection. The Pre-Cool cohort study, which started in 2011, is an example of a longitudinal study with a large group of participants (N = 3000). The present study is a longitudinal study, with a fairly large group of participants: different kinds of data about the language use of the participants were gathered. The data could be described qualitatively but with additionally specific quantitative analyses, by means of statistical measures. Gilmore (2016) considers a mixed methods approach most appropriate for classroom-based research projects, like our study, because in classroom-based research an attempt is made to measure changes in a complex construct, composed of multiple, interacting sub-components, and emerging across multiple layers of a complex learning context over an extended period. In our case the complex construct is the second language development, of which we assessed the interacting sub-components receptive vocabulary and narrative ability. Furthermore, this second language development is emerging across multiple layers of a complex learning context, namely the language classroom and we followed the participants over an extended period of two-and-a-half years. To summarize, the present study is longitudinal, with a descriptive, evaluative, and correlational design. To longitudinally describe the second language development of newly arrived migrant pupils, information was gathered using four primary data sources: (1) performance tests on receptive vocabulary development in Dutch and (2) performance tests on a productive narrative task. Data set (1) and (2) were both conducted with the 42 kindergarteners who participated in the study. Data set (3) includes observational data from