Page 104 - Second language development of newly arrived migrant kindergarteners - Frederike Groothoff

104 Chapter 5 uttered in a story are included, there is no maximum because pupils can use as much ISTs as they wish, in any part of the story. Analyzing the stories at both the microstructural and macrostructural levels enabled us to discuss the narrative ability development of newly arrived migrant kindergarteners in both the linguistic competence as well as the discourse competence areas of communicative competence. Table 5.7 summarizes all the six measurements in the present study. Table 5.7: Summary of the Six Measurements with the MAIN. Microstructure Lexical Diversity 1. Number of Different Words Macrostructure Statistical analysis Lexical Richness (NDW) 2. Guiraud Index Score (GUI) 3. Measure of Lexical Richness (MLR) 4. Story Structure (SS) 5. Story Complexity (SC) 6. Internal State Terms (IST) To answer our research questions, multilevel modeling is applied to take into account the hierarchical structure of the data due to repeated measures per participant. Measurements of a pupil’s outcome variables (level 1, N = 168) were temporally ordered within pupils (level 2, N = 42). For this purpose, multilevel modeling of repeated measures data procedures (MLwiN software version 2.36; Rasbash, Charlton, Browne, Healy, & Cameron, 2016) was carried out. See for further details on the statistical analysis Section 4.3.4. Intra-rater reliability Microstructure The transcription of the stories was done by one transcriber due to practical limitations. Nevertheless, we calculated the intra-transcriber reliability in order to see if the transcriptions could be considered reliable. We randomly selected 16 stories, 4 from each session. These 16 stories were transcribed twice by the same researcher, with a minimum of five months a part. The word-by-word agreement for the transcriptions was 89.9% (range 68.8–100%) with a Cohen’s kappa of .80, which is considered a substantial strength of agreement (Landis & Koch, 1977). The mean reliability in Session 3 was the highest (95.0%) and in Session 1 the lowest (82.9%). Mean Cohen’s kappa scores of the sessions ranged from .68 to .90, substantial to almost perfect agreement. This high intra-transcriber