Investigating the effects of a kindergarten vocabulary intervention on the word learning of English-language learners

Date of Completion

January 2008


Education, Bilingual and Multicultural|Education, Special|Education, Reading




The purpose of this dissertation study was to reanalyze the data from Year 03 of Project VITAL, an 18-week vocabulary intervention study, to determine how English-language learners (ELLs) responded to direct vocabulary intervention compared to English-only learners (EOLs) using the simple view of reading (Gough & Tunmer, 1986) as its theoretical framework. Participants included 122 kindergarteners from three elementary schools within three separate school districts in the northeastern United States who were assigned to either a treatment condition (ELLS, n = 31; EOLs, n = 49) or no-treatment condition (ELLs, n = 17; EOLs, n = 25). Interventionists were trained to deliver direct vocabulary instruction using one of 18 storybooks twice per week in 20-25 minute sessions. The research design of the study employed a pretest, posttest, quasi- experimental group design. Analyses consisted of two-by-two analyses of variance (2 x 2 ANOVAs), with two between-subjects factors (instructional condition x language status), and t-tests to examine the effects of vocabulary intervention on the word learning of ELLs and EOLs, and hierarchical multiple regression (HMR) to investigate word learning differences between ELLs and EOLs using pretest PPVT-III to predict posttest target-word and general, receptive vocabulary knowledge. Participants' knowledge of vocabulary was individually assessed with a researcher-developed measure of target-word knowledge (TWKM) and the Peabody Picture Vocabulary Test-III (PPVT-III; Dunn & Dunn, 1997). Results for 2 x 2 ANOVAs and t-tests indicated that EOLs significantly outperformed ELLs on dependent measures with evidence on an interaction effect on TWKM. HMR results indicated (a) centered pretest PPVT-III accounted for a statistically significant proportion of the variance in posttest measures for participants in the treatment condition and (b) language status did not explain any additional variance in posttest measures above pretest PPVT-III. Each of the two mediation models using the independent variable, language status; mediating variable, centered pretest PPVT-III; and dependent measures TWKM and PPVT-III resulted in full mediation. Major findings indicated that treatment ELLs and treatment EOLs did equally well on posttest if they had similar initial general, receptive vocabulary knowledge on PPVT-III. Implications, limitations, and directions for future research are discussed. ^