BackHome
Refresh: OffView Personal Schedule
Paper Summary
Share...

Direct link:

Assessing General Academic Word Knowledge of Diverse Learners

Fri, April 17, 12:25 to 1:55pm, Sheraton, Floor: Ballroom Level, Sheraton II

Abstract

Objectives

This paper describes the Test of Academic Vocabulary in English (TAVE), an assessment of general academic words indexed to school text, administered to English learners (ELs) and their non-EL peers across grades 3-8.

Theoretical Framework

Although norm-referenced vocabulary measures assess how students compare to peers, they provide minimal instructional information beyond student differences. Because well-developed vocabulary is crucial for reading comprehension, a grade-level vocabulary measure that reflects academic text and accounts for word attributes can shed light on high-utility, unfamiliar words. This information can inform prudent word selection through uncovering words that require more intensive instructional attention due to their difficulty.

Nation (2001) describes the “learning burden” of a word as “the amount of effort required to learn it” (p. 20) and presents the “burden” (in part) as a function of the learner—cognates tend to be more readily acquired by language learners (August, Carlo, Dressler, & Snow, 2005). Extending the “learning burden” to word-level considerations, Nagy and Hiebert (2010) cite form and semantic properties as influencers of word difficulty.

Methods
To develop a word source reflective of academic texts, The Educator’s Word Frequency Guide (WFG, Zeno, Ivens, Millard, & Duvvuri, 1995) and Living Word Vocabulary (LWV, Dale & O’Rourke, 1981) were combined. Meanings were coded for morphology and word-level properties that influence acquisition (August & Barr, 2010; Schwanenflugel, 2013). The final paper will detail these attributes, coding protocols, and procedures.
Six separate grade-level forms were constructed using empirical estimates of item difficulty calculated from pilot data. From predictive difficulty, four word complexity quartiles were created for each grade. Words were randomly selected from each grade-level quartile using methods to ensure proportionality with respect to grade-level morphological distribution of the word source.
Data Sources

The Gates MacGinitie Reading Test (GMRT), Test of Silent Word Reading Fluency, and TAVE were administered to 1,450 English learners (ELs) and non-EL across grades 3-8 in the Southwest. For each TAVE item students chose a word matching a definition and sentence.

Results

Psychometrics were explored through Item Response Theory (IRT) and between- subjects factors examined with ANCOVA.

The Test Information Function magnitude was five times larger than the standard error indicating low error along the measurement continuum. Student factor scores correlated highly with GMRT extended scales scores (r=.78). Regarding unidimensionality, GMRT was regressed on TAVE factor scores and the first residual factor. The residual factor was nonsignificant (β=.02) suggesting the TAVE measured a single factor.

Non-ELs outperformed ELs across grades. Although grade- level forms were vertically equated, students in upper grades had higher factor scores relative to lower grades. GMRT outcomes and magnitudes of between-subjects differences and word attribute findings will be reported in the final paper.

Significance
To our knowledge the TAVE is the only test indexing academic text as described in this paper. Given the high-level of vocabulary needed to comprehend increasingly complex texts in the CCSS, it is vital for practitioners to strategically target essential words for instruction. Tests such as the TAVE can inform word selection.

Authors