Monitoring Mastery of the Alphabetic Principle: Item-Based Learning
Joe Jenkins points out that screens in first and second grade that employ word reading tasks as predictors of end-of-year outcomes in reading have good specificity, that is they identify as not-at-risk individuals who later perform satisfactorily on a future criterion measure (i.e., true negatives). The way we linked word reading in the TPRI screen to achievement outcomes was to analyze our experimental word lists using Item Response Theory, then find the theta coefficient that predicted a specific cutpoint on the WJ Broad Reading--within half a year of grade equivalency. Words around that theta value were placed on the screen. Thus, we were not designing a phonics screener, but rather tapping into alphabetic knowledge associated with grade-level benchmark words. We are currently developing a decodability tool on our website (www.tpri.org) that takes oral reading errors from the TPRI passages and analyzes the phonetic coding and orthographic parsing represented by the errors. Although the narrative and informational passages are not considered part of the screen, all students read a passage for accuracy, fluency, and comprehension regardless of performance on the screen. Teachers who administer the TPRI using a PDA tap on the words misread while the student reads the story out of the booklet. When the teacher syncs the PDA with the website, each student's reading errors are analyzed using a relational database we developed in our IERI grant. We can compare individual error profiles with "normative" error profiles across beginning, middle, and end of year and decide whether instruction on a particular spelling pattern is warranted. We can also do this for students grouped by performance levels. The website could provide words that contain the pattern for additional student practice or refer the teacher to the lesson in the state-adopted basal where relevant instructional strategies and reading selections may be found (Foorman, Francis, Davidson, Harm, & Griffin, in press; Foorman, Santi, & Berger, in press).
I call such assessment-driven instruction "item-based" rather than "stage-based" because the instructional feedback we give to beginning readers builds up orthographic representations of specific items or words. Consequently, unlike older approaches to phonics, instruction should be based on what words the student already knows, not what stage he or she is in. Furthermore, automaticity is a characteristic of words, not readers (Rayner, Foorman, Perfetti, Pesetsky, & Seidenberg, 2001). Thus, the question becomes, "On what words is this child fluent?" rather than "Is this child a fluent reader?". By considering the kinds of words that beginning readers misread, we can start to develop a theory of learning to underlie our reading interventions. I looked at the words in TPRI stories misread more than 50% of the time by approximately 1200 first and second graders. First, I noticed difficulty with various types of syllable units: closed syllables (e.g., believe) and open syllables (e.g., later); consonant-le (e.g., scrambled); vowel team (e.g., Maria), and R-controlled in multisyllabic words (e.g., favorite, surprise) and in single syllable words (e.g., lures, hardly). Blends were challenging (e.g., cricket, dribbling), as well as vowel-consonant-e (e.g., faces, poles), and schwa (e.g., enemies). Errors were also frequently made on complex vowel team spellings (e.g., caught, feelers, true) and on less common orthographic patterns (e.g., gu in arguing and in lifeguard). Instructional strategies for parsing and deriving sound-spelling patterns in multisyllabic words, deriving schwa from "flexed" pronunciations, and inconsistent sound-spelling patterns in single-syllable words are typically not taught in the basals. As part of our response-to-treatment model, we need to build a theory of learning to read from item-based data.
Previous Page | Next Page
(Early Screening) | (RTI)
