test of phoneme identities: Predicting alphabetic insight in prealphabetic readers, The

Journal of Literacy Research, Sep 2000 by Murray, Bruce A, Smith, Kimberly A, Murray, Geralyn G

Children received a score indicating the trial number during which they met the criterion, representing their relative difficulty in learning to use phonetic cues to identify words. For example, a participant who gave 5 consecutive correct answers during Test 1 received a score of 1, and one who met the criterion during test 9 received a score of 9. Participants unable to meet the criterion after 17 lesson segments were assigned a score of ig. Thus, a high score indicates that much teaching and many trials were needed to attain alphabetic insight, and a low score indicates the student mastered the concepts easily.

Results

Descriptives

Means, standard deviations, and ranges for the principal measures of this study are presented in Table 3.

An initial question concerns the reliability of the Tpi; no test that is unreliable can be valid. The alpha split-half reliability was .91. The test-retest reliability, or correlation between pretest and posttest forms, was .83. These statistics show strong reliability with kindergarten students, a population for whom test reliability is often elusive.

Correlation

Table 4 shows the intercorrelations of the principal measures used in the study. Correlations with phonetic cue reading, a concurrent measure of alphabetic insight and rudimentary decoding, are shown in the final column. The two versions of the TPI, the Yopp-Singer, and the TO PA all correlate with the test of phonetic cue reading around r = .6. In other words, all three phoneme awareness measures rise or fall with early reading ability.

A different picture emerges when the correlations are computed only for the 34 prealphabetic readers who failed the test of phonetic cue reading. The results are shown in Table 5. In the absence of reading ability, the strongest correlate of phonetic cue reading is accuracy of letter naming (r = .58), and the second strongest is knowledge of nursery rhymes (r = .41). A new variable, Trials, has been added, a variable that relates only to prealphabetic readers, because only they undertook the lessons from which trials were counted. All correlations are strikingly reduced from the full sample. Both the Yopp-Singer and the ToPA are moderately correlated with phonetic cue reading, whereas neither version of the TPI is significantly correlated with early reading in a population of children who do not consistently read phonetic cues. This is consistent with the view that children who succeed on the Yopp-Singer and the TOPA tend to be the children who already have some dawning alphabetic insight.

Why wasn't the posttest version of the T PI significantly correlated with Trials? The TPI is fairly lengthy (38 verbal items, with administration time about 10 minutes), and participants had taken it only a few days earlier. Some prealphabetic readers, unable to recognize phonemes, may have resorted to guessing strategies. Table 6 displays the mean scores of beginning readers (those who scored at least 10 on the test of phonetic cue reading) and prealphabetic readers (those who scored below 10) in the measures of phoneme awareness and shows the results of t tests for independent samples. The scores of children with and without alphabetic insight were significantly different on all measures of phoneme awareness, with the children who scored at chance levels on the Test of Phonetic Cue Reading showing relatively low phoneme awareness. These prealphabetic readers found it difficult to determine, for example, whether /kl was heard in tar or car. Many seemed to adopt a guessing strategy given the forced-choice format, for example, choosing the second option for each item. Such guessing probably introduced more chance variation into the posttest scores.