| he instrument of metalinguistic awareness for the secondary reader which was developed and validated for the purpose of this study had five subtests dealing with phoneme-grapheme identification, nonwords, rhyming, ambiguity, sentence anomaly, and definitions respectively. The objectives were to design an instrument that: (1) was congruent with the solicited definition of metalinguistic awareness, (2) provided a diagnostic assessment of the developmental nature of metalinguistic awareness skills and insights into the acquisition of superordinate metalinguistic awareness skills, (3) looked at form rather than the content of language, (4) provided aural and visual redundancies for the items to ensure that metalinguistic awareness and not reading is assessed, (5) had a high concurrent validity coefficient when correlated with the Stanford Diagnostic Reading Test, and (6) had a high test-retest reliability coefficient.;The procedures which were followed included: (1) soliciting definitions of the term metalinguistic awareness from experts in the field, (2) developing an operational definition from the theoretical definition, (3) developing the instrument including an informal instrument, (4) determining content validity, (5) selecting the subjects, (6) conducting two pilot studies, (7) conducting a field study with a sample of 100 eighth-grade subjects--50 high-proficient readers and 50 low-proficient readers, (8) studying concurrent validity, and (9) determining test-retest reliability.;The findings were as follows: The content validity data showed a grand average of 94.39 percent for item acceptability by the judges. The chi square value was 3.776 (p ;The concurrent validity coefficient data were obtained by the correlation of Baker's Assessment Instrument of Metalinguistic Awareness and the Stanford Diagnostic Reading Test. There were 49 correlations. Forty-two were significant at the.0001 level. There were seven minimal validity correlation coefficients which ranged from.001 to.07. Of the 42 correlations which were significant, 15 were moderate and ranged from.5382 to.6488. The 27 high correlation coefficients ranged from.6525 to.8578.;All the test-retest reliability coefficients were significant at the.0001 level. The correlations were.42 for Subtest One,.88 for Subtest Two,.80 for Subtest Three,.90 for Subtest Four,.88 for Subtest Five,.80 for Subtest Six, and.96 for the Total Test.;The data for the analysis of variance showed that with the exception of Subtest One, the t-tests were significant at the.0001 level and ranged from 9.9357 to 19.1718. F ratios were significant for Subtest Three (F 4.85 p... |
