The validation of the THAM-2 (Test de Habilidades Metalingüísticas n. 2, Núñez Delgado, & Pinto 2015)

This article describes the statistical analyses performed for validating the THAM-2, a metalinguistic ability test for children aged from 9-14, translated from the original Italian TAM-2 (Pinto, Candilera & Iliceto 2003) into Spanish (Pinto, Titone & Gonzáles Gil 2000 ; Núñez Delgado & Pinto 2015) with partial adaptations. The test is composed of six subtests : Comprehension, Synonymy, Acceptability, Ambiguity, Grammatical function and Phonemic Segmentation. The scoring system is based on the distinction between Linguistic (measuring metalinguistic awareness at the implicit level) and Metalinguistic scores (measuring metalinguistic awareness at the explicit level). Three hundred and one Spanish-speaking children between 9 and 16 (mean age = 12.5 ; SD = 1.3) participated in the study. Based on the Spanish school system, the sample was subdivided into four grades from 4-5 primaria (4th and 5th grade) to 2° ESO (8th grade). All the participants took the THAM-2 and the Raven’s SPM3 (Raven et al. 1998). Skewness and Kurtosis showed that the data were normally distributed. ANOVAs evidenced relevant group differences as a function of two main factors : gender, with females outperforming males, and education level, where each school grade was associated with significant improvement in performances. Inter-rater agreement, as measured by Cohen’s Kappa (Cohen 1960) was highly satisfactory. Cronbach’s alpha, calculated on the ML scales, presented acceptable internal consistency estimates. Correlations between the THAM-2 and the SPM38 were significant, although moderate. The results of Confirmatory Factor Analysis (CFA, Brown 2006) overall supported the theoretical two-factor structure of the original test, the ML and the L factor, but with a strong asymmetry between the two factors. Only the ML emerged as a unitary construct, well measured by the all the ML scales, whereas the L factor was well measured only by two out of six scales. These results stress the metalinguistic character of this test, which lends itself to various types of studies in education, especially in correlation with metacognitive competences.