A quantitative reanalysis of data on the structure of L1 and L2 language ability in multitrait-multimethod studies

To gain a better understanding of the structure of L1 and L2 language ability, we conducted an extensive literature search to collect multitrait-multimethod (MTMM) studies. We specifically examined the unitary, correlated, uncorrelated, and higher-order structures of language ability to determine which of these best fit the data across studies. We used confirmatory factor analysis to reanalyze 58 correlation or covariance matrices from 39 studies, among which we retained 17 correlation or covariance matrices from 14 studies for the main analysis. Empirical evidence showed that the unitary and higher-order models best defined both L1 and L2 ability. The support for the unitary model was surprising but consistent with the findings of Davidson (1988). Moderator variable analyses failed to identify clear relationships between the examined models and moderator variables.