Mu, G. M. (2015). A meta-analysis of the
correlation between heritage language and ethnic identity. Journal of Multilingual and Multicultural Development, 36, 239-254.
doi:10.1080/01434632.2014.909446
In his expansive yet critical paper,
Mu intervened to the gap in research justifying cross-ethnic claims between
ethnic identity and heritage language (HL) proficiency (p. 240). He employed
meta-analysis, an advanced statistics for contrasting and combining quantitative
studies (p. 241), and treated 18 articles that had successfully met the
selection criteria (p. 242). Component models of this analysis (pp. 246-247)
revealed that there is a medium positive correlation in the differences of the
included studies and significant correlation of ethnic identity and HL
proficiency across different ethnic groups (p. 248).
This instrumentation of analysis,
however, has its own limitations. Methodological variations in sampling across
the analyzed studies raised a concern (p. 248). Collectively, convenience
sampling is administered but the participants of the study are the
differential, i.e. strong proportions of children, adolescents, and young adults.
Parental intervention and multi-ethnicities are notable considerations
altogether (p. 249). Furthermore, the studies’ operationalization, i.e.,
adapted frameworks, measurement, i.e., self-reported results and
instrumentation, and localization, i.e., spatial diversities, have been
discussed as deeply felt consideration for future research (pp. 249-250). Mu
also raised the development of the studies’ theoretical bases. He urged working
on mainstreaming as a response to theoretical overlaps (p. 250). He personally
proposed Bourdieu’s “sociological notion of habitus” to harmonize concurrent
approaches in ethnic identity and HL research (p. 251).
Mu can be commended with his attempt
to ‘standardize’ the results of the studies. Yet, his attempt to arrive at a transcending
standard authority affirming the correlation of ethnic identity and HL use is in
itself must be reconsidered, e.g., for weighted sample sizes. Gladly, this was
mentioned before resting the paper’s case (pp. 250-251).