ABSTRACT
Using a randomized control trial, this study examined the causal evidence of cross-language transfer of phonological awareness and letter knowledge (names and sounds) using data from multilingual 1st-grade children (N = 322) in Kenya. Children in the treatment condition received an 8-week instruction on phonological awareness and letter knowledge in Kiswahili. The comparison group received business-as-usual classroom instruction. Children in the treatment condition showed greater improvement in phonological awareness and letter-sound knowledge in Kiswahili and English (positive transfer; effect sizes from .37 to .95), whereas a negative effect was found in letter-name knowledge (interference; effect size, g = .27). No effects were found in reading, nor did the results vary by moderators (e.g., Kiswahili vocabulary). Path analyses revealed divergent patterns of results for different outcomes. Results provide causal evidence for cross-language transfer of phonological awareness and letter knowledge and offer important theoretical and practical implications.
Acknowledgments
We thank participating schools, school leadership, teachers, and children, as well as project staff.
Notes
1. The relation between phonological awareness and word reading appears to vary as a function of the characteristics of writing systems (Koda & Reddy, Citation2008). Between alphabetic languages (e.g., English and Spanish), moderate and significant correlations are reported, whereas weak or negative correlations are apparent between an alphabetic language (English) and nonalphabetic language (e.g., Chinese; Bialystok et al., Citation2005; Wang, Park, & Lee, Citation2006; Wang, Yang, & Cheng, Citation2009). These findings suggest that L1 phonological awareness more strongly influences L2 word reading in languages with similar orthographic characteristics (e.g., alphabetic writing systems) than in languages with different orthographic characteristics.
2. Given that children were also nested with small groups in the treatment conditions, ICCs attributable to small-group nesting were estimated as follows: .08 for Kiswahili letter-name knowledge, 0 for Kiswahili letter-sound knowledge, 0 for Kiswahili Oddity task, .05 Kiswahili Blending task, .13 for English letter-name knowledge, 0 for English letter-sound knowledge, .08 for English Detection task, and .06 for English Blending task.
3. The weak but positive correlations between letter-name knowledge in Kiswahili and in English (.18 ≤ rs ≤ .24; see and ) are not at odds with the negative result in because the results in accounted for differences in covariates, which included pretest scores.
4. In Bialystok et al.’s (Citation2005) study, correlations were varied across language groups: .07 for Hebrew–English bilingual children, .19 for Spanish–English bilinguals, and .33 for Chinese–English bilinguals. It is notable that the magnitudes for the latter two groups are positive and range from small to moderate. All these were not statistically significant in the Bialystok’s study, but this is likely due to the small sample sizes (29 ≤ ns ≤ 33).
5. For example, several digraphs in Kiswahili do not correspond to those in English (dh for /ð/). Discrepancies are particularly large for vowel letters, for which Kiswahili letters represent a single sound consistently, whereas English vowel letters represent multiple sounds (even the predominant sounds English vowel letters represent are different from the sound in Kiswahli; e.g., /æ/ and /e/ in English for letter a, which represents /a/ in Kiswahili).