Although, no overall differences were found between the English and Spanish versions of the tests, a significant interaction was found between language background and performance on the English versus Spanish phoneme awareness tests, F (4,39) = 7.681, p < .0001. As shown in Figure 2, English-dominant speakers score higher on the English tests than Spanish-dominant speakers do on the Spanish tests.
A significant difference was found for the specific type of phoneme segmentation test presented, whether Mann or Yopp-Singer, F (1,39) = 32.749, p < .0001. In general, the Yopp-Singer PSTs were easier than the Mann PSTs. No significant interactions were found between the two language versions of these tests and the type of test. The Mann is more difficult than the Yopp, regardless of whether the test language is Spanish or English. Further, no interactions were found between test type and any other variable.
Correlations between all of the variables appear in Table 2. Both Spanish and English versions of the phoneme segmentation tasks correlated with reading ability. Here we focus on the transfer between tests.
For the Mann tests, the results showed a significant correlation between the Spanish and English versions of the test, r (52) = .492, p < .0001. Children who did better on one version of the test tended to do better on the other. A significant correlation was found between the Mann tests for the control group, but also for the English-only children, r (9) = .710, p < .032 (p < .05). Spanish-speaking students performed better on the Spanish task than on its English equivalent. However, once again, there was a significant correlation for both versions of the Mann test for this Spanish-only group, r (10) = .843, p < .002.
For the Yopp-Singer tests, the results also showed a significant correlation between the Spanish and English versions of the test, r (52) = .334, p < .003.
Children who did better on one version of the test tended to do better on the other. English-dominant children performed better on the English task. No significant correlation was found between the Yopp-Singer tests for the English-only children, r(52) = .367, p >.05. Spanish-speaking students performed better on the Spanish task than on its English equivalent. A significant correlation for both versions of the Yopp-Singer test for this Spanish-only group, r (10) = .734, p < .016.
The current study investigated previous findings that a) related reading ability and performance on phoneme awareness tests, b) concluded that neither bilinguals (Walley in Bruck and Genesee 1995) or monolinguals (Bruck and Genesee 1995) were advantaged in these tasks, and c) found general transfer of phoneme awareness between languages, as suggested by Durgunoglu et al. (1993). Our discussion of the results is organized around these main issues.
As described in the literature and as demonstrated in the present study, reading ability and performance on phoneme awareness tests are related. Our data show that reading ability is related to phoneme awareness in the English version of the Mann PST, r (52) = .431, p < .002, on the Spanish version of the Mann test, r (52) = .475, p < .001, on the English version of the Yopp-Singer PST, r (52) = .578, p < .0001, and on the Spanish version of the Yopp-Singer test, r (52) = .604, p < .0001. In the original literature, Yopp (1988) found the Yopp-Singer PST predicted subsequent learning of new words, r ( 104) = .67, p < .01. Mann (1993), as well, found phoneme awareness and reading to be related, r (100) = .5786, p < .01. Thus we show that our analyses yield very similar correlation values. There is an association in the level of development or impairment of subjects' awareness of associations between phonemes reading ability.