This study was undertaken to determine the effect of linguistic background on scores obtained by native and non-native subjects on two dichotic speech tests in American English, the staggered spondaic word (SSW) and the dichotic consonant-vowel (CV) tests. Thirty subjects whose native languages were either Hindi or Arabic were tested. Test findings were compared to results obtained from 10 native English-speaking subjects. The non-native English speakers showed abnormally high error scores on the SSW test while the native speakers performed with no errors on that test. Performance on the dichotic CV test was comparable for both the native English and the Hindi subjects, while the Arabic subjects performed significantly worse than the two other groups. These data indicate that tests of speech perception using English must be interpreted with caution when administered to non-native English speakers, and that neither the SSW nor the dichotic CV tests may be assumed to be free of linguistic bias.