The Nonword Repetition (NWR) task is a test in which children repeat nonsense words. It is given to children to test their phonological working memories, which is assumed to underlie vocabulary development. NWR is said to be 'language-free,' and so designed to be suitable for testing children from diverse linguistic and cultural backgrounds. In this research project, we predicted that children would score differently based on the language in which the test is given. A total of 186 children with and without known language delays participated in the study. Within each language, we found that children with language delay did not have as high NWR accuracy as typically developing children. However, we found that our hypothesis was correct; NWR is sensitive to native language. That is, native Spanish-speaking children had higher accuracy on a Spanish NWR task than an English NWR task and native English-speaking children did better in English than Spanish NWR. We also found that the majority of the pronunciation errors in English were with final consonants in the longest nonsense words. The consonants /p/, /g/, and /b/ presented particular difficulty. Children would either omit the final consonant or substitute another consonant. These errors also likely indicate language-specific differences.