The loss of a childhood language, especially in adoptees, has attracted scholars' attention in the past, but a search for any memory traces has yielded conflicting results. In a psycholinguistic tradition known as the savings paradigm, a learn-and-relearn technique is employed to examine whether the relearning of lexical items once known, often in a second or foreign language, can lead to a rate of learning advantage for old (previously known) over new (previously unknown) words. The present study adopted this technique to examine remnants of a lost childhood language in TJ, an adoptee who did not know her linguistic background prior to her adoption at the age of 3 years. Delayed posttests provided evidence for the savings effect for old words: TJ showed better savings for words that were likely known in her childhood language at the age of 3. In contrast, the comparison group revealed no effect for old words over new ones.