In situations of language endangerment, the ability to understand a language tends to persevere longer than the ability to speak it. As a result, the possibility of language revival remains high even when few speakers remain. Nonetheless, this potential requires that those with high levels of comprehension received sufficient input as children for the activation of speech to occur in later life. In many areas of Australia, input to children of traditional Aboriginal languages is rarely monolingual, but rather often mixed with a contact variety of English. Thus, it is not clear whether children receive enough input to later become active speakers of the traditional languages. This paper reports on a study which tested the relationship between language comprehension and child language input. A vocabulary test of 40 items was administered to 52 Gurindji participants in five age groups. Participants were asked to listen to a Gurindji word and choose a corresponding picture. The test items were graded as high, medium or low frequency on the basis of their use in a corpus of Gurindji child-input speech. We found that age and frequency of use in child-directed speech significantly altered the chance of a correct response. (Contains 6 tables, 1 figure, and 5 notes.)