Abstract This study examined the influence of obligatory linguistic marking of the source of information on source memory. Turkish grammar requires speakers to indicate if an assertion is based on first hand knowledge or non-firsthand knowledge (hearsay or inference); English grammar does not require this distinction. We hypothesized that obligatory coding of source of evidence leads to a greater weighting of first hand relative to non-firsthand accounts of events (an “evidentiality effect”), resulting in better memory for first hand sources. In support of this hypothesis, across two experiments native Turkish speaking adults showed significantly better recognition and source memory for assertions coded with first hand than non-firsthand evidential markers. Further, among Turkish speakers who also knew English, those who learned English later had less accurate recognition and source memory for non-firsthand sources presented in English than those who learned English earlier, suggesting a carryover from the first language (Turkish). English monolingual speakers showed no difference in recognition or source memory as a function of source type, but showed better memory than Turkish speakers for non-firsthand sources. These findings provide the first empirical support for an evidentiality effect, suggesting that when marking the source of evidence is required by the grammar first hand sources are privileged in memory and non-firsthand sources are discounted.