Cognitive Development, 8, 403--430 (1993) Mother-Child Conversations About the Past: Relationships of Style and Memory Over Time Elaine Reese Catherine A. Haden Robyn Fivush Emory University This study investigated long-term consistency and change in maternal style for talk about the past and relationships of those styles with children's memory participa- tion. Nineteen white, middle-class mother-child dyads talked about shared past events at four time points: when children were 40, 46, 58, and 70 months of age. Across the four time points, individual mothers could be consistently classified as high elaborative (e.g., they elaborated on event information much more often than they repeated their requests) or low elaborative (e.g., they elaborated less often in relation to their repetitions). However, all mothers became more elabora- tive over time; children also remembered more over time. Cross-lagged correla- tions revealed a relationship between maternal elaborativeness at the early time points and children's later memory responding, but by the later time points, direction of influence between maternal elaborations and children's memory re- sponding had become bidirectional. These results are framed with respect to the importance of shared past event conversations for the development of children's autobiographical memory. The ability to talk about one's personal experiences is an important communica- tive skill. Children must learn which events are significant enough to be told, and they must also learn to tell these stories in an engaging and relevant manner. Moreover, stories about the past are most often told in conversation; thus, relat- ing past experiences is a collaborative effort which frequently involves at least one other "teller" (Blum-Kulka & Snow, 1992). For very young children who are just beginning to engage in conversations about the past, it is adults who provide This projectwas funded by a Spencer Foundation grant to Robyn Fivush. Parts of thisresearch were published as Emery Cognition ProjectReport No. 23 (December, 1992) and presented at the Biennial Meeting of the Society for Research in Child Development, N e w Orleans, L A (March, 1993). W e would liketo thank Janet Kuebli for her conceptual contributions to thisproject,and Liza Donadonan, Laura Underwood, Marcella Eppen, and Marci F,eldman for theirassistance in data collection,transcription, and coding. Specialthanks go to the mothers and childrenwho participated in the study. Correspondence and requests for reprintsshould be sent to Robyn Fivush, Department of Psy- chology, E m e r y University,Atlanta, G A 30322. Manuscript received November 24, 1992; revision accepted April 19, 1993 403