Preschool children's use of decontextualized language, or talk about abstract topics beyond the here-and-now, is predictive of their kindergarten readiness and is associated with the frequency of parents' own use of decontextualized language. Does a brief, parent-focused intervention conveying the importance of decontextualized language cause parents to increase their use of these conversations, and as a result, their children's? We examined this question by randomly assigning 36 parents of 4-year-old children into a decontextualized language training condition or a no-treatment control condition and used mixed effects modeling to measure change (from baseline) in parent and child decontextualized language at 4 subsequent home mealtimes during the following month (N = 174 interactions including the baseline). Compared with the control condition, training condition dyads significantly increased their decontextualized talk and maintained these gains for the month following implementation. Furthermore, trained dyads generalized the program content to increase their use of similarly decontextualized, yet untrained conversations. These results demonstrate that an abstract feature of the input is malleable, and introduces a simple, scalable, and replicable approach to increase a feature of child language known to be foundational for children's school readiness.