This article reviews research on the effects of conversational style on learning. Studies of conversational style have variously investigated "personalization" through changing instances of first-person address to second or third person, including sentences that directly address the learner; including more polite forms of address; and making the views and personality of the author more visible. Meta-analyses provided mixed support for a model of learning processes; statistically reliable average effects were found on self-reports of friendliness (d?=?0.46) and effective cognitive processing (d?=?0.62), but not learning assistance (d?=?0.16) and interest (d?=?0.15). Statistically reliable average effects on retention (d?=?0.30) and transfer (d?=?0.54) learning outcomes supported conversational-style redesigns across a range of potential moderators; the clearest apparent boundary condition for learning outcomes across the moderators under analysis was instructional time, with small, non-significant effects being found in studies longer than 35 min. Recommendations for future investigations are discussed.