Two experiments explored reader reaction to written errors that were either typographic or grammatical. Errors were embedded in short texts presented as email responses to a housemate ad. In the first experiment, readers evaluated the writer and message on several dimensions (e.g., Was the writer trustworthy? Did the email flow smoothly?). Those dimensions were divided into a “social” scale (e.g. “This student seems similar to me”) and an “academic” scale (e.g. “This email reads well”). Both kinds of error correlated with lower ratings on the academic scale while only grammatical errors correlated with lower ratings on the social scale. In the second experiment, readers were asked to edit the emails. In Experiment 1, paragraphs with either typographical or grammatical errors were both evaluated more negatively than fully correct paragraphs and the cost was mitigated by high levels of electronic communication, such as texting and using Facebook. In Experiment 2, typos were more likely to be corrected than either homophonous grammatical forms or hypercorrected forms. These results suggest that written errors, when they are salient, contribute to the social meaning of text. Furthermore, this contribution is modulated by at least some characteristics of the reader.