An investigation into the impact of predictive text use upon the literacy skills of primary school, secondary school and university cohorts was conducted over the course of a year. No differences in use of text abbreviations ("textisms") were found between predictive text users and nonusers. However, secondary school children who used predictive text made more genuine spelling errors than nonusers. Predictive text was related to use of some specific grammatical violations in school-age children's text messages but was not related overall to the tendency to make grammatical errors when texting. University students, however, made significantly fewer grammatical errors in their text messages when they used predictive text. Over the course of a year, predictive text use was variable for all age groups. Consistency of predictive text use was unrelated to grammatical understanding, spelling or orthographic processing for primary and secondary school cohorts. Predictive text use was negatively related to morphological awareness for adult participants.