While ethnographies of literacy have played an important role in the shift towards understandings of literacy as situated social practice, these understandings have not necessarily impacted on day-to-day development work. This article draws on data collected during two periods of ethnographic work on the literacy practices of participants in grassroots social movements engaging in struggles around housing in South Africa. In this focus on the quotidian tactics of the participants in such projects, mundane everyday texts (like hand-written lists, memos, bank cheques, plans, invoices and so on) were central to the carrying across and projecting of meanings into new contexts and important in the construction of agency for individuals (in the cases reported here, for three individual women). Through the use of multi-site, micro-ethnographic methods, a language of description was developed for identifying, reconstructing and analysing the sequences of events through which people acted to change their living conditions and make things happen. However, recontextualisation and projection of meanings did not require literate individuals, nor did it always require alphabetical texts; it could be accomplished by groups in which literacy was viewed as a distributed capacity or it could be carefully mediated by development workers with a focus on capacities rather than deficits, it could draw on a wider range of mediational means like physical occupations of sites, or building extensions. The research showed that a lack of attention in organisational procedures to the detailed politics of recontextualisation and projection of meanings in such trajectories indicated the reification of literacy and its use as a marker of status and stratification. On the other hand, when careful attention was paid to this detail, literacy became naturalised, as a pragmatics of engagement in textually-mediated practices, less implicated in gate-keeping and conflict. Some studies in the critical discourse tradition in a range of fields have explored ‘chains of discourse’ and make claims that discourse is recontextualised and resemiotised as it travels through contexts, tending towards legitimacy and authority, and this in turn leads to permanence and stability in infrastructures and environments. The article argues that in contexts of extreme poverty, conflict and lack of resources, such uni-directionality cannot be assumed.