The place of grammar in both first and second language learning curricula has been long contested, particularly in Anglophone countries, where since the Dartmouth Conference of 1966, countries such as the UK, the USA, Australia and New Zealand have had an uncertain stance towards grammar, in many cases, completely eschewing grammar for many years. This historical attitude towards grammar, and professional ambivalence at best, or antipathy at worst, towards the teaching of grammar has been well-documented (Kolln and Hancock 2005; Locke 2009; Myhill and Watson 2014) and will not be repeated here. But at the heart of this apparent rejection of grammar was the conviction that the explicit teaching of grammatical terminology had no discernible impact on young learners' capacities as language users. Indeed, several research studies confirmed this (Elley et al 1979; EPPI 2004). However, this debate has been framed principally by a curriculum focus on the merits or otherwise of its inclusion, rather than any evidence-based or well-theorised consideration of the issue. Our own research at the University of Exeter, which informs this article, has revealed that when writing and grammar share the same learning focus, explicit grammar teaching can be beneficial in improving learners' outcomes in writing. This article sets out to ground the debate within a theoretical framework and in particular to consider the value of metalinguistic talk within a pedagogical approach to the teaching of grammar which foregrounds the meaning-making relationships of grammatical choices in writing.