Miller’s (1956, The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review 63(2). 81–97) working memory (WM) capacity of around seven items, plus or minus two, was never found by usage-based linguists to be a recurrent pattern in language. Thus, it has not figured prominently in cognitive models of grammar. This is somewhat unusual, as WM has been considered a fundamental cognitive domain for information processing in psychology, so one might have reasonably awaited properties like ability constraints to be reflected in language use and structures derived from use. This article suggests that Miller’s (1956) number has not been especially productive in usage-based linguistics because it appears to have been an overestimate. A reviewed WM capacity has now replaced it within cognitive science, a “magic number four plus or minus one” (Cowan 2001, The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences 24(1). 87–185). This article proposes, using evidence from spoken language corpora and multiple languages, that a set of linguistic structures and patterns align with this revised capacity estimate, unlike Miller’s (1956), ranging from phrasal verbs, idioms, n-grams, the lengths of intonation units and some abstract grammatical properties of phrasal categories and clause structure.