Research findings indicate that synchrony between events in two different modalities is a key concept in early social learning. This longitudinal pilot study with 14 mother–child dyads is the first to support the idea that synchrony between action and language as a form of responsive behaviour in mothers relates to later language acquisition in their children. The study conducted a fine-grained coding of multimodal behaviour within the dyad during an everyday diapering activity when the children were three and six months old. When the children attained 24 months, their mothers completed language surveys; this data was then related to the dyadic measures in early interaction. The study proposes a ‘role-switching’ model according to which it is important for three-month-olds to be exposed to multimodal input for a great deal of time, whereas for six-month-old infants, the mother should respond to the infant’s attention and provide multimodal input when her child is gazing at her.