An experimental study was conducted on children aged 2;6–3;0 and 3;6–4;0 investigating the priming effect of two WANT‐constructions to establish whether constructional competition contributes to English‐speaking children's infinitival to omission errors (e.g., *I want ___ jump now). In two between‐participant groups, children either just heard or heard and repeated WANT‐to, WANT‐X, and control prime sentences after which to‐infinitival constructions were elicited. We found that both age groups were primed, but in different ways. In the 2;6–3;0 year olds, WANT‐to primes facilitated the provision of to in target utterances relative to the control contexts, but no significant effect was found for WANT‐X primes. In the 3;6–4;0 year olds, both WANT‐to and WANT‐X primes showed a priming effect, namely WANT‐to primes facilitated and WANT‐X primes inhibited provision of to. We argue that these effects reflect developmental differences in the level of proficiency in and preference for the two constructions, and they are broadly consistent with “priming as implicit learning” accounts. The current study shows that (a) children as young as 2;6–3;0 years of age can be primed when they have only heard (not repeated) particular constructions, (b) children are acquiring at least two constructions for the matrix verb WANT, and (c) that these two WANT‐constructions compete for production.