The study investigates the development of English multiword negation, in particular the negation of zero marked verbs (e.g. no sleep, not see, can't reach) from a usage-based perspective. The data was taken from a dense database consisting of the speech of an English-speaking child (Brian) aged 2;3–3;4 (MLU 2·05–3·1) and his mother. The focus of the study was the emergence and usage of negators in the child's and mother's speech (e.g. no, not, can't, won't, don't). Two analyses were conducted: firstly, the emergence and usage of all negators in Brian's speech and in the input were calculated in order to present an overall picture of negator usage across the sample. The findings indicate a gradual and systematic development of negator selection in Brian's speech which follows the trajectory no–not–'nt. The pattern of negator emergence was found to follow the frequency of negators in the input; that is negators used frequently in the input were the first to emerge in the child's speech. Secondly, a more fine-grained analysis of utterances containing negated zero marked verbs (neg V utterances) was conducted on both the child's and mother's speech. In the first instance the development of negator selection for all neg V utterances was calculated. The results indicated the same no–not–'nt cline as attested in the initial analysis. A function-based analysis of neg V utterances was also conducted which indicated that the speed of movement across the no–not–'nt cline varied from one function to the next. A function-based analysis of the input suggested that the speed at which Brian moved across the cline within a particular function could be traced to function-based frequency effects in the input. Thus the findings of the study indicate function-based, input-driven learning which is consistent with the usage-based approach. However the findings also indicate creative learning on the part of the child from the earliest stages of multiword negation.