Pine & Lieven (1993) suggest that a lexically-based positional analysi
s can account for the structure of a considerable proportion of childr
en's early multiword corpora. The present study tests this claim on a
second, larger sample of eleven children aged between I;0 and 3;0 from
a different social background, and extends the analysis to later in d
evelopment. Results indicate that the positional analysis can account
for a mean of 60 % of all the children's multiword utterances and that
the great majority of all other utterances are defined as frozen by t
he analysis. Alternative explanations of the data based on hypothesizi
ng underlying syntactic or semantic relations are investigated through
analyses of pronoun case marking and of verbs with prototypical agent
-patient roles. Neither supports the view that the children's utteranc
es are being produced on the basis of general underlying rules and cat
egories. The implications of widespread distributional learning in ear
ly language development are discussed.