Accounts of English initial consonant sequences suggest that not all sequen
ces are the same. Data from acquisition, speech errors and language games n
ecessitate unusual rules/constraints on subsyllabic structure to account fo
r C/j/ and /s/C(C) sequences. Acquisition evidence supports within- and acr
oss-speaker representational differences for these sequences; however, such
evidence for adults remains elusive. This study investigates the represent
ation of consonant sequences within and across adult speakers of English. T
hirty adults were trained in Pig Latin using words with singletons and true
clusters. Generalization to C/j/ and /s/C(C) was then measured. Results su
ggested differential patterning for consonant sequences within and across s
peakers based on consistent production patterns and errors. Differences occ
urred most often with /s/C(C) sequences and even more so for C/j/. Orthogra
phy also influenced production patterns. Results support differential repre
sentational structure within and across individual grammars, which is accou
nted for via constraint-based output-output correspondence between surface
and Pig Latin forms. (C) 2001 Elsevier Science B.V. All rights reserved.