Estimation of the effort required for a software project is difficult. Vari
ous means are used, but most rely on some expert assessment of the individu
al requirements and their implications. A method of supporting this assessm
ent for object-oriented developments is described. Lexical analysis of a dr
aft requirements specification can be used to identify individual objects w
hich will translate directly into the final implementation. These object co
unts can then be used to provide 'first-cut' effort estimates, using histor
ical information from previous projects. Experiments were conducted on a pr
oblem implemented by student project teams. The results show that the untra
ined domain-independent automated noun and technical term finding programs
used were no worse than the typical student group in deriving problem-space
objects, and that these object counts provided a reasonable indicator to t
he effort required. Further work in this area is discussed. (C) 2000 Elsevi
er Science Inc. All rights reserved.