It is shown that in the constituent quark model of Georgi-Manohar, the
dispersion relation that leads to the Adler-Weisberger sum rule for t
he axial vector coupling g(A) requires a subtraction constant. This fa
ct explains the discrepancy between the results of different recent es
timates of the 1/N(c) corrections to Weinberg's large N(c) result g(A)
= 1, where N(c) is the number of QCD colours. We also discuss a possi
ble scenario which shows that g(A) = 1 might not be a necessary conseq
uence of QCD in the large N(c) limit.