Turbulence modulation, namely the effect of dispersed particles on turbulence intensity, has attracted a lot of attention and been a long-term controversy in the multiphase flow community. So far, the mechanism for turbulence modulation has not been fully clear, and there is not a general theory and method to qualitatively determine whether particles enhance or weaken turbulence, let alone accurately quantitatively predict turbulence modulation. To understand the underlying physics and characterize turbulence modification, a novel dimensionless parameter is derived in this letter by using dimensional analysis of the momentum equation for turbulent flows laden with finite-size solid particles combined with the Buckingham Pi theory. A set of 238 previous relevant experimental data are then carefully collected and used to validate the parameter. It is found that turbulence modulation can be successfully classified by this parameter. Turbulence fluctuation is attenuated in the region with a low value of the new parameter and is enhanced otherwise. An empirical formula is proposed to quantitatively estimate the turbulence modulation through multivariate linear regression based on large amounts of previous experimental data. (C) 2016 Elsevier B.V. All rights reserved.