In this short paper, we demonstrate that the popular penalized estimation method typically used for variable selection in parametric or semiparametric models can actually provide a way to identify linear components in additive models. Unlike most studies in the literature, we are NOT performing...