Our study on sign lowering in ASL investigated which of the phonetic and variationist sociolinguistic perspectives better explain why ASL signs do not reach the target location of forehead, but instead lower to other locations below the forehead during naturalistic signed discourse. Based on the corpus study of 3000 ASL signs, findings revealed that lexical frequency appeared to be a driving mechanism in a gradient continuum of ASL lowering across locations (taking into account the phonetic perspective). A categorical lowering process was observed in a small set of ASL signs for two clusters of alternate locations per given ASL sign. Although our analysis partially supported both phonetic and variationist perspectives, we found that signers systematically avoided the eye area, suggesting that undershoot is a strategically planned, somatosensory, forward modeling mechanism. We argue that signers build their mental lexicon of ASL signs from the phonological distributions of ASL signs as they filter “good” and “bad” ASL forms either within acceptable ranges of undershoot for each sign or patterns that imitate the results of categorization processes (Russell et al, 2011).