According to Merchant’s (2001:62) Sluice-COMP Generalization (henceforth SCG), "in sluicing, no non-operator material may appear in COMP." In other words, the remnant, usually located in the CP periphery, cannot consist of non-operators (cf. Lobeck 1995). Therefore, the SCG predicts that only the wh-operator in (1), for example, is allowed to remain after the TP/FinP of the interrogative sentence is elided. However, Baltin (2010), using Gungbe as a case study, argues that in morphological focus marking languages, the focus marker, which is a non-operator, survives ellipsis, thus violating the SCG. We will call this the "Baltin’s Generalization" (henceforth BG) (see also Mendes & Kandybowicz  2023; Lipták & Aboh 2021; Aboh 2004). In this study, drawing on data from two morphological focus marking languages, Likpakpaanl (Mabia, SVO) and Yoruba (Kwa, SVO), we argue that both languages behave differently to SCG and BG.