In Malay/Indonesian, when certain predicates such as "want" are followed by a passive verb, an ambiguity arises about who has the desire and other attitudes in question. The attitude-holder can be either the surface subject or the passive agent. This article critically assesses the data and claims presented in three recent studies (Mike Berger 2019; Paul Kroeger and Kristen Frazier 2020; Helen Jeoung 2020) through consideration of additional data. It shows that the ambiguity is empirically robust, contrary to the doubts expressed by Jeoung, and that the restructuring analysis advocated by the latter two studies has problems with its primary evidence: alternate voice marking realization. Instead, the paper confirms the previous understanding of the construction, including a bi-clausal structure with a dyadic matrix predicate and the importance of voice marking. Methodologically, it demonstrates that linguistic evidence should come from multiple sources, that is, not from elicitation or texts alone but from both of these (and perhaps more).