We investigate how enterprise modelers see common types (e.g., actor, event, process) used in most modeling languages in terms of their semantic feature structure (e.g., is human, is material). We hypothesize that modelers have specific interpretations for some of these common types that affect their range of conceptually valid instantiations (e.g., actors should not be instantiated as human things). Based on two exploratory psychometric studies performed with enterprise modeling practitioners and computing science students we discuss the way these typical interpretations affect their $model(ing)$ semantics (e.g., results typically having to be modeled as well-described and non-natural entities, restrictions typically as logical necessities), and what consequences these findings have for modeling languages and the use and creation of models themselves, especially in an inherently collaborative effort like enterprise modeling. We conclude by arguing that insights into these conceptualizations are likely useful and should receive more attention and studies.