RMSS/FRP definitely requires speaking (or taking a penalty). In RM2, the ESF rules are really the only place where all the requirements are laid out in one place, and there is no mention of speech (or other vocalization). Now, properly speaking, a "spell" is the verbal (that is, composed of words or pseudo-words, and thus includes inscribed spells) and writing would be an odd choice in a tactical casting system, so one might claim that an oral component is involved. However, the use of "spell" in the game is clearly just a term of art for an action or set of actions that produces a magical effect (what is properly called a magical ritual, of which a spell may be the whole or merely a component or even absent). It seems most likely that it was merely assumed that Rolemaster "spells" were to be spoken and that the gaming community would share that assumption. I think that given such a strong assumption that they didn't even feel a need to state it, that it would be natural enough to not relax that (unstated) requirement with ESF rules (or the ESF rules were written just looking at the stated requirements and just never even thought about silent casting, since the need to speak was not mentioned).
The first Rolemaster Companion introduced "Magical Languages" to boost your spell-casting, which seems to accept the assumption that spells require language and also using Latin or Hebrew is better than English (or rather, the fantasy world equivalent of that notion).
So, I am reasonably confident that you were supposed to be uttering something while doing magic in RM2, but I am glad things were made clearer in RMSS. Finally, in RMU, the description of the modifiers to the spell-casting roll (and the obviousness of one's spell-casting) make it clear that both vocalizations and gestures are optional, but that it is easier to do magic with them than without them.
