Hans Bork
The class of Impersonal Passive verbs in Latin has been diversely treated in the grammatical literature, and the variety of methodologies used to identify Impersonal Passives has obscured our understanding of these constructions in authors such as Plautus. Paradigmatic examples of impersonal passive (hereafter IP) verbs are pugnatum est and itur, which are 0-argument passives that have been generated from intransitive (1-argument) base verbs, and which show neuter singular agreement. Thus IP verbs are similar to “active” 0-argument environmental verbs such as ningit and pluit, and Pinkster (1992) posits that Latin impersonal passives are in fact prototypically generated from the reduction of 1-argument verbs to 0-argument verbs through passivization. In contrast to this formal definition of IP verbs, some scholars classify as “impersonal” any verb of passive morphology that lacks an overt agent (e.g., valetur in Wackernagel (2009) and dicitur/videtur in Hofmann-Szantyr (1965) and Allen and Greenough (2001).) The coexistence of these two approaches in the grammatical literature makes for a large pool of potential IP verbs, many of which have conflicting or heterogeneous argument structures.
The status of IP verbs in Plautus is particularly difficult, due to the still-open question of whether Plautine grammar is wholly rule-based, or whether it is somewhat “experimental” in generating stylistically conditioned forms (e.g., nonce compounds and wordplay). Lindsay (1907) comments that “we see in Plautus a marked predilection for the 3 sing. Pass. [sic] used impersonally,” and goes on to list forms such as caletur, amatur, vivitur, nubitur and certum est as examples. Christenson (2000) makes a similar claim, and this opinion seems to be the norm among commentators. It is clear that the verbs in Lindsay’s list have diverse argument structures, however (e.g., amatur is from a 2-argument verb, caletur from a 1-argument verb, certum est is a collocation that takes a sentential complement), and there has been little systematic study of where and how often IP verbs occur in Plautus.
Modern syntactic typologies (e.g., Dixon (2010), Malchukov and Siewierska (2011), and Napoli (2013)) distinguish between raising/sentential complement verbs such as videtur or certum est and “true” impersonal constructions such as pugnatum est or itur (all found in Plautus). Unlike itur, which is a 0-argument verb derived from an intransitive base, verbs such as videtur and certum est are constructions that take clausal arguments, and thus are not impersonal (i.e., 0-argument) in argument structure. If a collection of putative IP tokens from Plautus are examined taxonomically, it becomes clear that the most abundant types are not 0-argument impersonals, but rather are of the certum est type. In fact, there are over 78 instances of the certum est collocation alone in Plautus, and all but one occur with the verbal constituents in precisely that order. Add to this the fact that certum est is almost never split by other constituents, and one can propose that the construction is in fact a grammaticalized, phonological word, rather than a productive passive formation. In contrast, there are several unusual IP verbs in Plautus that are derived from 2-argument verbs (e.g., amatur, bibitur, coquetur). These forms should not be allowed under the normal template of IP generation, and yet multiple examples exist; they are not productive, however, and it is likely that they are in fact deliberate syntactic distortions that originated from the established IP generational template. IP verbs such as itur and pugnatum est were used by speakers to designate abstract verbal states; Plautus created “irregular” IP forms by generalizing the rule {verb[+pass,-agent] = abstract verbal action} without regard to the argument of the verb. Such innovations would have been comprehensible, but not iterable, due to the larger constraints of IP generation rules within the speech community. Examination of the problem in this way thus provides insight into both Plautine stylistics, and the syntax of early Latin.