Agency
The notion of agency is also important in RRG, serving as the basis for a further subcategorization of predicate types (Foley & Van Valin 1984:40). In fact, Van Valin (1990:248ff) argues that, in Acehnese, it is agentiveness rather than lexical semantics that conditions split intransitivity. Mithun 1991, citing Foley & Van Valin 1984, considers as a prototypical agent ‘the participant which performs, effects, instigates, or controls the situation denoted by the predicate’, while a prototypical patient ‘does not perform, initiate, or control any situation but rather is affected by it in some way’ (1991:516). She further shows how languages can be more sensitive to different specific features of agency: e.g. Lakhota differentiates performers from non-performers regardless of control, while Central Pomo separates controllers from non-controllers regardless of performativity (1991:516–20). Therefore, verbs of performance without control, such as hiccough, snore, cry, vomit, yawn, etc. pattern differently in Lakhota and Central Pomo.
As for Cariban languages, the examples in Table 1 seem to indicate that agentiveness (both as performance/effect/instigation and control/volitionality) is not a good class indicator.
Table 1. Agentive and patientive verbs in Cariban languages.
Van Valin 1990:252 predicted that agentive activity verbs should belong to the SA class in all languages. However, clear cases of SO agentive activity verbs can be found in the Cariban data, e.g. Tiriyó eremina ‘sing’, emamina ‘play’, urakana ‘go for a walk’.
