If reference is that to which meaning “speaks” (metaphorically, because this is a multimodal grammar), agency is the patterning of action. Reference is the addressed; agency is the addressing. Three features of agency are event, role and conditionality. These are thoroughly named and analyzed in a linguistics of the clause, though the complexities at times confound. They are not so thoroughly analyzed in the other forms of meaning that are also of concern to us. Our focus is at a broader level of generality than linguistics, one that crosses multiple forms. Agency is constituted in events (predication by means of which entity into action fold into each other; and transactivity or the relations of entities-in-action to each other). Agents assume roles (as self, other, or thing). Different nuances of conditionality are established in the relations of entities and action (assertion, requirement, and possibility).


An event, as defined in this grammar, is manifest in agency as expressed in the relationships of entities to action, and entities-in-action to each other. An event happens when entities connect through action. Predication is a relation of a setting (the given, the found, a starting point), and the subsequent (what happens, effects, consequences). Transactivity is patterns of action, the kinds of relations established between entities in action.


We propose a classification roles into self, other, and thing, able to span forms of meaning. “Self” is quite different in speech than it is in text: the co-present self in speech compared to the distant, reporting self of writing. Here is another substantial difference which separates the grammars of speech and writing. So too with “other.” The co-present other of speech is different from the distant other of writing. We are going to include grammatical third persons within “other,” because we can distinguish the other of existence from the other of command in the grammar of conditionality. This “other” includes creatures whose sentience the sense maker recognizes—pets for instance, or wild animals, or insects. Finally we have things, insentient “it” and “they,” nevertheless capable of acting naturally and being acted upon.


Conditionality refers to modulations in relationships of agency between selves, others and things according to the qualities of these relationships: assertion, requirement, or possibility. These represent different kinds of agency, connecting selves, others and things. In grammars of text, these are called mood: indicative (entities acting in roles that have been, are, and will be); imperative (actions that must happen); and subjunctive (actions that might happen). In a language like English where there is minimal inflection, conditionality in text is marked by auxiliaries such as “is,” “has,” “can,” “may,” “might,” “will,” “ought,” “could,” “would,” and “should.” Grammarians and semantic theorists have come up with systems of conditionality that, for such ordinary aspects of human meaning, are alarmingly at variance with each other.