One of the fundamental parameters by which human beings conceptualize a situation is by the control cline: One of the participants in a situation has most control, the others have less or no control over it and over the other participants. This asymmetry is clearest in situations with two participants. This is represented in Diagram 1.

Control cline

control	full	none
role	agent	patient

Likewise, in coding a situation with two participants, almost all languages distinguish them by the criterion of control. Depending on the alignment of primary syntactic relations, the agent is coded as subject, ergative or active actant, while the patient is coded as object, absolutive or inactive actant. In most languages, a syntactic distinction along these lines is made in transitive clauses, as in a. The active and inactive role in intransitive clauses, as in a vs #b, are seldom overtly marked differently. Moreover, the formal schema of the transitive clause is grammaticalized in most languages and may then mislead one as to the control cline, as in b.

However, in all these cases, tests are applicable which yield a clear semantic difference. One relatively reliable test frame is embedding the clause in question below a control verb, as in , where the subject of the clause to be tested is inserted in the position marked by X and the rest is embedded under the matrix verb.

Embedding the #a sentences of and in the contexts of either a or #b (e.g. Linda tried to break the twig) is fine, while embedding the #b sentences in the same contexts (e.g. Linda refused to fall down the stairs) yields questionable results. By this criterion, the subject of the #a sentences of and has control, i.e. it is an agent, while the subject of the #b sentences has no control and is a patient. By the same token, the #a sentences designate actions or acts, while the #b sentences designate processes or events.

There are also tests on intentionality of the subject. One of them is the frame shown in , where the subject of the clause to be tested is replaced by you and the rest follows at the end of the frame.

Again, the result is fine with the #a sentences (e.g. What for did you work?), while it is weird with the #b sentences (e.g. What for did you suffer a stroke?). Since control involves intentionality, this again produces the same distinction among the sentences of and .

Most languages have productive processes that mark or change control relations in a clause. The most common construction that marks a participant as having highest control in a situation is the causative construction. German once had a causative derivation by root vowel modification, some of whose remnants are shown in – .

The mirror image of causativization is deagentivization alias anticausativization, illustrated by the conversion operation of f.

The #b examples suppress the actant which otherwise would have control in the situation, thus conveying a situation that happens by itself. The above set of examples alludes to shelves of literature which show that control and its absence are a chief structuring factor of the verbal grammar in languages all over the globe. One is entitled to conclude that the concepts transported by these linguistic structures play an important role in human cognition and communication about situations.