An active–stative language (active language for short), also commonly called a split intransitive language, is a language in which the sole argument ("subject") of an intransitive clause (often symbolized as S) is sometimes marked in the same way as an agent of a transitive verb (that is, like a subject such as "I" or "she" in English) but other times in the same way as a direct object (such as "me" or "her" in English).

The case or agreement of the intransitive argument (S) depends on semantic or lexical criteria particular to each language. The criteria tend to be based on the degree of volition, or control over the verbal action exercised by the participant.

For example, if one tripped and fell, an active–stative language might require them to say the equivalent of "fell me." To say "I fell" would mean that the person had done it on purpose, such as taking a fall in boxing. Another possibility is empathy; for example, if someone's dog were run over by a car, one might say the equivalent of "died her." To say "she died" would imply that the person was not affected emotionally.

If the core arguments of a transitive clause are termed A (agent of a transitive verb) and P (patient of a transitive verb), active–stative languages can be described as languages that align intransitive S as S = P/O∗∗ ("me fell") or S = A ("I fell"), depending on the criteria described above.

Active–stative languages contrast with accusative languages such as English that generally align S as S = A, and to ergative languages that generally align S as S = P/O.

Care should be taken when reasoning about language structure, specifically, as reasoning on syntactic roles (S=subject/ O=object) is sometimes difficult to separate from reasoning on semantic functions (A=agent/ P=patient). For example, in some languages, "me fell," is regarded as less impersonal and more empathic.

For most such languages, the case of the intransitive argument is lexically fixed for each verb, regardless of the actual degree of volition of the subject, but often corresponding to the most typical situation. For example, the argument of swim may always be treated like the transitive subject (agent-like), and the argument of sleep like the transitive direct object (patient-like). In Dakota, arguments of active verbs such as to run are marked like transitive agents, as in accusative languages, and arguments of inactive verbs such as to stand are marked like transitive objects, as in ergative languages. In such language, if the subject of a verb like run or swallow is defined as agentive, it will be always marked so even if the action of swallowing is involuntary. This subtype is sometimes known as split-S.

In other languages, the marking of the intransitive argument is decided by the speaker, based on semantic considerations. For any given intransitive verb, the speaker may choose whether to mark the argument as agentive or patientive. In some of these languages, agentive marking encodes a degree of volition or control over the action, with the patientive used as the default case; in others, patientive marking encodes a lack of volition or control, suffering from or being otherwise affected by the action, or sympathy on the part of the speaker, with the agentive used as the default case. These subtypes are sometimes known as fluid-S.

If the language has morphological case, the arguments of a transitive verb are marked by using the agentive case for the subject and the patientive case for the object. The argument of an intransitive verb may be marked as either.

Languages lacking case inflections may indicate case by different word orders, using adpositions, etc. For example, the patientive argument might precede the verb, and the agentive argument might follow the verb.

Cross-linguistically, the agentive argument tends to be marked, and the patientive argument tends to be unmarked. That is, if one case is indicated by zero-inflection, it is often the patientive.

Active languages are a relatively new field of study. Active morphosyntactic alignment used to be not recognized as such, and it was treated mostly as an interesting deviation from the standard alternatives (nominative–accusative and ergative–absolutive). Also, active languages are few and often show complications and special cases ("pure" active alignment is an ideal).

Thus, the terminology used is rather flexible. The morphosyntactic alignment of active languages is also termed active–stative alignment or semantic alignment. The terms agentive case and patientive case used above are sometimes replaced by the terms active and inactive.

(†) = extinct language

According to Castro Alves (2010), a split-S alignment can be safely reconstructed for Proto-Northern Jê finite clauses. Clauses headed by a non-finite verb, on the contrary, would have been aligned ergatively in this reconstructed language.

The reconstructed Pre-Proto-Indo-European language, not to be confused with the Proto-Indo-European language, its direct descendant, shows many features known to correlate with active alignment like the animate vs. inanimate distinction, related to the distinction between active and inactive or stative verb arguments. Even in its descendant languages, there are traces of a morphological split between volitional and nonvolitional verbs, such as a pattern in verbs of perception and cognition where the argument takes an oblique case (called quirky subject), a relic of which can be seen in Middle English methinks or in the distinction between see vs. look or hear vs. listen. Other possible relics from a structure, in descendant languages of Indo-European, include conceptualization of possession and extensive use of particles.