Alignment systems classify how agents (A), patients (P), and single arguments (S) of intransitive verbs are marked (through agreement, case, word order, or voice). Nominative-accusative systems treat S and A alike; ergative-absolutive systems treat S and P alike. Other systems are tripartite, horizontal, or hierarchical. Alignment is not universal but varies; it determines how semantic roles map to grammatical relations.
Compare alignment patterns across languages, using data on agreement, case-marking, and word-order to identify the alignment type. Examine which semantic and pragmatic factors motivate different alignments.
You've already worked through ergativity and the basic distinction between nominative-accusative and ergative-absolutive systems, and you've studied linguistic typology's broader project of mapping cross-linguistic variation. Alignment systems typology zooms out from those individual patterns to ask: across the world's languages, what are *all* the possible ways to group arguments, and what does that variation tell us about the relationship between grammar and meaning?
The core insight begins with three argument types. S is the single argument of an intransitive verb — the "she" in "she sleeps." A is the agent-like argument of a transitive verb — the "she" in "she kicked the ball." P is the patient-like argument of a transitive verb — the "ball" in "she kicked the ball." The logical question is: how many ways can these three types be grouped by grammatical marking? The options are mathematically constrained. If you treat S and A alike (and P differently), you get nominative-accusative alignment — the pattern familiar from English, Latin, and most European languages, where the subject of any verb (transitive or not) takes the same form. If you treat S and P alike (and A differently), you get ergative-absolutive alignment — found in Basque, many Australian languages, and numerous others, where the agent of a transitive clause is specially marked and the subject of an intransitive clause groups with the patient. If all three argument types are marked differently, you get a tripartite system. If they are all marked the same, you get a fully neutral system.
What makes alignment typology analytically powerful is that these patterns are not arbitrary: they reflect how different languages organize information about agency and affectedness. In a nominative-accusative language, the most "prominent" participant is the agent — S and A cluster together as the default grammatical role. In an ergative language, the absolutive (S and P) clusters as the default, decentering the agent. Different alignment choices encode different intuitions about which participant in an event is the most salient starting point for describing it. This is why typologists connect alignment to broader questions about event construal: languages differ not just in their morphology but in how they package events conceptually.
Real languages complicate the clean logical picture through split systems — the most important complication your prerequisite material introduced. A single language may show nominative-accusative alignment in some grammatical contexts and ergative-absolutive alignment in others. Georgian, for instance, shows nominative-accusative in the present tense and ergative-absolutive in the past. Many languages show ergative morphological case on nouns but nominative-accusative agreement on verbs. Splits often correlate with person hierarchies (first and second person arguments behaving differently from third person) or with tense/aspect distinctions (imperfective vs. perfective). These splits reveal that alignment is not a single, unified property of a language but a family of related phenomena that can partially decouple across grammatical subsystems — and understanding why splits occur where they do remains one of the productive open questions in typological linguistics.