Ergative-Absolutive Alignment

The S, A, P question

Every clause in every human language has at most three core arguments. Linguists label them with single letters:

• S — the sole argument of an intransitive verb. The woman slept.
• A — the more agent-like argument of a transitive verb. The woman saw the dog.
• P (sometimes O) — the more patient-like argument of a transitive verb. The woman saw the dog.

Every language has to decide which of these to group together morphologically. Three logically possible alignments dominate:

• Nominative-accusative: {S, A} share one form ("nominative"), P takes a different form ("accusative"). English, Spanish, Russian, Japanese.
• Ergative-absolutive: {S, P} share one form ("absolutive"), A takes a different form ("ergative"). Basque, Georgian, Inuktitut.
• Tripartite: S, A, and P each take a different form. Rare; Nez Perce, Wangkumara.

Stop and notice the strangeness of ergative grouping. Why would a language treat the woman in "the woman slept" the same way as the dog in "the woman saw the dog"? But about a sixth of the world's languages do exactly that — and from inside the system it feels just as natural as English does to its native speakers.

Worked example: Basque case marking

Basque is the textbook ergative language because the marking is overt and the system is consistent. Take three sentences:

Intransitive:
  Gizon-a    etorri da.
  man-ABS    came   AUX
  "The man came."

Transitive:
  Gizon-a-k  emakume-a ikusi du.
  man-ABS-ERG woman-ABS saw   AUX
  "The man saw the woman."

Transitive (reversed):
  Emakume-a-k gizon-a    ikusi du.
  woman-ABS-ERG man-ABS    saw   AUX
  "The woman saw the man."

The bare form gizona ("the man") is absolutive — used for S in the first sentence and for P in the second. When gizon is the agent of a transitive in sentence 3, it picks up the ergative suffix -k: gizonak. The auxiliary verb also agrees differently: da for intransitive (one absolutive argument), du for transitive (one absolutive plus one ergative). Both case and agreement participate in the ergative pattern.

Compare what English does with the same sentences. The man is identical in all three; English doesn't mark its arguments at all on full nouns, only on pronouns (he saw her vs she saw him), and even there the alignment is accusative — he is used for both "he came" and "he saw her".

Hindi-Urdu: split ergativity in action

Hindi-Urdu is the most famous split-ergative language. It uses ergative alignment in the perfective tenses and accusative alignment everywhere else. Compare:

Imperfective (accusative-aligned):
  Laṛkā    kitāb        paṛhtā          hai.
  boy.NOM  book.NOM/ACC read.IMPV.M.SG  is
  "The boy reads the book."

Perfective (ergative-aligned):
  Laṛke    ne   kitāb        paṛhī.
  boy.OBL  ERG  book.NOM/ABS read.PFV.F.SG
  "The boy read the book."

Two things change in the perfective. First, the agent laṛkā goes into oblique case and picks up the ergative postposition ne. Second — and this is the giveaway that the alignment has flipped — the verb agrees with kitāb ("book"), not with the agent: paṛhī is feminine singular because kitāb is feminine. In the imperfective the verb agrees with laṛkā; in the perfective it agrees with kitāb. The agreement target follows the absolutive.

This is a textbook case because Hindi-Urdu's ergative emerged historically from a passive participle: the modern perfective descends from an Old Indic construction roughly translatable as "the book is read by the boy." Over time the participle was reinterpreted as an active verb, but the original agent kept its oblique case, becoming the new ergative. Split ergativity isn't usually designed — it's left over from the route the language took.

Alignment typology at a glance

Alignment	Grouping	S marked as	A marked as	P marked as	Examples	Approx. share of languages
Nominative-accusative	{S, A} = nominative	NOM	NOM	ACC	English, Spanish, Russian, Japanese	~50%
Ergative-absolutive	{S, P} = absolutive	ABS	ERG	ABS	Basque, Inuktitut, Mayan	~16–25%
Split ergative (tense/aspect)	changes per tense	NOM/ABS	NOM/ERG	ACC/ABS	Hindi-Urdu, Pashto, Georgian (perfective)	large subset of "ergative" languages
Split ergative (animacy/person)	changes per argument	varies	varies	varies	Dyirbal (Australia), Yidiny	several Australian languages
Active-stative	S split: agentive vs patientive	NOM or ACC depending on volition	NOM	ACC	Lakota, Acehnese, some Caucasian	small minority
Tripartite	S, A, P all distinct	distinct	distinct	distinct	Nez Perce, Wangkumara	very rare
Direct (neutral)	none — no marking	unmarked	unmarked	unmarked	Mandarin (relies on word order)	~30%, mostly isolating

Morphological vs syntactic ergativity

"Ergative language" is often a morphological label only. Most ergative languages still treat {S, A} as a unit for the purposes of higher-level syntax — coordination, control, relativisation. Basque is morphologically ergative but its pivot for "subject" in coordination is, like English, {S, A}: "the man came and saw the woman" works the same way it does in English.

The poster child for syntactic ergativity — where {S, P} act as a true grammatical pivot — is Dyirbal, an Australian Aboriginal language famously analysed by R. M. W. Dixon. In Dyirbal, the equivalent of "the man came and was-hit" is grammatical (S of the first clause = P of the second), but "the man came and hit X" requires extra antipassive morphology to coerce the second clause into using P as pivot. Genuine deep syntactic ergativity of this kind is rare. Most Mayan languages also show partial syntactic ergativity, especially in wh-questions and relative clauses.

Ergative agreement

Ergativity can show up in case marking, in agreement, or in both. Inuktitut shows it in both. Mayan languages show it primarily in agreement morphology — verbs carry two pronominal markers, set A (used for transitive agents) and set B (used for transitive patients and intransitive subjects). For instance Tzotzil:

Intransitive:
  Ch-i-bat.
  ASP-1B-go
  "I'm going."   (1st person = set B / absolutive)

Transitive:
  Ch-a-j-mul-an.
  ASP-2B-1A-love-TR
  "I love you."   (1st person agent = set A / ergative; 2nd person patient = set B / absolutive)

Notice that the same first-person referent appears as i- (set B) in the intransitive and as j- (set A) in the transitive. The pivot is grammatical role, not lexical identity.

Why ergative alignment exists

From a discourse standpoint, both alignments are sensible solutions to the same problem. Almost every language clearly distinguishes A from P — agents and patients are too important to confuse. The interesting question is where to put S, which has no contrast partner.

• Nominative reasoning: S resembles A in being typically the topic, the most prominent referent. So group them.
• Ergative reasoning: S resembles P in being unmarked semantically — neither acts on anything. So group them.

Both strategies are coherent. The diachronic record shows that languages can drift between them, often via reinterpretation of passive constructions (Indo-Iranian) or of nominalised clauses. Polynesian languages have moved in both directions over the last few thousand years.

How to diagnose alignment

• Compare three minimal sentences: S only, A + P, just to see what marking each role attracts.
• Check the verb agreement. Does the verb agree with both arguments? Does intransitive agreement pattern with transitive subject or object?
• Check 1st/2nd person pronouns separately. Splits often cut between pronouns and full nouns.
• Check tense/aspect. Hindi-style splits show up only when you compare imperfective and perfective sentences.
• Probe coordination and relativisation. Even a morphologically ergative language is unlikely to be syntactically ergative; testing pivots is the only way to know.

Common pitfalls

• Calling ergatives passives. Ergative is the active, default voice. Don't confuse the historical origin with the synchronic function.
• Translating "ergative" as "subject." The ergative argument is not the subject — it's the agent of a transitive. The intransitive subject is in the absolutive.
• Assuming all ergative languages behave alike. Basque is morphologically ergative but syntactically accusative. Dyirbal is both. Hindi is split. The category covers wide variation.
• Looking only at case marking. Some languages (Mayan, parts of Salish) encode ergativity solely through agreement morphology, not noun marking.
• Forgetting the antipassive. Just as accusative languages have a passive that demotes A, ergative languages often have an antipassive that demotes P, putting the agent into absolutive — a useful tool when {S, A} need to act as a syntactic pivot.

Why ergative-absolutive alignment matters

• Typology. Alignment is one of the most-cited parameters in cross-linguistic classification.
• Theoretical syntax. Generative theories of case and agreement must accommodate ergative patterns.
• Historical linguistics. Alignment shifts trace deep contact and reanalysis events.
• Field documentation. Many endangered languages are ergative; misanalysing alignment can wreck a grammar.
• Cognitive science. Ergativity is a clean test case for whether grammar shapes the conceptualisation of agency (it largely doesn't).
• L2 pedagogy. Teaching Basque, Georgian, or Hindi to Anglophones requires explicit attention to the unfamiliar grouping.