Morphosyntactic Alignment - Semantics and Grammatical Relations

Semantics and Grammatical Relations

Transitive verbs have two core arguments, labelled A (the more active or in-control) and O, which in a language like English are subject (A) and object (O). Intransitive verbs have a single core argument, labelled S, which in English (but not in all languages) is also a subject.

(The label P is sometimes used in place of O. Note that while the labels S, A, O, and P originally stood for "subject", "agent", "object", and "patient", the concepts of S, A, and O/P are distinct both from the traditional grammatical relations "subject" and "object", and from the thematic relations "agent" and "patient": an A or S need not be an agent or subject, an O need not be a patient, even in a language where they usually are.)

Generally, not all three of these labelled arguments (S, A, O) are distinct. In most languages, two are treated one way (such as the subject in English, which covers S and A) and the third is distinct (as the object in English, which covers O). That is, the labels are abstractions. They help illustrate the similarities and differences between languages, but are not real in all but a handful. There is no reality to S vs A in English, for example, only between S/A and O; in other languages, A will be realized, but not S vs O.

The best-known system is the English type, with a distinct O (object). These are called nominative–accusative languages, or just accusative languages, after the nominative and accusative cases, which are how A/S and O are distinguished in Latin. The best known of the other systems is the ergative system, named after the ergative case, which is how A is marked in many languages (such as Inuit) which distinguish A from S/O. Yet other systems are found in many Native American languages and in the Philippines. Many Australian languages will sometimes be one, and sometimes another, depending on the situation.

The principal types, or alignments—so called because S is said to "align" with either A (as in English) or O (as in Inuit)—are summarized here. The distinction between the two most common systems is expanded upon in the next section. For further information, including the rarer systems, see the individual articles.

1. Nominative–accusative (or accusative) alignment treats the S argument of an intransitive verb like the A argument of transitive verbs, with the O argument distinct (S = A; O separate) (see nominative–accusative language). In a language with morphological case marking, an S and an A may both be unmarked or marked with the nominative case, while the O is marked with an accusative case (or sometimes an oblique case also used for dative or instrumental case roles) as occurs with nominative -us and accusative -um in Latin: Julius venit "Julius came"; Julius Brutum vidit "Julius saw Brutus". Languages with nominative–accusative alignment can detransitivize transitive verbs by demoting the A argument, and promoting the O to be an S (thus taking nominative case marking); this is called the passive voice. A majority of the world's languages have accusative alignment.
   An uncommon subtype is called marked nominative or nominative–absolutive. In such languages, the subject of a verb is marked for nominative case, but the object is unmarked, as are citation forms and objects of prepositions. Such alignments are only clearly documented in northeastern Africa (e.g. in Cushitic languages) and the southwestern United States (Yuman languages).
2. Ergative–absolutive (or ergative) alignment treats an intransitive argument like a transitive O argument (S=O; A separate) (see ergative–absolutive language). An A may be marked with an ergative case (or sometimes an oblique case also used for genitive or instrumental case roles), while the S argument of an intransitive verb and the O argument of a transitive verb are left unmarked or sometimes marked with an absolutive case. Ergative–absolutive languages can detransitivize transitive verbs by demoting the O and promoting the A to an S, thus taking the absolutive case; this is called the antipassive voice. About a sixth of the world's languages have ergative alignment. The best known are probably Inuit and Basque.
3. Fluid (or semantic) alignment (see active–stative languages) treats the arguments of intransitive verbs in the same way as the A argument of transitives (like English) in some cases, and like transitive O arguments (like Inuit) in other cases (S_a=A; S_o=O). For example, in Georgian, Mariamma imğera "Mary (-ma) sang", Mariam shares the same narrative case ending as in the transitive clause Mariamma c'erili dac'era "Mary (-ma) wrote the letter (-i)", while in Mariami iq'o Tbilisši revolutsiamde "Mary (-i) was in Tbilisi up to the revolution", Mariam shares the same case ending (-i) as the object of the transitive clause. Thus the arguments of intransitive verbs are not uniform in its behavior.
   The reasons for treating intransitive arguments like A or like O usually has a semantic basis. The particular criteria vary from language to language, and may either be fixed for each verb, or chosen by the speaker according to the degree of volition, control, or suffering of the participant, or to the degree of sympathy that the speaker has for the participant.
4. Austronesian alignment, also called Philippine-type alignment, is found in the Austronesian languages of the Philippines, Borneo, Taiwan, and Madagascar. These languages have both accusative-type and ergative-type alignments in transitive verbs. They are traditionally (and misleadingly) called "active" and "passive" voice, because the speaker can choose to use either one, rather like active and passive voice in English. But because they aren't true voice, terms such as "agent trigger" or "actor focus" are increasingly used for the accusative type (S=A), and "patient trigger" or "undergoer focus" for the ergative type (S=O). (The terms with "trigger" may be preferred over those with "focus" because these are not focus systems either—morphological alignment has a long history of confused terminology). Patient-trigger alignment is the default in most of these languages. For either alignment two core cases are used (unlike passive and antipassive voice, which have only one), but the same morphology is used for the "nominative" of the agent-trigger alignment and the "absolutive" of the patient-trigger alignment, so there is a total of just three core cases: common S/A/O (usually called nominative, or less ambiguously direct), ergative A, and accusative O. Some Austronesianists argue that these languages have four alignments, with additional "voices" that mark a locative or benefactive with the direct case, but most maintain that these are not core arguments and thus not basic to the system.
5. A very few languages make no distinction whatsoever between agent, patient, and intransitive arguments, leaving the hearer to rely entirely on context and common sense to figure them out. This S/A/O case is called direct, as it sometimes is in Austronesian alignment.
6. Some others, called tripartite languages, use a separate case or syntax for each argument, which are conventionally called the accusative case, the intransitive case, and the ergative case. The Nez Perce language is a notable example.
7. Certain Iranian languages, such as Rushani, distinguish only transitivity (in the past tense), using a transitive case for both A and O, and an intransitive case for S. This is sometimes called a double-oblique system, as the transitive case is equivalent to the accusative in the non-past tense.

These direct, tripartite, and transitive are all quite rare. The alignment types other than Austronesian can be shown graphically like this:

In addition, in some languages both nominative–accusative and ergative–absolutive systems may be used, split between different grammatical contexts. This is called split ergativity. The split may sometimes be linked to animacy, as in many Australian Aboriginal languages, or to aspect, as in Mayan languages. A few Australian languages, such as Diyari, are split between accusative, ergative, and tripartite alignment, depending on animacy.

A popular idea, introduced in Anderson (1976), is that some constructions universally favor accusative alignment while others are more flexible. In general, behavioral constructions (control, raising, relativization) are claimed to favor nominative–accusative alignment, while coding constructions (especially case constructions) do not show any alignment preferences. This idea underlies early notions of ‘deep’ vs. ‘surface’ (or ‘syntactic’ vs. ‘morphological’) ergativity (e.g. Comrie 1978; Dixon 1994): many languages have surface ergativity only, i.e. ergative alignments only in their coding constructions (like case or agreement) but not in their behavioral constructions, or at least not in all of them. Languages with deep ergativity, i.e. with ergative alignment in behavioral constructions, appear to be less common.