Valency (linguistics)


In linguistics, valency or valence is the number of arguments controlled by a predicate, content verbs being typical predicates. Valency is related, though not identical, to subcategorization and transitivity, which count only object arguments – valency counts all arguments, including the subject. The linguistic meaning of valency derives from the definition of valency in chemistry. The valency metaphor appeared first in linguistics in Charles Sanders Peirce's essay "The Logic of Relatives" in 1897, and it then surfaced in the works of a number of linguists decades later in the late 1940s and 1950s. Lucien Tesnière is credited most with having established the valency concept in linguistics. A major authority on the valency of the English verbs is Allerton, who made the important distinction between semantic and syntactic valency.

Types

There are several types of valency:
  1. impersonal it rains
  2. intransitive she sleeps
  3. transitive she kicks the ball
  4. ditransitive she gave him a book
  5. tritransitive I bet her a dollar on a horse
The term valence also refers to the syntactic category of these elements. Verbs show considerable variety in this respect. In the examples above, the arguments are noun phrases, but arguments can in many cases be other categories, e.g.
Many of these patterns can appear in a form rather different from the ones just shown above. For example, they can also be expressed using the passive voice:
The above examples show some of the most common valence patterns in English, but do not begin to exhaust them. Other linguists have examined the patterns of more than three thousand verbs and placed them in one or more of several dozen groups.
The verb requires all of its arguments in a well-formed sentence, although they can sometimes undergo valency reduction or expansion. For instance, to eat is naturally divalent, as in he eats an apple, but may be reduced to monovalency in he eats. This is called valency reduction. In the southeastern United States, an emphatic trivalent form of eat is in use, as in I'll eat myself some supper. Verbs that are usually monovalent, like sleep, cannot take a direct object. However, there are cases where the valency of such verbs can be expanded, for instance in He sleeps the sleep of death. This is called valency expansion. Verb valence can also be described in terms of syntactic versus semantic criteria. The syntactic valency of a verb refers to the number of dependent arguments that the verb can have, while semantic valence describes the thematic relations associated with a verb.

Compared with subcategorization

Tesnière 1959 expresses the idea of valence as follows :
Tesnière used the word actants to mean what are now widely called arguments. An important aspect of Tesnière's understanding of valency was that the subject is an actant of the verb in the same manner that the object is. The concept of subcategorization, which is related to valency but associated more with phrase structure grammars than with the dependency grammar that Tesnière developed, did not originally view the subject as part of the subcategorization frame, although the more modern understanding of subcategorization seems to be almost synonymous with valency.

Changing valency

Most languages provide a means to change the valency of verbs. There are two ways to change the valency of a verb: reducing and increasing.
Note that for this section, the labels S, A, and P will be used. These are commonly used names given to arguments of a verb. S refers to the subject of an intransitive verb, A refers to the agent of a transitive verb, and P refers to the patient of a transitive verb.
These are core arguments of a verb:
Non-core arguments are called obliques and are typically optional:
Reducing valency involves moving an argument from the core to oblique status. The passive voice and antipassive voice are prototypical valency reducing devices. This kind of derivation applies most to transitive clauses. Since there are two arguments in a transitive clause, A and P, there are two possibilities for reducing the valency:
There are some problems, however, with the terms passive and antipassive because they have been used to describe a wide range of behaviors across the world's languages. For example, when compared to a canonical European passive, the passive construction in other languages is justified in its name. However, when comparing passives across the world's languages, they do not share a single common feature.
R. M. W. Dixon has proposed four properties of passives and antipassives.
  1. They apply to underlying transitive clauses and form a derived intransitive.
  2. The underlying P of the passive and A of the antipassive become S.
  3. The underlying A of the passive and P of the antipassive go into the periphery and are marked by a non-core case/preposition/etc. These can be omitted, but there's always the option of including them.
  4. There is some explicit marking of the construction.
He acknowledges that this excludes some constructions labeled as "passive" by some linguists.
Other ways to reduce valency include the reflexives, reciprocals, inverse constructions, middle voice, object demotion, noun incorporation, and object incorporation.

Valency-increasing

This involves moving an argument from the periphery into the core. Applicatives and causatives are prototypical valency increasing devices.

In syntactic theory

Valence plays an important role in a number of the syntactic frameworks that have been developed in the last few decades. In generalized phrase structure grammar, many of the phrase structure rules generate the class of verbs with a particular valence. For example, the following rule generates the class of transitive verbs:
H stands for the head of the VP, that is the part which shares the same category as the VP, in this case, the verb. Some linguists objected that there would be one such rule for every valence pattern. Such a list would miss the fact that all such rules have certain properties in common. Work in government and binding takes the approach of generating all such structures with a single schema, called the X-bar schema:
X and Y can stand for a number of different lexical categories, and each instance of the symbol ′ stands for a bar. So A′, for instance, would be a kind of AP. Two bars, used here for the complements, is thought by some linguists to be a maximal projection of a lexical category. Such a schema is meant to be combined with specific lexical rules and the projection principle to distinguish the various patterns of specific verbs.
Head-driven phrase structure grammar introduces a handful of such schemata which aim to subsume all such valence related rules as well as other rules not related to valence. A network is developed for information related to specific lexical items. The network and one of the schemata aims to subsume the large number of specific rules defining the valence of particular lexical items.
Notice that the rule and the schema deal only with non-subject complements. This is because all of the above syntactic frameworks use a totally separate rule to introduce the subject. This is a major difference between them and Tesnière's original understanding of valency, which included the subject, as mentioned above.
One of the most widely known versions of construction grammar also treats the subject like other complements, but this may be because the emphasis is more on semantic roles and compatibility with work in cognitive science than on syntax.