In linguistics, incorporation is a phenomenon by which a grammatical category, such as a verb, forms a compound with its direct object (object incorporation) or adverbial modifier, while retaining its original syntactic function. The inclusion of a noun qualifies the verb, narrowing its scope rather than making reference to a specific entity.

Incorporation is central to many polysynthetic languages such as those found in North America, Siberia and northern Australia. However, polysynthesis does not necessarily imply incorporation (Mithun 2009), and the presence of incorporation does not imply that the language is polysynthetic.

Examples of incorporation English Although incorporation does not occur regularly, English uses it sometimes: breastfeed, and direct object incorporation, as in babysit. Etymologically, such verbs in English are usually back-formations: the verbs breastfeed and babysit are formed from the adjective breast-fed and the noun babysitter respectively. Incorporation and plain compounding may be fuzzy categories: consider backstabbing, name-calling, axe murder.

Oneida The following example from Oneida (Iroquoian) illustrates noun incorporation.

In this example, the verbal root hninu appears with its usual verbal morphology: a factive marker (FACT), which very roughly translates as past tense, although this is not quite accurate; an agreement marker (1.SG), which tells us that the verb agrees with 1st person singular (the speaker); and an aspect marker, punctual (PUNC), which tells us that this is a completed event. The direct object ne kanaktaʼ follows the verb. The function of the particle ne is to determine the bed: in the example, I bought this specific bed. The word for bed consists of a root nakt plus a prefix and a suffix. The notion of the root is important here, but the properties of the prefix and suffix do not matter for this discussion.

In the following sentence, the bed is unspecified. Unspecified nouns can be incorporated, thus creating a general statement. In this example: I bought a bed (and not a specific bed). In a broader sense, depending on context, it can even mean that I am a bed buyer, as in: I am a trader of beds, buying beds is my profession.

In this example, the root for bed nakt has incorporated into the verbal construction and appears before the verbal root. Two other incidental changes are noticed here. First, the agreement marker in the first example is k and in the second example is ke. These are two phonologically-conditioned allomorphs. In other words, the choice between using k and ke is based on the other sounds in the word (and has nothing to do with noun incorporation). Also, there is an epenthetic vowel a between the nominal and verbal roots. This vowel is inserted to break up an illegal consonant cluster (and also has nothing to do with noun incorporation).

Panare The next example, from Panare, illustrates the cross-linguistically common phenomenon that the incorporated form of a noun may be significantly different from its unincorporated form. The first sentence contains the incorporated form u' of "head", and the second its unincorporated form ipu:

Chukchi Chukchi, a Chukotko-Kamchatkan language spoken in North Eastern Siberia, provides a wealth of examples of noun incorporation. The phrase təpelarkən qoraŋə means "I'm leaving the reindeer" and has two words (the verb in the first person singular, and the noun). The same idea can be expressed with the single word təqorapelarkən, in which the noun root qora- "reindeer" is incorporated into the verb word.

Mohawk Mohawk, an Iroquoian language, makes heavy use of incorporation, as in: watia'tawi'tsherí:io "it is a good shirt", where the noun root atia'tawi "upper body garment" is present inside the verb.

Cheyenne Cheyenne, an Algonquian language of the Great Plains, also uses noun incorporation on a regular basis. Consider nátahpe'emaheona, meaning "I have a big house", which contains the noun morpheme maheo "house".

Chinese (Mandarin) Chinese makes extensive use of verb-object compounds, which are compounds composed of two constituents having the syntactic relation of verb and its direct object. For example, the verb 'sleep (VO)' is composed of the verb 'sleep (V)' and the bound morpheme object 'sleep (N)'. Aspect markers (e.g. PERFECTIVE), classifier phrases (e.g. THREE + CL + hours), and other elements may separate the two constituents of these compounds, though different verb-object compounds vary in degree of separability.

Turkish The verb etmek in Turkish always has an incorporated noun object: it cannot occur without one. For example, the noun yardım means "help"; the verbal complex yardım etmek means "to help", with the person being helping occurring in the dative case, e.g. bana yardım etti "s/he helped me", with the first person singular pronoun ben "I" in the dative case bana "to me". The verb kaybetmek evolved from kayıp etmek (both mean "to lose"); kayıp means "loss"; kayıp olmak "to be lost" evolved into kaybolmak.

Noun incorporation Sapir (1911) and Mithun (1984) define noun incorporation (NI) as "a construction in which a noun and a verb stem combine to yield a complex verb". Due to the wide variation in how noun incorporation presents itself in different languages, however, it is difficult to create an agreed upon and all-encompassing definition. As a result, most syntacticians have focused on generating definitions that apply to the languages they have studied, regardless of whether or not they are cross-linguistically attested.

In many cases, a phrase with an incorporated noun carries a different meaning with respect to the equivalent phrase where the noun is not incorporated into the verb. The difference seems to hang around the generality and definiteness of the statement. The incorporated phrase is usually generic and indefinite, while the non-incorporated one is more specific.

In Yucatec Maya, for example, the phrase "I chopped a tree", when the word for "tree" is incorporated, changes its meaning to "I chopped wood". In Lahu (a Tibeto-Burman language), the definite phrase "I drink the liquor" becomes the more general "I drink liquor" when "liquor" is incorporated. The Japanese phrase 目を覚ます me o samasu means "to wake up" or literally to wake (one's) eyes. But when the direct object is incorporated into the nominal form of the verb, the resulting noun 目覚まし mezamashi literally means "waking up", as in 目覚まし時計 mezamashidokei meaning "alarm clock."

This tendency is not a rule. There are languages where noun incorporation does not produce a meaning change (though it may cause a change in syntaxas explained below).

Noun incorporation can interact with the transitivity of the verb it applies to in two different ways. In some languages, the incorporated noun deletes one of the arguments of the verb, and this is shown explicitly: if the verb is transitive, the derived verb word with an incorporated noun (which functions as the direct object) becomes formally intransitive and is marked as such. In other languages this change does not take place, or at least it is not shown by explicit morphology. A recent study found out that across languages, morphosyntactically highly transitive verbs and patientive intransitive verbs are most likely to perform noun incorporation.

Incorporation looks at whether verb arguments, its nominal complements, exist on the same syntactic level or not. Incorporation is characterized as a stem combination meaning it combines independent lexical items into a modal or auxiliary verb to ultimately form a complex verb. The stem of the verb will be the determiner of the new category in which the incorporation belongs and the noun which was incorporated drops its own categorical features and grammatical markings, if employed. This is done by the movement of the incorporated noun to its new position in syntax. When participating in noun incorporation, it allows for the speaker to represent an alternative expression to further explain and shift focus to the information being presented (Mithun 1984).

Although incorporation exists in many languages, incorporation is optional and non-obligatory. Incorporation is restricted to certain noun categories; namely on the degree to which they are animate or alive or suppletive forms.

If a language participates in productive compounding it does not allow for incorporation. An example of a compounding language is German. Respectively, if a language participates in incorporation it does not allow for productive compounding.

The most common type of NI is where the incorporated noun acts as the notional subject of the clause. This can be observed in Onondaga, Southern Tiwa and Koryak.

Types In 1985, Mithun introduced a four-type system to define the functionality and progression of noun incorporation in a language. This system is important as many discuss this, and it is widely applied to explain the differences in NI in languages. The four types are:

Lexical compounding: involves a verb incorporating a nominal argument. The resulting compound usually describes a noteworthy or recurring activity. The noun in these compounds are not commonly marked for definiteness or number. Manipulation of case roles: The second type uses the same process to manipulate case roles, incorporating the argument into the verb to allow for a new argument to take its place. Manipulation of discourse structure: The third type uses noun incorporation to background old or established information. A speaker might explicitly mention an entity once, for example, and thereafter refer to it using an incorporated verbal compound. This kind of noun incorporation is usually seen in polysynthetic languages. Classificatory incorporation: The fourth and final type proposed by Mithun involves the development of a set of classificatory compounds, in which verbs are paired with generic nouns to describe properties of an entity, rather than the entity itself.

According to Mithun, languages exhibiting any of these types always display all of the lower types as well (that is, a language exhibiting type 3 would also display type 1 and 2, for example). This seems to imply a pattern of progression, as Mithun describes in her 1984 paper on the evolution of noun incorporation. It is argued that it is necessary to distinguish at least two types of noun incorporation.

Accounts

A large field of inquiry addresses whether NI is a syntactic process (verb and noun originate in different nodes and come together through syntactic means), a lexical process (word-formation rules that apply in the lexicon dictate NI), or a combined process (entailing the investigation of which aspects of noun incorporation can be productively created through general syntactic rules and which must be specified in the lexicon). This will vary from one case to another, as some languages, primarily those labelled as polysynthetic, allow for incorporated structures in a wide variety of sentences, whereas in others (such as English) this incorporation is more limited. Theories of morphology-syntax interaction and the debate between syntactic and lexical accounts of NI strive to be restrictive enough to account for the stable properties of NI in a unified way, but also account for language-specific variations. While this section discusses the influential syntactic and combined approaches to NI, highly-influential lexical accounts, such as Rosen's (1989) paper, do exist.