Phrase (linguistics)

In linguistics, a phrase is a syntactic unit that is closed, ie "syntactically saturated", in contrast to units that are still missing additions. So it is a special case of a constituent . In the context of the X-bar theory , a phrase is also referred to as “maximum projection (of a head )”. The term sentence element , which is common in the Germanic tradition , usually denotes phrases, but is more narrowly defined because it is only applied to units that can be moved within the sentence.

The exact definition of the phrase differs depending on which grammar theory is used. In some constituent grammars , for example, individual words are counted as phrases if they do not require any further additions, while dependency grammars only recognize a syntactic unit as a phrase if it consists of more than one word.

Examples

The following sentence serves as a starting point:

„Nach der zu langen Party sind wir an den Fluss gefahren.“

The following combinations of words in a sentence are called phrases by most grammars:

too long - adjective phrase (AP)

to the river - prepositional phrase (PP)

after the (...) party - prepositional phrase (PP)

the too long party - noun phrase (NP)

In each of these phrases, the head of the phrase is italicized. The head or nucleus of a phrase determines the syntactic category of the entire phrase. If the head of a phrase is a noun, the phrase is a noun phrase ; if the head of a phrase is an adjective, the phrase is an adjective phrase , etc. The most controversial cases of phrase formation are the status of sentences as phrases and the status of the verb phrase . There is consensus that at least verbs in the infinitive can form phrases; these can be identified by normal constituent tests such as the changeover sample:

to the river run - verb phrase (VP)

Phrases are not always parts of sentences . For example, in the above sentence, the AP too long is not a part of a sentence. The essential feature of a part of a sentence is that it can be moved. Since the AP can not be moved for too long , it is not a part of a sentence: * We drove to the lake very quickly after the party for too long . It should also be noted here that all examples consist of at least two words. Some grammars - e.g. B. those who build on the X-bar theory - would also recognize the word we as a phrase, since we as the subject of the sentence forms a self-contained unit.

Phrases in phrase structure grammars

The phrase concept is mainly associated with the phrase structure grammars (= constituent grammars ). Sentences can be decomposed, and many of the constituents that result in the course of the decomposition are phrases. These phrases are specified in the structure tree with… P, e. B.

This structure tree shows only one of the possible analyzes of the set. In the present context it is relevant that each of the above phrases is identified as a phrase in the tree by the ... P. But there is also an additional word combination that is specified in the tree here (but not further above) as a phrase, namely the finite verb phrase “we drove to the lake very quickly”.

Possible dissent about this tree could concern the individual words. As mentioned above, some constituent grammars of the sentence would regard several of the individual words as phrases. In the tree here, on the other hand, only those word combinations are considered to be phrases that consist of more than one word, which corresponds more to the traditional usage of the word phrase.

Phrases in the dependency grammar

While the phrase concept originally comes from the phrase structure grammars (= constituent grammars), it can also be applied to dependency grammar structures. In the phrase structure grammar tree above, each subtree is a phrase made up of more than one word. Seen in this way, the trees of the dependency grammar also contain phrases:

This tree again shows only one of the possible dependency grammatical analyzes of the sentence. Any subtree that consists of more than one word is a phrase. There are six phrases in this tree, and these six phrases are the same as those mentioned above. One of the differences between the trees, however, is that the dependency tree does not label the phrases with ... P (here the P stands for preposition). However, a finite verb phrase is missing; the word combination “we drove to the lake very quickly” is not a phrase here because it does not form a complete subtree.

literature

Vilmos Ágel , Ludwig Eichinger, Hans-Werner Eroms, Peter Hellwig, Hans Heringer, Hennig Lobin (eds.): Dependenz and Valenz: An international handbook of contemporary research. Berlin. Walter de Gruyter 2003/2006.
Leonard Bloomfield: 1933. Language. New York. Henry Holt 1933.
Noam Chomsky: Syntactic Structures. The Hague / Paris. Mouton 1957.
Timothy Osborne, Michael Putnam, Thomas Groß: Bare phrase structure, label-less trees, and specifier-less syntax: Is Minimalism becoming a dependency grammar? The Linguistic Review 28, 315-364, 2011.
Rulon S. Wells: Immediate Constituents, in: Language 23, 81-117, 1947.
Lucien Tesnière: Elements de syntaxe structurale. Paris. Klincksieck 1959.

Remarks

↑ Constituent grammar is primarily associated with the works of Leonard Bloomfield (1933), Rulon Wells (1947), and the young Noam Chomsky (1957). The dependency grammar is based on the theory of Lucien Tesnière (1959); see also Ágel et al. (2003/6).
↑ The tree here corresponds to the development of constituent grammar as it was practiced around the 1970s. Many of the modern constituent trees would look different (e.g. only start with binary branches).
↑ Dependency trees like the one here are a frequent occurrence in several dependency grammars. See, for example, Osborne et al. (2011).
↑ A fundamental difference between the structures of the dependency grammar and those of the constituent grammar is the presence or absence of a finite VP, which is considered a constituent. See Tesnière (1959: 103-105).

[1] Constituent grammar is primarily associated with the works of Leonard Bloomfield (1933), Rulon Wells (1947), and the young Noam Chomsky (1957). The dependency grammar is based on the theory of Lucien Tesnière (1959); see also Ágel et al. (2003/6).

[2] The tree here corresponds to the development of constituent grammar as it was practiced around the 1970s. Many of the modern constituent trees would look different (e.g. only start with binary branches).

[3] Dependency trees like the one here are a frequent occurrence in several dependency grammars. See, for example, Osborne et al. (2011).

[4] A fundamental difference between the structures of the dependency grammar and those of the constituent grammar is the presence or absence of a finite VP, which is considered a constituent. See Tesnière (1959: 103-105).