Dependency grammar

Dependency refers to a Lucien Tesnière (* 1893, † December 6, 1954 ) established, in the approach but also in the Middle Ages by Thomas of Erfurt to-find form of the grammar. She explores the hierarchical structure of sentences based on interdependencies ( Dependenzstruktur ). Dependence therefore means the dependence of a word ( the word or ruled the dependens ) by another word ( the reigning word or the rain). A dependenzielle grammar differs from the phrase structure grammar ( = Konstituentengrammatik ) by the absence phrasaler node in the analyzed structure.

History

The dependency grammar starts with the approaches already living in the 5th or 4th century BCE Indian grammarian Panini. So goes the Konstituentengrammatik, which was developed from the ancient propositional logic of Aristotle later, by many centuries ahead. Ibn Mada may have been the first in Europe, who used " dependency " in the modern sense. Mada was born in 1119 in Cordoba, studied in Seville and Ceuta, and died in 1195 in Seville. In the German -speaking world it was in the Middle Ages the Dependenzbegriff in the grammar of Thomas of Erfurt, the rector of St. Severi schools and St. Jacob was. In the early modern dependency and constituency appear as terms to have existed side by side. Constituency came mainly from the ancient logic that was spread mainly by the Port-Royal logic in modern grammars of various European languages.

Modern Dependenzgrammatiken but are mainly due to the grammatical theory Lucien Tesnières. Tesnière was French, Berlitz and professor of linguistics at the Universities of Strasbourg and Montpellier. His main work Éléments de syntaxe structurale was published in 1959, five years after his death. Developed by Tesnière concept of dependence seems but in the 1960s, regardless of Tesnière to have been recognized by others. The dependency grammar has a strong tradition in German speaking countries where the principle of the German V2 and the relatively free word order can be insightful examined on the basis of dependency. For several years, interest in the growing dependency grammar. This is mainly to be explained by that more and more the benefits of Dependenzprinzips for machining and translation of texts is detected. The first international conference for dependence in linguistics took place only in September 2011.

Dependence vs. constituency

Dependence is a one-to - one relation. Each element in the sentence (word or morph) corresponds to exactly one node in the structure of the sentence. Constituency, however, is a one-to -one or - more - relation. There, each element ( word or morph) corresponds to at least one node in the set and often more than one. The difference is easy to recognize by simple tree structures:

In Dependenzbaum left there are two words and two nodes. In Konstituenzbaum right, however, there are two words, but three nodes. Constituency assumes that the number of nodes, the number of set elements (words) exceeds at least one. A second example highlights the difference:

"Sub" stands for " subordinator " and SubP for subordinator - phrase ( = subordinate clause ). The Dependenzbaum left contains four words and exactly four nodes, while the right Konstituenzbaum contains four words, but six nodes. Noteworthy on Konstituenzbaum is the absence of a finite verb phrase (as constituent ), which excludes the subject. In this regard, the Konstituenzbaum is a direct translation of the Dependenzbaums ( dependence → constituency ). Dependence as a principle of hierarchical organization of elements, it avoids to recognize a finite VP constituent as.

The Konstituentenbegriff is important for the difference between dependency and constituency. Both dependency and constituency recognize constituents. Each complete tree and every complete subtree of a tree is considered to be constituent. Thus, there are only four constituents in the second Dependenzbaum above, but six constituents in the second Konstituenzbaum. In this respect, it is of great importance whether the grammar notes for the adoption of the relatively large number of constituents provides that postulates the constituency, and let if find evidence for the assumption of a finite VP constituent.

Some important differences between dependency and constituency are summarized in the following table:

Finite verb phrase?

The principle of constituency comes from the ancient logic, where a statement in a subject and a predicate is divided. This binary classification is taken from the Port-Royal logic, applied directly to rates and has been in several grammars of several languages ​​basis of sentence analysis. It appears in the early work of Noam Chomsky in the form S → NP VP: one set (S ) consists of a noun phrase (NP ) and a verb phrase (VP). Tesnière but has turned against this division of the sentence by arguing that this binary division of the logic in the syntax did not belong. Instead of the binary division looked Tesnière the verb as the center and root of the sentence. If one accepts the binary division, constituency is, of course, as an organizational principle of the sentence. But if one rejects this division and sets in place of the central role of the verb as an organizational center of the sentence, dependence is quite natural.

This difference between dependence and constituency concerns directly the existence of a finite VP constituent. Most Konstituenzgrammatiken assume that there is such a constituent, especially for English. Dependenzgrammatiken, however, can not detect such a constituent. In this regard, recourse to the so-called dependency grammar Konstituententests ( = constituency tests). These tests provide both for English and for the German little evidence of the existence of a finite VP constituent, eg

Trying to topikalisieren the finite verb phrase, fails clearly. Whether the finite verb phrase can be replaced by a proform is not easy to detect because it is not clear whether doing this can be considered as a proform for said nothing. The demonstrative pronoun that alone should be the Proform (Hans did not say anything ). And the appearance of said nothing in response fragment is questionable, since the answer with the subject is clearly preferable: He said nothing.

Compare this data with the results of the tests when they are applied to a infinite verb phrase:

The topicalization of the verb phrase is possible to say anything. The combination do to be regarded as Proform is now plausible, and the fragment can say nothing very well occur in response to the given question. With respect to these data one can conclude that infinite verb phrases are constituents. This fact is responsible for the dependency grammar but not a problem, because, as the Konstituentengrammatik, well- considered infinite verb phrases as constituents, eg

The cord (string) say nothing is a complete subtree and is therefore regarded as constituent.

Minimum structures

The dependenzielle one-to- one relation limits the number of nodes in the structure of a sentence to fit the number of elements in the set. From the small number of nodes implies that dependency structures are comparatively minimal. A Dependenzbaum usually contains half of the nodes and edges of the corresponding Konstituenzbaumes. Because of this minimal structures, it is less difficult to make decisions regarding the structure of sentences. So it is easier to set up a plausible Dependenzanalyse of sentences. The following Dependenzbäume are examples of structures that hold many Dependenzgrammatiker is right, because the hierarchies of dependencies shown are supported by empirical observations:

With such structures, most Dependenzanalysen match - only when discontinuities are the ghosts. The agreement stems from the fact that not rise occurring, difficult questions in Konstituentengrammatiken. For example, there is the difficult question not whether there should be a finite VP constituent, because dependence is opposed to such constituent. In addition, the further difficult question does not depend on whether the structures ( deep) should be relatively flat or more layered. Dependency structures can be strongly layered structures not to, because the number of nodes is limited to the number of words.

Dependenzgrammatiken and phrase structure grammars

Both Dependenzgrammatiken and phrase structure grammars are widely used. In English-speaking countries, however, dominate the phrase structure grammars. The following grammars are phrase structure grammars, because they are based on the concept of constituency:

Dependenzgrammatiken are more strongly represented in Europe. The following grammars are based on Dependence:

Was in German-speaking Europe and is particularly a lot of research in the field of Depedenzgrammatik. The work of Ágel, Angel, Eroms, Happ, Heringer, Lobin, Weber among many others have the dependency grammar very advanced.

Valenzgrammatik?

The dependency grammar is sometimes referred to as " dependency and Valenzgrammatik ". This designation is, however, problematic because the valence is considered to be part of many Grammar theory and can not be considered as a separate grammar. Valence is the ability of the predicate, the number and the function of the constituents in the vicinity of the predicate ( that is, the argument of the predicate ) to be determined. That it " dependence and Valenzgrammtik " gives this designation is understandable, as the annex and valence are terms that are due to Tesnières theory of syntax and grammar. The term " valence " is but more plausible than " Valenzgrammatik ". So there is the dependency grammar and valence theory, the two are related historically close. However, the valence theory is an autonomous region, ie it exists, even if one rejects dependence and in their place puts constituency. Most of phrase structure grammars vacate the valence of predicates long played a central role in the theory. However, phrase structure grammars are not Dependenzgrammatiken.

Notes