Attachment Theory (Linguistics)

As a binding theory ( English Binding Theory ) is a portion of the linguistics , in particular the generative Linguistics , referred to the distribution of anaphora , pronouns and free referential noun phrases attempts to describe and explain. Binding describes the referential dependence of a pronominal element on a so-called antecedent. For example, the typical function of reflexive pronouns is to speak anaphorically. They must then have an antecedent with which they are co-referencing within a certain range of the same (sometimes complex) sentence . Pronouns, on the other hand, usually cannot have such an antecedent in the same area of ​​the sentence. Freely referring noun phrases, such as proper names , do not accept any expression other than antecedent.

history

The attachment theory has its origin in the transformation grammar and the extended standard theory of the 1960s - 70s. Through Chomsky, she found in the form of Government and Binding the first attempt, which was very important for linguistic research in the years that followed, to describe binding phenomena through a closed partial theory of a larger theoretical framework. They are traced back to a few structural conditions to which not only overt elements, but also the empty elements assumed in this syntactic theory, such as traces and empty pronouns, are subject. Since its inception, however, it has been questioned and revised in most central aspects and is no longer believed to be correct in its original form. In particular, the expansion of the empirical basis to include languages ​​other than English has shown that the principles formulated by Chomsky are inadequate. In addition, the assumption of the complementary distribution of anaphors and pronouns, which are implied by the bonding principles A and B, cannot be maintained due to the observation of relevant exceptions. Attachment phenomena occur in every language and therefore also play an important role in other syntactic theories such as the head-driven phrase structure grammar ( HPSG ) or construction grammar .

Bonding phenomena (examples)

The following examples are intended to illustrate which phenomena the attachment theory tries to explain. Indices indicate the intended reference of the elements marked with them. Co-indexed elements have the same referent, i.e. they designate the same entity . The asterisk indicates when the reading indicated by the index is not possible.

1. Peter _i like him _{* i / j} / himself _{i / * j} .
2. Peter _i is pleased that Johan _j him _{i / * j} / is _{* i / j} likes.
3. Johan _j's father _{i thinks} he _{* i / j} / himself _{i / * j is} pretty great.
4. Johan _i / Er _i likes Johan _{* i / j.}

Obviously, different conditions apply in 1. and 2. for the co-indexing of the pronominal elements him or himself with the names Peter and Johan . The personal pronoun him seems not to be allowed to be co-indexed with an element in the same (sub) sentence. In the case of the reflexive pronoun, just the opposite is apparently the case. It must be co-indexed with an element in the same (sub) sentence to have a grammatical reading. Proper names must also not be co-indexed by an element in the same sentence. From 4. you can see that the linear sequence is not decisive for which element can function as the antecedent of the two pronominal elements, but that other structural relationships must apply here.

Government and Binding

The following presentation is essentially based on the binding theory as formulated within the theoretical framework of Government and Binding.

binding

Binding is the relation between a reflexive pronoun or a pronoun, the bound element, and its antecedents, the binding element, with which they are co-referent. First of all, the term bond should be precisely defined.

α binds β iff. (i) and (ii) apply:

(i) α is co-indexed with β

(ii) α c-commanded β

Accordingly, the relevant structural relation under which the link can take place is the C command. This explains why is not in Ex. 4 Johan can be bound. Johan is embedded in the noun phrase Johan’s father and does not c-command the reflexive. The father, on the other hand, fulfills this condition, so that a reading that is co-indexed with him is grammatical.

Binding principles

Based on this concept of binding, three binding principles are formulated. The presentation here should initially follow the reasoning of traditional attachment theory as formulated in Government and Binding. There a distinction is made between three types of nominal elements that are subject to these principles.

Anaphors (bond principle A)

The first group of nominal elements whose behavior needs to be explained are anaphors. In this context, reflexive and reciprocal pronouns are called anaphors . In addition, in Government and Binding, traces of moving arguments (A-movement) are considered anaphors. It applies to them that they must be bound by an antecedent within a certain area in the sentence, sometimes called the "local area" or "rectification domain". From examples 1 and 2 it could be concluded that the minimal sentence containing the anaphor is considered to be the binding area of ​​the antecedent. However, if one looks at the two following examples, it becomes clear that this can only be correct under certain conditions.

5. Peter _i heard of [ _i / * it _i sing a song] 6. Peter _i see [the man _j him _i / * is _i rush to the aid]

In 5. the anaphor is contained in a sentence-valued infinitive. It is nevertheless bound by the subject of the superior sentence. In 6th, however, the anaphor cannot be bound by the subject of the matrix sentence . The contrast arises from the fact that in 5. the element that gives the anaphor its case, also called Regens - in this case the verb hear - is not contained in the embedded sentence with it, as in 6.. The relevant area for attachment should therefore be taken to be the minimal sentence that contains the anaphor itself and the element that gives it its case, rain. For English, this local area would be too narrow in the sense that not only the sentence can count as such, but also a NP.

7. John _i likes [ _NP stories about himself _i ] 8. John _i likes [ _NP Bill _j 's stories about himself _{* i / j} ]

In 7. the NP contains the rain of the anaphor - the preposition about - and it is bound by the subject of the sentence. However, this is no longer possible as soon as Bill steps between the anaphor and the subject. Bill can be interpreted here as the subject of the NP. The concept of domains must therefore be generalized on the one hand with regard to the syntactic category, which can be considered a binding area, but on the other hand it must also be expanded to include the observation that it must form a minimal, complete functional complex that contains all grammatical functions related to the head are compatible with the projection. In 7th the sentence is therefore the minimal complete functional complex, which contains both the anaphor and its rain, and thus also the area relevant for the binding of the anaphor, whereby in 8th this is the NP. This covers all the terms that are necessary to understand the bond principle A.

Principle A: Anaphers must be bound within their local area.

Pronouns (binding principle B)

Personal pronouns are referred to as pronouns in this context . In 1. - 3. we see that the pronoun must not be bound by an antecedent in the area relevant for binding, while the anaphor must be. The pronoun is therefore said to be free within the local area . An element is free iff. it is not bound, ie iff. at least one of the two conditions for binding is not met. A pronoun can therefore have an antecedent who commands it with a C command, but this must not be within the local area. We have thus discussed the terms that are necessary to understand the bond principle B.

Principle B: Pronouns must be free within their local area.

Reference and variable binding

According to principle B, pronouns must be locally free, but can have an antecedent outside the local area. It is important to distinguish between two types of bond. On the one hand, a pronoun can be bound by a NP outside of the maximum projection of the sentence:

9. Johan _i wakes up early. He usually makes _i is first of all a coffee.

10. * [No boy] _i wanted to go home. He _i had way too much fun.

In such cases one speaks of co-reference. The binding of the pronoun in 9th by Johan is not affected by any syntactic restriction. The only condition for the successful binding is that the binding NP refers, which is why the sentence pair in 10. is not well-formed. On the other hand, a pronoun from a nichtreferierenden quantified NP, as no boy 11 ., Are bound, provided it is in a certain syntactic configuration to her. Because in such cases the pronoun is interpreted as a semantic variable that is bound by a quantified expression , this type of binding is called variable binding and it is differentiated from the co-reference.

11. [No boy] _i thought he had sung _i badly.

12. * He _i found that [no boy] _i had sung badly.

13. The note [every boy] _i got him was _i too bad.

The contrast between 11th and 12th suggests that the syntactic configuration relevant for variable binding could be a C command. But it becomes clear from sentences like 13. that this syntactic relation is too narrow to explain variable binding. Variable binding and co-reference can thus be differentiated on the basis of the referential properties of the binder and the syntactic configuration in which it must and does not have to appear with the pronoun. The necessary conditions for co-referencing are, on the one hand, the referentiality of at least one of the NPs that are co-referencing, and on the other hand, the fulfillment of principle B or C. In the case of variable binding, the referentiality of the binding element does not matter, but a structural configuration must exist which is similar to the C command.

R-expressions (bond principle C)

Freely referring expressions are referred to as R expressions. There are therefore all nominal elements that do not belong to either the anaphors or the pronouns. Freely referencing expressions cannot have an antecedent who C-commands them.

14. Johan _i thinks that Peter thinks that Johan * _{i is} smart.

This leads to the formulation of the last binding principle C:

Principle C: R-expressions are free everywhere.

Language differences: empirical problems of traditional attachment theory

Traditional attachment theory, by tracing attachment phenomena back to three principles and at the same time claiming to have universal, i.e. cross-lingual validity, makes very strong statements that could not withstand the expansion of the empirical basis to languages ​​other than English in this form. Already in the discussion about the expansion of the local area to categories other than the sentence (see above), it was indicated that individual languages ​​differ with regard to what can be considered a binding area. The translation of 8. into German (15.), for example, does not produce such a clear grammatical judgment for many native speakers as does 8. for English.

15. Johan _i likes Peters _j stories about himself _{i / j.}

For many speakers, the anaphor contained in the NP can apparently also be bound by Johan, although the NP is a so-called fully functional complex. Similar differences can also be observed in relation to the sentence as a local area.

Long anaphora bond

The cross-lingual comparison has shown that with regard to the sentence, the local area occurs in different "sizes" in different languages. In Icelandic, the subject of a matrix sentence can bind a reflexive pronoun across the boundaries of finite sub-clauses, provided the verbs of these intervening sentences are in the subjunctive (16.). Long ties are possible in Norwegian if the intervening sentences are in the infinitive (17.).

16. Jón _i segir [að Maria elski sig _i ]

J _i . says M. love _CONJ to _i

"John says that Maria loves him."

17. Jon _i bad oss ​​snakke om seg _i.

J _i . asked us to talk about yourself _i.

"John asked us to talk about him.

Various reactions are possible to this fact. Theory can try to parameterize rules for anaphors , that is, a language-specific range of variation is defined for the exact application conditions. Or reflexive pronouns could be seen as ambiguous, namely between a function as an anaphor or as a logophoric pronoun .

Anaphora inventory

A clear distinction between anaphors and pronouns is already possible in English, at least not without further ado. Pronouns in the genitive , such as his or her , can behave like pronouns as well as anaphors.

18. John _i hates his _i shoes. 19. John _i thinks that Susan hates his _i shoes.

In 18., contrary to principle B, the pronoun is bound by an antecedent in the local area, so it behaves like an anaphor. In 19. his again behaves like a pronoun. Since the English anapher himself does not have a genitive form, such examples can be explained within the theory discussed here with the fact that the mental lexicon has two entries for the word his because the anaphor himself does not have a genitive form himself's . However, if one takes other languages ​​into consideration, it becomes clear that dividing nominal elements into three groups is not sufficient for some of them. Some languages ​​have morphologically different complexes anaphors for which different conditions apply. In Dutch, Icelandic and Norwegian, for example, anaphors can be distinguished that contain a morpheme -zelf or sjalfan (Icelandic) and selv (Norwegian) as well as the monomorphematic zich (Icelandic: sig, Norwegian: seg ). The latter allow long binding in these languages, while the former must be bound locally.

20. John _i snakker om seg selv _i . 21. John _i bad oss ​​snakke om seg _i / * seg selv _i .

Individual evidence

1. ^ Robert Lees, Edward Klima: Rules for English Pronominalization . In: Language . No. 39 , 1963, pp. 17-28 .
2. ^ Noam Chomsky: Conditions on transformations. In: Stephen Anderson, Paul Kiparsky (eds.): A Festschrift for Morris Halle . Holt, Rinehart and Winston, New York 1973, pp. 232-286 .
3. ^ Noam Chomsky: Conditions on Rules of Grammar . In: Linguistic Analysis . No. 2 , p. 303-351 .
4. ↑ Robert Fiengo: Semantic Conditions on Surface Structure . MIT, 1974.
5. ^ Tanya Reinhart: The Syntactic Domain of Anaphora . MIT, 1976.
6. ^ Noam Chomsky: Lectures on Government and Binding . Foris, Dordrecht, S. 1981 .
7. ^ Noam Chomsky: A Minimalist Program for Linguistic Theory . In: Kenneth Hale, Samuel Jay Keyser (Eds.): The View from Building 20: Essays in honor of Sylvain Bromberger . MIT Press, Cambridge, MA 1993, pp. 1-52 .
8. ^ ^{A b} Kenneth Wexler, M. Rita Manzini: Parameters and Learnability in Binding Theory . In: Thomas Roeper, Edwin Williams (Ed.): Parameter Setting. Studies in Theoretical Psycholinguistics . No. 4 . Springer, Dordrecht 1987, p. 41-76 .
9. ^ ^{A b} Lars Hellan: Containment and Connectedness Anaphors . In: Jan Koster, Eric Reuland (Eds.): Long-Distance Anaphora . Cambridge University Press, Cambridge 1991, pp. 27-48 .
10. ↑ Jan Koster, Eric Reuland: Long-distance anaphora: An overview. In: Jan Koster, Eric Reuland (Eds.): Long-Distance Anaphora . Cambridge University Press, Cambridge 1991, pp. 1-25 .
11. ^ Tanya Reinhard, Eric Reuland: Reflexivity . In: Linguistic Inquiry . No. 24 , 1993, pp. 657-720 .
12. ↑ Eric Reuland: Primitives of Binding . In: Linguistic Inquiry . No. 32 , 2001, p. 439-492 .
13. ↑ Eric Reuland: Anaphora and Language Design . MIT Press, Cambridge, MA.
14. ↑ Johan Rooryck, Guido Vanden Wyngaerd: Dissolving Binding Theory . Oxford University Press, Oxford 2011.
15. ↑ C.-T. James Huang, C.-C. Jane Tang: The Local Nature of the Long-Distance Reflexive in Chinese . Ed .: Jan Koster, Eric Reuland. Cambridge edition. Cambridge University Press, 1991, pp. 263-282 .
16. ^ Carl Pollard: Remarks on Binding Theory . In: Stefan Muller (Ed.): Proceedings of the 12th International Conference on Head-Driven Phrase Structure Grammar . CSLI Publications, Lisbon 2005, pp. 561-577 .
17. ^ Paul Kay: Anaphoric Binding in Construction Grammar . In: Proceedings of the Twentieth Annual Meeting of the Berkeley Linguistics Society: General Session Dedicated to the Contributions of Charles J. Fillmore . 1994, p. 283-299 .
18. ^ Günter Grewendorf: Minimalist Syntax . A. Francke Verlag, Tübingen / Basel, p. 26 .
19. ↑ Chris Barker: Quantificational binding does not require c-command . In: Linguistic Inquiry . No. 43 , 2012, p. 614-633 .