Locality (linguistics)

Last updated

In linguistics, locality refers to the proximity of elements in a linguistic structure. Constraints on locality limit the span over which rules can apply to a particular structure. Theories of transformational grammar use syntactic locality constraints to explain restrictions on argument selection, syntactic binding, and syntactic movement.

Contents

Where locality is observed

Locality is observed in a number of linguistic contexts, and most notably with:

  1. Selection of arguments; this is regulated by the projection principle
  2. Binding of two DPs; [lower-alpha 1] this is regulated by binding theory
  3. Displacement of wh-phrases; this is regulated by wh-movement

Selection

The projection principle requires that lexical properties — in particular argument structure properties such as thematic roles — be "projected" onto syntactic structures. Together with Locality of Selection, which forces lexical properties to be projected within a local projection (as defined by X-bar theory [1] :149), the projection principle constrains syntactic trees. Syntactic trees are represented through constituents of a sentence, which are represented in a hierarchical fashion in order to satisfy locality of selection through the restraints of X-bar theory. [2] In X-bar theory, immediate dominance relations are invariant, meaning that all languages have the same constituent structure. However, the linear precedence relations can vary across languages. For example, word order (i.e. constituent order) can vary with and across languages. [1]

Locality of selection
Every argument that α selects must appear in the local domain of α.

If α selects β, then β depends on α. If α selects β, and if locality of selection is satisfied, then α and β are in a local dependency. If α selects β, and if locality of selection is not satisfied, then α and β are in a non-local dependency. The existence of a non-local dependency indicates that movement has occurred.

From the perspective of projection, an element can be "stretched" to occupy the following projection levels:

minimal (X)

intermediate (X')

maximal (X max)

These occupying elements appear valid for all syntactically relevant lexical and functional categories. [3]

Head-complement selection

Locality projection of the head X-bar Theory locality.svg
Locality projection of the head

According to locality of selection, the material introduced in the syntactic tree must have a local relationship with the head that introduces it. This means that each argument must be introduced into the same projection as its head. Therefore, each complement and specifier will appear within the local projection of the head that selects it.

For example, the contrast between the well-formed (1a) and the ill-formed (1b) shows that in English, an adverb cannot intervene between a head (the verb study) and its complement (the DP the report).

(1)  a.  John carefully [Vstudies] [DPthe report].      b. *John [Vstudies] carefully [DPthe report]. [1] :192

In structural accounts of the contrast between (1a) and (1b), the two sentences differ relative to their underlying structure. The starting point is the lexical entry for the verb study, which specifies that the verb introduces two arguments, namely a DP which bears the semantic role of Agent, and another DP which bears the semantic role of Theme.

lexical entry for study: V, <DPAGENT,DPTHEME>

In the tree for sentence (1a), the verb, studies, is the Head of the VP projection, the DPTHEME,the report, is projected onto the Complement position (as sister to the head V), and the DPAGENT , John, is projected onto the Specifier (as sister to V'). In this way, (1a) satisfies Locality of Selection as both arguments are projected within the projection of the head that introduces them. The adverb phrase, AdvP carefully attaches as an unselected adjunct to VP; structurally this means that it is outside of the local projection of V as it is sister to and dominated by VP. In contrast, in the tree for sentence (1b), the introduction of the AdvP carefully as sister to the verb study violates Locality of Selection; this is because the lexical entry of the verb study does not select an AdvP, so the latter cannot be introduced in the local projection of the verb.

1 a) "Carefully John studies the report" 1 b) *"John studies carefully the report" Syntactic Trees demonstrating Locality of Selection.png
1 a) "Carefully John studies the report" 1 b) *"John studies carefully the report"

Morphological selection

Locality can also be broken down into a morphological perspective, by analyzing words with some, or many affixes. A speaker who can make sense of a word with many morphemes (e.g. affixes) must know: how the morpheme is pronounced and what kind of morpheme it is, (free, prefix, suffix). If it is an affix, then the speaker also must know what the affix c-selects. The speaker must also know that the c-selected element must be adjacent to the affix, amounting to the requirement that branches of a tree never cross. Crossing branches is not included in the lexicon, and it is a general property of how linguistic structures are grammatically structured. This is true because lexical entries do no impose a requirement on a part of word structure that it is not sister to. This relates to the fact that affixes cannot c-select for an element which is not a sister. Additionally, the speaker must know what kinds of thing results after c-selection. These key aspects that a speaker must know can be observed in the lexical entries below, with the example "denationalization".

1234
nation:free
-al:suffixc-selects Nto form an A
-ize:suffixc-selects Ato form a V
de-:prefixc-selects Vto form a V
-ation:suffixc-selects Vto form an N

Lexical selection

When meeting selection requirements, the semantic content of the constituent selected by the head must be taken into consideration. For example, the thematic role of the constituent that is selected, and the properties of the head which selects it. Take, for example, the verb head elapse, which selects for a DP subject.

E.g. a) *[DP Johnagent] elapsed.

b) [DP Timeagent, can elapse] elapsed. [4]

However, while [DP John] is syntactically in subject position, it gives an ungrammatical sentence as [DP John] cannot elapse, it has no thematic quality to elapse, and as such cannot meet the lexical selection requirements of [VP elapse]. However, [DP Time] in subject position does have this thematic quality and can be selected by [VP elapse].

Lexical selection is specific to individual word requirements, these must abide to both Projected Principle and Locality requirements.

Detecting selection

Crucially, selection determines the shape of syntactic structures. Selection takes into consideration not only lexical properties but also constituent selection, that is what X-Bar Theory predicts as appropriate formulations for specific constituents.

Covariation

One way to determine which syntactic items relate to each other within the tree structure is to examine covariances of constituents. For example, given the selectional properties of the verb elapse, we see that not only does this verb select for a DP subject, but is specific about the thematic role this DP subject must have. [2]

Case

In English, case relates to properties of the pronoun, nominative, accusative, and genitive. Case can be selected by heads within the structure, and this can affect the syntactic structure expressed in the underlying and surface structure of the tree. [2]

EPP properties

EPP properties, or Extended Projection Principle, is located in certain syntactic items, which motivate movement due to their selection requirements. Such can be found most commonly in T, which in English, requires a DP subject. This selection by T creates a non-local dependency, and leaves behind a "trace" of the moved item. [2]

Binding

Binding Theory refers to 3 different theoretic principles that regulate DP's (Determiner Phrase). [3] In consideration of the following definitions of the principles, the local domain refers to the closest XP with a subject. If a DP(1) is bound, this means it is c-commanded and co-indexed by a DP(2) that is sister to the XP dominating over DP (1) .To contrast, if it is free, then the DP in question must not be c-commanded and co-indexed by another DP.

Principle A

Principle A for locality in Binding Theory refers to the binding of an anaphor and its antecedent which must occur within its local domain. Principle A states that anaphors must be bound in their local domain, and that DP's must be in a local relation. The local domain is the smallest XP containing a DP, in order to satisfy Binding Theory, the DP must c-command the anaphor and have a subject. [1] Therefore, the antecedent must be in the same clause that contains the anaphor if it is to abide to Binding Theory.

An anaphor is considered to be free when it is not c-commanded or co-indexed. [5] A node is c-commanded if a sister node of the first node dominates it, (i.e. node X c-commands node Y if a sister of X dominates Y). A node is co-indexed if the DPs in question both are indexed by a matching subscript letter, as seen in the DPs of (2) a. and (2) b.

In English, Principle A governs over anaphors, which include lexical items like reflexives, (e.g. myself, yourself...etc.), and reciprocals, (e.g. each other, etc.). These items must refer back to a previous item in the constituent in order to satisfy its semantic meaning, and in turn, abide to Principle A.

The following examples show the application of Binding Theory, Principle A, in relation to reflexives:

(2) a. Mary revealed [DP John]i to [DP himself]i.     b. *Mary revealed [DP himself]i to [DP John]i. [1] :162
Example (2a) "Mary revealed John to himself" Locality (linguistics) 2a.png
Example (2a) "Mary revealed John to himself"

Example (2a) is predicted to be grammatical by Principle A of binding theory. The anaphor, [DP himself]i, and antecedent, [DP John]i, are selected within the same local domain. TP is the smallest XP that contains the anaphor and DP subject (in this case, the subject is the antecedent). Given that the antecent, [DP John]i, is governed by VP, which is sister to PP, and PP is the maximal node dominating over [DP himself]i, the anaphor, [DP John]i can therefore c-command [DP himself]i. As co-indexation is already established by the matching subscript letter i, this sentence is grammatical and abides to Principle A.

Example (2b) "Mary revealed himself to John" Locality (linguistics) 2b.png
Example (2b) "Mary revealed himself to John"

However, in example (2b), the anaphor [DP himself]i has within its local domain antecedent, [DP Mary], which would serve as a candidate for binding. However, [DP himself]i is co-indexed to [DP John]i, which is a pronoun. Two factors have gone wrong here. Firstly, as shown in Principle B below, pronouns must be free in their local domain, and as [DP John]i is being bound locally by [DP himself]i, this does not abide to Binding Theory and is considered ungrammatical. Secondly, and most important to this section, Principle A establishes that an anaphor must be bound locally. [DP himself]i is not c-commanded by any local DP, nor any DP, in fact, [DP himself]i is c-commanding [DP John]i instead.

As discussed previously, the local DP which could bind [DP himself]i, [DP Mary]. However, while [DP Mary] can c-command [DP himself]i, and can be co-indexed to complete binding, this sentence would still be ungrammatical. This is because, in English, anaphors and their antecedent must agree in gender. As such, attempting to fix 2b by binding [DP himself]i with [DP Mary] would still render an ungrammatical sentence.

This is exemplified below:

Attempted correction of (2b)

i) *[DP Mary]i revealed [DP himself]i to [DP John]

ii) [DP Mary]i revealed [DP herself]i to [DP John]

The following examples show the application of Binding Theory, Principle A, in relation to reciprocals:

(3) a. John heard [DP their]i criticisms of [DP each other]i.     b. *[DP They]i heard John's criticisms of [DP each other]i. [1] :167
(3) a. John heard [DP their]i criticisms of [DP each other]i. Final tree 3a wiki.png
(3) a. John heard [DP their]i criticisms of [DP each other]i.


(3) b. *[DP They]i heard John's criticisms of [DP each other]i. They heard john's criticisms of each other.png
(3) b. *[DP They]i heard John's criticisms of [DP each other]i.

Example (3) follows the same explanations to example (2).

As predicted by Binding Theory, Principle A, (3a) is grammatical because the anaphor [DP each other]i is bound within the same domain as the antecedent [DP their]i. However, example (3b) is ungrammatical because the anaphor is bound by the antecedent non-locally, which goes against Principle A which specifies local binding. Further, Principle A would predict that in fact it is [DP John] which could bind [DP each other]i, however, similarly to example 2b, anaphors not only have to agree with gender with the antecedent that binds them, but also number. Given that [DP John] is a singular entity, and [DP each other] refers to multiple, this co-indexation cannot occur, rendering this sentence ungrammatical.

To summarize, it must be noted that anaphors must agree in gender, number, and also person with their antecedent, in a local domain.

Principle B

Principle B states: pronouns must be free in their local domain, and predicts that some DP's are non-locally bound to other DP's.

Take for example, these two sentences:

(4) a. *[DP Lucy]i admires [DP her]i

b. [DP Lucy]i thinks that I admire [DP her]i

In 4a), when the [DP Lucy], is co-indexed with, and c-commands, [DP her], this violates principle B. This is because [DP her] has a c-commanding antecedent in its local domain (i.e. [DP Lucy]) this shows that the pronoun is bound in its domain. As such, pronoun [DP Lucy], which also abides to principle B, also cannot be bound locally, and contributes to the sentences problems with abiding to Principle B.

In (4b) Principle B is obeyed, this is because while there is co-indexation and a c-commanding relation between [DP Lucy] and [DP her], both DPs are free in their local domains. Remember that local domain is determined by the smallest XP containing a subject. In the case of [DP Lucy], the local domain pertains to the head which dominates it, of which [DP Lucy] is the subject, while for [DP her], it would be the smallest XP containing a subject, which is [DP I].

Principle B does not state anything regarding whether a pronoun requires an antecedent. It is permissible for a pronoun to not have an antecedent in a sentence. Principle B simply states that if a pronoun does have a c-commanding antecedent, then it must be outside of the smallest XP with a subject that has the pronoun, i.e. outside the domain of the pronoun. [1] Further, both Principle A and B predict that pronouns and anaphos must occur in complementary distribution.

Principle C

The following examples show the application of Binding Theory, Principle C which states: R-expressions cannot be bound, and certain DP's, such as R-expressions are never related to other DP's. [1]

In English, R-expressions refer to Quantified Expressions, [2] (e.g. every, all, some...etc.) and Independently Referential Expressions, [2] (e.g. this, the, my, a, pronouns)

(5) a. *[DP She]i said that [DP Lucy]i took the car

b) After you spoke to [DPher]i, [DPLucy]i took the car

c) The builder of [DPher]i house visited [DPLucy]i [2]

It is important to note that 5a, can be distinguished from 5b and 5c the differences in structural relations between the pronoun and the name. In 5a, "she" c-commands "Lucy", but this does not occur in 5b and 5c. These observations can be described by the preliminary observation that non-pronominals cannot be bound, i.e., non-pronominals cannot be c-commanded by a co-indexed pronoun. Compared to Principle A and Principle B, this requires goes all the way up to the root node, since it is not limited to any domain.

In addition to these principles, it is required that pronouns and reflexives agree with their antecedent in gender. For example, regardless of the consideration of locality, a sentence such as "[DPJohn]i likes [DPherself]i", it would be ungrammatical because the two co-indexed entities do not agree in gender. Pronouns and reflexives also have to agree with their antecedent in number and person.

Small clauses and Binding Theory

Small clauses show that different categories can have subjects, which is supported by Binding Theory. The internal structure of a small clause is determined, typically, by a predicate or a functional element, and are considered as being projections of a functional category. [6]

Given this, Binding Theory can predict the internal structure of a small clause, depending on which Principle is present within the structure.

Take, for example, the following data:

5.1 a)*[DP Mary]j considers [John proud of [DP herself]j] [2]
5.1 b)[DP Mary]j considers [John proud of [DP her]j]] [2]

This data suggests that [AP proud] has a subject, [DP John], and that the anaphor that it has as a complement, [DP herself]/[DP her], has a domain local domain that extends to the node which dominates [DP Mary], as it c-commands and binds the anaphor. [2]

As such, the underlying structure that is suggested is the following:

5.1 a) Mary considers John.png
5.1 a)
5.1 b) 5.1 b) .png
5.1 b)

Binding Theory correctly predicts that 5.1 a) will be an ungrammatical construction given Principle A which requires the anaphor to be bound locally. As well as correctly predicting 5.1 b) as grammatical, given Principle B, which states that pronouns cannot be locally bound. Both instances are represented, respectively, within the generated structures.

Syntactic dependencies

Syntactic dependencies of all types are confined to a limited portion of structure. [7] Referential and filler gap-dependencies remain a divide in locality principles. Few theories which have succeeded in unifying these two types of dependencies undel locality principles. While there is no agreed-upon theory, general observations are seen. Absolute and relative barriers are a great divide in locality theory and have yet to be formally unified under a single theory. [7]

Absolute barriers do not allow movement beyond it. (WH-island, Subjacency conditions and Condition on Extraction Domain)

Relative barrier is the idea that syntactic dependencies between a filler and a gap are blocked by the intervention of a closer element of the same type

Movement

Movement is the phenomenon that accounts for the possibility of a single syntactic constituent or element occupying multiple, yet distinct locations, depending on the type of sentence the element or constituent is in. [8] Movement is motivated by selection of certain word types, which require their Projection Principles be met Locally. In short, Locality predicts movement of syntactic constituents.

Raising to subject: surface and underlying tree structure

When comparing surface structure to what selection predicts, there appears to be an anomaly in the word order of the sentence and the production of the tree. Within the underlying structure, at times referred to as deep structure, there exist Deep Grammatical Relations, which relate to the manifestation of subject, object and indirect object. [9] Deep grammatical relations are mapped onto the underlying structure, (deep structure). These are expressed configurationally in relevance to particular languages, and are seen represented in the surface representation of the syntactic tree. This surface representation is motivated by selection, locality, and item specific features which allow for movement of syntactic items.

Take for example the following sentence:

  • [DP He] [VP seems] to[VP run] slowly

Given the word order of the sentence, we would expect the tree to have violated Locality of Selection and Projection Principle guidelines. Projection Principle specifies what the head selects for, and Locality of selection ensures that these are established in the local domain of the head which selects it.

As such, we would expect the following ungrammatical tree:

*Seems he runs slowly He seems to run slowly.png
*Seems he runs slowly

This tree represents local dependencies of selection. [VP run] selects for a DP subject and can have an AdvP complement, this is satisfied. However, [VP seems] also requires a DP subject, which is unsatisfied. Lastly, T has an EPP feature, as discussed above, which selects for a DP subject. It's these selectional properties which motivate movement of certain syntactic items. In this particular tree, it is the DP which is motivated to move, in order to satisfy the selectional properties of [VP seems] and T's EPP feature.

The following surface tree is expected, which follows the word order of the sentence provided:

He seems to run slowly Wiki power 2 raising png.png
He seems to run slowly

In the surface representation, we see that DP movement is motivated by Locality of selection, movement is marked by brackets <>, (or at times arrows following the movement). The movement leaves behind a trace of the DP which still satisfies selection, however, the selection is now a non-local dependency.

Raising to object

Wh-movement

In wh-movement in English, an interrogative sentence is formed by moving the wh-word (determiner phrase, preposition phrase, or adverb phrase) to the specifier position of the complementizer phrase. This results in the movement of the wh-phrase into the initial position of the clause. [1] This is seen in English word order of questions, which show Wh components as sentence initial, though in the underlying structure, this is not so.

The wh-phrase must also contain a question word, due to the fact that it needs to qualify as meeting the +q feature requirements. The +q feature of the complementizer (+q= question feature) results in an EPP:XP+q feature: This forces an XP to the specifier position of CP. The +q feature also attracts the bound morpheme in the tense position to move to the head complementizer position; leading to do-support. [1] :260–262

Wh-movement violations

There are seven types of violations that can occur for wh-movement. These constraints predict the environments in which movement generates an ungrammatical sentence: Movement does not occur locally.

Wh-island constraint
The Wh-island constraint
If CP has the feature +q, movement of a wh-phrase to a position outside of the clause cannot occur. [1] :271

This definition tells us that if the specifier position of CP is occupied or if a C is occupied by a +q word, movement of a wh-phrase out of the CP cannot occur. [1] :271 In other words, a CP that has a wh-phrase in its [spec, CP] that is filled with another wh-phrase that is not the one that was extracted, but from higher in the tree. The movement of the wh-phrase is being obstructed by another wh-phrase. [1]

(6) a. [DP Who]i do you wonder [DP e]i bought what?     b. *[DP What]i do you wonder who bought [DP e]i?
6 (a) "who do you wonder bought what" Wonder(a).png
6 (a) "who do you wonder bought what"
6(b) "What do you wonder who bought" Wonder (b).png
6(b) "What do you wonder who bought"

Example (6b) illustrates the wh-island constraint. The embedded clause contains a complementizer with the feature +q. This causes the DP "who" to move to the specifier position of that complementizer phrase. Movement of the complement DP "what" cannot occur since the specifier position of CP is filled. Therefore, movement of the wh-word "what" generates an ungrammatical sentence, while movement of the wh-word "who" is allowed (specifier position of the embedded CP is not occupied).

Adjunct island condition
Adjunct island condition
If an adjunct contains a CP, movement of an element inside the CP to a position outside of the adjunct is not allowed. [1] :273
(7) a. [PP Where]i did he go [PP e]i before they finished the food?     b. *[DP What]i did he go home before Mary finished [DP e]i?

Example (7b) demonstrates the adjunct island condition. We can see that the wh-word, "what", occurs within the complementizer phrase that appears in the adjunct. Therefore, movement of the DP out of the adjunct will generate an ungrammatical sentence. Example (7a) is grammatical because the trace of the PP (prepositional phrase) "where" is not within the adjunct, therefore, movement is allowed. This demonstrates the prohibition of extraction from inside an adjunct and the condition that states that no element in a CP inside an adjunct may move out of this adjunct.

Sentential subject constraint
The sentential subject constraint
Movement of an element that appears within the CP subject cannot occur. [1] :273

A sentential subject is a subject that is a clause, not the subject of a sentence. Therefore, a clause that is a subject is called a sentential subject. The Sentential Subject Constraint is violated when an element moves out of a CP that is in the subject position.

(8) a. [DP Who]i did that Bill threw out the cheese annoy [DP e]i?     b. *[DP What]i did that Bill threw out [DP e]i annoy you?

Example (8b) displays the sentential subject condition. The subject of the verb in this sentence is a complementizer clause. The DP "what" that appears within the CP subject moves to the specifier position of the main clause. The sentential subject constraint predicts that this wh-movement will result in an ungrammatical sentence since the trace was within the CP subject. Example (8a) is grammatical because the DP "who" does not have a trace within the CP subject, therefore, allowing movement to occur.

Coordinate structure constraint
The coordinate structure constraint
An element within a conjunct cannot undergo movement out of the conjunct. [1] :278
(9) a. [DP What and [rice]i did you eat [DP e]i?     b. *[DP What]i did you eat [DP ei and [rice]]? [1] :267

Example (9a) is grammatical since the DP complement is moving as a whole to the specifier position of the matrix clause; nothing is extracted from the larger DP. Example (9b) is an example of the coordinate structure constraint. The DP "what" originally occurs within the DP conjunct, therefore, this constraint predicts that an ungrammatical sentence will result due to the extraction of an element within the conjunct. [1] :278

Complex NP constraint
The complex noun phrase constraint
Extraction of an element that is a complement or adjunct of a NP is not allowed. [1] :274
(9) a. [DP Whose book]i did you buy [DP e]i?     b. *[D Whose]i did you buy [D e]i book?

Example (8a) is a grammatical because the DP complement of the verb moves as a whole to the specifier position of the main clause. Example (8b) displays the complex noun phrase constraint. The NP complement D, "whose", is extracted and moved to the specifier position of the main clause. The complex noun phrase constraint predicts that this wh-movement will result in an ungrammatical sentence since extraction of an element within the complex NP is not allowed.

Subject condition
The subject condition
Movement of a DP out of the subject DP of the verb is not allowed. [1] :277
(10) a. A picture of which students appeared in the newspapers?      b. *[DP Which students]i did [DP a picture of [DP e]i] appear in the newspaper? [1] :277

Example (10a) does not display any wh-movement. Therefore, the sentence is grammatical since nothing is extracted from the subject DP. Example (10b) contains wh-movement of a DP that is within the subject DP. The subject condition tells us that this type of movement is not allowed and the sentence will be ungrammatical. [1] :277

Left branch constraint
The left branch constraint
Extraction of a DP subject within a larger DP cannot occur. [1] :278
(11) a. You are eating [DP [DP whose] cake].       b. *[DP Whose]i are you eating [DP [DP ei] cake]? [1] :278

In example (11a), there is no wh-movement, therefore the left branch constraint does not apply and this sentence is grammatical. In example (11b), the DP "whose" is extracted from the larger DP "whose cake". This extraction under the left branch constraint is not allowed, therefore, the sentence is predicted to be ungrammatical. This sentence can be made grammatical by moving the larger DP as a unit to the specifier position of CP. [1] :278

(12) c. [DP Whose cake]i are you eating [DP ei]? [1] :278

In example (12c), the whole subject DP structure undergoes wh-movement, which results in a grammatical sentence. This suggests that pied-piping can be used to reverse the effects of the violations or extraction constraints. [1] :278


(13) *[DP What]i did you wonder who ate ei? [1] :271

Example (13) is an example of a wh-island violation. There are two TP bounding nodes that appear between the DP "what" and its trace. The subjacency condition postulates that wh-movement cannot undergo when the elements are spread too far apart. [1] When two positions are separated by only one bounding node, or no bounding node at all, they are considered subjacent. [1] Therefore, according to the subjacency condition, movement will result in an ungrammatical sentence. [1] :271

See also

Notes

  1. A determiner phrase (DP) is what is traditionally known as a noun phrase (NP). Thus a small dog, for instance, would be a DP.

Related Research Articles

In linguistics and grammar, a pronoun is a word or a group of words that one may substitute for a noun or noun phrase.

In linguistics, X-bar theory is a model of phrase-structure grammar and a theory of syntactic category formation that was first proposed by Noam Chomsky in 1970 reformulating the ideas of Zellig Harris (1951), and further developed by Ray Jackendoff, along the lines of the theory of generative grammar put forth in the 1950s by Chomsky. It attempts to capture the structure of phrasal categories with a single uniform structure called the X-bar schema, basing itself on the assumption that any phrase in natural language is an XP that is headed by a given syntactic category X. It played a significant role in resolving issues that phrase structure rules had, representative of which is the proliferation of grammatical rules, which is against the thesis of generative grammar.

Government and binding is a theory of syntax and a phrase structure grammar in the tradition of transformational grammar developed principally by Noam Chomsky in the 1980s. This theory is a radical revision of his earlier theories and was later revised in The Minimalist Program (1995) and several subsequent papers, the latest being Three Factors in Language Design (2005). Although there is a large literature on government and binding theory which is not written by Chomsky, Chomsky's papers have been foundational in setting the research agenda.

In linguistics, the minimalist program is a major line of inquiry that has been developing inside generative grammar since the early 1990s, starting with a 1993 paper by Noam Chomsky.

In linguistics, binding is the phenomenon in which anaphoric elements such as pronouns are grammatically associated with their antecedents. For instance in the English sentence "Mary saw herself", the anaphor "herself" is bound by its antecedent "Mary". Binding can be licensed or blocked in certain contexts or syntactic configurations, e.g. the pronoun "her" cannot be bound by "Mary" in the English sentence "Mary saw her". While all languages have binding, restrictions on it vary even among closely related languages. Binding has been a major area of research in syntax and semantics since the 1970s and, as the name implies, is a core component of government and binding theory.

In linguistics, wh-movement is the formation of syntactic dependencies involving interrogative words. An example in English is the dependency formed between what and the object position of doing in "What are you doing?" Interrogative forms are sometimes known within English linguistics as wh-words, such as what, when, where, who, and why, but also include other interrogative words, such as how. This dependency has been used as a diagnostic tool in syntactic studies as it can be observed to interact with other grammatical constraints.

In generative grammar and related frameworks, a node in a parse tree c-commands its sister node and all of its sister's descendants. In these frameworks, c-command plays a central role in defining and constraining operations such as syntactic movement, binding, and scope. Tanya Reinhart introduced c-command in 1976 as a key component of her theory of anaphora. The term is short for "constituent command".

In linguistics, the projection principle is a stipulation proposed by Noam Chomsky as part of the phrase structure component of generative-transformational grammar. The projection principle is used in the derivation of phrases under the auspices of the principles and parameters theory.

The theta-criterion is a constraint on x-bar theory that was first proposed by Noam Chomsky as a rule within the system of principles of the government and binding theory, called theta-theory (θ-theory). As theta-theory is concerned with the distribution and assignment of theta-roles, the theta-criterion describes the specific match between arguments and theta-roles (θ-roles) in logical form (LF):

In generative grammar and related approaches, the logical form (LF) of a linguistic expression is the variant of its syntactic structure which undergoes semantic interpretation. It is distinguished from phonetic form, the structure which corresponds to a sentence's pronunciation. These separate representations are postulated in order to explain the ways in which an expression's meaning can be partially independent of its pronunciation, e.g. scope ambiguities.

In linguistics, a small clause consists of a subject and its predicate, but lacks an overt expression of tense. Small clauses have the semantic subject-predicate characteristics of a clause, and have some, but not all, properties of a constituent. Structural analyses of small clauses vary according to whether a flat or layered analysis is pursued. The small clause is related to the phenomena of raising-to-object, exceptional case-marking, accusativus cum infinitivo, and object control.

<span class="mw-page-title-main">Empty category principle</span>

In linguistics, the empty category principle (ECP) was proposed in Noam Chomsky's syntactic framework of government and binding theory. The ECP is supposed to be a universal syntactic constraint that requires certain types of empty categories, namely traces, to be properly governed.

A reciprocal pronoun is a pronoun that indicates a reciprocal relationship. A reciprocal pronoun can be used for one of the participants of a reciprocal construction, i.e. a clause in which two participants are in a mutual relationship. The reciprocal pronouns of English are one another and each other, and they form the category of anaphors along with reflexive pronouns.

<span class="mw-page-title-main">Empty category</span> Linguistics concept

In linguistics, an empty category, which may also be referred to as a covert category, is an element in the study of syntax that does not have any phonological content and is therefore unpronounced. Empty categories exist in contrast to overt categories which are pronounced. When representing empty categories in tree structures, linguists use a null symbol (∅) to depict the idea that there is a mental category at the level being represented, even if the word(s) are being left out of overt speech. The phenomenon was named and outlined by Noam Chomsky in his 1981 LGB framework, and serves to address apparent violations of locality of selection — there are different types of empty categories that each appear to account for locality violations in different environments. Empty categories are present in most of the world's languages, although different languages allow for different categories to be empty.

In generative linguistics, PRO is a pronominal determiner phrase (DP) without phonological content. As such, it is part of the set of empty categories. The null pronoun PRO is postulated in the subject position of non-finite clauses. One property of PRO is that, when it occurs in a non-finite complement clause, it can be bound by the main clause subject or the main clause object. The presence of PRO in non-finite clauses lacking overt subjects allows a principled solution for problems relating to binding theory.

A resumptive pronoun is a personal pronoun appearing in a relative clause, which restates the antecedent after a pause or interruption, as in This is the girli that whenever it rains shei cries.

In linguistics and grammar, Avalency refers to the property of a predicate, often a verb, taking no arguments. Valency refers to how many and what kinds of arguments a predicate licenses—i.e. what arguments the predicate selects grammatically. Avalent verbs are verbs which have no valency, meaning that they have no logical arguments, such as subject or object. Languages known as pro-drop or null-subject languages do not require clauses to have an overt subject when the subject is easily inferred, meaning that a verb can appear alone. However, non-null-subject languages such as English require a pronounced subject in order for a sentence to be grammatical. This means that the avalency of a verb is not readily apparent, because, despite the fact that avalent verbs lack arguments, the verb nevertheless has a subject. According to some, avalent verbs may have an inserted subject, which is syntactically required, yet semantically meaningless, making no reference to anything that exists in the real world. An inserted subject is referred to as a pleonastic, or expletive it. Because it is semantically meaningless, pleonastic it is not considered a true argument, meaning that a verb with this it as the subject is truly avalent. However, others believe that it represents a quasi-argument, having no real-world referent, but retaining certain syntactic abilities. Still others consider it to be a true argument, meaning that it is referential, and not merely a syntactic placeholder. There is no general consensus on how it should be analyzed under such circumstances, but determining the status of it as a non-argument, a quasi-argument, or a true argument, will help linguists to understand what verbs, if any, are truly avalent. A common example of such verbs in many languages is the set of verbs describing weather. In providing examples for the avalent verbs below, this article must assume the analysis of pleonastic it, but will delve into the other two analyses following the examples.

A bound variable pronoun is a pronoun that has a quantified determiner phrase (DP) – such as every, some, or who – as its antecedent.

<span class="mw-page-title-main">Sloppy identity</span> Concept in linguistics

In linguistics, sloppy identity is an interpretive property that is found with verb phrase ellipsis where the identity of the pronoun in an elided VP is not identical to the antecedent VP.

Logophoricity is a phenomenon of binding relation that may employ a morphologically different set of anaphoric forms, in the context where the referent is an entity whose speech, thoughts, or feelings are being reported. This entity may or may not be distant from the discourse, but the referent must reside in a clause external to the one in which the logophor resides. The specially-formed anaphors that are morphologically distinct from the typical pronouns of a language are known as logophoric pronouns, originally coined by the linguist Claude Hagège. The linguistic importance of logophoricity is its capability to do away with ambiguity as to who is being referred to. A crucial element of logophoricity is the logophoric context, defined as the environment where use of logophoric pronouns is possible. Several syntactic and semantic accounts have been suggested. While some languages may not be purely logophoric, logophoric context may still be found in those languages; in those cases, it is common to find that in the place where logophoric pronouns would typically occur, non-clause-bounded reflexive pronouns appear instead.

References

  1. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Sportiche, Dominique; Koopman, Hilda; Stabler, Edward (2014). An Introduction to Syntactic Analysis. West Sussex: Wiley Blackwell. p. 284. ISBN   978-1-4051-0017-5.
  2. 1 2 3 4 5 6 7 8 9 10 Dominique., Sportiche (2013-09-23). An introduction to syntactic analysis and theory. Koopman, Hilda Judith., Stabler, Edward P. Hoboken. ISBN   9781118470480. OCLC   861536792.{{cite book}}: CS1 maint: location missing publisher (link)
  3. 1 2 Boeckx, Cedric (2008). Bare Syntax. Oxford New York: Oxford University Press. ISBN   978-0-19-953424-1.
  4. Sportiche, Dominique. (2013-09-30). An introduction to syntactic analysis and theory. Koopman, Hilda Judith,, Stabler, Edward P. Chichester, West Sussex. ISBN   9781118470473. OCLC   842337755.{{cite book}}: CS1 maint: location missing publisher (link)
  5. Koster, Jan (1981). Locality Principles in Syntax. USA: Foris Publications. p. 178. ISBN   90-70176-06-8.
  6. Citko, Barbara (October 2011). "Small Clauses: Small Clauses". Language and Linguistics Compass. 5 (10): 748–763. doi:10.1111/j.1749-818X.2011.00312.x.
  7. 1 2 The Cambridge Handbook of Generative Syntax. Dikken, Marcel den, 1965-. Cambridge. 2014-05-14. ISBN   9781107341210. OCLC   854970711.{{cite book}}: CS1 maint: location missing publisher (link) CS1 maint: others (link)
  8. Haegeman, Liliane; Guéron, Jacqueline (1999). English Grammar: A Generative Perspective. Malden, Massachusetts: Blackwell Publishers Inc. ISBN   0-631-18839-8.
  9. Culicover, Peter W. (1984). Locality in linguistic theory. Wilkins, Wendy K. Orlando, Fla.: Academic Press. ISBN   0121992802. OCLC   9557971.