University of Groningen Binding and Bulgarian Schürcks-Grozeva, Lilia Lubomirova IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2003 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Schürcks-Grozeva, L. L. (2003). Binding and Bulgarian Groningen: s.n. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date: 19-10-2018
Chapter 2 The Classical Binding Theory 2.1. Introduction The major goal of this study is to investigate the binding phenomena in Bulgarian with respect to the classical Binding Theory, check its validity and come up with a new account if that validity is questioned. In this chapter, the development of the Binding Theory will be explored which basically covers work by Chomsky after 1973. The classical Binding Theory (Chomsky 1981, 1986) expresses a system for assigning intended coreference in all languages. It is the part of Government and Binding Theory which deals with indexing relationships between NPs. The two definitions of the classical Binding Theory are given in (1) and (2) below: (1) Binding Theory: (Chomsky 1981:188) Principle A: An anaphor is bound in its governing category. Principle B: A pronominal is free in its governing category. Principle C: An R-expression is free. (2) Binding Theory: (Chomsky 1986:166) Principle A: An anaphor is bound in a local domain. Principle B: A pronominal is free in a local domain. Principle C: An R-expression is free (in the domain of the head of its maximal chain). Chomsky (1982:78-89) suggests that the three types of NPs, namely, the anaphors, the pronouns and the R-expressions, are not syntactic primitives. Each of them can rather be represented in terms of the features [ anaphoric] 33
and [ pronominal]. Thus the three types of NPs have the following feature matrices: (3) a. anaphors [+anaphoric, -pronominal] b. pronouns [-anaphoric, +pronominal] c. R-expressions [-anaphoric, -pronominal] There is another theoretically possible combination of features, that is, [+anaphoric, +pronominal]. Chomsky (1982:78) claims that no overt element can exist with this feature characteristics. An explanation follows that such an element would have to be ungoverned according to Principles A and B and would hence violate the Case Filter. Only empty categories (e.g. PRO) can contain both anaphoric and pronominal features. It is the NP classification based on the features [ anaphoric] and [ pronominal] which poses the most serious problems for the binding phenomena in Bulgarian. These problems are addressed in detail in Chapter 3. In fact, the reasoning presented above, namely, that overt items with the feature matrix [+anaphoric, +pronominal] are not allowed because they would be ungoverned, does not hold in the minimalist framework. In it Chomsky (1993, 1995) abandons the notion 'government'. The later developments of the classical Binding Theory does not change much the originally postulated Principles A, B, and C. A survey of this post-classical Binding Theory research is given in Chapter 4. The critical review of the different stages and emphases of the Binding Theory exposed in this chapter, aims at bridging towards the morphological Binding Theory proposed in Chapter 4. On the basis of the binding phenomena in Bulgarian, the basic notions of the classical Binding Theory will be rethought and viewed on feature-based strategies in Chapters 4 and 5. 2.2. Some preliminaries The first steps towards the Binding Theory were introduced in Chomsky (1973). The theory was revised and refined throughout the 1970s and the 1980s. 34
In Chomsky (1973), it is pointed out that there is one domain relevant to both disjoint reference effects and anaphoric dependence. This domain is characterized in terms of two constraints, the Tensed-S Condition (TSC) and the Specified Subject Condition (SSC). These are given in (4) and (5), respectively: (4) Tensed-S Condition: (Chomsky 1973:238) No rule can involve X, Y in the structure... X... [ =... Y...]... where = is a tensed sentence (5) Specified Subject Condition: (Chomsky 1973:239) No rule can involve X, Y in the structure... X... [ =... Z... - WYV...]... where Z is the specified subject of WYV in = (4) excludes a passive like (6a) but allows (6b): (6) a. *John is believed [is honest] b. John is believed [to be honest] The embedded finite clause is = for (6a). While introducing traces, Chomsky (1973) does not yet treat the trace of NP movement as an anaphor. In order to understand the application of (4), i.e., the Tensed-S Condition, let us consider the sentences in (7): (7) a. The candidates each hated the other(s) b. The candidates each expected the other(s) to win c. The candidates each expected that the other(s) would win Following Dougherty (1970), Chomsky assumes a rule of each-movement according to which each is moved into the determiner position in the other(s). Having in mind this, the sentences in (7) should be transformed into those of (8): (8) a. The candidates hated each other b. The candidates expected each other to win c. *The candidates expected that each other would win 35
Only (8a) and (8b) are allowed. (8c) is blocked, as predicted by the Tensed- S Condition. To illustrate how the Specified Subject Condition (5) works, we regard the examples: (9) a. The candidates each expected [ S PRO to defeat the other] b. The candidates expected to defeat each other (10) a. The men each expected [ S the soldier to shoot the other] b. *The men expected the soldier to shoot each other (11) a. The men each saw [ NP pictures of the other] b. The men saw pictures of each other (12) a. The men each saw [ NP John s pictures of the other] b. *The men saw John s pictures of each other By specified subject in SSC (5), it is meant a subject NP that contains either lexical items or a pronoun that is not anaphoric. In the definition (5), = can be either NP or S. The sentences in (9) and (10) show the application of SSC (5) where = = S and those in (11) and (12) illustrate the application of this condition where = = NP. The rule of each-insertion works for (9a) and (11a), but it is blocked by the SSC (5) in (10b) and (12b). Another problem in Chomsky (1973) concerns non-coreference, the effects of which are very similar to those of Condition B in Chomsky (1981). This is his Rule of Interpretation (RI), formulated below: (13) Rule of Interpretation: (Chomsky 1973:241)... a rule of interpretation RI applying to the structure NP-V-NP (among others) seeks to interpret the two NPs as nonintersecting in reference, and where this is impossible (as in the case of first and second person pronouns), it assigns strangeness, marking the sentence with *. In Chomsky (1976), basically the same rule exists, but with a new name, Disjoint Reference (DR). This rule assigns disjoint reference to a pair (NP, pronoun).... It is also constrained by TSC (4) and SSC (5). When addressing first and second person pronouns in (13), Chomsky means the Unlike Person Constraint discussed by Postal (1966, 1969). This is formulated as a rule that assigns the feature * (deviant) to a sentence 36
PRO i -V-PRO j -X, where PRO i and PRO j are both first person or both second person. Therefore, sentences like (14) cannot exist: (14) a. *I watched me b. *I watched us leaving (in the mirror) c. *We watched me leaving If we consider the pair of sentences in (15) with the third person, we will see that they are permitted, but their interpretation is restricted: (15) a. He saw him b. The soldiers shot the officers (among them) Thus in (15a), the two pronouns are interpreted as different in reference and in (15b), the two NPs are interpreted as non-intersecting in reference, that is, the officers are not included among the soldiers doing the shooting. In addition, Chomsky argues that RI (13), like passive and each-movement, is constrained by TSC (4) and SSC (5). As long as = is between the two NPs, RI will be blocked. Thus the sentences in (16) do not possess the strangeness effect of those in (14): (16) a. I believe I will fail b. We believe I will fail Lasnik (1976) argues that there is no rule of coreference assignment responsible for the coreferential interpretation of pronouns in examples like (17): (17) a. John said that he would leave b. John said that he and Bill would leave In other words, no stipulated coreference or coindexing rules are needed for non-reflexive pronouns. If a pronoun is not marked as disjoint in reference from a given NP, its referential interpretation is free. Whether the pronoun is assigned the same or different reference as the given NP is a pragmatic rather than syntactic matter. In On Binding (OB), Chomsky (1980) formulates for the first time the revision of what is known as the classical Binding Theory. The notions of bound and free are introduced (Chomsky 1980:10): 37
a. An anaphor = is bound in > if there is a category c-commanding it and coindexed with it in >. b. Otherwise, = is free in >. In OB, it is assumed that all NPs are base-generated with referential indices. There are three further indexing procedures, given in (19), consistent with the General Conditions on Anaphora in (20): (19) General Conditions: (Reinhart 1983:136) a. A non-pronominal NP must be interpreted as noncoreferential with any NP that c-commands it. (This entails that a pronoun must be interpreted as noncoreferential with any full NP it c- commands.) b. A reflexive or reciprocal pronoun (an R-pronoun) must be interpreted as coreferential with (and only with) a c-commanding NP within a specified syntactic domain (e.g. its minimal governing category). c. A non-r-pronoun must be interpreted as noncoreferential with any c-commanding NP in the syntactic domain specified for (19b). (20) Indexing Procedures: (Reinhart 1983:138) a. All R-pronouns are obligatorily coindexed with any c- commanding NP, which means that their base-generated referential index is replaced with that of the c-commanding NP. b. All other NPs (including non-r-pronouns) are negatively coindexed or contra-indexed with all c-commanding NPs. Formally, this is achieved by adding to the referential index of the c-commanding NP, which is called anaphoric index (meaning contra-anaphoric index). c. An index-cancelling procedure then applies to all pronouns (R or non-r) equally cancelling coindexing and contra-indexing in the appropriate syntactic environment. (19a) is the anaphora condition proposed originally in Reinhart (1976), while (19b) and (19c) paraphrase the reciprocal rule and the disjoint reference rule from Chomsky (1973). The system illustrated in (20) establishes a three-valued coindexation system, that is, two NPs may be coindexed, contra-indexed or free (neither of the first two). 38
2.3. Chomsky (1981) In Chomsky (1981), the indexing system is simplified by being reduced to only two relations: coindexed (or bound) and free. The fact that pronouns show disjoint reference remained unsolved in the framework of the Government and Binding Theory (GB) in Chomsky (1981). 13 A core notion in GB theory appears to be the concept of Government: (21) Government: (Chomsky 1981:163) = governs > if and only if (i) = = X o (ii) = c-commands > and if C c-commands > then C either c-commands = or is c-commanded by >. A further precision of this notion follows in (22): (22) [ >... C... =... C... ], where (i) = = X o. (ii) where B is a maximal projection, if B dominates C then B dominates =. (iii) = c-commands C. (Chomsky 1981:165) C-command in (21) and (22) is understood as originally proposed by Reinhart (1976:32, 1983:18): 13 Note that in OB framework three possible referential relations among NPs existed: disjoint reference (indicated by anaphoric index); bound anaphora (indicated by identical referential indices); and finally free reference. In GB system every NP has exactly one index, that is, the referential index in OB terms. In other words, two integers are either the same or different. Concerning these two possibilities, Chomsky proposes that... pronouns are proximate if they are coindexed with some other element and obviative if not coindexed with any other element.... For further discussion see Lasnik (1976, 1981). 39
(23) C-command: Node A c (constituent)-commands node B iff the branching node most immediately dominating A also dominates B. Chomsky (1981:166) restates c-command in the following way: (24) = c-commands > if and only if: (i) = does not contain >. (ii) Suppose that C 1,..., C n is the maximal sequence such that (a) C n = =. (b) C i = = j. (c) C i immediately dominates C i+1. Then if @ dominates =, then either (I) @ dominates >, or (II) @ = C 1 and C 1 dominates >. Having adopted a very simple indexing theory, where indices are freely assigned and only referential indices exist (in the sense of OB), Chomsky proceeds with the basic notions of the Binding Theory. Case (i) and (ii) of (25) define the terms bound and free. X can be replaced by A or A' in (i), (ii) and (iv) of (25). In (iii) X and Y may be independently replaced by A or A'. 14 (25) (i) = is X-bound by > if and only if = and > are coindexed, > c-commands =, and > is in an X-position. (ii) = is X-free if and only if it is not X-bound. (iii) = is locally bound by > if and only if = is X-bound by >, and if C Y-binds = then either C Y-binds > or C = >. (iv) = is locally X-bound by > if and only if = is locally bound and X-bound by >. (Chomsky 1981:184-185) The domain in which an anaphor must be bound and pronominal free constitutes the governing category of the respective item. The Binding Theory, like the Case Theory, is developed within the theory of Government: 14 The possibility that an element may be locally bound by two different elements is excluded, i.e., that it may be both locally A-bound and locally A'-bound. 40
(26) Governing Category: (Chomsky 1981:188) = is the governing category for > if and only if = is the minimal category containing > and a governor of >, where = = NP or S. The nominal expressions have been subdivided into three basic categories: anaphors, pronominals and R-expressions (lexical NPs). Anaphors are NPs which have no capacity for inherent reference. There are two types of anaphors: lexical anaphors (reciprocals and reflexives) and NP-trace. Thus the three-way division is realized in the form of the following binding principles: (27) Binding Theory: (Chomsky 1981:188) (A) An anaphor is bound in its governing category (B) A pronominal is free in its governing category (C) An R-expression is free It is pointed out that binding is only A-binding, that is, binding by an NP in an argument position. Principle A of the Binding Theory (27) excludes anaphors from nominative subject position: nominative Case, like other Cases, is assigned under Government and since there is no NP in an S to c- command the subject, a nominative anaphor will always be free in the minimal S which contains it. This accounts for the ungrammatical examples in (28): (28) a. *John 1 believes that himself 1 is happy b. *They 1 believe that each other 1 are happy c. *John 1 is believed t 1 is happy In (28a) and (28b), himself and each other are lexical anaphors, while in (28c), we see the trace of NP movement. Pronouns are predicted to be free exactly in the same contexts where anaphors are bound. Compare the sentences (29) and (30): (29) *John 1 believes [him 1 to be happy] (30) John 1 believes that he 1 is happy 41
It is clear from these examples that the Nominative Island Condition (NIC) effects follow from Principles A and B of the Binding Theory (27). 15 If we consider next (31) and (32), the Specified Subject Condition (SSC) (see (2)) effect will be evident as well: (31) *John 1 believes [Mary to like himself 1 ] (32) John 1 believes [Mary to like him 1 ] For both himself and him in (31) and (32), the GC is the complement clause. Principle A rules out himself in (31) since it is not bound in its GC. Principle B, on the other hand, allows (32) since him is free in its GC. The fact that the governing category (GC) is irrelevant for R-expressions is illustrated in (33): (33) a. *John 1 thinks John 1 likes Mary b. *He 1 thinks John 1 likes Mary Chomsky further mentions that Binding Theory has one problematic feature at the conceptual level, namely, that it does not give an answer to the question (34): (34) Why are NP and S the two governing categories? Chomsky has been concerned with structures such as (35) since his work in 1973: (35) The men 1 think [that [[pictures of [each other] 1 ] will be on sale]] In (35), each other is free in its GC, the NP pictures of each other. In Chomsky (1981), a similar problem is presented: (36) a. *We 1 heard [their stories about each other 1 ] b. We 1 heard [some stories about each other 1 ] 15 The Nominative Island Condition (NIC) is defined in OB in the following way: A nominative anaphor cannot be free in S' (Chomsky 1980:36) The NIC aims at prohibiting the following types of constructions: (i) *The predict [that each other is suspicious] 42
In both sentences (36a) and (36b), each other is governed by about. According to the definition for GC in (26), the bracketed NP will be the GC for the anaphor which correctly rules out (36a) but incorrectly does so for (36b). In order to account for this kind of contrast, the notion governing category in (26) is reformulated by (37): (37) > is a governing category for = if and only if > is the minimal category containing =, a governor of =, and a SUBJECT accessible to =. (Chomsky 1981:211) The notion SUBJECT is introduced which includes standard subject and AGR. AGR is the SUBJECT in a finite clause, standard subject (NP of S) is the SUBJECT of an infinitival clause and finally NP of NP (if present) is SUBJECT of an NP. Thus in (36b), the object NP does not have SUBJECT and the GC is the matrix clause. In (36a), the NP contains a SUBJECT and consequently is a GC for the anaphor where a violation of Principle A is observed, while bound pronouns being allowed in these examples. Chomsky defines the well-formedness condition in (38) and accessibility in (39), respectively: 16 (38) *[ C... @...], where C and @ bear the same index. (39) Accessibility: (Chomsky 1981:212) = is accessible to > if and only if > is in the c-command domain of = and assignment to > of the index of = would not violate the Licensing Condition (given below in (56)). The condition expressed in (38) guarantees that a SUBJECT will not be accessible to a given anaphor if the same SUBJECT is coindexed with a category containing the anaphor. Let us consider (35) once again below, where AGR is taken into account: 17 (40) The men 1 think [ that [[pictures of each other 1 ] 2 AGR 2 will be on sale]] 16 (38) is often referred to as i-within-i condition. 17 The subject of a finite clause is always coindexed with AGR, expressed by superscripts in Chomsky (1981). 43
If we assign the index of AGR to each other, namely, 2, the effect of (38) will obtain, since the NP pictures of each other has the index 2. Consequently, the embedded clause does not constitute the GC for each other, and hence the latter is free to look for an antecedent in the matrix. Both the OB and GB frameworks presuppose total complementarity between anaphors and pronouns. This results in the incorrect conclusion that (39) will exclude a bound pronoun in the same circumstances: (41) The men 1 think [ that [[pictures of them 1 ] 2 AGR 2 will be on sale]] The governing category for them in (41) is the matrix clause and it is bound and not free in that clause, in violation of Principle B. Chomsky accounts for this phenomenon later in his work Knowledge of Language (1986). 2.4. Chomsky (1982) and Chomsky (1986) In Some Concepts and Consequences of the Theory of Government and Binding (1982:78-89), Chomsky proposes a featured-based classification of NPs with a view to anaphors, pronominals and R-expressions. This classification relies on the feature combinations of two oppositions, [ anaphoric] and [ pronominal]. Four categories of expressions are obtained then: (42) a. [+anaphoric, -pronominal] b. [-anaphoric, +pronominal] c. [+anaphoric, +pronominal] d. [-anaphoric, -pronominal] For overt categories with lexical content, examples of the above feature combinations are: (43) a. [+anaphoric, -pronominal] anaphor himself, each other b. [-anaphoric, +pronominal] pronominal him, he c. [+anaphoric, +pronominal]? d. [-anaphoric, -pronominal] R-expressions John, Mary 44
Overt categories with the characteristics [+anaphoric, +pronominal] are ruled out. As Chomsky (1982:78) puts it, such elements must be ungoverned by virtue of Principles A and B of the Binding Theory, hence violating the Case Filter. 18 It is only the empty category PRO which satisfies these requirements: (44) a. Mary was happy [PRO to see her kids] b. [PRO to criticize him] would be too late In (44a), PRO behaves like an anaphor, i.e., it is [+anaphoric], whereas in (44b) it resembles a pronominal, it is [+pronominal]. This dual character of PRO is the reason for its classification with the feature matrix [+anaphoric, +pronominal]. 19 In Knowledge of Language (1986), Chomsky introduces the term local domain. Turning to the definition of the local domain first, it is based on the notions of governing category (GC) and Complete Functional Complex (CFC). A governing category is defined in (45): 20 (45) Governing Category for =: (Chomsky 1986:169) A governing category is a maximal projection containing both a subject and a lexical category governing = (hence, containing =). Chomsky (1986) claims next that a governing category must be complete with respect to the realization of grammatical functions. He then introduces the notion of Complete Functional Complex (CFC): (46) Complete Functional Complex: (Chomsky 1986:169) A governing category is a complete functional complex (CFC) in the sense that all grammatical functions compatible with its head are realized in it. Hence, CFCs contain all the arguments which are compatible with the head of the category. CFSs also have complements and a subject. The local domain for binding an anaphor or a pronominal is defined as the Minimal Governing Category (MGC) for that anaphor or pronominal: 18 But as pointed out in the introduction to this chapter, this line of reasoning is unsustainable due to the lack of the notion 'government' in the minimalist account. 19 After proposing the four types of NPs in (42), Chomsky (1982) suggests that Principle C should be eliminated from the Binding Theory. 20 Compare earlier definitions of GC in (26) and (37). 45
(47) Local Domain: (Chomsky 1986:169) The local domain for an anaphor or pronominal = is the least CFC containing a lexical governor of = - the minimal governing category of = (MGC(=)). The general conclusion from these definitions is that both S and NP can constitute a Minimal Governing Category (MGC) if they contain a subject. We therefore obtain a substantial unification of the binding theory, incorporating both SSC and NIC, by defining the local domain as the minimal governing category. (Chomsky 1986:169) The Binding Theory, then, acquires the following form in (48): (48) Principle A: An anaphor is bound in a local domain. Principle B: A pronominal is free in a local domain. Principle C: An R-expression is free (in the domain of the head of its maximal chain). (Chomsky 1986:166) One case of a local domain is given by the Specified Subject Condition (SSC), illustrated in the examples (49) and (50): (49) The men i expected [ S the boys j to see them k ] (50) The men i expected [ S the boys j to see [each other] l ] Remember that the SSC states that pronouns are free and anaphors are bound in the domain of the nearest subject. Hence in (49), them may not be bound by the boys but may be bound by the men. In (50), each other must be bound by the boys. In other words, k is distinct from j but may be identical with i, and l must be identical with j. Once again, the SSC defines the domain of the closest c-commanding subject as a local domain. For the Nominative Island Condition (NIC) to be observed, we see from the sentences below that anaphors are not allowed in the subject position of a tensed clause, while bound pronominals are: (51) a. They expected [each other to win] b. They would prefer [ C for [ S each other to win]] c. *They expected [that each other would win] 46
(52) a. *They i expected [them i to win] b. *They i would prefer [ C for [ S them i to win]] c. They i expected [ that they i would win] In (51a) and (51b), the anaphor can occur in the subject position of infinitivals and the bound pronoun is excluded in (51c). It is clear from (52) that exactly the opposite picture appears: the bound pronoun cannot be in the subject position of infinitivals but it is grammatical in the subject position of a tensed sentence. As has already been observed, there is no full complementarity between anaphors and pronouns. The examples in (53) and (54) demonstrate this lack of complementarity: (53) The children like [each other s friends] (54) The children like [their friends] Contrary to expectations, the anaphor each other and the pronoun their can be bound by the children in (53) and (54). These configurations suggest that the GCs for anaphors and pronouns are different. As the Binding Theory states in (48), Principle A is satisfied by (53), but Principle B is violated in (54): the embedded noun phrase counts as MGC(=) when = is the anaphor or the pronoun, because it contains a subject (the anaphor or the pronoun itself) and a governor, that is, the head noun. Thus in (53), the clause is the relevant governing category for the anaphor while the NP is the relevant governing category for the pronoun in (54). Chomsky (1986) reformulates the Binding Theory taking into account the considerations above. There is an expression E with the indexing I, where an indexing is understood an association of indices with phrases of E: We say that the indexing I and the pair (=, >) are compatible with respect to the binding theory in the local domain > under the indexing I. (Chomsky 1986:171) Next BT-Compatibility is defined which extends the binding domain in case the element in consideration could not fulfill its binding theoretic requirement in the MGC: 47
(55) BT-Compatibility: (Chomsky 1986:171) I is BT-compatible with (=, >) if: A. = is an anaphor and is bound in > under I. B. = is a pronominal and is free in > under I. C. = is an R-expression and is free in > under I. The licensing condition (56) is added for a category = governed by a lexical category C in the expression E with indexing I. In fact, this condition expresses the revised content of the Binding Theory (48): (56) Licensing Condition: For some > such that (i) or (ii), I is BT-compatible with (=, >): (i) = is an R-expression and (a) if = heads its chain or (b) otherwise (a) > = E. (b) > is the domain of the head of the chain of =. (ii) = is an anphor or pronominal and > is the least CFC containing C for which there is an indexing J BT-compatible with (=, >). The revised formulation in (56) accounts for the problematic examples in (53) and (54). 21 As for the anaphor each other in the subject position of NP, there is no indexing within the domain NP that would be BT-compatible: being in the subject position of the NP, nothing within this NP can bind the anaphor. Therefore, NP is not the least CFC in which there is an indexing that is BT-compatible. The least CFC for which this condition is met must be the next higher CFC, that is, S. On the other hand, the pronominal in the subject position of the NP, must be free in its binding domain. The least CFC in which this requirement is satisfied is the NP itself since it constitutes the minimal governing category where the pronominal can be free. 21 It is argued in Lasnik (1989:86) that the licensing condition (56) holds strictly for governed =. The line of reasoning includes the analysis of sentences containing PRO: (i) *The children like [PRO friends] (ii)*the children thought that [[pictures (of) PRO[ were on sale] The analysis of these sentences in terms of (56) can be the following: the object NP is the least CFC containing PRO and a governor for PRO; there is a potential indexing, namely any indexing, under which PRO is free in that domain and the licensing condition for pronominals is satisfied; if PRO is coindexed with the children, the licensing condition for the anaphor will also be fulfilled; but there must be exactly one domain in which the licensing requirements are met. Consequently, the analysis presented above is wrong since it rests on two different >s, namely, the object NP and the matrix S. 48
Consequently, the binding domain for the pronominal is the NP, it is not extended to S as for the anaphor. In this chapter, the successive modifications in Binding Theory covered in a pre-minimalist theory by Chomsky (up to 1986) have been summarized. The developments of the theory can be viewed as an evolutionary refinement of one basic approach, rather than a series of drastic changes. 49