• Dutch1
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
11.3.6.Reconstruction
quickinfo

Normally, wh-movement is semantically or functionally motivated, which is especially clear in the case of wh-questions and topicalization constructions: Wh-movement in question (465a) is needed to create the operator-variable configuration in (465a'), while topicalization in example (465b) results in a special information-structural configuration, such as the topic-comment structure in (465b'). The traces indicated by t in the primeless examples in (465) are traditionally motivated by the fact that the displaced elements wat'what' and dit boek'this book' also perform the syntactic function of direct object; they indicate the designated argument position that is assigned the thematic role of theme as well as accusative case by the transitive main verb kopen'to buy'.

465
a. Wati heeft Peter ti gekocht?
  what  has  Peter  bought
  'What has Peter bought?'
a'. ?x (Peter has bought x)
b. Dit boeki heeft Peter ti gekocht.
  this book  has  Peter  bought
  'This book, Peter has bought.'
b'. [topicDit boek] [comment heeft Peter gekocht].

Of course, there are theories in which thematic roles and/or case are assigned in the surface position of the wh-phrase but there are empirical reasons for assuming that these elements are semantically interpreted in the position of their trace, a phenomenon that has become known as reconstruction; we refer the reader to Subsection IIB for the origin of this technical notion. This section will mainly illustrate reconstruction effects by means of the binding properties of wh-moved elements; see Barrs (2001) for a similar review for English, subsection I will therefore start by providing some theoretical background on binding. Given that reconstruction facts are easiest to demonstrate by means of topicalization, Subsection II will start with a discussion of this structure; reconstruction in questions and relative clauses is discussed in, respectively, III and IV. As the discussion of topicalization, wh-movement and relativization suffices to sketch a general picture of the issues involved, we will not discuss reconstruction in wh-exclamative and comparative (sub)deletion constructions (which have in fact not played a major role in the descriptive and theoretical literature on the phenomenon so far).

readmore
[+]  I.  Binding

Most research on binding is based on the empirical observation that referential personal pronouns such as hem'him' and (complex) reflexive personal pronouns such as zichzelf'himself' are in complementary distribution; this is illustrated for Dutch in the primeless examples in (466), in which coreferentiality is indicated by italics. The primed examples show that referential non-pronominal noun phrases normally cannot be used if a referential or a reflexive personal pronoun is possible; these examples are excluded on the reading that Jan and de jongen refer to the same individual.

466
a. Ik denk [dat Jan zichzelf/*hem bewondert].
  think   that  Jan  himself/*him  admires
  'I think that Jan admires himself.'
a'. * Ik denk [dat Jan de jongen bewondert].
  think   that  Jan  the boy  admires
b. Jan denkt [dat ik hem/*zichzelf bewonder].
  Jan thinks   that  him/himself  admire
  'Jan thinks that I admire him.'
b'. * Jan denkt [dat ik de jongen bewonder].
  Jan thinks   that  the boy  admire

Data like (466) are accounted for by binding theory, which has found its classic formulation in the so-called binding conditions proposed in Chomsky (1981), which we provide in a somewhat loose formulation as (467).

467
Binding conditions
a. Reflexive and reciprocal personal pronouns are bound in their local domain.
b. Referential personal pronouns are free (= not bound) in their local domain.
c. Referential noun phrases like Jan or de jongen'the boy' are free.

These conditions are extensively discussed in Section N5.2.1.5, but we will repeat some core issues here that are needed for our present purposes. A noun phrase is said to be bound if it is coreferential with a c-commanding antecedent. The term c-command refers to an asymmetric syntactic relation between the constituents in a sentence, which can be made more precise by means of the hierarchy in (468), in which A > B indicates that A c-commands B and everything that is embedded in B.

468
C-command hierarchy:
subject > direct object > indirect object-PP > PP-complement > adjunct

We can thus say that, under the intended coreferential readings, the direct objects in the (a)-examples in (466) are bound by the subject noun phrase Jan of the embedded clause, and that the embedded nominal direct objects in the (b)-examples are bound by the subject noun phrase Jan of the main clause; recall that A > B in (468) indicates that A c-commands B and everything that is embedded in B. Now consider again the three binding conditions in (467), which are normally referred to as conditions A, B and C. The fact that the primed examples in (466) are ungrammatical on the intended readings shows that c-command does not suffice to license binding: binding condition C expresses this by saying that a referential non-pronominal noun phrase cannot have a c-commanding antecedent at all. Binding conditions A and B further express that reflexive/reciprocal and referential personal pronouns differ with respect to the syntactic domain in which binding is possible, that is, in which they must/can have a c-commanding antecedent. If we assume for the moment that the relevant domain is the minimal clause in which we find the bound element, the data in (466a&b) follow: in (466a) the antecedent Jan is within the local domain of the pronoun, and binding conditions A and B predict that a reflexive pronoun can, but a referential pronoun cannot be bound by Jan; in (466b) the antecedent Jan is not within the local domain of the pronoun, and binding conditions A and B predict that a referential pronoun can, but a reflexive pronoun cannot be bound by Jan. This derives the complementary distribution of the referential and reflexive personal pronouns illustrated in (466a&b).
      The crucial thing for our discussion of reconstruction is that it is normally assumed that the c-command hierarchy in (468) is not a primitive notion, but derived from the hierarchical structural relations between the elements mentioned in it. It suffices for our present purpose to say that the subject of a clause c-commands the direct object of the same clause because the former is in a structurally higher position than the latter; in the overall structure of the clause given in (469), which is extensively discussed in Chapter 9, the subject occupies the specifier position of TP immediately following the C-position, while the object occupies some lower position within XP.

469

If c-command should indeed be defined in terms of structural representations, wh-movement affects the c-command relations between the clausal constituents: after wh-movement of the object into the specifier of CP, the object will c-command the subject in the specifier of TP. We therefore expect wh-movement to alter the binding possibilities, but the following subsections will show that this expectation is not borne out; the wh-moved phrase normally behaves as if it still occupies its original position.

[+]  II.  Topicalization

That wh-movement does not affect binding relations can be easily demonstrated by means of topicalization. We will start with a presentation of the core data, which shows that the binding possibilities are computed from the original position of the topicalized phrase. After this, we will briefly compare reconstruction effects with so-called connectivity effects found in contrastive left-dislocation constructions.

[+]  A.  The data

If the binding conditions were calculated from the landing site of wh-movement, topicalization of a reflexive pronominal direct object is expected to bleed binding. Example (470b) shows, however, that with respect to binding the topicalized reflexive pronoun zichzelf behaves as if it is still in the position indicated by its trace; coreferentiality is again indicated by italics.

470
a. Jan bewondert zichzelf het meest.
  Jan  admires  himself the most
  'Jan admires himself the most.'
b. Zichzelfi bewondert Janti het meest.
  himself  admires  Jan  the most
  'Himself Jan admires the most.'

That topicalization does not bleed binding can also be illustrated by means of the examples in (471), in which a reciprocal possessive pronoun is embedded in a direct object; topicalization of this object does not affect the binding possibilities. Note in passing that, contrary to reciprocal and referential personal pronouns, reciprocal and referential possessive pronouns are not in complementary distribution given that elkaars can readily be replaced by hun'their'; we refer the reader to Section N5.2.2 for detailed discussion.

471
a. Zij bewonderen [elkaars moeder] het meest.
  they  admire  each.otherʼs mother  the most
  'They admire each otherʼs mother the most.'
b. [Elkaarsmoeder]i bewonderen zijti het meest.
  each.otherʼs mother  admire  they  the most
  'Each otherʼs mother they admire the most.'

Another case showing that topicalization does not bleed binding is illustrated by the examples in (472), which allow a bound-variable reading of the possessive pronoun zijn'his'; according to this reading every person x admires his own parents: ∀x (x:person) admire (x, x's parents). This reading only arises if the quantifier binds (hence: c-commands) a referential pronoun and we might therefore expect that topicalization in (472b) would make this reading impossible, but this expectation is not borne out.

472
a. Iedereen bewondert zijn (eigen) ouders het meest.
  everyone  admires  his own parents  the most
  'Everyone admires his (own) parents the most.'
b. Zijn (eigen) oudersi bewondert iedereen ti het meest.
  his own parents  admires  everyone  the most
  'His (own) parents everyone admires the most.'

      If the binding conditions were calculated from the landing site of wh-movement, topicalization of a referential (pronominal) direct object is expected to enable it to function as the antecedent of the subject of its clause, but example (473b) shows that this is not the case: with respect to binding the objects hem and die jongen again behave as if they are still in the position indicated by their trace.

473
a. * Jan bewondert hem/die jongen het meest.
  Jan  admires  him/that boy  the most
b. * Hemi/Die jongeni bewondert Janti het meest.
  him/that boy  admires  Jan  the most

A plausible hypothesis would of course be that example (473b) is unacceptable because the subject Jan is bound by the topicalized phrase and thus violates binding condition C. This hypothesis is, however, refuted by the fact that the matrix subject Jan in (474b) can be coreferential with the topicalized pronoun hem'him': the example is perhaps somewhat marked compared to example (466b) but this seems to be a more general property of long topicalization; see the discussion in Section 11.3.3, sub II. This again leads to the conclusion that wh-movement does not affect binding possibilities.

474
a. Jan denkt [dat ik hem/*die jongen het meest bewonder].
  Jan thinks   that  him/that boy  the most  admire
  'Jan thinks that I admire him the most.'
b. (?) Hem/*Die jongen denkt Jan [dat ik ti het meest bewonder].
  him/that boy  thinks  Jan   that  the most  admire
  'Him Jan thinks that I admire the most.'

      Reconstruction is sometimes also illustrated in the literature by means of examples such as (475a), in which a bound nominal phrase is embedded in a complementive.

475
a. Jan is [AP trots [PP op zichzelf/*hem/*die jongen]].
  Jan is  proud  of himself/him/that boy
b. [AP trots [PP op zichzelf/*hem/*die jongen]] is Jan niet.
  proud  of himself/him/that boy  is Jan not

Some linguists do not accept (475b) as a convincing example of reconstruction as they assume that the subject originates as the external argument of the AP: on the assumption that the moved phrase is a full small clause that contains an NP-trace of the subject Jan, this trace serves as an antecedent for the nominal phrase.

476
a. Jani is [APti trots [PP op zichzelf/*hem/*die jongen]].
  Jan is  proud  of himself/him/that boy
b. [APti trots [PP op zichzelf/*hem/*die jongen]]j is Janitj niet.
  proud  of himself/him/that boy  is Jan not

However, even if the representations in (476) are the correct ones, reconstruction is still needed because it is generally assumed that NP-traces are subject to binding condition A as well: like reflexive pronouns, they must be bound by their antecedent (= the moved phrase) within their local domain.
      For VP-topicalization constructions like (477b) more or less the same holds: some linguists who assume that the subject is base-generated in the lexical projection of the verb do not accept it as a convincing example of reconstruction since they assume that the topicalized VP also contains the NP-trace of the subject Jan, which can serve as an antecedent. But even if this is true, reconstruction is still needed given that NP-traces are generally assumed to be subject to binding condition A as well.

477
a. Jani heeft [VPti zichzelf/*hem/*die jongen beschreven].
  Jan  has  himself/him/that boy  described
  'Jan has described himself/him/that boy.'
b. [VPti zichzelf/*hem/*die jongen beschreven]j heeft Janitj.
  himself/him/that boy  described  has  Jan

If NP-traces must indeed be bound, VP-topicalization constructions of the type in (478) also provide evidence in favor of reconstruction. Under the standard assumption that the clause-initial position can be filled by phrases only (and not by heads), the theme argument must have been extracted from the VP by NP-movement (nominal argument shift of the type discussed in Section 13.2) before the VP is topicalized. The VP thus contains a trace of the theme argument and reconstruction is needed in order for the trace to be bound by the moved noun phrase mijn huis'my house'; see Section 11.3.3, sub VIC, for more discussion.

478
a. Ze hebben mijn huis nog niet geschilderd.
perfect tense
  they  have  my house  yet  not  painted
  'They havenʼt painted my house yet.'
a'. [VPti Geschilderd]j hebben ze mijn huisitj nog niet.
  painted  have  they  my house  yet  not
  'They havenʼt painted my house yet.'
b. Mijn huis wordt volgend jaar geschilderd.
passive
  my house  is  next year  painted
  'My house will be painted next year.'
b'. [VPti Geschilderd]j wordt mijn huisi volgend jaar tj.
  painted  is   my house next year
  'My house will be painted next year.'

      The examples so far all involve topicalization of arguments, complementives, and VP, and we have seen that such cases exhibit reconstruction effects: binding possibilities are computed from the base position of the moved phrase. This does not seem to hold for adjuncts, however, as is clear from the contrast between the two examples in (479); if the adverbial clause in (479b) were interpreted in the same position as the adverbial clause in (479a), we would wrongly expect coreference between Jan and hij to be blocked by binding condition C in both cases. This contrast has given rise to the idea that examples such as (479b) are actually not derived by wh-movement, but involve base-generation of the adjunct in clause-initial position; that this is possible is then attributed to the fact that adjuncts are not selected by the verb and can consequently be generated externally to the lexical projection of the verb.

479
a. * Hij ging naar de film [omdat Jan moe was].
  he  went  to the movie  because  Jan  tired  was
b. [Omdat Jan moe was], ging hij naar de film.
  because  Jan  tired  was  went  he  to the movie
  'Because Jan was tired, he went to the movie.'

Note in passing that the lack of reconstruction cannot be demonstrated on the basis of binding condition B, as referential pronouns embedded in adverbial clause can always be coreferential with the subject of a matrix clause; this is shown in (480).

480
a. Jan ging niet naar de film [omdat hij moe was].
  Jan  went  not  to the movie  because  he  tired  was
  'Jan didnʼt go to the movie because he was tired.'
b. [Omdat hij moe was], ging Jan niet naar de film.
  because he tired was  went  Jan  not  to the movie
  'Because he was tired, Jan didnʼt go to the movie.'

      A similar lack of reconstruction can be observed in the examples in (481); cf. Van Riemsdijk & Williams (1981). In this case an argument is topicalized but the contrast between the two examples shows that the reconstruction effect is lacking: contrary to what would be expected if the topicalized phrase were interpreted in the position of its trace, the referential noun phrase Jan embedded in the relative clause can be coreferential with the pronoun hij in (481b). It is of course not possible to appeal to an argument-adjunct asymmetry in this case, but it has been suggested that the (optional) relative clause is an adjunct that can be generated after the object has undergone wh-movement; see Barss (2001) and Sportiche (2006) for details.

481
a. * Hij wil [het boek [dat Jan gekocht heeft]] aan Marie geven.
  he  wants   the book   that Jan  bought  has  to Marie  give
b. [Het boek [dat Jan gekocht heeft]]i wil hijti aan Marie geven.
  the book  that  Jan bought  has ` wants  he  to Marie  give
  'The book that Jan has bought, he wants to give to Marie.'

The examples in (482) show again that the lack of reconstruction cannot be demonstrated on the basis of binding condition B, as referential pronouns embedded in a relative clause can be coreferential with the subject of a matrix clause.

482
a. Jan wil [het boek [dat hij gekocht heeft]] aan Marie geven.
  Jan wants   the book   that  he  bought  has  to Marie  give
  'Jan wants to give the book that he has bought to Marie.'
b. [Het boek [dat hij gekocht heeft]]i wil Janti aan Marie geven.
  the book   that  he  bought  has  wants  Jan  to Marie  give
  'The book that he has bought, Jan wants to give to Marie.'

The discussion of the data in this subsection has shown that a reconstruction effect obligatorily occurs if some argument, complementive or verbal projection is topicalized. Reconstruction effects are absent if an adverbial clause occupies the clause-initial position or if the topicalized phrase is modified by a relative clause.

[+]  B.  Reconstruction versus connectivity effects

Because wh-movement has a clear semantic import, the standard (but not uncontroversial) assumption is that it precedes the semantic interpretation of the clause. The fact that for the purpose of the binding theory formulated in (467) topicalized phrases behave as if they still occupy the position indicated by their traces has led to theories according to which wh-movement is at least partly undone before the semantic interpretation of the syntactic representation takes place; the technical term for this is Reconstruction. A more recent approach, which makes reconstruction superfluous, is Chomsky's (1995:ch.3) copy theory of movement, according to which movement is a copy-and-paste operation that leaves a phonetically empty copy (a copy that is not pronounced in the actual utterance) of the moved constituent in its original position. For convenience, we will follow general practice by maintaining the notion of reconstruction as a purely descriptive term. The core finding that all theories try to explain is that binding of nominal arguments should be formulated in terms of A-positions, that is, argument positions to which thematic roles, agreement features and/or case are assigned; movement into A'-positions (positions such as the clause-initial position that may also be occupied by non-arguments) does not affect the binding possibilities. We refer the reader to Barrs (2001), Sportiche (2006) and Salzmann (2006) for critical reviews and discussions of the various theoretical implementations of this insight.
      The standard view seems to be that reconstruction effects are syntactic in nature, but there are grounds for doubting that these effects are part of syntax proper. In order to show this we have to make a brief digression on contrastive and hanging-topic left-dislocation; see Section 14.2 for a more extensive discussion. Left dislocation is characterized by the fact that there is some phrase preceding the clause-initial position, which is associated with a resumptive element elsewhere in the clause. The two types of left-dislocation constructions differ in the form and position of the resumptive element: hanging-topic left-dislocation constructions have a resumptive pronoun in the form of a referential pronoun such as hem'him, which is' located in the middle field of the clause, as in (483a); contrastive left-dislocation constructionshave a resumptive pronoun in the form of a demonstrative pronoun such as die'that, which is' located in clause-initial position, as in (483b). Observe that we indicate the relation between the left-dislocated phrase and the resumptive pronoun by means of indices (just like the relation between a moved phrase and its trace).

483
a. Jani, ik heb hemi niet gezien.
hanging-topic LD
  Jan  have  him  not  seen
  'Jan I havenʼt seen him.'
b. Jani, diei heb ik ti niet gezien.
contrastive LD
  Jan  dem  have  not  seen
  'Jan I havenʼt seen him.'

At first sight, the examples in (484) seem to show that left dislocation differs from topicalization in that it does affect the binding possibilities. Van Riemsdijk & Zwarts (1997) and Vat (1997) suggest, however, that the unacceptability of the examples in (484) is due to the fact that resumptive pronouns are referential pronouns which are subject to binding condition B of the binding theory by themselves. In order to satisfy the binding conditions on the reflexive zichzelf'himself' the resumptive pronouns hem'him' and die'that' must take the subject Jan as a local antecedent, which results in a violation of binding condition B. Observe that the binding conditions for the resumptive pronoun die in (484b) should be computed from its original object position indicated by its trace in object position.

484
a. * Zichzelfi, Jan bewondert hemi het meest.
hanging-topic LD
  himself  Jan  admires  him  the most
  Intended meaning: 'Jan admires himself the most.'
b. * Zichzelfi, diei bewondert Janti het meest.
contrastive LD
  himself  dem  admires  Jan  the most
  Intended meaning: 'Jan admires himself the most.'

Violations of binding condition B induced by the resumptive pronouns themselves can be avoided if the reflexive/reciprocal pronoun is more deeply embedded in the topicalized phrase, as in the examples in (471). Their left-dislocation counterparts in (485) show that the two types of left dislocation exhibit different behavior in such cases; while the hanging-topic construction is rated as ungrammatical in Van Riemsdijk & Zwarts (1997) and Vat (1997), the contrastive left-dislocation construction is fully acceptable. The fact that the left-dislocated phrase can be interpreted in the position of the trace of the wh-moved demonstrative die has become known as the connectivity effect.

485
a. * [Elkaarsmoeder]i, zij bewonderen haari het meest.
hanging topic LD
  each.otherʼs mother  they  admire  her  the most
  'Each otherʼs mother they admire the most.'
b. [Elkaarsmoeder]i, diei bewonderen zijti het meest.
contrastive LD
  each.otherʼs mother  dem  admire  they  the most
  'Each otherʼs mother they admire the most.'

Connectivity effects also arise in the left-dislocation counterparts of the topicalization construction in (472b) with a bound variable reading. Van Riemsdijk & Zwarts (1997) and Vat (1997) show that there is again a contrast between hanging-topic and contrastive left-dislocation.

486
a. * [Zijn (eigen) ouders]i, iedereen bewondert zei het meest.
hanging-topic LD
  his own parents  everyone admires  them  the most
b. [Zijn (eigen) ouders]i, diei bewondert iedereenti het meest.
contrastive LD
  his own parents  dem  admires  everyone  the most

For completeness' sake, consider the contrastive left-dislocation constructions in (487), which show again that the acceptability judgments on the contrastive left-dislocation constructions are more or less the same as in the corresponding topicalization constructions in (473b) and (474b).

487
a. * Hemi/Die jongeni, diei bewondert Janti het meest.
  him/that boy  dem  admires  Jan  the most
b. (?) Hemi/*Die jongeni, diei denkt Jan [dat ik ti het meest bewonder].
  him/that boy  dem  thinks  Jan   that  the most   admire
  'Him, Jan thinks that I admire the most.'

      The discussion above has shown that contrastive left-dislocation constructions exhibit connectivity effects which closely resemble the reconstruction effects found in topicalization constructions. Given this similarity, it is temping to provide a single theoretical account of the two types of effect. This might lead to the conclusion that there is some kind of matching effect in the sense that the demonstrative pronoun die simply takes over certain semantic properties of the left-dislocated phrase and transmits these to the position of its trace; however, this would go against the current idea that reconstruction effects follow from the copy theory of movement: the claim that movement is a copy-and-paste operation that leaves an actual copy of the moved constituent in its original position.
      Alternatively, one might attempt to show that left-dislocated phrases are base-generated within the clause they are attached to and find their surface position by (a series of movements including) wh-movement. If such an analysis is feasible, we could maintain that reconstruction effects result from the copy-and-paste operation proposed by the copy theory of movement; see Grohmann (2003:ch.4) and De Vries (2009) for detailed proposals. This would immediately account for the differences in connectivity effects established in this subsection between hanging-topic and contrastive left-dislocation constructions: hanging-topic constructions have a resumptive pronoun in the middle field of the clause, and we can therefore safely conclude that they do not involve wh-movement, and we consequently expect connectivity effects to be absent. There are, however, two potential problems for this approach. First there does not seem to be independent evidence for assuming that left-dislocated phrases have ever occupied a clause-internal position. Second, this approach should provide a reasonable account for the fact that left-dislocated phrases may strand prepositions, while topicalized phrases (and wh-moved phrases in general) are normally not able to do that; see the contrast between the (a)- and (b)-examples in (488).

488
a. *? Dat boek heb ik lang naar gezocht.
topicalization
  that book  have  long  for  looked
a'. * Wat heb je lang naar gezocht?
question formation
  what  have  you  long  for  looked
b. Dat boek, daar heb ik lang naar gezocht.
contrastive LD
  that book  that have I long for looked
  'that book, I have looked for it a long time.'

We will return to the question as to whether reconstruction and connectivity effects can be given a (more or less) unified treatment in the discussion of relativization in Subsection IV below.

[+]  III.  Wh-movement

Section 11.3.1.1, sub II, discussed the hypothesis that the obligatoriness of wh-movement in wh-questions follows from the fact that it is instrumental in deriving an operator-variable chain in the sense of predicate calculus. It has also shown that this hypothesis runs into problems with examples like (489a&b), in which the moved wh-phrase is complex: the resulting syntactic representations cannot be directly translated into the desired semantic representations in the primed examples, as only a subpart of the wh-moved phrase corresponds to the question o