32
101 Scope-marking Strategies in the Acquisition of Long Distance wh-Questions in French and Dutch Celia Jakubowicz 1 , Nelleke Strik 2 1 CNRS-FRE 2929, Université Paris Descartes, Institut de Psychologie 2 CNRS-UMR 7023, Université Paris 8 and LPP, CNRS-FRE 2929 1 Introduction It is well known that Long Distance (LD) wh- questions are rare in corpora of spon- taneous speech of young children. For instance, Stromswold (1995) shows that only 200 out of 13120 object and subject questions (that is 1.5%) produced by 12 English- Abstract This paper reports the results of an elicited production task of Long Distance (LD) wh-questions conducted with typically developing French- and Dutch- speaking children aged four and six, and adult control groups for each language. It is shown that besides input-convergent wh-questions, in both languages children use nontarget strategies to express scope. While in both French and Dutch children produce Partial Movement and wh-copying questions, only French children use Partial Movement without an overt scope-marker in the left periphery of the matrix clause. We argue that our results are consistent with the Derivational Complexity hypothesis put forward by Jakubowicz (2004, 2005). Moreover our results confirm that parametric choices regarding the position of both the wh-word and the verb in the sentence are already set at the ages considered here. Key words derivational complexity language acquisition scope-marking strategies syntax wh-interrogatives LANGUAGE AND SPEECH, 2008, 51 (1 & 2), 101 –132 Language and Speech ‘Language and Speech’ is © Kingston Press Ltd. 1958 – 2008 Acknowledgements. We would like to thank the principals (Mr. Denat and Mrs. Halimi), teachers and children of the French Elementary School and Kindergarten (52 – 56 rue de Turenne, Paris 3, France) and the principal (Mr. Dijkstra), teachers and children of the Dutch Elementary School and Kindergarten Schoter Duijn (Den Helder, the Netherlands) for their collaboration in this research. In addition, we would like to thank Catherine Rigaut for her assistance in all aspects of the project, Marlies van der Velde for her help in recording the French protocol and in translating the French version into Dutch. Thanks also to the audience of the Latsis Colloquium (Geneva, January 2006) for helpful suggestions to Nelleke Strik. Finally we are very much grateful to Jacqueline van Kampen and Rosalind Thornton for their comments and observations of the first version of this paper, to Luigi Rizzi and Laurie Tuller for insightful comments on the present version and to Ulrich Frauenfelder for important material support. This work was supported by the three years Grant N° 04 5 601 of the Fondation National de la Science, Agence Nationale de la Recherche, France and a doctoral fellowship from the “Prins Bernhard Cultuurfonds” the Netherlands). Address for correspondence. Nelleke Strik, CNRS-UMR 7023 Université Paris 8, 2 rue de la Liberté, 93526 Saint-Denis Cedex, France; e-mail: < [email protected] >.

Scope-marking strategies in the acquisition of long distance wh-questions in French and Dutch

Embed Size (px)

Citation preview

Language and Speech

C. Jakubowicz, N. Strik 101

Scope-marking Strategies in the Acquisition of Long Distance wh-Questions in French and Dutch

Celia Jakubowicz1, Nelleke Strik2

1 CNRS-FRE 2929, Université Paris Descartes, Institut de Psychologie2 CNRS-UMR 7023, Université Paris 8 and LPP, CNRS-FRE 2929

1Introduction

It is well known that Long Distance (LD) wh-questions are rare in corpora of spon-taneous speech of young children. For instance, Stromswold (1995) shows that only 200 out of 13120 object and subject questions (that is 1.5%) produced by 12 English-

Language and Speech

Abstract

This paper reports the results of an elicited production task of Long Distance (LD) wh-questions conducted with typically developing French- and Dutch-speaking children aged four and six, and adult control groups for each language. It is shown that besides input-convergent wh-questions, in both languages children use nontarget strategies to express scope. While in both French and Dutch children produce Partial Movement and wh-copying questions, only French children use Partial Movement without an overt scope-marker in the left periphery of the matrix clause. We argue that our results are consistent with the Derivational Complexity hypothesis put forward by Jakubowicz (2004, 2005). Moreover our results confirm that parametric choices regarding the position of both the wh-word and the verb in the sentence are already set at the ages considered here.

Key words

derivational complexity

language acquisition

scope-marking strategies

syntax

wh-interrogatives

LANGUAGE AND SPEECH, 2008, 51 (1&2), 101–132

Language and Speech ‘Language and Speech’ is © Kingston Press Ltd. 1958 – 2008

Acknowledgements. We would like to thank the principals (Mr. Denat and Mrs. Halimi), teachers and children of the French Elementary School and Kindergarten (52 – 56 rue de Turenne, Paris 3, France) and the principal (Mr. Dijkstra), teachers and children of the Dutch Elementary School and Kindergarten Schoter Duijn (Den Helder, the Netherlands) for their collaboration in this research. In addition, we would like to thank Catherine Rigaut for her assistance in all aspects of the project, Marlies van der Velde for her help in recording the French protocol and in translating the French version into Dutch. Thanks also to the audience of the Latsis Colloquium (Geneva, January 2006) for helpful suggestions to Nelleke Strik. Finally we are very much grateful to Jacqueline van Kampen and Rosalind Thornton for their comments and observations of the first version of this paper, to Luigi Rizzi and Laurie Tuller for insightful comments on the present version and to Ulrich Frauenfelder for important material support. This work was supported by the three years Grant N° 04 5 601 of the Fondation National de la Science, Agence Nationale de la Recherche, France and a doctoral fellowship from the “Prins Bernhard Cultuurfonds” the Netherlands).

Address for correspondence. Nelleke Strik, CNRS-UMR 7023 Université Paris 8, 2 rue de la Liberté, 93526 Saint-Denis Cedex, France; e-mail: < [email protected] >.

Language and Speech

102 Acquisition of LD wh-questions in French and Dutch

speaking children are LD questions.1 These children are aged between 2;0 and 6;0 and their data are available in the CHILDES database (MacWhinney, 1991).

However, in elicited production experiments children as young as three are able to produce LD wh-questions, as originally shown by Thornton (1990). Interestingly, besides target-consistent questions like (1), English-speaking children optionally use scope-marking strategies that are not always available in their input, such as Partial Movement of the wh-word to the left periphery of the embedded sentence (henceforth PM) as in (2a) and wh-copying in (2b). (See also McDaniel, Chiu, & Maxfield, 1995; de Villiers, Roeper, & Vainikka, 1990)

(1) Who do you think is in the box?

(2) a. What do you think who is in the box?

b. Who do you think who is in the box?

During the last 10 years, similar phenomena have been observed in other languages like Dutch (van Kampen, 1997; van Kampen & Evers, 1995), French (Dos Anjos, 2004; Jakubowicz, 2004, 2005, to appear; Lancien, 2003; Oiry, 2002; Oiry & Demirdache, 2006; Strik, 2002, 2003, 2006), Spanish, and Basque (Gutierrez, 2006) in studies conducted with typically developing children, L2 learners and children with Specific Language Impairment (SLI). A third type of nontarget scope-marking strategy showed up in the French studies (and very rarely in Spanish), namely Partial Movement of the wh-word to the left periphery of the embedded clause, without an overt scope-marker in the left periphery of the matrix clause (see (3)). Following Fanselow (2006), we will call this construction Simple Partial Movement (henceforth SPM).

(3) Tommy, tu penses quoi que Lala préfère? (Arno 4;11.18)Tommy, you think what that Lala prefers

‘Tommy, what do you think that Lala prefers?’

Different explanations have been proposed to account for the nontarget strate-gies used by the children (see the references cited above). Note however that these explanations have been suggested on the basis of a rather small number of data in each of the studies. Furthermore in certain studies no information is given concerning the proportion of nontarget LD strategies and / or the number of elicited items. For example, in Crain and Thornton (1998) the number of items elicited from each partici-pant remains implicit and it seems to be the case that this number is not the same for each participant (cf. Table 22.1, p.198). In other studies, where the above-mentioned information is present, a post hoc analysis of the experimental protocols suggests that the way in which LD wh-questions were elicited may not have been felicitous (enough). For example, in the studies conducted by Oiry (2002) and Strik (2002) with

1 Van Kampen (1997), who studies spontaneous and partly elicited utterances of two Dutch-speaking children, reports that the first child produces 49 LD wh-questions between the age of 7;1.7 and 8 ;11.14 years and the second five LD wh-questions between the age of 4 ;7.2 and 5 ;11.18 years.

Language and Speech

C. Jakubowicz, N. Strik 103

French-speaking children using approximately the same experimental protocol, the proportion of produced LD wh-questions is never more than 50%, and the number of non- target LD questions is very low. An improved version of this protocol used in the studies of Dos Anjos (2004), Jakubowicz (2004, 2005), Lancien (2003) and Strik (2003, 2006),2 similarly shows that LD wh-questions are in general avoided by typically developing children and children with SLI (about 50% of LD questions for the 6-year-old typically developing children and considerably less for the 3- and 4-year-olds and the children with SLI). Furthermore among the LD wh-questions, nontarget structures represent a small number, except for the children with SLI.

One of the aims of the present study is to see whether creating a more appropriate context of elicitation increases the number of LD wh-questions produced by the children. Additionally, by using exactly the same experimental protocol in French and in Dutch, we want to make a systematic comparison of the acquisition of LD wh-questions in these two languages that differ with respect to the position of both the wh-word and the (inflected) verb in the sentence.

This paper is organized as follows: First we present the principal typological differences of LD wh-questions in French and Dutch. Second we present the hypoth-esis adopted in this paper namely, the Derivational Complexity hypothesis, and the predictions for our study. After a description of the methodology we report the main results. These are followed by a discussion and a conclusion.

2Typology of LD wh-questions in adult French and Dutch

There exist several strategies to make a LD wh-question in French. In the most “standard” or normative one the wh-word is fronted and there is subject-verb inversion or V-to-C movement (see (4a) for a locative adjunct question and (4b) for a subject question). Adopting the Minimalist Program (e.g., Chomsky, 2001), we assume that the left periphery of each wh-question contains a wh-feature and that the wh-word moves to the left periphery in order to check this feature leaving an unpronounced copy in its base position and in the left periphery of each embedded clause (see the following section).

(4) a. Oùi disj-tu (disj) (oùi) que Marie va (oùi)? 3

where say you that Marie goes

‘Where do you say that Marie is going?’

2 This protocol as well as the former one was elaborated by members of the team ‘Acquisition et dysfonctionnement du langage’ under the direction of C. Jakubowicz at the LPE, CNRS, Université Paris 5. The 2002 protocol was created by C. Jakubowicz, C. Rigaut, and N. Strik, the second one also included the participation of A. de Vanssay and S. Lancien.

3 Elements in brackets represent unpronounced copies of moved constituents. Only copies in relation to wh-movement are indicated.

Language and Speech

104 Acquisition of LD wh-questions in French and Dutch

b. Quii pensesj-tu (pensesj) (quii) qui (quii) lit des histoires? who think you that reads stories

‘Who do you think is reading stories?’

Second, the wh-word can be fronted without V-to-C movement (see (5), which is the equivalent of (4)). Note that in (5) a and b there is a bare wh-word, while in (5) a’ and b’ the wh-word is followed by est-ce que. In this paper we do not go into the details of the est-ce que construction.

(5) a. Oùi tu dis (oùi) que Marie va (oùi)? where you say that Marie goes

a’. Oùi est-ce que tu dis (oùi) que Marie va (oùi)? where is it that you say that Marie goes

b. Quii tu penses (quii) qui (quii) lit des histoires? who you think that reads stories

b’. Quii est-ce que tu penses (quii) qui (quii) lit des histoires? who is it that you think that reads stories

A third strategy is to leave the wh-word in-situ (see (6)), a strategy which, although attested, appears to be infrequent. Some authors do not accept the existence of wh-in-situ in LD questions (e.g., Cheng & Rooryck, 2000), but others do (Obenauer, 1994, among others). Indeed, wh-in-situ LD questions were produced, although very infrequently, by adults and children in the French studies mentioned above.

(6) a. Tu dis que Marie va où? you say that Marie goes where?

b. Tu penses que qui lit des histoires? you think that who reads stories

A fourth strategy, that to our knowledge has not been previously described in the literature on LD wh-questions in French, consists of LD questions with a cleft wh-phrase in the left periphery of the embedded sentence (see (7)). This strategy has been found in previous research conducted in French (see the references mentioned above) and is considered to be grammatical (according to our French informants).

(7) a. Tu dis que c’est oùi que Marie va (oùi)? you say that it is where that Marie goes?

b. Tu penses que c’est quii qui (quii) lit des histoires? you think that it is who that reads stories

We now turn to Dutch, a V2 language. In standard Dutch LD wh-questions, the wh-word is in the left periphery of the matrix clause and there is V-to-C movement (see (8)).

(8) a. Waari zegj je (zegj) (waari) dat Marie (waari) heengaat? where say you that Marie goes

‘Where do you say that Marie is going?’

Language and Speech

C. Jakubowicz, N. Strik 105

b. Wiei denkj je (denkj) (wiei) dat (wiei) verhalen leest? who think you that stories reads

‘Who do you think is reading stories?’

Furthermore, according to recent studies (Schippers, 2006a,b, see also van Kampen, 1997 and van Kampen & Evers, 1995), PM and wh-copying questions (see (9) and (10) respectively) are observed in several Dutch dialects and in colloquial Dutch.

(9) Wat zegj je (zegj) waari Marie (waari) heengaat?what say you where Marie goes

(10) Wiei denkj je (denkj) wiei (wiei) verhalen leest?who think you who histories reads

A comparison between French and Dutch interrogatives shows that the variety of LD strategies is bigger in French than in Dutch. The differences between the two languages are reported in Table 1.

Table 1Main typological differences between French and Dutch in LD wh-questions*

French Dutch

V-to-C optional obligatory

wh-in situ yes no

wh-cleft yes no

PM no yes

wh-Copy no yes

* Recall that PM and wh-copying are possible only in particular dialectal varieties and registers of Dutch (see above). For wh in situ in French see above and note (17).

3The Derivational Complexity hypothesis4

The idea that language development is constrained by economy considerations (in the sense of Chomsky, 1995, and further work) has been in the literature since the middle of the nineties. Specifically it was considered that movement is costly and that young children avoid it unless it is obligatorily required. For example, van Kampen (1997, see also van Kampen & Evers, 1995) suggested that children first master constructions that require less movement and more generally, that in contrast to adults who prefer a more dense information packaging in PF, children prefer a PF that is closer to LF. For Hulk and Zuckerman (2000) and Zuckerman (2001), if the language they are acquiring allows different options, young children will initially use the more economical one,

4 This section was entirely written by Celia Jakubowicz, who assumes full responsibility for the ideas developed in it (see Jakubowicz, 2005, to appear).

Language and Speech

106 Acquisition of LD wh-questions in French and Dutch

that is, the option that involves less movement (see also Hamann (2006) and Rizzi (2000) among others). The idea that children prefer more economical derivations was also proposed by Jakubowicz (2002, 2003), Jakubowicz and Nash (2001), and Soares (2003, 2006), who claimed that young children prefer Merge over Move. Adopting Chomsky’s (2001) idea that the computation proceeds through the iteration of Merge (External Merge and Internal Merge)5 and that Internal Merge (or Movement) is driven by uninterpretable feature checking, Jakubowicz (2004, 2005) proposed that (ab)normal language build up is affected by developmental constraints such as the capacity of working memory, that are sensitive to the computational complexity of the derivation.6 Furthermore, this author suggested that computational complexity can be estimated by a metric where complexity is precisely defined. Her general hypothesis, the Derivational Complexity hypothesis (henceforth DCH) that we adopt here claims that during language development of typically developing children and children with SLI, less complex derivations are input convergent (i.e., correctly spelled out and “pronounced” at the interfaces) before more complex ones. Complexity is measured by the Derivational Complexity Metric (henceforth DCM) which includes the following two clauses:

(11) Derivational Complexity Metric (Jakubowicz, 2005)

A. Merging αi n times gives rise to a less complex derivation than merging αi (n + 1) times.

B. Internal Merge of α gives rise to a less complex derivation than Internal Merge of α+β.

The DCM claims that in deriving a wh-question, the child is sensitive to the number of times that a copy of the wh-element must be merged to satisfy a compu-tational requirement and to the number of constituents that may (or must) undergo Internal Merge. If as proposed by the DCH, children start with less complex deriva-tions, according to clause A of the DCM, all other things being equal, wh-questions involving no IM of the wh-element, or a smaller number of IM will emerge before. Furthermore, and according to clause B of the DCM, children will initially prefer structures where only one constituent (the wh-element)7 undergoes IM in the left periphery of the clause.

5 According to Chomsky’s (2001) copy theory of movement, the movement operation involves successive applications of Merge. See Jakubowicz (2004) for a former formulation of derivational complexity in terms of Merge and Move.

6 This idea is based on Chomsky’s (2005a) proposal according to which three factors enter into the development of language, namely the genetic endowment (or UG), experience and developmental principles not specific to the Language Faculty. As remarked by one of the reviewers, the idea that the capacity of working memory might be one of these principles can be related to former proposals that in order to alleviate processing, children may use simpler constructions instead of more complex ones (Bloom, 1990; Brown, 1973; Brown & Hanlon, 1970, among others).

7 Unless a second constituent is obligatorily required by typological properties of the target language, for example V2 in Dutch.

Language and Speech

C. Jakubowicz, N. Strik 107

The DCH, which is couched in the Minimalist framework, can be thought of as a modern version of the “Derivational Theory of Complexity” of the 60’s.8 As this theory does, the DCH correlates representational and computational complexity with behavioral, dependent variables. It applies to different conditions of language acquisition (L1, L2, SLI, etc.) and to adult processing as well. Focusing on the former, as pointed out by one of the reviewers, the DCH predicts stages in children’s acquisition of long distance questions, with the less complex UG-compatible structures being called up before target consistent fronted wh-questions.9

What does the DCH predict for French and Dutch? Let us first consider French. The examples presented in the previous sections show that in wh-in-situ construc-tions, the wh-word is spelled out / pronounced in the domain of its phase (e.g., vP for an object question) while in cleft, SPM and PM structures it is spelled out in the left periphery of the embedded clause. In contrast, in target-consistent LD wh-fronted questions, the wh-word is spelled out in the left periphery of the matrix clause, while in LD copying constructions it is spelled out twice: once in the left periphery of the matrix and once in that of the embedded clause.

Although a precise syntactic analysis of all these strategies is beyond the scope of this paper, for concreteness we will here adopt the following assumptions: First, along the lines of Newmeyer (2003) and Chomsky (2005b), we will assume that the wh-phrase has an interpretable interrogative feature [iWh] and that C, in the left periphery has a uninterpretable wh-feature [uWh] with or without the EPP property.10 If [uWh] has the EPP property, the wh-word reaches the right position in the left periphery by internal merge for interpretation. If [uWh] does not have the EPP property, the wh-phrase is associated with such a position by some other operation, a choice function in the sense of Reinhart (1998) or by the Agree operation as defined in Chomsky (2001). Adopting these ideas we will here consider that in the case of wh-in-situ, the [uWh] feature of the matrix C has not the EPP property; then the uninterpretable feature in C is checked by the corresponding interpretable feature of the wh-word by the means of long distance Agree.11,12 As a consequence of this hypothesis and focusing on what is intrinsic to wh- constructions, it follows that French wh- in-situ involves only EM.13

8 The “Derivational Theory of Complexity” (the DTC), built on Chomsky’s (1957) model, was at the source of an experimental program which tested the hypothesis that the number of opera-tions that the grammar uses to generate a sentence correlates with behavioral data such as the reaction times of speakers processing the sentence in some psycholinguistic task. (see Berwick & Weinberg, 1984; Marantz, 2005, and references in there).

9 Thanks to Rosalind Thornton for raising this point in her review of a previous version of this paper.

10 The EPP property attracts the wh-phrase to the edge of C. Note that Chomsky (2005b) ascribes this function to the edge-feature EF that is automatically available for a lexical item (LI).

11 Alternatively, matrix scope of the wh-word can be established through the choice function proposed by Reinhart (op. cit).

12 Note that for wh-in-situ an abstract Q-feature operator in CP was argued for by Cheng and Rooryck (2000), Mathieu (2004), and Hamann (2006).

13 A possible variant of the in situ analysis may involve a shorter movement to the edge of vP, with

Language and Speech

108 Acquisition of LD wh-questions in French and Dutch

(12) [CP [uWh] [TP tu penses [C que [TP Marie aime [vP (Marie) (aime) qui?]]]]]14

you think that Mary loves who

‘Who do you think that Mary loves?’

In contrast, in target consistent LD questions where the wh-element is spelled out in the left periphery of the matrix clause, the wh-word undergoes IM: the [uWh] feature of the matrix C has the EPP property, and the wh-word moves to this position to check this feature, leaving an unpronounced copy in its base position and in the left periphery of each embedded clause.15 Furthermore, we adopt the idea that C has a Tense feature [T] which establishes a relation between the CP domain and the TP domain (Evers, 1981; Gueron & Hoekstra, 1988; Pesetsky & Torrego, 2001). Following Pesetsky and Torrego (2001) and Platzack (1998) we also assume that the matrix C has an uninterpretable tense feature [uT] and we propose that in French [uT] may or may not have a EPP feature. If the [uT] feature in C does not have the EPP property, [uT] is checked by the [iT] feature of T by the operation Agree (see (13)). If [uT] in C has the EPP feature, [uT] is checked by IM of the Verb in C (see (14)).16

(13) [CP Qui [uWh] [uT] [TP tu penses [CP (qui) que [TP Marie aime [vP (Marie) (aime) (qui)?]]]]]

Who you think that Mary loves

(14) [CP Qui penses [uWh] [uT] [TP tu (penses) [CP (qui) que [TP Marie aime [vP (Marie) (aime) (qui)?]]]]]

Who do you think that Mary loves?

As for the fourth target consistent long distance wh-strategy, namely questions with a cleft in the left periphery of the embedded clause as in (15), for concreteness, we will provisionally adopt — and adapt to French wh-clefts — the analysis of canonical clefts proposed by Soares (2006) for European Portuguese.

a subsequent Agree relation established with a feature in the C system. See Belletti (2006) and Kato (2003) for an analysis concerning the position of the wh-word in wh-in-situ questions.

14 Checking and deletion of uninterpretable features followed by IM of the goal are indicated by a line crossing the uninterpretable feature(s). Checking through Agreement is moreover indicated by a line between the elements entering into the Agree relation. Copies of moved constituents are in brackets.

15 The presence of intermediate copies in the left periphery of each embedded clause reflects succes-sive cyclic movement and is supported by a large body of empirical evidence from a variety of sources. Intermediate copies are required by locality constraints like the subjacency condition in the Principles and Parameters theory. However, their presence is problematic for the Minimalist Program, where all movement must be triggered by a matching feature. Given that in sentences like (12) the embedded C is [-wh], it is not obvious how movement of the [iwh] element qui to the [-wh] subordinated C can be forced. This problem is beyond the scope of this paper.

16 The option of attracting T must be contingent on the specification [+wh] of C: otherwise, a declarative C could also trigger T to C. Thanks to Luigi Rizzi for this observation.

Language and Speech

C. Jakubowicz, N. Strik 109

(15) a. Tu penses que c’est qui [que Marie aime (qui)]? You think that it is who that Mary loves?

‘Do you think that it is who that Mary loves?’

Specifically we assume that the C of a cleft question has a [uWh] feature with the EPP property. This probe attracts the wh-word (qui) endowed with its [iWh] feature to the edge of C, where it is internally merged. The verb être, (be) the exple-tive pronoun ce, (it) and then the matrix clause are merged to the cleft clause. In the preceding section we have seen that although the clefted wh-constituent is spelled out in the left periphery of the cleft clause, it has matrix scope. Given this fact, we will here consider that the matrix C has a [uWh] feature with no EPP property and that this feature is checked by the clefted wh-word through long distance Agree, as in the case of wh-in-situ.

(16) [CP [uWh] [TP tu penses [CP que c’est [CP qui [uWh] [iT] que [TP Marie aime [vP

(Marie) (aime) (qui)?]]]]]]

you think that it is who that Mary loves

Returning to the DCH, if as claimed by the DCM, in deriving a wh-question the child is sensitive to the number of times that a copy of the wh-word must be merged to satisfy a computational requirement and to the number of constituents that may (or must) undergo Internal Merge, among the available target consistent strategies to build a long distance question in French, we expect, (i) that those involving no IM of the wh-word (but see note (13), or a smaller number of IM will emerge first and (ii) that children will initially prefer structures where only one constituent (the wh-word) undergoes IM in the left periphery of the clause. In other words, to the extent that the strategy of leaving the wh-word in-situ is target consistent,17 we expect wh-in-situ long distance questions like (11) to emerge before long distance cleft questions such as (16), and wh-fronted long distance questions without V-to-C as (13) to emerge before wh-fronted long distance questions with V-to-C as (14).

Furthermore, as mentioned in the introduction, besides target-consistent wh-ques-tions, in previous research typically developing French-speaking children and children with SLI optionally used scope-marking strategies that are not available in their input such as wh-copying, PM and SPM constructions. Consider the SPM question in (17).

17 As mentioned in the typology section above, wh-in-situ in LD appears to be infrequent and may be restricted to certain dialects or registers of French. This state of affairs raises important ques-tions regarding parameter setting: (1) Are these constructions available in the input to children generally or only for some dialects? (2) In the case of nonavailability, will children consider that movement is obligatory in LD questions, regardless of the fact that it is optional in root questions? These issues await more complete understanding of this construction (thanks to Laurie Tuller for bringing these issues to our attention). On the assumption that order parameters are very precociously fixed (Jakubowicz, 2007; Pierce, 1992; Rizzi, 2007, among others), if wh in-situ is not target-consistent (or is not in the child’s input), we expect that typically developing children aged three years or less having already fixed the wh-position(s) allowed in their language should not produce wh-in-situ long distance questions.

Language and Speech

110 Acquisition of LD wh-questions in French and Dutch

(17) [CP [uWh] [TP tu penses [CP qui [uWh] [iT] que [TP Marie aime [vP (Marie)

(aime) (qui)?]]]]]

you think that it is who that Mary loves

Based on the prosody of these questions and following a suggestion by A. Belletti (p.c.; see also Belletti, 2006) we will assume that in French a SPM question is in fact a reduced cleft.18 If such is the case, SPM questions are derivationally less complex than long distance cleft questions; according to the DCH they should emerge before clefts but also before PM questions and wh-copying. Consider the PM question in (18) where besides the wh-word in the left periphery of the embedded clause, there is a wh-word in the left periphery of the matrix.

(18) [CP Qu’est-ce que [TP tu penses [CP qui [TP Marie aime [vP (Marie) (aime) (qui)?]]]]]

What you think who Mary loves?

Two main types of analysis have been given to PM questions: the direct depen-dency approach and the indirect dependency approach19 (Lutz, Mueller, & von Stechow, 2000). According to the first approach, the wh-word in the left edge of the matrix clause qu’est-ce que (what) is an expletive that marks the scope position of the wh-phrase in the embedded C qui (who); a chain is formed between the position occupied by the scope marker and the one occupied by the wh-phrase at some level of representation (see Bayer, 1996; Cheng, 1997; McDaniel, 1989; van Riemsdijk, 1982, among others). According to the indirect dependency approach (Dayal, 1994, 2000), the wh-scope marker is not an expletive; it is an ordinary wh-argument that quantifies over propositions rather than over individuals. On this view, the subordinated CP is syntactically an appositive clause adjoined to the matrix and is a question over individuals. As remarked by one of the reviewers, two wh-chains are assumed on this view, one for the embedded clause and one for the matrix clause.

Focusing on the way in which the DCM is phrased, we will here adopt the direct dependency approach which is the classical one and yields a clear developmental prediction.20 Indeed, the direct dependency approach involves IM of the wh-word to the left periphery of the subordinate clause but EM of the expletive in the left edge of the matrix CP, besides coindexation between this one and the wh-word at the left edge of the subordinate clause.21

18 We leave open the question of whether the reduction takes place before or after spell-out.

19 See Lutz, Mueller, & von Stechow, (2000) for a third type of analysis called the mixed approach.

20 See Oiry and Dermidache (2006) who adopt Dayal’s approach not only for French PM questions but also for wh-copying. As remarked by one of the reviewers, Jacqueline van Kampen, Dayal’s analysis is inadequate for wh-copying questions with verbs that do not select a [+wh] CP (Où tu penses où Marie va?, ‘Where do you think where Marie is going?’).

21 In contrast the indirect dependency approach à la Dayal, involves IM of the wh-word to the left periphery of each clause plus coindexation between the wh-word of the matrix and the embedded CP. Thus, the indirect dependency approach gives rise to more IM operations than

Language and Speech

C. Jakubowicz, N. Strik 111

Now, if we compare PM and the SPM strategy from the DCH perspective, given that the latter involves less derivational complexity than the former, it should emerge first. Indeed, as it can be seen in (17) above, in the SPM strategy there is no wh-word in the left periphery of the matrix clause. In contrast there is one in the PM construction that undergoes EM (according to the direct dependency approach).

Finally consider the wh-copying construction in (19):

(19) [CP Qui [uWh] [uT] [TP tu penses [CP qui [TP Marie aime [vP (Marie) (aime) (qui)?]]]]]

who you think who Mary loves

As mentioned in the introduction, the wh-copying strategy of asking a long distance question is used by typically developing children in a variety of languages and by children with SLI. This strategy is also attested in the adult speech of a number of languages including present-day German, Frisian, Dutch, Afrikaans, and Passamaquoddy (See Felser, 2004, and references in there, Bruening, 2006; Schippers, 2006a, 2006b, among others). Several arguments have been given to establish that wh-coping and PM are different strategies of asking long distance questions and we will adopt this idea here (see Bruening, op. cit., and Felser, op. cit., among others). Specifically we will assume that the wh-copy strategy exemplified in (19) is essentially identical to the long distance movement chain in (13), except that two copies of the wh-phrase are pronounced, as proposed by Fanselow and Cavar (2001), Fanselow and Mahajan (2000), and Felser (op. cit.).

In what follows we will ignore the question of why it is only the copies in Spec CP that are pronounced (and not those in argument position) and why this double pronunciation is allowed in some languages and not in others.22 To explain the use of the wh-copying strategy in child language, we will adopt certain considerations by Chomsky (2005a) concerning certain differences between the semantic and the phonetic interfaces. Chomsky suggests that in the mapping to the semantic inter-face there is processing of each copy, that these operations are universal, “hence instantaneous and costless” (Chomsky, 2005a, p.12). At the phonetic interface, in contrast, two desiderata enter in conflict: (i) ease of processing and (ii) minimization of computation. Processing will be eased if all copies remain. But minimization of computation requires deletion of all but one copy, “so the phonological component can forget all but one copy.” Unless certain special conditions require some residuals of the lowest copies to be pronounced (see Chomsky op. cit. and references therein),

a derivation based on the direct dependency approach. Moreover, as remarked by Luigi Rizzi (p.c.), at this point it is not clear why, under the indirect dependency approach, PM should involve less derivational complexity, hence be favored over long distance movement. In other words, while in the direct dependency approach the conclusion holds (one application of IM in PM vs. two in long distance movement), in the indirect dependency approach it is far from obvious: There are two IM of the two different wh elements, so clause B of DCH would seem to favor long wh movement (involving IM of a single element), and here a tension would arise between clauses A and B of the DCM, a case with which we have not explicitly dealt with here.

22 As for adult language, Felser (2004) suggests that languages that have split phrases like the “was-für” split in German allow the wh-copying strategy. We will not discuss this suggestion here.

Language and Speech

112 Acquisition of LD wh-questions in French and Dutch

generally only one copy per wh-question is to be phonetically realized. This fact takes Chomsky to the conclusion that language is designed for minimizing computation, with utility for communication being only a secondary factor.

Coming back to the strategy of wh-copying in child language, and based on the ideas just exposed, we will speculate that the wh-copying strategy for asking long-distance questions may help keep the wh-word alive in working memory, particularly in cases where the number of phases that the wh-word needs to pass through on its way to the left edge of the matrix CP exceeds the limits of processing resources / working memory capacity. Assuming that this capacity grows over time and is therefore more limited in young children and children with SLI (Jakubowicz, 2005, to appear), we expect that the wh-copying strategy should emerge almost at the same time as target consistent LD wh-fronted questions (as those in (13)), because, as we have seen, they involve the same derivation (as they are both LD questions).

Finally note that to the extent that, in languages where it is not legitimate, the wh-copying strategy is mainly used by young children (or children who have limited resources of working memory) in the conditions just described, this fact does not contradict Chomsky’s conclusion that language is designed for minimizing computation.

To summarize: Previous research on French has shown that young typically developing children and children with SLI may use six or seven different UG compat-ible strategies for asking LD questions: target consistent LD wh-fronted questions with or without V-to-C, cleft questions, wh-in-situ questions, PM, SPM and wh-copying questions. To evaluate the complexity involved in each of these constructions, the DCM was applied to the syntactic analysis of these questions proposed above. On this basis, the DCH allows us to predict the following order of emergence of LD questions in French: (i) LD wh-in-situ questions (if target-consistent), (ii) SPM questions, (iii) PM questions, (iv) cleft questions, (v) wh-copying and wh-fronted questions without V-to-C, and (vi) wh-fronted questions with V-to-C.

Turning now to Dutch, in the typological section we have seen that Dutch is a V2 language and that there are three different ways of asking a LD question: standard wh-fronted questions, PM, and wh-copying. In standard LD questions, the wh-word is spelled out at the left periphery of the matrix CP; in PM questions, the wh-word is spelled out in the left periphery of the embedded clause and a default wh-word (wat ‘what’) appears in the left periphery of the matrix CP; finally, in wh-copying, the wh-word is spelled out twice — once in the left periphery of the matrix clause and once in that of the embedded clause. As already noted, in these three strategies for asking a wh-question, V-to-C is obligatory.

For concreteness, we will adopt here for Dutch the same syntactic assumptions we proposed for French.23 Namely, we will consider that the derivation of standard LD wh-questions in Dutch can be schematized as the counterpart of French fronted LD questions with V-to-C, as in (20).

23 It is clear that a more in-depth analysis is necessary for these constructions, and more generally, for wh-movement in Dutch. See Strik (in progress).

Language and Speech

C. Jakubowicz, N. Strik 113

(20) [CP Wie denk [uWh] [uT] [TP je (denk) [CP (wie) dat [TP Marie kust [vP (Marie) (wie) (kust)]]]]]

‘Who do you think that Mary kisses?’

For PM constructions, presented in (21), as in French, we will consider that this strategy is best analyzed by the direct dependency approach (see above and note (21)).

(21) [CP Wati denk [TP je (denk) [CPi wie [uWh] [iT] [TP Marie kust [vP (Marie)(wie) (kust)?]]]]]

‘What do you think who Mary kisses?’

Now, if we compare PM and standard wh-fronted constructions from the DCH perspective, given that PM involves less derivational complexity, it should emerge first.

Finally consider the wh-copying construction exemplified in (22).

(22) [CP Wie denk [uWh] [uT] [TP je (denk) [CP wie [uWh] [iT] [TP Marie kust [vP

(Marie) (wie) (kust)?]]]]]

‘Who do you think who Mary kisses?’

As we did for French we will assume that this strategy is essentially identical to the long distance movement chain in (20) except as two copies of the wh-phrase are pronounced. For French we proposed that the wh-copying strategy may favor processing particularly in the presence of limited resources of working memory. In Dutch, the use of this strategy by children may vary according to weather wh-copying is or not present in their input. Recall that according to recent studies (see the Typology section) wh-copying (and also PM) is observed in several Dutch dialects and in colloquial Dutch. If the strategy is available in the child’s input, its use by children may be determined by the same reasons it is for adults (see note 22). If the strategy is not available in the input, it may as in French help keep the wh word alive in working memory in the same cases as in French, that is, when the number of phases that the wh word needs to pass through on its way to the left edge of the matrix CP exceeds the limits of processing resources / working memory capacity. As remarked by Luigi Rizzi (p.c.), it is likely that in these cases the pronunciation of the trace helps an immature production system “refresh” the element involved in a long distance dependency.

Returning to the DCH, recall that clauses A and B of the DCM say that wh questions involving a smaller number of IM or a smaller number of constituents that undergo IM in the left periphery of the matrix CP emerge first because they are derivationally less complex. Based on the analysis proposed above we expect the following order of acquisition of LD questions in Dutch: PM questions, w-copying and wh fronted questions. Let us now turn to the Methodology and Results sections.

4MethodologyTo test our predictions we used an elicited production task of root and LD wh-ques-tions, inspired by the work of Thornton (1990) and the protocols used in our former work (see references mentioned in the introduction). Two important differences with

Language and Speech

114 Acquisition of LD wh-questions in French and Dutch

respect to our former protocol were introduced: (i) the use of the verb say (dire in French and zeggen in Dutch) in the matrix clause, and (ii) a dissociation between the characters who answer a root wh-question and the character who the child asks to report what the first ones said.24 Thus, the task consists of asking questions to Nina, a robot puppet on a computer screen, in the context of a story. When she wants to answer a (root) question, Nina is often interrupted by Billy, her little brother, and Lala, her little sister, but these two (partially) speak in their own robot language which is incomprehensible for humans. Nina is naturally able to understand what her little brother and sister say in their robot language. So the investigator invites the child to ask Nina to help them and to tell what her little brother and sister said. The fact that the child has to ask a character a question about two other characters is expected to facilitate the use of LD wh-questions. The experimental protocol includes four types of LD questions: object, subject, adjunct where and adjunct why questions. Given that there are six items per condition a total number of 24 LD wh-questions is elicited from each participant. Below we present an example of each type of test item as it is introduced by the investigator, in both French and Dutch:

(23) a. Object LD question

French: Demande à Nina ce que Lala a dit que Lapin boit. Dutch: Vraag aan Nina wat Lala zei dat Haas drinkt.

‘Ask Nina what Lala said that Rabbit is drinking.’

b. Subject LD question

French: Demande à Nina qui est-ce que Lala a dit qui boit de l’eau.25

Dutch: Vraag aan Nina wie Lala zei die water drinkt.26

‘Ask Nina who Lala said is drinking water.’

c. Adjunct where LD question

French: Demande à Nina où Billy a dit que le poisson nage. Dutch: Vraag aan Nina waar Billy zei dat de vis zwemt.

‘Ask Nina where Billy said that the fish is swimming.’

d. Adjunct why LD question

French: Demande à Nina pourquoi Billy a dit que Grenouille se promène dans la forêt.

24 Thanks to Alec Marantz for suggesting that such a dissociation may improve LD productivity by the children and to Claire Beyssade, Paul Egré and Léa Nash for helpful comments.

25 In principle, in standard French, a LD subject question should be Demande a Nina qui Lala a dit qui boit de l’eau. We used the construction with ‘qui est-ce que’ after discussing with French linguists who did not prefer / accept LD subject questions with ‘qui’. This choice was not without consequences for the children (see the Results section).

26 Note that, in Dutch, subject questions were elicited with die (also used in relative constructions) instead of the standard complementizer dat. According to Schippers (2006a) a.o. die is a dialectal variant of dat with subject extraction and it is the form used in the dialect spoken by the Dutch investigator.

Language and Speech

C. Jakubowicz, N. Strik 115

Dutch: Vraag aan Nina waarom Billy zei dat Kikker in het bos wandelt.

‘Ask Nina why Billy said that Frog is walking in the forest.’

Both in French and in Dutch, two groups of 12 children each and one adult control group participated in the task. The participants were all monolingual French or Dutch native speakers. The age range, mean age and Standard Deviation (SD) of each group are reported in Table 2.

Table 2Number and age of French and Dutch participants*

Participants n Age Mean Age SD

4N-Fr 12 4;0-4;7 4;4 2

4N-D 12 4;2-4;11 4;7 3

6N-Fr 12 6;5-6;9 6;6 1.4

6N-D 12 6;7-6;11 6;9 1.6

Ad-Fr 12 21-34 27;6 3.6

Ad-D 12 24-29 26;6 1.3

* The ‘N’ in 4N and 6N refers to ‘normal’, given that the children of our study are “normally or typically developing.”

5Results

Table 3 reports the mean number of LD wh-questions and SDs produced in each group in each of the languages. As can be seen, children and adults produce a relatively high number of LD wh-questions, and, in both languages, although especially in Dutch, 4-year-olds perform less well than the older children. Moreover, in French, as well as in Dutch, LD wh-questions include target-consistent and nontarget responses which we discuss in the following sections starting with French.

Table 3Mean number of LD wh-questions and SDs collapsed across wh-and response type as a function of language and age*

French Dutch

Participants N /288 Mean number SD N /288 Mean number SD

Adults 263 21.9 2.9 232 19.3 5.2

6N 254 21.2 2.5 247 20.6 3.6

4N 239 19.9 4.3 189 15.8 5.1

* ‘N/288’ indicates the total number of test-items presented to the participants of each age-group. The numbers below in the same column report the total number of LD wh-questions out of 288, produced in each group.

Language and Speech

116 Acquisition of LD wh-questions in French and Dutch

Unexpected responses, namely non-LD wh-questions, produced instead of LD ones were principally of two types: (a) simple root questions as in (i):

(i) Investigator: Demande à Nina qui est-ce que Lala a dit qui boit de l’eau. ‘Ask Nina who it is that Lala said drank water’

Subject: Qui boit de l’eau? ‘Who is drinking water?’— and (b) root questions with a parenthetical expression, as in (ii):

(ii) Subject : D’après Lala, qui boit de l’eau?

‘According to Lala, who is drinking water?’

While the former were produced by the children, the latter were principally produced by the adults (specifically, 8 questions of type (ii) were observed in the adult group in French and 20 in Dutch).

5.1 French

Figure 1 shows the mean number of LD questions produced in each age group (collapsed across the four types of wh-words). Three response types are represented in this figure: (i) target-consistent wh-fronted LD questions (‘LD wh-fronted’), (ii) wh-in-situ LD questions (‘LD wh-In-situ’), and (iii) other LD questions (‘LD wh “Other”’). This last category includes different types of LD scope-marking strategies: the wh-cleft construction, PM, SPM and wh-copying (see below for a more detailed characterization of these strategies).

Figure 1

Mean number of different types of LD questions (collapsed across wh-type) in French

We can see that in the three groups wh-fronted LD questions are the most frequent. In contrast, wh-in-situ is almost inexistent in the child groups and rare in the adult group. LD wh “Other” responses are produced more frequently by 4N and 6N children than by adults. However, they are much less numerous than target-consistent wh-fronted LD questions.

Language and Speech

C. Jakubowicz, N. Strik 117

Figures 2a, b, and c report the distribution of the three classes of responses across the four types of questions we elicited: object (‘LDO’), subject (‘LDS’), adjunct where (‘LDA-où’) and adjunct why (‘LDA-pourquoi’) LD questions.

Figure 2a, b, c Mean number of LD wh-Fronted, LD wh-In-situ and LD wh “Other” responses as a func-tion of wh-type for each age group in French

Language and Speech

118 Acquisition of LD wh-questions in French and Dutch

Consider first the LD wh-fronted questions in Figures 2a, b, and c. A common pattern shows up, namely object questions are among the most frequently produced questions in the three groups, whereas subject questions are among the least frequently produced ones, particularly in the adult and the 6N groups. Furthermore note that the category LD wh-fronted questions includes different constructions: questions with and without V-to-C movement and also questions with est-ce que ‘is it that’ or c’est quoi que ‘it’s what that’ in the left periphery of the matrix clause. Examples of these different constructions for an adjunct where LD question are given below.27

(24) a. Où a dit Lala que le poisson nage? where has said Lala that the fish swims

b. Où Lala a dit que le poisson nage? where Lala has said that the fish swims

c. Où est-ce que Lala a dit que le poisson nage? where is it that Lala has said that the fish swims

d. C’est où que Lala a dit que le poisson nage? It is where that Lala said that the fish is swimming?

‘Where did Lala say that the fish is swimming?’

Questions with est-ce que, like in (24c), were mostly produced in the case of object questions (qu’est-ce que ‘what is it that’). For the other types of wh-questions, we mostly observed fronting without est-ce que, as in (24b). Importantly, V-to-C movement was attested only in adults, and amounted to only 17% of the LD questions produced in this group.

As already pointed out, wh-in-situ is very rare. As can be seen in Figures 2a and 2b, this strategy is relatively more frequent for the adjunct where questions in the adult group (see (25)) than for the other wh-types.

(25) Lala a dit que le poisson nage où?Lala has said that the fish swims where

‘Where did Lala say that the fish is swimming?’

Consider now the category LD wh “Other” questions in Figures 2a, b, and c. In contrast to wh-fronted questions, children (4N and 6N) produce more of this class of responses for subject questions and where questions than for object questions. Let us have a closer look in Table 4 at the different strategies included in this response class.

As can be seen, PM questions are generally more frequent than the remaining classes of responses of the category LD wh “Other” produced in each group. PM questions

27 As for object LD questions, note that contrary to other French wh-words, V-to-C movement is obligatory for que.

(i) a. Qu’a dit Billy que Grenouille mange? what has said Billy that Frog eats

b. *Que Billy a dit que Grenouille mange? what Billy has said that Frog eats

Language and Speech

C. Jakubowicz, N. Strik 119

are more frequently produced by children than by adults. Although this nontarget construction is more productive for subject questions (see (26a)), it also appears in the case of adjunct where and adjunct why questions (see (26b) and (26c) respectively).

(26) a. Qu’est-ce que Billy a dit qui boit de l’eau? 28

what is it that Billy has said who drinks water

‘Who did Billy say is drinking water?’

b. Qu’est-ce que Lala a dit où le poisson nage? what is it that Lala has said where the fish swims

‘Where did Lala say that the fish is swimming?’

c. Qu’est-ce que Lala a dit pourquoi Grenouille part? what is it that Lala has said why Frog leaves

‘Why did Lala say that Frog is leaving?’

Table 4Frequency (and percentages) of different types of LD wh “Other” responses in French*

Partic. Type of Questions wh-Cleft SPM PM wh-Copying

Ad

LDO /69 2 (3%) -

LDS /60 6 (10%) 6 (10%)

LDA-où /71 1 (1%) 3 (4%)

LDA-pourquoi /63 1 (2%) 2 (3%) 3 (5%)

6N

LDO /60 3 (5%) -

LDS /61 3 (5%) 2 (3%) 20 (33%)

LDA-où /71 4 (6%) 8 (11%)

LDA-pourquoi /62 7 (11%)

4N

LDO /64 2 (3%) - 4 (6%)

LDS /60 13 (22%) 1 (2%)

LDA-où /64 6 (9%) 3 (5%) 15 (23%)

LDA-pourquoi /51 1 (2%) 3 (6%) 2 (4%)

Total /756 11 (1%) 22 (3%) 66 (9%) 22 (3%)

* The percentages have been calculated on the basis of the number of expected (LD wh) responses (and not the number of items elicited), provided for each type, in column 2.

28 In the group of adults, 8 out of 12 PM responses have the wh-word que in the left periphery of the matrix clause (see (i)) and the remaining four have qu’est-ce que (see (17)). In contrast, in the child groups, all the PM responses are constructed with qu’est-ce que.

(i) Qu’a dit Billy qui boit de l’eau? what Billy has said who drinks water

‘Who did Billy say is drinking water?’

Language and Speech

120 Acquisition of LD wh-questions in French and Dutch

The SPM strategy is produced by the children and very infrequently by the adults, and it is relatively more frequent for adjunct where questions (see (27)).

(27) Lala a dit où que le poisson nage?Lala has said where that the fish swims

‘Where did Lala say that the fish is swimming?’

Wh-copying appears in the 4N group only and similarly to SPM, it is more frequent for adjunct where (see (28)) than for the remaining types of LD questions. Recall that in contrast to the wh-Cleft, SPM and PM strategies, in wh-copying, the wh-word reaches the left periphery of the matrix clause. Given that this strategy is not target-consistent, it seemed more adequate to us to distinguish it from the target-consistent LD wh-Fronted responses. However, if we consider wh-copying and wh-fronted responses together, the 4-year-olds perform like the 6-year-olds with respect to adjunct where questions. That is they produce more LD wh-Fronted responses for object, adjunct where and adjunct why questions than for subject questions.29

(28) Où Lala a dit où le poisson nage?where Lala has said where the fish swims

‘Where did Lala say that the fish is swimming?’

As for object questions with qu’est-ce que as in (29), it is difficult to determine whether this type of response is wh-copying question or PM. Although this response type is in fact ambiguous, we (arbitrarily) counted the four attested examples as wh-copying questions.

(29) Qu’est-ce que Billy a dit qu’est-ce que Grenouille mange?30

what is it that Billy has said what is it that Frog eats

‘What did Billy say Frog eats?’

29 For the 4-year-olds, adding wh-copying to the LD wh-Fronted category raises the mean number of the entire category from 3.3 (SD 2) to 4.3 (SD 1.7).

30 Although, in standard object LD questions with the wh-word que, the segmental content of the wh-word and the complementizer is the same, sentences like (i) cannot be taken as instances of wh-copying.

(i) Qu’a dit Billy que Grenouille mange? what has said Billy that Frog eats

‘What did Billy say that Frog is eating?’

Similar considerations apply to object LD questions with stylistic inversion in the matrix and embedded clause like (ii).

(ii) Qu’a dit Billy que mange Grenouille? what has said Billy what eats Frog

‘What did Billy say that Frog is eating?’

(Note that sentences of this type are not produced in our study.)

Language and Speech

C. Jakubowicz, N. Strik 121

Finally, note that target-consistent wh-cleft questions appear essentially in the adult group, particularly for subject questions (see (7b), Typology section).

To sum up, French-speaking children and adults produce a high number of target-consistent wh-fronted LD questions. Nontarget wh-scope-marking strategies are much less frequent. They are mostly produced in the case of subject and adjunct where questions and, as expected, they are relatively more frequent in the child groups than in the adult group.

5.2 Dutch

Consider Figure 3, which represents the mean number of LD questions produced in each age group (collapsed across the four types of wh-words). Two response types were found in Dutch: (i) target-consistent wh-fronted LD questions (‘LD wh-Fronted’) and (ii) other LD questions (‘LD wh “Other”’). The LD wh “Other” category contains PM and wh-copying questions.

Figure 3

Mean number of different types of LD questions (collapsed across wh-type) in Dutch

Figure 3 shows that adults produce a large number of wh-fronted questions and few LD wh “Other” questions. On the contrary, both 6- and 4-year-old children produce considerably more LD wh “Other” questions than wh-fronted questions. However, 6-year-old children produce more target-consistent wh-fronted questions than 4-year-old children.

Figures 4a, b, and c report the distribution of the two classes of responses across the four types of questions elicited: object (‘LDO’), subject (‘LDS’), adjunct where (‘LDA-waar’) and adjunct why (‘LDA-waarom’) LD questions.

Language and Speech

122 Acquisition of LD wh-questions in French and Dutch

Figure 4a, b, cMean number of LD wh-Fronted and LD wh “Other” responses per wh-type for each age group in Dutch

Language and Speech

C. Jakubowicz, N. Strik 123

Consider first the wh-fronted questions in Figures 4a, b, and c. For the adults, adjunct where and object questions are more frequent than adjunct why and subject questions. As for the children, wh-fronted questions are remarkably more frequent for the adjunct why questions than for the other wh-types. In all of the wh-fronted LD questions, there is V-to-C movement, as expected given the age of the children.

Consider now the LD wh “Other” responses. These are rare for the adults, but rather frequent for both the 4- and 6-year-old children. Table 6, which show the percentages of PM and wh-copying by group and wh-type, allows us to have a closer look at the LD wh “Other” response type.

Table 5Frequency (and percentage) of different types of LD wh “Other” responses in Dutch

Partic. Type of Questions PM wh-Copying

Ad

LDO /63 - 4 (6%)

LDS /45 3 (7%)

LDA-waar /68 1 (1%) 1 (1%)

LDA-waarom /56 5 (9%)

6N

LDO /65 - 52 (80%)

LDS /55 24 (44%) 6 (11%)

LDA-waar /63 23 (37%) 26 (41%)

LDA-waarom /64 6 (9%) 5 (8%)

4N

LDO /58 - 55 (95%)

LDS /42 12 (29%) 11 (26%)

LDA-waar /48 22 (46%) 19 (40%)

LDA-waarom /41 5 (12%) 2 (5%)

Total /668 101 (15%) 181 (27%)

Table 5 shows that wh-copying is more frequent than PM in Dutch, and that this frequency varies according to wh-type. For the two child groups, in particular, wh-copying appears more frequently for object and where questions than for the other two wh-types. Furthermore, wh-copying is more frequent for object questions than for where questions. However, as in French, wh-copying in object questions (see (30)) could also be analyzed as PM questions. If so, wh-copying would essentially be limited to where questions.

(30) Wat zei Billy wat Kikker eet? what said Billy what Frog eats

‘What did Billy say Frog eats?’

Examples of copying of other wh-types are provided in (31).

Language and Speech

124 Acquisition of LD wh-questions in French and Dutch

(31) a. Waar zei Lala waar de vis zwemt? where said Lala where the fish swims

‘Where did Lala say the fish is swimming?’

b. Wie zei Billy wie water drinkt? who said Billy who water drinks

‘Who did Billy say drinks water?’

c. Waarom zei Lala waarom Kikker weggaat? why said Lala why Frog leaves

‘Why did Lala say that Frog leaves?’

If, like we did for French, we consider the wh-copying and wh-fronted responses together in order to obtain a response type where the wh-word is in the left periphery of the matrix clause, we get a completely different picture. In so doing the number of LD wh-Fronted responses increases considerably and the results come closer to those of the French-speaking children.31

Turning now to PM, this strategy is frequent in subject and adjunct where LD questions (see (32a) and (b), respectively) in both child groups, while it is far less frequent in the adjunct why questions (see (32c)). As is the case for wh-copying, PM is not frequent in the adult group.

(32) a. Wat zei Billy wie water drinkt? what said Billy who water drinks

‘Who did Billy say drinks water?’

b. Wat zei Lala waar de vis zwemt? what said Lala where the fish swims

‘Where did Lala say the fish is swimming?’

c. Wat zei Lala waarom Kikker weggaat? what said Lala why Frog leaves

‘Why did Lala say that Frog leaves?’

Notice that PM appears to be inexistent for object questions in Dutch (as in French) only because these cases were counted as wh-copying (see discussion above).

To summarize, in Dutch, wh-fronted LD questions are frequent in the adult group, while they are not so frequent in the two child groups. For the children, PM and wh-copying questions are more frequent with all wh-types, except adjunct why. As predicted, Dutch children systematically apply V-to-C movement in their wh-fronted and wh-copying questions.

31 For instance, for the 4-year-olds, adding wh-copying to the LD wh-Fronted category raises the mean number of object questions from 0.25 (SD 0.9) to 4.8 (SD 1.5) and the mean number of adjunct where questions from 0.6 (SD 1.4) to 2.2 (SD 2.6). For the 6-year-olds, the mean number of object questions raises from 1.1 (SD 1.7) to 5.4 (SD 1.4) and the mean number of adjunct where questions from 1.2 (SD 1.7) to 3.3 (SD 2.5).

Language and Speech

C. Jakubowicz, N. Strik 125

6Discussion

Three main goals were set in this study:

(i) To see whether in comparison with previous work, a more felicitous context of elicitation increases the number of LD questions produced by children (see the Introduction and Methodology section).

(ii) To systematically compare the acquisition of LD questions in two languages that differ with respect to the position of the wh word and the (inflected) verb in the sentence, and

(iii) To determine whether order of emergence and frequency of different strate-gies used by the children for asking a LD question can be explained by the DCH according to which less complex UG-compatible derivations emerge first.

Regarding Point (i), Table 3 in the Results section shows that the methodology used in this study significantly increased the production of LD questions. This fact is particularly clear for French. While in former elicitation studies the proportion of LD wh questions never passed beyond 50%, in this study percentages reached about 83% and 88% for the 4- and 6- year-old children respectively. Although because of the absence (to our knowledge) of quantified data of this type in earlier Dutch studies, a similar comparison cannot be done, the same table shows that in the child groups the proportion of LD question was quite high: about 65% for the 4-year-olds and about 86% for the 6-year-olds. It could be argued that in French, children’s success was due to the fact that for who, where, and why questions, the child can just imitate the investigator’s utterance in order to produce a correct wh-fronted LD question without V-to-C movement.32 However in the case of French object questions, the child cannot simply repeat part of the investigator’s utterance; nevertheless performance on these questions was not inferior to that of the other question types (see French Examples (23) and Figure 2). Similarly, in Dutch, the idea that the relatively high proportion of LD questions is due to the characteristics of the test item cannot be adequate given that the Dutch participants must apply V-to-C (which is not present in investigator’s utterance; see Dutch examples in (23)).

Consider now point (ii). Although it is clear that since the age of 4, both French and Dutch children are able to produce standard target-consistent LD questions where the wh-word is in the left periphery of the matrix CP, certain differences and a few similarities appear in this comparative French-Dutch study regarding the emergence and frequency of different types of LD wh questions. We will here focus on two series of differences and two series of similarities.

First, in French, adults and children produce a high number of target-consistent wh fronted questions and there is very little frequency variation for this type of response between the three groups. The other two target consistent strategies, clefts, and wh in-situ, are very rare: Clefts (3% of the total of LD questions) and wh in-situ questions (about 10%) appear in the adult group; in the 6-year-olds they do not

32 Thanks to Jacqueline van Kampen for raising this point.

Language and Speech

126 Acquisition of LD wh-questions in French and Dutch

exceed 1% and they are inexistent in the 4-year-olds. As for the nontarget strategies, SPM, PM and wh-copying, these are relatively more frequent in the child groups. However, they are much less frequent than target- consistent wh-fronted questions. And wh-copying appears only in the 4-year-olds group. A completely different picture emerges in Dutch. On the one hand, there is a strong variation between adults and children. Adults mostly produce target-consistent LD wh-fronted questions and very few PM and wh-copying questions. An opposite pattern shows up in the child groups. Furthermore, target consistent wh-fronted questions are more frequent in the 6-year-olds than in the 4-year-olds. Finally, Dutch children produce a large number of PM and wh-copying questions; and, in contrast to French, wh-copying questions also appear in the group of 6-year-olds.

Second, another difference between the French and Dutch children is that the latter systematically use V-to-C movement (as expected on the basis of the fact that V-to-C is very precociously fixed), while French children do not. Only French adults produce questions with V-to-C movement (in about 17% of their wh-fronted ques-tions). Furthermore, while French participants produce clefts and SPM questions, these wh-scope marking strategies are absent in Dutch. Let us now turn to what is common to both languages.

First, in French as well as in Dutch, for the majority of adjunct “why” questions, the wh-word is spelled out in the left periphery of the matrix clause. In French, “why” gives rise to very few SPM, PM, and wh-copying questions, which are nontarget in this language. The same holds for PM and wh-copying questions in Dutch, where these strategies are used in colloquial speech (see Table 5). Furthermore, in the Dutch child groups, the most frequent standard wh-fronted questions are precisely those with adjunct ‘why’.

Second, in both languages PM questions are more frequent than wh-copying questions. At first sight, however, the opposite pattern appears in Table 5 in Dutch,33 namely, percentages are higher for wh-copying than for PM questions. Note however

33 In contrast to Dutch, where the PM strategy is used for subject, where and object questions, in French the PM strategy is particularly frequent for subject questions. This could be due to the way subject questions were elicited, namely using the wh word qui est-ce que (who is it that) (see Example (23b)). This invited a lot of children to repeat the qui est-ce que wh-form in their responses. Actually, many children used qu’est-ce que instead of qui est-ce que. One now obtains a PM question (see (i)) but notice that this question also could be a wh-fronted subject question with a feature error in the wh-word (the default inanimate form qu’est-ce que used instead of the animate form qui est-ce que).

(i) Qu’est-ce que Billy a dit qui boit de l’eau? what is it that Billy has said who drinks water

‘Who did Billy say is drinking water?’

Although we created criteria to determine whether (some of) the presumed PM subject ques-tions were in fact wh-fronted questions with a feature error, for some children it was not so easy to decide if their subject questions were PM or wh-fronted questions and the final choice was rather arbitrary. Possibly, some of the PM questions are actually wh-fronted subject ques-tions with a feature error in the wh-word.

Language and Speech

C. Jakubowicz, N. Strik 127

that object questions (see 30 above), which are not decidable between wh-copying and PM questions, form the majority of these questions (see Table 5). If we do not take into consideration object questions, in French as well as in Dutch, PM is more frequent than wh-copying, and also in both languages in the child groups, wh-copying is more frequent for “where” than for “who” and “why” questions.

Let us now turn to point (iii). Can emergence and frequency of the different types of LD questions considered here be explained by the DCH?

First consider French. If as claimed by clause A of the DCM, in building a wh question the child is sensitive to the number of times that a copy of the wh element must be merged to satisfy a computational requirement, we expect that she / he opts for wh in-situ, SPM, PM, cleft questions and wh-copying in preference to LD structures with wh in the left periphery of the matrix CP without V-to-C. Such preference was of course not observed. In our opinion, however, this does not implicate that the DCH or the DCM are wrong. In fact, as we will see below, some of our data are indeed predicted by the DCM. Coming back to the massive presence of LD wh fronted questions in the French data, we agree with one of the reviewers (R. Thornton) who remarks that other variables may interact with the DCH / DCM. For example, it might be that once processing resources / working memory reaches a certain threshold, the DCH no longer has any observable effects. This could explain the results for the French 4- and 6-year-old children who produce wh-fronted LD questions avoiding, nevertheless V-to-C, that is, limiting to one the number of constituents which undergo IM in the matrix CP (in accordance with clause B of the DCM). In this respect a study with 2- and 3-year-old children might indicate whether or not younger children with limited working memory resources will begin with less complex UG compatible structures, as predicted by the DCH.34 Another variable that may explain the absence of certain types of questions in our data can be their availability in the child’s input. Specifically we think about the availability of LD wh in-situ and LD clefts that may be restricted to certain dialects or registers in French. In this respect we refer the reader to note 17, where we consider possible consequences of this situation regarding parameter setting.

As indicated above, some of our findings are compatible with the DCH. Rizzi (1990) observes that in comparison with other wh operators pourquoi (why) has a special status. For example, pourquoi does not occur in wh in-situ root questions in French (see (33)). Based on this and other facts in French and Italian, Rizzi (op. cit.) proposes that why does not undergo movement but is directly merged in the CP.

(33) a. *Tu as peint la voiture pourquoi? You painted the car why?

a’. Pourquoi tu as peint la voiture?

34 In this respect it is interesting to note that SLI children aged 6 to 13 years with limited working memory resources produce more SPM and PM questions than LD wh-fronted questions, but very few wh in-situ constructions (See Jakubowicz, 2005, to appear). See, moreover, Soares (2006) for a very interesting study of the emergence of left periphery constructions in European Portuguese children aged 1;2 to 4;6 based on the predictions of a slightly modified version of the DCH presented here.

Language and Speech

128 Acquisition of LD wh-questions in French and Dutch

b. Tu as peint la voiture comment? You painted the car how?

b’. Comment tu as peint la voiture?

Coming back to our findings, the fact that French children, and even more frequently, Dutch children produce more LD wh fronted questions with pourquoi and waarom (why) than with the other wh words, gives strong evidence for clause A of the DCM, which predicts better performance for wh questions involving no IM of the wh-element or a smaller number of IMs.

Furthermore, recall that according to clause B of the DCM, children initially prefer structures where only one constituent (the wh-element, but see note 7) undergoes IM to the left periphery of the clause. The fact that no V-to-C was observed in any of the LD wh-fronted questions produced by the French children is consistent with clause B. Although it could be argued that the child’s (or the adult’s) choice of structure reflects register— recall that V-to-C is in fact optional in French — and has nothing to do with the DCH, this argument would not be able to explain why Dutch children produce remarkably fewer standard wh-fronted questions than the French children. Recall that in Dutch V-to-C movement is obligatory. As expected, Dutch children respect this rule and systematically apply it in their wh-fronted LD questions. French children, in contrast, do not need to apply it. This means that Dutch children have an extra syntactic operation to perform in order to form a target-consistent wh-fronted LD question. Less complex derivations or constructions like PM and wh-copying are then called for to avoid complexity.

To sum up, the following facts lend support to the DCM:

(i) in accordance with clause A, “why” questions in both languages are more frequent than the other wh questions considered here;

(ii) in both languages — although marginally in French — PM questions are more frequent than wh-copying questions;

(iii) In both languages wh-copying is more frequent for “where” questions than for the remaining wh questions (it is likely that from its initial merge position in the subordinate clause until the left periphery of the matrix CP, the number of successive cyclic steps is larger for the locative adjunct than for the other wh-words considered in this study);

(iv) In accordance with clause B of the DCM, French children avoid V-to-C. In Dutch, where V-to-C is obligatory, children avoid IM of the wh-constituent, preferring instead PM constructions that involve fewer IM operations, or wh-copying, which facilitates processing to the extent that, for the production system, it refreshes the wh word involved in a long distance dependency.

Finally, we have seen that in contrast to French, no SPM or LD cleft questions are produced in Dutch. Although in French the number of these constructions is very low, their presence might be related to Fanselow’s generalization (2006), according to which (a) if a language tolerates SPM, it also tolerates wh-elements in-situ and allows full movement; and (b) if a construction is grammatical with SPM, it can also be constructed with a wh-word in-situ. Given that Dutch is not a wh in-situ language, to

Language and Speech

C. Jakubowicz, N. Strik 129

the extent that Fanselow’s generalization is adequate, the fact that Dutch participants do not produce SPM and LD cleft questions can be due to the absence of wh in-situ in Dutch.

7Conclusion

Although at first sight, the data reported here seem to give little support for the DCH / DCM, a closer look at the phenomena shows, on the contrary, that more facts support, rather than contradict, the hypothesis that less complex derivations emerge first. Although this is a very encouraging conclusion for the DCH, it is undeniable that several issues raised in this paper await more complete understanding.

First, from an empirical point of view it is important to determine whether the less complex derivations are more frequent in the younger child (in comparison with the 4- and 6-year old children tested here).

Second, we would like to better understand whether and how derivational complexity manifests itself in different acquisition conditions (L1, L2, SLI, etc.) and in adult processing, if it is restricted to production or whether and under which conditions it also concerns the perception and comprehension of wh-questions.

Third, it seems crucial to understand whether the DCM presented here accounts for the development of other left periphery and nonleft periphery structures, or whether it needs to be modified.

Furthermore, other ways (derivational and representational) besides the DCM have been proposed for evaluating complexity (see, e.g., Rizzi, 2000 and Soares, 2006). Are these proposals equivalent? Do they give rise to the same predictions? Are certain of these proposals more accurate than others and if so, why?

Finally, there is the question of the relation between the DCH and adult gram-mars, and whether the DCH applies to adult grammars as well as to child grammars. We want to claim that the DCH can be thought of as a modern version of the “deri-vational theory of complexity” of the 60’s. As in this theory, the DCH correlates representational and computational complexity with behavioral, dependent variables and applies to different conditions of language acquisition and to adult processing as well. Actually, several recent “on-line” studies show that there exists a relation between the length and complexity of the linguistic computation necessary to generate a representation and the length of the time taken by the participant to perform any task involving the representation (see Marantz op. cit., and references there in). Thus, our answer is that the DCH applies to adult’s use of language, but to show how it applies and under which conditions, it is necessary to use on line experimental tasks and materials which are appropriate for adults.

manuscript received: 04. 03. 2006

manuscript accepted: 05. 10. 2007

Language and Speech

130 Acquisition of LD wh-questions in French and Dutch

ReferencesBAYER, J. (1996), Directionality and logical form: On the scope of focusing particles and wh-in-

situ. Dordrecht: Kluwer.BELLETTI, A. (2006). Clefts and wh in-situ: Some notes. Presentation Lisbon COST-Meeting,

Workshop on wh.BERWICK, R., & WEINBERG, A. (1984). Grammars and models of language use. The gram-

matical basis of linguistic performance. Cambridge, MA: MIT Press.BLOOM, P. (1990). Subjectless sentences in child language. Linguistic Inquiry, 21, 491 – 504.BROWN, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.BROWN, R., & HANLON, C. (1970). Derivational complexity and order of acquisition in child

speech. In J. R. Hayes (Ed.), Cognition and the development of language. New York: Wiley.BRUENING, B. (2006). Differences between wh-Scope-Marking and wh-Copy Constructions in

Passamaquoddy. Linguistic Inquiry, 37, 25 – 49.CHENG, L. (1997). On the typology of wh-questions. New York and London: Garland Publishing, Inc.CHENG, L., & ROORYCK, J. (2000). Licensing wh-in-situ. Syntax, 3(1), 1 – 19.CHOMSKY, N. (1957). Syntactic structures. Den Haag: Mouton.CHOMSKY, N. (1995). Bare phrase structure. In G. Webelhuth (Ed.), Government and binding

theory and the minimalist program (pp.385 – 439). Oxford, Cambridge, MA: Blackwell.CHOMSKY, N. (2001). Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale, a life in language

(pp.1 –52). Cambridge, MA: MIT Press.CHOMSKY, N. (2005a). Three factors in language design. Linguistic Inquiry, 36, 1 – 22.CHOMSKY, N. (2005b). On phases. Ms., MIT. Cambridge, Mass.CRAIN, S., & THORNTON, R. (1998). Investigations in universal grammar, A guide to experiments

on the acquisition of syntax and semantics. Cambridge, MA: MIT Press.DAYAL, V. (1994). Scope marking as indirect wh-dependency. Natural language semantics, 2, 137 – 170.DAYAL, V. (2000). Scope marking: Cross-linguistic variation in indirect dependency. In U. Lutz, G.

Müller & A. von Stechow (Eds.), Wh-scope marking (pp.195 – 230). Amsterdam: John Benjamins.De VILLIERS, J., ROEPER, T., & VAINIKKA, A. (1990). The acquisition of long-distance

rules. In L. Frazier & J. deVilliers (Eds.), Language processing and language acquisition (pp. 257 – 297). Dordrecht: Kluwer.

EVERS, A. (1981). Two functional principles to “Move V.” In W. Abraham (Ed.), Groninger Arbeiten zur Germanistischen Linguistiek (GAGL), 19, 96 –110. Rijksuniversiteit Groningen.

DOS ANJOS, C., (2004). L’acquisition des questions à longue distance chez des enfants normaux et chez des enfants atteints de Dysphasie de Développement. Quel rôle la théorie de l’esprit joue-t-elle?. Mémoire de maîtrise. Université Paris 5, Boulogne-Billancourt.

FANSELOW, G. (2006). Partial movement. In M. Everaert & H. van Riemsdijk (Eds.), The Blackwell companion to syntax, Vol.III (pp.437–492). Malden/Oxford/Carlton: Blackwell Publishing.

FANSELOW, G., & CAVAR, D. (2001). Remarks on the economy of pronunciation. In G. Mueller & W. Sternefeld (Eds.), Competition in syntax (pp.107 – 150). Berlin: Mouton de Gruyter.

FANSELOW, G., & MAHAJAN, A. (2000). Towards a minimalist theory of wh-expletives, wh-copying and successive cyclicity. In U. Lutz, G. Müller, & A. von Stechow (Eds.), Wh-scope marking (pp.195 – 230). Amsterdam: John Benjamins.

FELSER, C. (2004). Wh-copying, phases and successive cyclicity. Lingua, 114, 543 – 574.GUERON, J., & HOEKSTRA, T. (1988). T-chains and the constituent structure of auxiliaries.

In A. Cardinaletti, G. Giusti, & G. Cinque (Eds.), Constituent structure (pp.35 – 100). Dordrecht: Foris.

GUTIERREZ, J. (2006). Acquiring long distance wh-questions in L1 Spanish: A longitudinal investigation. In V. Torrens & L. Escobar (Eds.), The acquisition of syntax in romance languages (pp.251 – 287). Amsterdam: John Benjamins.

Language and Speech

C. Jakubowicz, N. Strik 131

HAMANN, C. (2006). Speculations about early syntax: The production of wh-questions by normally developing French children and French children with SLI. Catalan Journal of Linguistics, 5, 143 – 189.

HULK., A., & ZUCKERMAN, S. (2000). The interaction between input and economy: Acquiring optionality in French wh-Questions. In S. C. Howell et al. (Eds.), BUCLD 24 Proceedings (pp.438 – 449). Somerville, MA: Cascadilla Press.

JAKUBOWICZ, C. (2002). Functional categories in (Ab)Normal language acquisition. In I. Lasser (Ed.), The process of language acquisition. Berlin: Peter Lang Verlag.

JAKUBOWICZ, C. (2003). Hypothèses psycholinguistiques sur la nature du déficit dysphasique. In Y. Brun & Ch.L. Gérard (Eds.), Les dysphasies. Paris: Masson.

JAKUBOWICZ, C. (2004). Is movement costly? The grammar and the processor in language acquisition. Presentation, JEL (Journée d’Etudes Linguistiques), Nantes.

JAKUBOWICZ, C. (2005). The language faculty: (Ab)Normal development and interface constraints. Presentation, GALA 2005, Siena.

JAKUBOWICZ, C. (2007). Grammaire Universelle et Trouble Spécifique du Langage. In J. Bricmont & J. Franck (Eds.), Noam Chomsky (pp.164 – 175). Paris: Les Editions de l’Herne.

JAKUBOWICZ, C. (to appear). Measuring derivational complexity: New evidence from typically-developing and SLI learners of L1 French. In C. Jakubowicz & P. Schulz (Eds.), Specific Language Impairment (SLI) across languages: Properties, and possible loci. Special issue for Lingua.

JAKUBOWICZ, C., & NASH, L. (2001). Functional categories and syntactic operations in (Ab)Normal language acquisition. In Brain and Language, 77, 321 – 333.

KAMPEN, J. van (1997). First steps in wh-movement. PhD, Universiteit Utrecht.KAMPEN, J. van, & EVERS, A. (1995). How greedy is the child? Long distance movement in

small steps. In M. Verrips & F. Wijnen (Eds.), ASCLD 3 Papers from the Dutch-German colloquium on Language Acquisition 1994 (pp.91 – 108). Amsterdam: Instituut voor Algemene Taalwetenschap.

KATO, M. (2003). The interpretation and derivation of wh-in-situ constructions in Brazilian Portuguese. Presentation, UNICAMP.

LANCIEN, S. (2003). Où crois-tu que j’ai caché le playmobil ? L’acquisition des questions à longue distance chez des enfants normaux et chez des enfants atteints de Dysphasie de Développement. Mémoire de maîtrise, LPE Université Paris 5, Boulogne-Billancourt.

LUTZ, U., MUELLER, G., & von STECHOW, A. (2000). Wh-scope marking. Amsterdam: John Benjamins.

Mac WHINNEY, B. (1991). The CHILDES project: Tools for analyzing talk. Hillsdale, NJ: Lawrence Erlbaum Associates.

MARANTZ, A. (2005). Generative linguistics within the cognitive neuroscience of language. The Linguistice Review, 22, 429 – 445.

MATHIEU, E. (2004). The mapping of form and interpretation: The case of optional wh-movement in French. Lingua, 114, 1090 –1132.

McDANIEL, D. (1989). Partial and multiple wh-movement. Natural language and linguistic theory, 7, 565 – 604.

McDANIEL, D., CHIU, B., & MAXFIELD, T. (1995). Parameters for wh-movement types: Evidence from child English. Natural Language and Linguistic Theory, 13, 709 – 753.

NEWMEYER, F. (2003). “Interpretable” features and feature-driven A-bar movement. In C. Beyssade, O. Bonami, P. Cabredo-Hofherr, & F. Corblin (Eds.), Empirical issues in syntax and semantics. Paris : Presses Universitaires de Paris Sorbonne.

OBENAUER, H.-G. (1994). Aspects de la Syntaxe A-Barre, Effets d’Intervention et Mouvements des Quantifieurs. Thèse d’Etat, Université Paris 8.

OIRY, M. (2002). Acquisition des Questions à Longue Distance. Mémoire de maîtrise, Université de Nantes.

Language and Speech

132 Acquisition of LD wh-questions in French and Dutch

OIRY, M., & DEMIRDACHE, H. (2006). Evidence from L1 acquisition for the syntax of wh — scope marking in French. In V. Torrens & L. Escobar (Eds.), The acquisition of syntax in romance languages (pp.289 – 315). Amsterdam: John Benjamins.

PESETSKY, D., & TORREGO, E. (2001). T-to-C movement: Causes and consequences. In M. Kenstowicz (Ed.), Ken Hale: A life in language (pp.355 – 426). Cambridge, MA: MIT Press.

PIERCE, A. (1992). Language acquisition and syntactic theory: A comparative analysis of French and English child grammar. Dordrecht: Kluwer.

PLATZACK, C. (1998). A visibility condition for the C-domain. Working papers in Scandinavian syntax, 61, 53 – 100. Lund: Department of Scandinavian Languages.

REINHART, T. (1998). Wh-In-Situ in the framework of the Minimalist Program. Natural Language Semantics, 6, 29 – 56.

RIEMSDIJK, H. van (1982). Correspondence effects and the empty category principle. Tilburg Papers in Language and Literature, 12. Tilburg: Tilburg University.

RIZZI, L. (1990). Relativized minimality. Cambridge, MA: MIT Press.RIZZI, L. (2000). Remarks on early null subjects. In M. A. Friedemann & L. Rizzi (Eds.), The

acquisition of syntax (pp.269 – 292). London: Longman.RIZZI, L. (2007). L’acquisition de la langue et de la faculté de langage. In J. Bricmont & J. Franck

(Eds.), Noam Chomsky (pp.147 – 157). Paris: Editions de l’Herne.SCHIPPERS, A. (2006a). Variation in the CP domain: Long-distance dependencies in the Dutch

dialects. Unpublished ms., Rijksuniversiteit Groningen.SCHIPPERS, A. (2006b). Wie wat or waar. Unpublished ms., Rijksuniversiteit Groningen.SOARES, C. (2003). The C-domain and the acquisition of European Portuguese: The case of

wh-questions. Probus, 15(1), 147 – 176.SOARES, C. (2006). La syntaxe de la périphérie gauche en portugais européen et son acquisition.

Thèse de doctorat, Université Paris 8.STRIK, N. (2002). L’Acquisition de Questions Wh Objet et Sujet chez les Enfants Francophones

Agés de 4 et de 5 ans, Mémoire de Maîtrise, Universiteit van Amsterdam and LPE, Université Paris 5, Boulogne-Billancourt.

STRIK, N. (2003). Où tu as caché ton sac? Qu’est-ce que tu penses que je lis? Acquisition des Questions Wh chez les Enfants Francophones de 3 à 6 Ans, Mémoire de DEA, Université Paris 8 and LPE, Université Paris 5, Boulogne-Billancourt.

STRIK, N. (2006). L’acquisition des phrases interrogatives chez les enfants francophones. Psychologie Française, 52, 27 – 39.

STRIK, N. (in progress). Syntaxe et acquisition des phrases interrogatives en français et en néer landais: une étude contrastive, PhD, Université Paris 8.

STROMSWOLD, K. (1995). The acquisition of subject and object wh-questions. Language Acquisition, 4, 5 – 48.

THORNTON, R. (1990). Adventures in long distance moving: The acquisition of complex wh-questions. PhD, University of Connecticut.

ZUCKERMAN, S. (2001). The acquisition of “optional” movement. Groningen: University Library Groningen.