Dummy Auxiliaries in First and Second Language Acquisition 9781614513476, 9781614515579

Citation preview

Elma Blom, Ineke van de Craats, Josje Verhagen (Eds.) Dummy Auxiliaries in First and Second Language Acquisition

Studies on Language Acquisition 49

Editor Peter Jordens

De Gruyter Mouton

Dummy Auxiliaries in First and Second Language Acquisition edited by Elma Blom Ineke van de Craats Josje Verhagen

De Gruyter Mouton

ISBN 978-1-61451-557-9 e-ISBN 978-1-61451-347-6 ISSN 1861-4248 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. 쑔 2013 Walter de Gruyter, Inc., Boston/Berlin Printing: Hubert & Co. GmbH & Co. KG, Göttingen 앝 Printed on acid-free paper 앪 Printed in Germany www.degruyter.com

Contents Introduction Elma Blom, Ineke van de Craats and Josje Verhagen

1

Dummy auxiliaries in monolingual first language acquisition Superfluous ‘do’ and comparison of spell-outs Carson T. Schütze

11

Dummy auxiliaries in Dutch first language acquisition Shalom Zuckerman

39

Semantic dummy verbs in child Dutch Bart Hollebrandse, Margreet van Koert and Angeliek van Hout

75

There is a dummy ‘is’ in early first language acquisition Manuela Julien, Ineke van de Craats and Roeland van Hout

101

Lightverbhood in child language: Evidence from Cypriot Greek Kleanthes K. Grohmann and Evelina Leivada

141

Dummy auxiliaries in simultaneous and successive child bilingualism Reflections on dummy ‘do’ in child language and syntactic theory Ute Bohnacker

171

Dummy verbs in first and second language acquisition in German Solveig Chilla, Stefanie Haberzettl and Nadja Wulff

209

Dummy auxiliaries in children with SLI ˗ a study on Dutch, in monolinguals and bilinguals Jan de Jong, Elma Blom and Antje Orgassa

251

Dummy auxiliaries in adult second language acquisition From dummy auxiliary to auxiliary in Moroccan adult learners’ production and comprehension of Dutch Josje Verhagen

281

Dummy verbs and the acquisition of verb raising in L2 German and French Sarah Schimke

307

vi

Contents

Broader perspectives on dummy auxiliaries Dummies and auxiliaries in the acquisition of L1 and L2 Dutch Peter Jordens Child use of auxiliary + infinitive in Dutch: Acquisition device or reflection of the input Leonie Cornips

341

369

Dummy auxiliaries in Dutch dialects, L1 and L2 acquisition Sjef Barbiers

395

Index

417

Introduction Elma Blom, Ineke van de Craats and Josje Verhagen

Researchers working on first, bilingual or second language acquisition have reported that learners often use auxiliary verbs with little or no meaning of their own in their speech and tend to use such auxiliary constructions in contexts where native adult speakers would not. For an illustration of this phenomenon, consider the following two examples from English and Dutch, which have been taken from chapters in this volume: (1)

She does want eyes on her back. she do.3SG want.INF eyes on her back ‘She wants eyes on her back.’ (child L2; Bohnacker, p. 171)

(2)

De meneer Blauw is springen. the mister Blue is.3SG jump.INF ‘Mister Blue jumps.’ Target: Meneer Blauw springt. (adult L2,Verhagen, p. 283)

When pronouncing the sentence in (1), most proficient speakers of English will have the tendency to emphasize does. However, learners of English use sentences as in (1) without emphatic stress on does. In Dutch, the auxiliary is is typically used with participial verb forms; sentences as in (2) in which is selects an infinitival main verb are only allowed as an ‘absentive’. That is, to refer to an activity that is taking place at another location than the present location of the speaker. Yet, learners of Dutch use sentences as in (2) without obeying this semantic restriction. In this volume, elements as does in (1) and is in (2) will be referred to as ‘dummy auxiliaries’. Dummy auxiliaries possess two intriguing properties, which become even more striking when they co-occur: They appear to be a learner’s own invention and they do not seem to add a specific meaning to the sentence. Why would a learner make up a word without a meaning? Do all learners do this, or are dummy auxiliaries more prevalent in particular learners? Are dummy auxiliaries really meaningless? And are they truly new inventions or is our concept of the target language too limited? Several researchers have noted the occurrence of dummy auxiliaries and raised

2

Elma Blom, Ineke van de Craats and Josje Verhagen

these questions. However, to date, dummy auxiliaries have seldom been the main focus of research. This volume provides an overview of state-of-the-art research on dummy auxiliaries in learner and speaker varieties. Its chapters cover various areas of linguistic investigation: typical and atypical first language acquisition, bilingual first language acquisition, child and adult second language acquisition, dialectal variation, and situations of language contact. Also, contributors to the volume have situated their descriptions and explanations in different theoretical frameworks, including generative syntactic or semantic-pragmatic approaches, whereas others have taken a more theoryneutral approach. Data reported are naturalistic and experimental, and language production and comprehension are discussed. Finally, several target languages are studied such as Cypriot Greek, Dutch, English, French, and German. Taken together, this volume represents our current knowledge of dummy auxiliaries, both in breadth and depth. The notion ‘dummy auxiliary’ Before turning to the different chapters in this volume, we would like to discuss the notion of ‘dummy auxiliary’. In previous research, dummy auxiliaries have also been described as light verbs, dummies, dummy verbs, fillers, placeholders, general all-purpose verbs, or simply as auxiliaries. For this volume, we prefer the term ‘dummy auxiliaries’ because we feel it covers both relevant semantic and structural properties of the constructions under investigation, whereas other notions tend to focus on either semantic or structural properties: ‘Dummy’ expresses that the element is semantically empty and ‘auxiliary’ is added because the elements in question typically co-occur with a thematic verb in an auxiliary-like construction (Crystal 2008). That said, we need to emphasize that both the notion of ‘dummy’ and that of ‘auxiliary’ should be used with caution. As few studies have systematically investigated the semantics of dummy auxiliaries (but see Jordens, this volume; Verhagen, this volume) and most studies have focused on (semi-)spontaneous production rather than comprehension (but see Verhagen and Zuckerman, this volume), it may be the case that auxiliaries express subtle meanings that have gone unnoticed so far. The assumption that elements such as those in (1) and (2) are auxiliaries may also require rethinking, as claimed by Jordens (this volume). Jordens argues that there is an early stage in L1 and L2 acquisition of Dutch that is characterized by the absence of grammatical relations, which casts doubt on whether it is justified to call the above elements ‘auxiliaries’. In the context of this volume,

Introduction

3

we should note that not all chapters are restricted to ‘auxiliary-like’ elements. Chilla, Haberzettl and Wulff discuss both dummy auxiliary and dummy main verbs in German. Studying child data from Cypriot Greek, Grohman and Leivada explore various aspects of light verb use of one particular verb, and in their contribution they focus on sentences in which this verb takes nominal instead of verbal complements. The thirteen chapters in this volume are divided into four parts, each covering a different area of investigation. Part I contains contributions that focus on dummy auxiliaries in monolingual first language acquisition. Part II deals with dummy auxiliaries in (both simultaneous and successive) bilingual children. In addition, one study (De Jong, Blom and Orgassa) in this part compares effects of (child) bilingualism to effects of language impairment. Part III continues with two chapters on second language acquisition by adult learners. Finally, Part IV, includes contributions that take a broader perspective by generalizing over various learner populations. These contributions shed new light on commonly held assumptions and critically review some fundamental notions that are important to the debate on dummy auxiliaries. Part I: Monolingual first language acquisition The studies in this first part build on previous research and extends this research in various ways. Focusing on English monolingual toddlers, Schütze (Chapter 1) proposes a generative analysis according to which differences between children and adults in their use of dummy, or ‘superfluous’ do, are the result of economy considerations. Economy is a notion that appears throughout the volume (Barbiers; Bohnacker; De Jong, Blom and Orgassa; Hollebrandse, Van Koert and Van Hout; Zuckerman). According to Schütze, superfluous do is the result of not lowering T(ense) to V(erb). Children fail to lower T to V because they cannot reliably compare derivations. This comparison is driven by the economy principle that structures with fewer overt morphemes are preferred to structures with more overt morphemes. In adults, but not in children, this leads to obligatory T to V lowering. Consequently, children use superfluous do, and differ from adults in this respect. The spell-out of T will of course differ from language to language and seems to be influenced by the environmental input, as becomes manifest in Zuckerman’s contribution (Chapter 2). Dutch-acquiring children overuse dummy auxiliaries that are used in the dialect of their region: either doen (‘do’) or gaan (‘go’), depending on whether children are exposed to a

4

Elma Blom, Ineke van de Craats and Josje Verhagen

southern or a northern variety of Dutch. Based on a comprehension experiment, Zuckerman proposes that children who overuse the dummy auxiliary gaan do so because they misinterpret gaan-structures in the adult speech as denoting ongoing, present tense events. Like Schütze, Zuckerman assumes that structural economy is crucial, but in Zuckerman’s view it is not the number of overt morphemes that is relevant, but the type of syntactic operation (following, e.g., Van Kampen 1997) of the lexical verb. The somewhat counterintuitive consequence of this approach is that a sentence with more words is more economical than the sentence with fewer words. In using dummy auxiliaries, children avoid the more costly movement operation of the lexical verb. Like Zuckerman, Hollebrandse et al. (Chapter 3) focus on the auxiliary gaan (‘go’), but in older children. The authors present results from an elicited production experiment that show that even at five years of age, some Dutch children produce gaan in contexts where adults do not. The children use dummy gaan to refer to past (ging), present (gaat) or future (gaat) events, and some children use the same form of the verb in all three temporal conditions. The authors propose that children stick to dummy gaan, not only for syntactic reasons, but also for morphological and semantic ones. Hollebrandse et al. argue that gaan is a morphologically easy-toretrieve form that is used to perform existential closure over the event. Julien, Van de Craats and Van Hout (Chapter 4) focus on the Dutch dummy auxiliary is (‘is’). This form of the copula zijn (‘be’) is frequently used in combination with a lexical verb ˗ most often in infinitival form ˗ by adult L2 learners of Dutch, as shown in the contribution of Verhagen (Chapter 9), but not by Dutch L1 children. The authors conclude that is as a dummy has similar functions as gaat (‘goes’) and doet (‘does’).The same form is or its language-specific equivalent is also observed in L2 English by Fleta (2003), L2 German (Chilla, Haberzettl and Wulf, Chapter 6; Schimke, Chapter 7), and L2 French (Schimke, Chapter 7). Julien et al. suggest that the dummy auxiliary construction (Aux+INF) with is in child Dutch is modeled after the highly frequent copula form is. Grohman and Leivada’s contribution (Chapter 5) is the first study to investigate dummy verbs in children acquiring Cypriot Greek. The authors focus on the verb kamno (‘do/make’) combined with lexical material – most often a noun – to express predication. They administered a picturebased elicitation task to children with priming of either a light verb construction or a lexical verb. It was found that children by four years of age already extensively produce light verb constructions even when primed with lexical verbs. The experiment itself, however, yields various methodo-

Introduction

5

logical problems when used with children of various ages because it is not clear whether a light verb construction is due to linguistic development or to priming effects. Part II: Simultaneous and successive child bilingualism The second part of this volume contains three contributions that focus on bilingual children. Bohnacker (Chapter 6) reports on dummy do in an Icelandic-English bilingual child. After providing a detailed overview of previous research on dummy do, Bohnacker meticulously goes over the various explanations for dummy do, concluding that dummy do as observed in her data presents an example of optionality and free variation that presumably needs to be unlearned by evaluation of the input. It is particularly interesting to compare her account to Schütze’s (i.e., do as an allomorph of indicative mood, located in the functional projection Mood), who offers a different explanation on this point. Also, Bohnacker makes explicit how flexibly the notion of economy has been interpreted in the literature on dummy auxiliaries. According to some researchers economy predicts use of lexical verbs instead of dummy auxiliaries, whereas others derive the opposite prediction. Bohnacker concludes that both views of economy are equally stipulative. Chilla et al. (Chapter 7) deal with the question of whether the use of dummy auxiliaries differs for monolingual (L1) and sequential bilingual (L2) children acquiring German. They find that dummy use in L1 children and child L2 learners with an age of onset (contact with the L2) of three years is rare. It is only for children who start learning German as an L2 after the age of six that dummy use constitutes a distinct stage of acquisition. Data of the latter group indicate that the use of dummy verbs increases with slower development and later input, which is confirmed by the results of the next study (De Jong et al.). De Jong et al. (Chapter 8) investigate dummy auxiliaries in TurkishDutch bilinguals with and without Specific Language Impairment (SLI), in comparison with monolingual controls. Previous research has indicated that (sequentially) bilingual children and children with SLI may show comparable patterns in their use of verb inflection. One question raised in this chapter is whether the same holds true for dummy auxiliary use. It is revealed that the SLI groups use more dummy auxiliaries than the groups with typical development. Moreover, effects of impairment overrule effects of bilingualism. A few children with SLI use dummy auxiliaries in dependent clauses, a pattern also noted by Hollebrandse et al. (Chapter 3) in typically

6

Elma Blom, Ineke van de Craats and Josje Verhagen

developing younger children. In children with SLI, economy at the morphological level could play a role, whereas in other children syntactic economy may be driving dummy auxiliary use. Part III: Adult second language acquisition This part focuses on adult L2 learners. Verhagen (Chapter 9) deals with the acquisition of Dutch by Moroccan Arabic speakers and presents results from elicited-production tasks and a comprehension experiment. The aim of the comprehension study is to examine whether Moroccan learners of Dutch use the dummy auxiliary is as a marker of aspect or a structural device, as claimed in previous studies. The results suggest that learners do not associate is with perfective aspect, but with present-tense ongoing events. This result converges with earlier findings for Moroccan learners of Dutch. In addition, Verhagen finds that the use of is positively correlates with the use of copula forms, which supports the idea that is may be modeled after the copula as was also suggested for L1 Dutch by Julien et al. (Chapter 4). Schimke (Chapter 10) presents production and elicited-imitation data from Turkish learners of French and German. While dummy auxiliaries are found in both L2 French and L2 German, L2 French is also characterized by the frequent use of presentational constructions like c’est or il y a (‘there is’) and forms of have (‘hat’/‘a’), even when sentences in an elicitedimitation task are imitated. Schimke’s main findings are that dummy verbs are related to the acquisition of verb raising in German (similar to what is claimed for L2 Dutch by Verhagen in Chapter 9, and for L1 Dutch by Julien et al. in Chapter 4), whereas dummy auxiliaries are more frequent in French than in German. To explain these cross-linguistic differences, Schimke proposes an account in terms of a difference in transparency of form-function mappings between French and German (and Dutch). Part IV: Broader perspectives The final part of the book includes three chapters that take a broader perspective. These chapters are, unlike the other chapters, not restricted to a specific population but are relevant to various (learner) populations. In Chapter 11, Jordens proposes an account of the acquisition of Dutch grammatical structure with a special role for auxiliary verbs. Jordens takes a viewpoint that differs from that of most other authors in this volume. In his approach, elements such as is and gaat (‘goes’) are lexical elements with a semantic/pragmatic function. These elements should therefore not be seen as auxiliaries, a function which develops at a later stage, when learners

Introduction

7

have acquired functional categories. According to Jordens, in the first socalled lexical stage, doet (‘does’) has the default function of ‘control’ (and only occurs in combination with agentive verbs) and is or 0 (zero) express assertion. This idea displays a certain similarity with Schütze’s (Chapter 1) interpretation of do as an allomorph of indicative mood and 0 as a counterpart, notwithstanding the marked differences between the two approaches. Cornips (Chapter 12) compares data from various speech corpora and argues that the use of dummy auxiliaries by L1 and child L2 learners of Dutch is not specific to language learners, but also occurs in the speech of adult native speakers. Cornips argues that we might underestimate the possibilities provided by the target language, in which auxiliaries may have a broader usage than previously assumed, at least in Dutch where doen and gaan actually sometimes do occur in dummy-like constructions. On the basis of these observations, Cornips argues that we need to be cautious in assuming that dummy auxiliaries reflect a developmental step, because their occurrence may in fact mirror the language that learners are exposed rather than being the learners’ own invention. A similar, but weaker claim was made by Zuckerman for L1 children (Chapter 2), who combines an input-based and a structural approach to dummy auxiliaries. In the final chapter of this volume, Barbiers (Chapter 13) compares the use of dummy auxiliaries in acquisition contexts to dummy constructions in dialectal variation. He finds that syntactic patterns observed in dummy auxiliary use in Dutch L1 and L2 are not paralleled in Dutch dialects (e.g., constructions with is and gaat plus an infinitive) and concludes that despite a superficial similarity, dummy auxiliaries in Dutch dialects have a distinct syntactic status from dummy auxiliaries in L2 learners’ production of Dutch. Barbiers addresses the question of whether dummy auxiliaries in Dutch dialects result from the tendency to be economic preferring ‘cheaper’ derivations over more ‘costly’ ones. However, he concludes that although economy may be an explanation in L1 and L2 acquisition, such an analysis cannot easily be applied to dialectal variation. In conclusion of the introduction to this volume, we would like to mention that the chapters in this volume originate from the workshop “Dummy auxiliaries in (a)typical first and second language acquisition” that was held at Radboud University Nijmegen, the Netherlands, in the summer of 2010. The workshop was a great success, which would not have been possible without the presenters and audience’s contribution to two very fruitful days. We also thank Mouton, and in particular Peter Jordens who, as one of the

8

Elma Blom, Ineke van de Craats and Josje Verhagen

editors of the Mouton Studies On Language Acquisition (SOLA) series, encouraged us to publish this volume. Finally, we would like to thank all reviewers for their constructive criticism and helpful comments. References Crystal, David 2008 A dictionary of linguistics and phonetics. Oxford: Blackwell. Fleta, Teresa 2003 Is-insertion in L2 grammars of English: A step forward between developmental stages? In Proceedings of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002), Helmut Zobl and Helen Goodluck, (eds.), 85˗96. Somerville, MA.: Cascadilla Proceedings Project. Van Kampen, Jacqueline 1997 First steps in Wh-movement. Ph.D. dissertation, Utrecht University.

Dummy auxiliaries in monolingual first language acquisition

Superfluous ‘do’ and comparison of spell-outs Carson T. Schütze

1. Introduction The ‘dummy’ auxiliary do is obligatory in English in a number of environments when no other auxiliary is present, including clauses containing sentential negation, questions with subject-auxiliary inversion, clauses whose truth is being emphasized, V(erb) P(hrase) ellipsis, and VP topicalization.1,2 (1)

a. b. c. d. e.

Mary does not like John. Does Mary like John? Mary DOES like John. Mary does TOO like John. Mary likes John, and Sue does too. Mary thought she would like John, and like John she does.

In the absence of one of these triggers, and in the presence of another auxiliary or the copula, dummy do is ungrammatical: (2)

a. b. c.

*Mary does like John. [no emphasis on does] *Mary does not be dancing. *Mary does (not) be happy.

Children acquiring (standard) English sometimes produce utterances like (2a), a phenomenon I dub ‘superfluous do,’ which constitute a rare case of a child error due not to omission but to production of an unnecessary element. The purpose of this paper is to document this phenomenon and propose an analysis of it. In particular, two familiar learning puzzles must be addressed. First, why do children produce utterances that are ungrammatical in their target grammar? Second, how do these children eventually stop producing such utterances, given the usual assumption of no negative evidence? An additional wrinkle to be dealt with is the fact that children generally do not produce errors of the type in (2b) and (2c).

12

Carson T. Schütze

The analysis is based on the account of do-support presented in Schütze (2004). The central claim of that approach, assuming Late Insertion, is that do is an allomorph of an otherwise phonologically empty Mood head, whose selection is triggered by the presence of an affix in need of a host word. The claim about acquisition is that children’s grammars are adult-like except for the fact that lowering of T(ense) to V(erb) is not obligatory. 2. Child data 2.1. Definition of superfluous do I define superfluous do as any production of auxiliary do that would be ungrammatical in (modern standard) adult English solely because a main verb should have been inflected instead. The typical environment where this can occur is a nonemphatic positive declarative, as in (2a).3 Children’s superfluous do has often been discussed in the context of tense doubling errors, illustrated in (3). (3)

a. b.

He didn’t sang. Did he ate it?

However, such cases do not fit my definition of superfluous do because what is wrong with them is not the presence of do (these are obligatory do contexts), but the presence of a (second) inflection on the main verb. There are also (rarer) cases that seem to display both doubled inflection and do in a non-obligatory context, as in (4). (4)

a. b.

He does likes it. I did made it.

These do not fit my definition of superfluous do either, because the presence of do is not the sole source of ungrammaticality, and I suggest that the underlying cause is plausibly different. Specifically, in (4) it is clear that T has been combined with the main V, unlike in (2a); what has gone wrong is that T is being pronounced in both its pre- and post-movement positions (on a theory in which Affix Hopping is construed as movement of some sort). Errors like (3) and (4) will not be further discussed in this paper. To anticipate the analysis, what is unique to the superfluous do error in (2a) is that T has not combined with V at all, although it “could have” doneso.The

Superfluous ‘do’ and comparison of spell-outs 13

rest of this section presents evidence that superfluous do as just defined is a feature of the speech of (at least some) children acquiring English. 2.2. Monolingual English data 2.2.1. Spontaneous production Several researchers have documented for monolingual English children the spontaneous production of superfluous do (Davis 1987; Hollebrandse and Roeper 1996; Zukowski 1996). Examples from one child’s speech are given in (5); Roeper (1991) attests that these do’s were not stressed and the contexts were not emphatic. (5)

a. b. c. d. e. f. g. h. i. j. k.

A witch did look like it has slippers. I did wear Bea’s helmet. I do have juice in my cup. Who did take this off? I did paint this one and I did paint this one…and I did paint this one. You did make my bed a little fan. And Daddy did say, “I want eat cereal.” I did paint yellow right here. I did put the brush in. I did paint it. A doggie did walk with Dorothy and the doggie did hurt itself. The doggie did hurt on the street. She Dorothy did get to Oz. A witch did look like it has slippers. (Tim 2;11–3;0, Roeper corpus)

Most researchers have concluded, sometimes on the basis of very dense longitudinal recordings, that the productive superfluous do phase is quite short-lived. For example, Zukowski (1996) analyzed Brian MacWhinney’s son Ross’s data in CHILDES and found out from MacWhinney that a stretch of apparently superfluous do’s were indeed unstressed and noncontrastive; these were concentrated between 2;11.07 and 3;3.27. Crucially, these errors are not part of a more general pattern in which do is widely overused; that is, examples like (6) are not attested (see also Stromswold 1990). (6) Unattested error types

14

Carson T. Schütze

a. b. c. d. e.

#He does ran. #He did runs. (Roeper 1991) #John doesn’t can play alto-sax. (Hollebrandse and Roeper 1996) #He does is. #He did was.

Two generalizations emerge here. First, we almost never find mismatches of tense between do and an inflected main verb, which indicates that children know how the forms of do relate to the underlying morphosyntactic features. Second, do does not occur in contexts where an auxiliary/modal/copula should have been the finite verb, cf. (2b) and (2c). One potential exception to this pattern, noted by Roeper (1991), involves the infinitival form of be, as illustrated in (7). (7)

a. b.

It does be. Do clowns be a boy or a girl.

However, Roeper observes errors such as (8) in the same set of corpora.4 (8)

He bes here.

Thus, it is possible that children who are producing sentences like (7) are doing so because they have misanalyzed be as a main verb that does not undergo V-to-T raising (the process by which is comes to be to the left of not in He is not in Standard English) and therefore shows the same distribution as, say, walk. This is not implausible, in light of the total lack of phonological overlap between the infinitival form be and its inflected counterparts ˗ the child may not initially recognize that these are forms of the same verb, and for them be could mean something slightly different than is. If this is so, then the aforementioned generalization can be maintained: children overextend do only in environments where a main verb should have been finite, not environments where an auxiliary/modal/copula should have been the finite verb. It should be acknowledged that it would be preferable if the phenomenon of superfluous do and its apparent restrictions could be quantitatively documented in a large corpus such as those available on CHILDES (MacWhinney 2000). Unfortunately none of those corpora contains sufficient prosodic information to be able to rule out the possibility that a given

Superfluous ‘do’ and comparison of spell-outs 15

occurrence of do might have been emphatic. Such a study would therefore require an analysis of original audio recordings. Until that is done, we can at least take the data reviewed here as suggestive. 2.2.2. Experimental elicitation Researchers have also reported superfluous do in elicitation studies. Foley, Pactovis, and Lust (in prep.) reported the examples in (9) from an elicited repetition study. (9)

a. b. c.

Model: Barney moves his penny and Ernie does too. Child: Grober does move his penny and and and Ernie does too. (3;04) Model: Kermit washes his face and Oscar does too. Child: Kermit does wash his face and Ostar does too. (3;04) Model: Ernie touches the ground and Grover does too. Child: Ernie touch the ground and Grover does touch the ground. (3;08)

A potential worry in interpreting these data is that the occurrence of do in the model sentence might be priming the child’s production of it. Still, we might hope that priming would not be able to induce children to systematically produce structures that are ungrammatical for them, though Snyder (2007, Chapter 6) suggests otherwise. Thornton (2010) conducted an elicited production study of which the goal was to encourage use of VP ellipsis. One of the three children studied, Georgia (2;3.16), produced superfluous do just in environments where VP ellipsis would be grammatical for adults; these pre-dated her first actual elisions of VP. Examples are given in (10) and (11). (10) Context: Discussion of which family members like goldfish-shaped crackers. Experimenter: Does your daddy eat goldfish? Georgia: No. Only Georgia does eat goldfish. (11) Context: Feeding the dog and cat puppets toy food. Experimenter: I like corn. What about you? Georgia: Hmm. Georgia likes corn. Experimenter: What about the dog? Georgia: He likes corn.

16

Carson T. Schütze

Experimenter: Georgia: Experimenter: Georgia:

And what about the cat? Mmm! She likes corn! Everyone likes corn. Mommy does like corn. And Georgia likes corn. Everybody likes corn. And daddy likes corn. Everybody likes corn!

Thornton suggests that Georgia is mastering the environment for licensing VP ellipsis before actually doing the elision. 2.3. Bilingual English data 2.3.1. English/German Knipschild (2007) reports data from a child named Joshua (ages 2;4.22– 3;1.19), who has both English and German as L1s. Joshua appears to have obligatory V-to-T raising of main verbs (across not) in his English through 2;9.4, presumably under the influence of German; in the second half of the recordings, the rate of V-to-T raising declines to under 60%, and shortly before that he starts using superfluous do (almost exclusively the form did, almost no instances of do or does). Examples are given in (12). (12) a. b. c. d.

I did do it in nanny and granddad’s garden. I did watch it. I did see that. Greedy did want all the food.

(2;9.27) (2;10.08) (2;10.18) (2;11.0)

His rate of superfluous do peaks at about 15%, then declines to 1% by the final session. I suggest that this represents a stage at which he is unlearning V-to-T raising, but has not yet mastered T-to-V lowering; superfluous do allows him to do neither. See also Tracy (1995). (I cannot rule out the possibility that Joshua has been exposed to superfluous tun in German (see Section 2.4), which means examples like those in (12) could in principle be due to influence from German.)

Superfluous ‘do’ and comparison of spell-outs 17

2.3.2. English/Norwegian Jensvoll (2003) reports data from a child Emilie (age 3;8) who is acquiring both English and Norwegian as L1s. She reports that in elicited answers to questions containing do-support (e.g., What did the boy do yesterday?) the child frequently used did instead of past tense inflection (i.e. superfluous do), occasionally used did in combination with a past tense main verb (i.e. tense doubling), and almost never used a correct past tense main verb alone. Superfluous do responses constitute about 70% of Emilie’s nonadultlike responses, while tense doubling errors constitute fewer than 5%. Jensvoll reports, “She did show signs of the fact that she was in the process of learning the -ed suffix … However, it seemed as though she was applying her did-construction as a default, because this construction is simpler than marking past tense on the verb itself.” I do not know whether Norwegian ever displays superfluous do, so I cannot assess whether crosslinguistic influence might be involved here. 2.3.3. English/Icelandic Bohnacker (1999, this volume) reports data from the child Katla, who learned Icelandic as her L1 but received mostly English input starting at age 1;3, and eventually behaved as balanced bilingual. An important fact about Icelandic, in contrast to German, is that it has no auxiliary uses of do whatsoever (even for VP topicalization, where it is possible in Mainland Scandinavian), so there are no issues of potential transfer. Unlike Joshua (Section 2.3.1), Katla shows no transfer of V-to-T raising: Her English main verbs always follow negation. From ages 2;4 to 3;0, she uses do in 80% of obligatory contexts with negation but in no obligatory contexts for emphasis, Yes/No questions, or (fronted) WH-questions.5 This is unlike her production of be and have, where omissions occur less than 50% of the time. At 3;0 a change occurs:6 Do in obligatory non-negative contexts rises to the level of be and have, and ‘oversupplied’ do appears as well. However, Katla never combines do with modals, auxiliaries, or the copula, consistent with the generalization in Section 2.2.1. This progression is detailed in Table 1.

18

Carson T. Schütze

Table 1. Katla’s production of auxiliary do as a function of age and context (Bohnacker this volume: Table 2) Age 1;6.0–2;9.14 2;10.15–3;0.29 3;1.10–3;3.11 3;4.07–3;6.07 3;7.01–4;7.04

Instances of oversupplied do 0 8

All instances of auxiliary do 96 81

Percentage of do’s that are oversupplied 0% 10%

115 6 0

254 109 350

45% 6% 0%

Bohnacker is aware of the possibility that do might be being primed by a preceding adult utterance (particularly if it was a question), but she asserts that “only 12% of Katla’s oversupplied do might arguably [be] copied or repeated from the adult lead question” (1999: 61). Oversupplied instances of do include all three forms of the verb, as shown in Table 2. Table 2. Katla’s production of oversupplied do (2;10.15–3;6.07) as a function of the form of do (Bohnacker 1999) Form of do

Instances

Percentage

do + main V does + main V

53 14

41% 11%

did + main V Other

57 5

44% 4%

129

100%

Total

Table 3 shows that most of the oversupplied do’s count as superfluous do by my definition (the first row); tense-doubled uses are a small minority. Bohnacker also observes no restriction on the type of verb or subject that occurs with oversupplied do.

Superfluous ‘do’ and comparison of spell-outs 19 Table 3. Katla’s production of oversupplied do (2;10.15–3;6.07) as a function of the form of the main verb (Bohnacker 1999) Form of following verb Uninflected main V Irregular past main V Affixed main V (-s, -ed) Total

Instances

Percentage

115 11 3 129

89% 9% 2% 100%

Bohnacker (this volume) reports that oversupplied do during the peak period highlighted in Table 1 was produced in 25% (115/456) of all contexts for finite simplex thematic verbs in declaratives. 2.4. Other Germanic languages It has been observed that children acquiring Dutch, German, and Swiss German also overuse (the counterpart of) do relative to the prescriptive norms of the adult language. For Dutch, sources include Blom and De Korte (2008), De Haan (1987), Hollebrandse and Roeper (1996), Jordens (1990), Van Kampen (1997), Wijnen and Verrips (1998), and Zuckerman (2001). For German, see Boser et al. (1992). For Swiss German, see Penner (1992) on Bernese and Schönenberger (2001) on Lucernese. It is unclear to what extent these productions can be considered errors (i.e., nontargetlike), however: Although tun/doen-periphrasis is proscribed in the standard languages, 7 it is widely used colloquially in spoken German (Erb 2001) and Swiss German (Glaser and Frey 2006; Schönenberger and Penner 1995), and in Dutch it is used at least in speech to children and in some dialect regions (Blom and De Korte 2008). It is therefore unclear whether there are any cases where we can say with certainty that a child produces tun/doenperiphrasis spontaneously in these languages, i.e., without ever having heard it in the input. Data from Zuckerman and colleagues (e.g., Bastiaanse et al. 2002; Zuckerman, this volume) on differences between children in Groningen versus Limburg suggest that input may indeed be a crucial trigger. There are also important differences between adult tun/doen and English do: There are no environments (other than VP topicalization) where tun/doen are obligatory, and they can never be stressed. Given how different the situation in English is, I will therefore refrain from making any claims about the potential extension of my account to these languages.

20

Carson T. Schütze

3. Background on adult English 3.1. Early Modern English Superfluous do is attested for English throughout the 16th century and persisted into the 18th (Visser 1969; Warner 1993). Its equivalence to inflected main verb counterparts is indicated by this contemporaneous grammarian’s comment: “I do is a verbe moche comenly used in our tonge to be put before other verbes, as it is all one to say ‘I do speake…’ and suche lyke, and ‘I speake…’” (Palsgrave 1530). Likewise, contemporary translators would put the periphrastic do construction in correspondence with a Latin sentence that gave no indication of emphasis. The point is underscored by Ellegård (1953): “do + the infinitive was functionally synonymous with the finite full verb” (p. 151); he concludes that we can “dispose effectively of any hypothesis that the do-form at this time conveyed a special shade of meaning, differentiating it from the simple form. When do is practically always used, it cannot fill any such function and is absolutely nothing but a mark of tense” (p. 167). Importantly, the optionality of do is attested within the production of individual authors. I conclude from this that there cannot be any principle of grammar that absolutely bars superfluous do; to the extent that it is impossible in modern English (on which see Section 3.2 below), this represents at most a parametric choice, i.e., it represents one among a small set of options that a given language can choose among. Interestingly, the historical record reports no instances of periphrastic do used with be or auxiliary have. The fact that the same restriction holds for children who are producing superfluous do reinforces the idea that children are exercising an option provided by UG (Universal Grammar). 3.2. Modern English dialects The point is further made by work on contemporary regional dialects of English. Superfluous do is found currently in the English of Southwestern England (Klemola 1998), where it is not a habitual marker but is in apparent free variation with inflected main verbs and seems to be compatible with any verb other than be. For other modern dialects, there are claimed to be particular shades of meaning associated with superfluous do8 (e.g., habituality) and/or restrictions on the verb types it can combine with (e.g., eventives but not statives); the situation looks much like what Erb (2001)

Superfluous ‘do’ and comparison of spell-outs 21

reports for German dialects: The particulars vary considerably from one dialect to the next, suggesting that they are a response to the availability of alternative forms (tun-periphrasis and inflected main verbs) but do not represent inherent grammatical properties of those forms. In other words, a general dispreference for homophony makes it likely that some particular distributional or semantic restrictions will be imposed on one of the variants, but this does not reflect the intrinsic properties of that variant. If so, these modern English dialects provide further evidence that superfluous do is a legitimate option in UG. And like in Early Modern English and child English, one distributional restriction that is consistent across dialects is that periphrastic do does not occur with modals, be and auxiliary have. That is, examples of the sort in (13) are unattested. (13) a. b. c. d.

#Do you be feeling all right? (cf. Are you feeling all right?) #I don’t have seen Jane. (cf. I haven’t seen Jane.) #Do you be a singer? (cf. Are you a singer?) #What does he will buy? (cf. What will he buy?)

3.3. Standard English In this section I consider whether standard (i.e., non-regional varieties of English have superfluous do. One candidate for this is regularly found in British English in situations where American English would simply use VP ellipsis (cf. Quirk et al. 1985): (14) a. b. c. d.

Q: Will you be attending the meeting this evening? A: I may do. Q: Has John already left? A: He might have done. Q: Why don’t you sit quietly? A: I AM doing.9 Bob says he is going to join the Labour Party. It will be interesting to see whether he DOES do.

The highlighted instances of do in (14) are not performing any of the ‘supportive’ functions outlined in Section 1. However, I claim they do not constitute superfluous uses of do by my definition either. The reason is that, as most of the literature agrees, syntactically these are not auxiliary but rather main verb instances of do, given their nonfinite forms, their co-occurrence with modals and other auxiliaries, and the fact that they can co-occur with a tensed (auxiliary) do, as in (14d). 10

22

Carson T. Schütze

A second candidate for adult superfluous do has received much less attention in the literature. I encounter it most often in public address announcements on airline flights, as exemplified in (15). 11 (15) a. b. c. d. e.

If you do have your cell phone with you, you may use it at this time. While you’re seated we’d like to request that you do keep your seatbelts fastened. If you do have any questions about safety, please ask a flight attendant. We do welcome you aboard. Please do understand.

The do in these examples does not seem to receive any pitch accent,12 nor does it seem to presuppose that the asserted content is under dispute, which would seem to be a precondition for verum focus (focus on the truth/falsity of the asserted statement). 13 I take it to fall under the description given by Quirk et al. (1985: Section 18.56) of uses of do that are not contrastive but express ‘emotive emphasis’; their examples are given in (16), where /˘/ denotes a fall-rise accent. 14 (16) a. b.

He did PRO˘Mise to go. I did TE˘LL you.

Regarding the semantic import of this use, they say that “the speaker (in a style that is sometimes felt to be rather gushing and extravagant) is conveying enthusiasm: A personal conviction about the truth of what is predicated.” (see also Nevalainen and Rissanen 1986). Although it seems unlikely that children hear much of this usage anyway, it is evidently not expressing a neutral statement, so it cannot provide a model for superfluous do. Assuming the above is on the right track, we can conclude that children acquiring standard English are not exposed to truly nonemphatic superfluous do, and therefore, unlike perhaps all such cases in the acquisition of Dutch, German, and Swiss-German, it represents an innovation on their part, and there must be something in their grammars that differs from those of adults. I turn now to exploring what this difference could be.

Superfluous ‘do’ and comparison of spell-outs 23

4. Analysis 4.1. Overview I propose that the grammar of children producing superfluous do is identical in relevant respects to that of Early Modern English speakers (cf. Section 3.1), i.e., it is a possible grammar under Universal Grammar, and my analysis will apply to both systems. The essence of the proposal is that T lowering (to V) is optional in these systems, whereas it is obligatory (except when blocked) in adult modern English, and this is the only property that differentiates the languages. I therefore begin with the account of adult modern English. I defer to Section 4.4 the question of why English children exhibit this optionality, and how they come to be speakers of modern English. 4.2. Account of adult modern English My account commences with the observation that, as far as inflectional suffixation is concerned, dummy do is completely regular. That is, the 3SG present form ends in the regular suffix /-z/ and the past tense form ends in the regular suffix /-d/. Do is ‘special’ only inasmuch as it exhibits stem allomorphy: /du/ becomes /dʌ/ in 3SG present contexts (and before the perfect participial suffix /-n/) and /dɪ/ in past tense contexts. Such allomorphy is familiar from other verbs, e.g., /se/ (‘say’) becomes /sɛ/ and /hæv/ (‘have’) becomes /hæ/ before 3SG /-z/, and many verbs show stem allomorphy in past tense and participle forms. What I conclude from this is that do is a stem unto itself, which combines with Tense and therefore cannot itself be a Tense head. What kind of stem is do? I claim in its auxiliary uses it is the same kind of stem as the English modals. These too cannot themselves be realizations of Tense: For one thing, they contribute meanings that are not temporal, but just as importantly, we can see that in some cases they combine with Tense just as do does. Specifically, could, would, and might can be used as the past tense forms of can, will, and may, respectively, e.g., in sequence of tense contexts (17); they also end in the past tense suffixes /d/ and /-t/.15 (17) a. b.

He said he would come, so why isn’t he here? Mary can run a 6-minute mile. When she was younger she could run a 5-minute mile.

24

Carson T. Schütze

c.

He said he might come, but he doesn’t seem to be here.

I conclude that the INFL complex (the set of functional projections that have supplanted the traditional monolithic INFL head) must include a head that hosts the stem do and the modal stems, all of which can combine with T by suffixation. 16 Following various authors (e.g., Culicover 1999; Erb 2001; Pollock 1997; Roberts 1993), I call this head M, intending to invoke Mood and/or Modality. This unification directly captures the observation that dummy do is excluded from all the environments where modals are excluded: VP small clauses, to-infinitivals, mandative subjunctives, Mad Magazine sentences (Akmajian 1984), and gerunds. (18) a. b. c. d. e.

*I made Mary can walk again. *I want Mary to can walk again. *It is vital that Mary can walk again.17 *What?? Mary can walk again??18 Impossible!! *Mary canning walk is surprising.

(19) a. b. c. d. e.

*I made Mary do (not) walk. *I want Mary to do (not) walk. *It is vital that Mary do (not) walk. *What?? Mary do (not) walk?? Impossible! *Mary doing (not) walk is surprising.

Note that be and auxiliary have differ from do in that they are possible in all of the aforementioned environments; this can be verified by substituting be able to for can in (18). From this we can conclude that be and have are not M heads. This conclusion is independently desirable because they can co-occur with modals, and although do generally does not co-occur with be and auxiliary have, in imperatives they can be found together, consistent with the claim that they belong to distinct categories. (20) a. b. c.

Do be careful! Don’t be late! Do have showered before you go to that interview!

I therefore claim that be and have are always of category V. Having motivated a functional head M that hosts modals and do and that is distinct from T, the next issue is where it fits in the overall clausal structure. I propose (21) for finite clauses. 19

Superfluous ‘do’ and comparison of spell-outs 25

(21) [CP [MP [TP [ΣP [v/VP The canonical subject position in this structure is Spec-MP; subjects may traverse Spec-TP to check case.20 ΣP is the locus of sentential negation and emphatic positive polarity, following Laka (1990). I claim that do is the marked allomorph of the M feature complex whose unmarked allomorph is Ø; this null allomorph occurs in nonemphatic, nonnegated declaratives and imperatives, as in the structures in (22). For the sake of concreteness, we can assume that the meaning associated with it is ‘indicative’. 21 (22) a. b.

C–Q Mary MINDIC -s like John CIMP pro2SG MINDIC -ØT sit down

Mary likes John. Sit down!

What conditions this allomorphy? I assume a Late Insertion approach to the syntax–morphology interface, as in Distributed Morphology (Halle and Marantz 1993), that is, phonological forms are inserted into the structure post-syntactically as realizations of abstract feature bundles. I propose that do is the allomorph selected in the presence of a morphologically dependent element needing a host. The relevant elements include the INFL affixes (-s and -ed and their allomorphs and present tense non-3SG-Ø). Thus, the circumstances in which the do allomorph of MINDIC will surface are those in which T does not get affixed to a main verb, auxiliary, or modal. In adult standard English the circumstances under which this will happen are those in which some element blocks the combination of T with a main verb. I follow Bobaljik (1996) in assuming that T combines with main verbs in English by a head lowering process that requires string adjacency. 22 Elements that block this lowering are thus the class of marked Σ heads (sentential negation, verum focus, and the emphatic positive polarity morphemes too and so) as well as an overt DP (in the case of a question with subjectauxiliary inversion).23 Thus, in adult modern English MINDIC will be pronounced as Ø in the examples in (22), because the T affixes will have lowered onto the following main verb. By contrast, MINDIC will be pronounced as do in the examples in (23), because T lowering to V is blocked. (23) a. b.

C–Q Mary MINDIC -s {not/too/so} like John. Mary does not/too/so like John. MINDIC+-s+C+Q Mary tM t-s like John? Does Mary like John?

26

Carson T. Schütze

The third scenario to consider is one in which a contentful modal is present: this will be an M head distinct from MINDIC, therefore do cannot be inserted. 24,25 4.3. Child English (and Early Modern English) All that we need to add to this system to derive child English and Early Modern English is the possibility that T need not combine with a main verb to its right even when nothing blocks this combination; in other words, T lowering to V is optional. In a clause with no auxiliaries (e.g., (22a)), failure to lower T will leave T as an affix without a host, and do will be chosen as the allomorph of MINDIC automatically. How do we derive the impossibility, in both child and adult English, of sentences like (13) above? What has gone wrong in these examples is that V has failed to raise to T when it should have. As long as we ensure that V raising to T is obligatory (when possible), 26 such examples will not be generated because T will not lack a host, and there is no problem ensuring this, since by hypothesis V raising to T and T lowering to V are distinct operations. An example is given in (24). (24) C–Q Mary MINDIC be+-s (not) tV sick. Mary is (not) sick. Not only are they distinct operations, they are distinct kinds of operations. T-to-V lowering is an operation of the Spell-Out component, while V-to-T raising in an operation of the ‘narrow syntax,’ i.e., the computation that builds a Logical Form from the numeration (the set of feature bundles to be assembled). We can assume that Spell-out operations are triggered in distinct ways (e.g., not by feature checking) and thus their learning follows a different path. 4.4. The source of optionality I now return to the question of why English-acquiring children have the option of not lowering T to V. I claim this is because the obligatoriness of T-to-V lowering can only be enforced by a comparison of derivations, something children have repeatedly been proposed to have difficulties with. This is so because the derivation yielding superfluous do converges, that is,

Superfluous ‘do’ and comparison of spell-outs 27

all its feature-checking requirements are met. In particular, if (22a) is sent to PF in its given form, without T having lowered to V, M INDIC will be spelled out as do, there will be no stranded affixes, and the result will be Mary does like John. In other words, I am claiming that whatever drives T lowering to V, it cannot be a requirement for convergence27 (e.g., a strong feature, i.e., one that must be checked prior to spell-out). This is because the very same T affixes are perfectly happy not lowering in the derivations in (23). Hence, superfluous do must be ruled out by the fact that another derivation based on the same numeration blocks it, in other words, a sort of economy constraint. Specifically, I propose a constraint that says that if two convergent derivations based on the same numeration differ in the number of overt morphemes (vocabulary items, in Distributed Morphology parlance) needed to spell them out, the derivation using fewer overt morphemes blocks the one using more morphemes.28 In this case, the competing derivations could both start out like (22a): In one, the -s affix lowers, yielding Mary likes John, while in the other, the affix does not lower and the result is Mary does like John. Since the latter contains an overt morpheme, do, beyond those that the former contains, the latter is ruled out. The reason that the same initial set of features can wind up containing different numbers of overt morphemes is that some morphemes (in this case MINDIC) may have both phonologically empty and overt allomorphs. Analogous principles (without implementational details) have been proposed in this domain before, e.g., Pollock (1989: 420, footnote 49) suggests “Perhaps there is an ‘Avoid Do’ principle in the grammar of Modern English falling under some version of Chomsky’s (1981) ‘Avoid Pronoun’ principle, itself conceivably the by-product of some more general ‘least effort’ principle”; Emonds (1994: 168) proposes that “The most economic realization of a given deep structure minimizes insertions of free morphemes (‘Use as few words as possible.’)”; see also Zuckerman (2001) and Arnold (1995a, b). My claim for Early Modern English is that this economy condition is not part of the grammar. My claim for child English is that the condition may well be part of the grammar but it cannot be reliably enforced because comparisons among derivations tax the abilities of the child’s processing system. Such limitations have been proposed in several domains, e.g., in so-called delay of Principle B effects (e.g., Chien and Wexler 1990; Grodzinsky and Reinhart 1993), scalar implicatures (Chierchia et al. 2001; Gualmini et al. 2001), and stress shift and focus sets (Reinhart 2004). Thus, I am proposing that the child’s grammar does not differ at all from that of the adult; only her ability to implement it does. In contrast, Warner (1993:

28

Carson T. Schütze

89) suggests a learning account: “Children overgeneralize unaccented do, and it is surely one of the few cases where frequency makes absence salient, so that an additional restriction [blocking superfluous do] could be learned later”. He thus appears to assume a sort of indirect negative evidence. My account would require some explanation for how and why the economy principle becomes part of children’s grammars in the first place. 5. Comparison with previous approaches In the previous literature there have been two notable proposals concerning how superfluous do arises in child English. I summarize them here in order to highlight where I differ from them. 5.1. Hollebrandse and Roeper Hollebrandse and Roeper (1996: 265) propose that in adult English and for the child, “do-support is a spell-out operation in the phonology and therefore is not [a] syntactic operation of insertion”. More specifically, “doinsertion is the spellout of the tense morpheme in phonology and therefore we call it Tense-spellout…a core feature of grammar, not an exceptional repair strategy.” Spelling out Tense as do is taken to be easier for the child than a complex combination of head movement and node relabeling operations required by the adult grammar to inflect a main verb. Technically it is more economical, in that it requires fewer syntactic operations. “The do is simple because it is a direct spellout of Tense. Therefore, the child can recognize it directly and no decomposition of a complex form like painted is necessary.”29 Putting aside the technical details of their node relabeling operation, the gist seems to be that producing do is an alternative to the adult way of expressing INFL features (by affixation on a main verb), and the child produces do using a rule that adults also use. At that level of abstraction, I concur. However, I have argued that do and Tense must be separate elements that receive separate spell-outs, so in my system do facilitates the spell-out of Tense but does not constitute it. Also, it is left unstated in the proposal how the child comes to favor a less economical option over a more economical one in acquiring the adult system.

Superfluous ‘do’ and comparison of spell-outs 29

5.2. Van Kampen Van Kampen (1997) makes the following proposal: “Let us assume that in child language do-insertion is free in all tense positions… More precisely, let us assume that in English child language children insert finite auxiliaries directly in the I0 position… The learning task for the English children will now consist in narrowing down the superset of do-insertions to a subset.” The way this should happen is that they should discover that “any string adjacency of do and its dependent lexical verb… should result in a deletion of do at PF.” She proposes that what must be learned is a ‘restructuring rule’ that combines I and V into a ‘compound’ under string adjacency. 30 The rule is inspired by Ross’s (1972) ‘do-gobbling’ rule. It is part of adult English grammars.31 The compounding rule is used in order to avoid the need for a lowering rule. Presumably there is a stage in acquisition when this rule is being applied optionally, and one may wonder whether this approach predicts a stage before the rule is acquired when use of do would be obligatory (such a stage has not been documented). One may also want to know how the child comes to posit free insertion of do in the first place. Putting technicalities aside, the gist seems to be that children produce superfluous do because they are not combining INFL with main verbs the way the adult grammar requires: “I propose that do is used by the Language Acquisition Device [for Dutch and English] as a temporary ‘least effort’ solution in order to avoid movement” (p. 49). At this level of abstraction, I agree. However, I do not posit in the child’s grammar any properties of do that differ from those in the adult language, and I assume that the process by which INFL combines with a main verb is PF movement, not restructuring/compounding. I also do not rely on children to make any discovery in order to arrive at the adult system. 6. Concluding remarks My proposal for children’s production of superfluous do is that their grammar is identical to the adult grammar of (modern) English, but that the latter includes an economy condition requiring comparison of derivations, and children cannot reliably carry out this comparison. This economy condition is apparently not universal, inasmuch as it was not active in Early Modern English, which also allowed free use of finite do as an alternative to inflecting main verbs. However, I assume that the general form of the

30

Carson T. Schütze

condition ˗ a preference for fewer overt morphemes in spelling out the result of a given numeration ˗ is made available by UG; perhaps its applicability to particular morphemes must be learned, and I have not addressed how this is accomplished. (In this respect my proposal seems no worse off than its predecessors.) The key to facilitating an analysis in terms of this economy condition has been the proposal that sentences with do and counterparts with inflection on the main verb do not differ in their numerations. This in turn has been made possible by an analysis on which do is the realization of a functional head (M) distinct from Tense and is in an allomorph relationship with a phonologically null version of that head. Notes 1. 2. 3.

4. 5.

6. 7. 8.

I would like to thank two anonymous reviewers for their helpful comments on a previous version of this paper. Note that the too in (1c) and the too in (1d) are different words with different meanings. Too in (1c), which must receive focus, emphasizes the truth of the statement. Too in (1d) means ‘as well’/‘also’. In principle there are two other environments that would fit this definition: nonemphatic matrix positive subject questions (i) and nonemphatic imperatives (ii): (i) Who does eat that? (ii) Do eat that! I am unaware of any systematic discussion of children’s use of superfluous do in these environments, and (5d) is the only case I have come across, so my focus shall rest on nonemphatic positive declaratives. Roeper explicitly predicts, but does not verify, that sentences like (7) and (8) are produced by the same children at the same age. I assume this would turn out to be so. Do in negation environments emerging earlier than do in other environments may be generally true of (L1) English acquisition. In the past it was often claimed this is because don’t is an unanalyzed negative marker, i.e., it is not seen as containing do; for extensive arguments against this claim see Schütze (2010). Bohnacker also argues against this unanalyzed don’t approach, in part because Katla is also producing doesn’t (but not didn’t). This is shortly after returning from a month-long trip to Iceland. This situation contrasts with that for English: prescriptivists do not tell English speakers not to overuse do, because this is an error that adult native speakers of standard English do not make. A referee suggests viewing superfluous do as a case of verb doubling, in which case the fact that it comes with complex shades of meaning would fit the general pattern for such constructions (Barbiers et al. 2008).

Superfluous ‘do’ and comparison of spell-outs 31 9. 10. 11. 12. 13.

14. 15. 16. 17. 18. 19. 20. 21.

22.

23. 24. 25.

Apparently, not all British English speakers are comfortable with this participial use of do. Notwithstanding these differences, Thoms (2011) attempts to provide a unified account of British do and standard do-support. Heard on two United Airlines flights, 22 June 2010. Thanks to Barbara Partee for originally drawing such cases to my attention. In contrast, Banks (1994) claims that flight attendants’ speech is characterized by unexpected stress on auxiliaries, including do. For what it is worth, I have a strong intuition that this phenomenon does not occur in environments where another auxiliary is present that should have been finite, as in (2b) and (2c). That is, I suspect you will not hear things like the following on your next airplane trip: (i) a. We do be here primarily for your safety. b. We do have begun our final descent. c. We do will be on the ground shortly. A referee suggests that Please, do come in! fits into this category. This is not to deny that would, could, etc. also have modal uses. I assume those are distinct lexical items whose homophony is due to historical reasons. 3SG T happens to have the allomorph -Ø in the context of a modal, alongside its regular allomorph /-z/, as the ungrammaticality of (i) shows. (i) *John cans run. Ungrammatical on the intended reading ‘It is vital that Mary be able to walk again.’ Ungrammatical on the intended irrealis reading ‘Mary be able to walk again??’ The M head is either absent or radically empty in nonfinite clauses, which cannot contain modals or do, as seen in (18) and (19). For simplicity I assume there is no Agree operation that would allow agreement to be checked without a Spec-Head relation being created. This might seem to be at odds with the claim that this element occurs in questions and imperatives. At least two remedies are possible. First, ‘imperative’ and ‘interrogative’ may be Force heads (in contrast with ‘declarative’) that are compatible with indicative Mood. Second, this may simply be the Elsewhere Mood head, underspecified for whatever differentiates these clause types. In order to account for sentences like John often sings, Bobaljik assumes that adverbs are not visible at the point in the derivation when T-lowering applies, possibly because as adjuncts they are late-merged. That is, at the point in the derivation where T lowers, adjuncts have not yet been merged into the structure and hence do not block adjacency. A referee notes that this correctly predicts the ungrammaticality of *Mary probably too likes John. The T affix will have the modal as a host to its left. A referee asks how we can ensure that MINDIC is pronounced as do in imperatives like Don’t be late! and does not violate the economy condition, which would seem to favor the ungrammatical Be not late! First, I follow the litera-

32

26.

27.

28.

29. 30.

Carson T. Schütze ture in assuming that auxiliaries cannot raise to T in imperatives, for reasons no one seems to understand. T must therefore combine with the verb by lowering when possible. Negation blocks such lowering, in imperatives as elsewhere, so the do allomorph of MINDIC will have to be inserted to support the T affix (which happens to be -Ø in imperatives). Second, the economy condition cannot come into play because it compares only derivations based on the same numeration, and Be not late! would require a different numeration in which a strong feature caused be to raise. The referee also asks what makes interrogatives different from imperatives, such that we find Aren’t you late? and not *Don’t you be late? I appeal to the independent fact that auxiliaries do raise to T in interrogatives. Having done so they block out the do-form of MINDIC because they serve as the morphological host for T, and MINDIC defaults to Ø. The ungrammatical version would involve violating the conditions on vocabulary insertion of do. To my knowledge there are no satisfactory proposals for why be and auxiliary have must raise to T just in case there is no modal present in the clause, but empirically it is clear that this requirement is enforced somehow. Under accounts (such as mine) in which modals are of a category distinct from T, it must be required that T adjoin to M when M is a modal. When that adjunction occurs, T will no longer be available to host V because it has formed a word with M. Convergence is the condition that a derivation meets when all features that required checking have been checked. It may or may not correspond to a grammatical sentence because economy constraints may still rule the derivation out. This is plausibly the same constraint whereby synthetic forms tend to block periphrastic ones, e.g. *more smart is blocked by smarter, on the assumption that more consists of mo+ər, i.e. the comparative suffix plus a stem that constitutes a distinct vocabulary item. Arnold (1995a) also suggests this connection. I am not certain what is meant by ‘decomposition’ here; it seems to suggest that difficulties in analyzing the input are being invoked. Evers and Van Kampen (1995), who describe the same proposal, note the apparent problem that adverbs can intervene between the positions of I and V, e.g., John often sings. They propose that this is consistent with their compounding approach because the adverbs in question are simple heads, and hence can form a complex head I+Adv+V. A second apparent problem they note is that if do-insertion is free, nothing seems to prevent generation of strings like John does not be happy, where negation presumably blocks the compounding rule. They suggest it is a selectional property of do that it cannot take auxiliaries as complements. (This would have to involve NEG being transparent to selection.) But that is not consistent with the behavior of do in imperatives, where we find Do not be late!

Superfluous ‘do’ and comparison of spell-outs 33 31. Evers and Van Kampen (1995: 34) make a further claim: “Suppose that the string adjacency of verbal-like elements in an extended projection invariably leads to a restructuring in a syntactic compound” (emphasis added). It sounds as if they might intend this to be a universal principle.

References Akmajian, Adrian 1984 Sentence types and the form-function fit. Natural Language and Linguistic Theory 2: 1–23. Arnold, Mark D. 1995a Case, periphrastic do and the loss of verb movement in English. Ph.D. dissertation, University of Maryland. 1995b Notions of economy in language change: The spread of periphrastic ‘do’. In Proceedings of the North East Linguistic Society 25, Volume Two: Papers from the Workshops on Language Acquisition and Language Change, Jill N. Beckman (ed.), 121–134. Amherst, MA: GLSA. Banks, Stephen P. 1994 Performing public announcements: The case of flight attendants’ work discourse. Text and Performance Quarterly 14: 253–267. Barbiers, Sjef, Olaf Koeneman, Marika Lekakou, and Margreet van der Ham (eds.) 2008 Microvariation in Syntactic Doubling. Bingley, UK: Emerald Group. Bastiaanse, Roelien, Gerard Bol, Sofie van Mol, and Shalom Zuckerman 2002 Verb movement and finiteness in language impairment and language development. In Clinical Linguistics: Theory and Applications in Speech Pathology and Therapy, Elisabetta Fava (ed.), 119–130. Amsterdam: John Benjamins. Blom, Elma, and Siebe de Korte 2008 De verwerving van het Nederlands: dummies en Verb Second. [The acquisition of Dutch: dummies and Verb Second.] Nederlandse Taalkunde 13: 133–159. Bobaljik, Jonathan David 1994 What does adjacency do? In The Morphology-Syntax Connection, Heidi Harley and Colin Phillips (eds.), MIT Working Papers in Linguistics 22: 1–31. Bohnacker, Ute 1999 Children, tense and auxiliary do. Newcastle and Durham Working Papers in Linguistics 5: 41–74.

34

Carson T. Schütze

Boser, Katharina, Barbara Lust, Lynn Santelmann, and John Whitman 1992

The syntax of CP and V-2 in Early Child German (ECG): The Strong Continuity Hypothesis. In Proceedings of NELS 22, Kimberly Broderick (ed.), 51–66. Amherst, MA: GLSA. Chien, Yu-Chin, and Kenneth Wexler 1990 Children’s knowledge of locality conditions in binding as evidence for the modularity of syntax and pragmatics, Language Acquisition 1: 225–295. Chierchia, Gennaro, Stephen Crain, Maria Teresa Guasti, Andrea Gualmini, and Luisa Meroni 2001 The acquisition of disjunction: Evidence for a grammatical view of scalar implicatures. In The Proceedings of the 25th Annual Boston University Conference on Language Development, Anna H.-J. Do, Laura Domínguez, and Aimee Johansen (eds.), 157– 168. Somerville, MA: Cascadilla Press. Chomsky, Noam 1981 Lectures on Government and Binding. Dordrecht: Foris. Culicover, Peter W. 1999 Syntactic Nuts: Hard Cases, Syntactic Theory, and Language Acquisition. Oxford: Oxford University Press. Davis, Henry Thomas 1987 The acquisition of the English auxiliary system and its relation to linguistic theory. Ph.D. dissertation, University of British Columbia. De Haan, Ger 1987 A theory-bound approach to the acquisition of verb placement in Dutch. In Formal Parameters of Generative Grammar: OTS Yearbook III, Ger de Haan and Wim Zonneveld (eds.), 15–30. University of Utrecht. Ellegård, Alvar 1953 The Auxiliary ‘do’: The Establishment and regulation of its use in English. Stockholm: Almqvist & Wiksell. Emonds, Joseph E. 1970 Root and structure-preserving transformations. Ph.D. dissertation, MIT. 1994 Two principles of economy. In Paths towards Universal Grammar: Studies in honor of Richard S. Kayne, Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi, and Raffaella Zanuttini (eds.), 155–172. Washington, D.C.: Georgetown University Press. Erb, Marie Christine 2001 Finite auxiliaries in German. Ph.D. dissertation, Tilburg University .

Superfluous ‘do’ and comparison of spell-outs 35 Evers, Arnold, and Jacqueline van Kampen 1995 Do-insertion and LF in child language. In OTS Yearbook 1994, Jan Don, Bert Schouten, and Wim Zonneveld (eds.), 25–41. Utrecht: Research Institute for Language and Speech, Utrecht University. Foley, Claire, Julie Pactovis, and Barbara Lust in prep. Links between LF and PF: New evidence from first language acquisition of VP Ellipsis. Ms., Cornell University. Glaser, Elvira, and Natascha Frey 2006 Doubling phenomena in Swiss German dialects. Paper presented at Workshop on Syntactic Doubling, Amsterdam, March 2006. [Downloadable from http://www.dialectsyntax.org/] Grodzinsky, Yosef, and Tanya Reinhart 1993 The innateness of binding and coreference, Linguistic Inquiry 24: 69–101. Gualmini, Andrea, Stephen Crain, Luisa Meroni, Gennaro Chierchia, and Maria Teresa Guasti 2001 At the semantics/pragmatics interface in child language. In Proceedings of SALT XI, Rachel Hastings, Brendan Jackson, and Zsofia Zvolenszky (eds.), 231–247. Ithaca, NY: CLC Publications. Halle, Morris, and Alec Marantz 1993 Distributed Morphology and the pieces of inflection. In The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger, Ken Hale and Samuel J. Keyser (eds.), 111–176. Cambridge, MA: MIT Press. Hollebrandse, Bart, and Thomas Roeper 1996 The concept of DO-insertion and the theory of INFL in acquisition. In Proceedings of the Groningen Assembly on Language Acquisition Held at the University of Groningen, 7–9 September 1995, Charlotte Koster and Frank Wijnen (eds.), 261–271. Groningen, The Netherlands: Centre for Language and Cognition. Jensvoll, Maja 2003 The acquisition of past tense in English/Norwegian bilingual children: Single versus dual mechanisms. In Proceedings of the 19th Scandinavian Conference of Linguistics: Acquisition, Anne Dahl, Peter Svenonius, and Marit Richardsen Westergaard (eds.). Nordlyd 31: 545–557. Jordens, Peter 1990 The acquisition of verb placement in Dutch and German. Linguistics 28: 1407–1448.

36

Carson T. Schütze

Klemola, Juhani 1998 Semantics of DO in South-Western dialects of English. In DO in English, Dutch and German: history and present-day variation, Ingrid Tieken-Boon van Ostade, Marijke van der Wal, and Arjan van Leuvensteijn (eds.), 25–51. Münster: Nodus. Knipschild, Bettina 2007 Verb placement, DO-insertion and object shift: A case study of a bilingual German/English child. Ph.D. dissertation, University of Essex. Laka, Itziar 1990 Negation in syntax: On the nature of functional categories and projections. Ph.D. dissertation, Cambridge, MA: MIT. MacWhinney, Brian 2000 The CHILDES project: Tools for analyzing talk. 3rd edition. Mahwah, NJ: Lawrence Erlbaum. Nevalainen, Terttu, and Matti Rissanen 1986 Do you support the do-support? Emphatic and non-emphatic do in affirmative statements in present-day spoken English. In Papers from the Third Scandinavian Symposium on Syntactic Variation, Sven Jacobson (ed.), 35–50. Stockholm: Almqvist and Wiksell. Palsgrave, John 1530 L’esclaircissement de la Langue Françoyse, François P. Génin (ed.). Paris: Imprimerie Nationale, 1852. Penner, Zvi 1992 The ban on parameter resetting, default mechanisms, and the acquisition of V2 in Bernese Swiss German. In The Acquisition of Verb Placement: Functional Categories and V2 Phenomena in Language Acquisition, Jürgen M. Meisel (ed.), 245–281. Dordrecht: Kluwer. Pollock, Jean-Yves 1989 Verb movement, Universal Grammar, and the structure of IP. Linguistic Inquiry 20: 365–424. 1997 Notes on clause structure. In Elements of Grammar: Handbook in Generative Syntax, Liliane Haegeman (ed.), 237–279. Dordrecht: Kluwer. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Roberts, Ian G. 1993 Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht: Kluwer.

Superfluous ‘do’ and comparison of spell-outs 37 Roeper, Thomas 1991 How a marked parameter is chosen: Adverbs and do-insertion in the IP of child grammar. In Papers in the Acquisition of WH: Proceedings of the University of Massachusetts Roundtable, May 1990, Thomas L. Maxfield and Bernadette Plunkett (eds.), 175– 202. University of Massachusetts Occasional Papers Special Edition. 1999 Universal bilingualism. Bilingualism: Language and Cognition 2: 169–186. Ross, John R. 1972 Doubl-ing. In J. Kimball (Ed.), Syntax and Semantics (Vol. 1, pp. 157–186). New York: Seminar Press. Schönenberger, Manuela 2001 Embedded V-to-C in Child Grammar: The Acquisition of Verb Placement in Swiss German. Dordrecht: Kluwer. Schönenberger, Manuela, and Zvi Penner 1995 Probing Swiss-German clause structure by means of the placement of verbal expletives: Tun “do” insertion and verb doubling. In Topics in Swiss German Syntax, Zvi Penner (ed.), 291–330. Bern: Peter Lang. Schütze, Carson T. 2004 Synchronic and diachronic microvariation in English do. Lingua 114: 495–516. 2010 The status of nonagreeing don’t and theories of root infinitives. Language Acquisition 17: 235–271. Snyder, William 2007 Child Language: The Parametric Approach. Oxford: Oxford University Press. Thoms, Gary 2011 From economy to locality: do-support as head movement. http://ling.auf.net/lingBuzz/001198. Thornton, Rosalind 2010 Verb phrase ellipsis in children’s answers to questions. Language Learning and Development 6: 1–31. Tracy, Rosemarie. 1995 Child languages in contact: The simultaneous acquisition of two languages (English/German) in early childhood. Habilitationsschrift, University Tübingen. Van Kampen, Jacqueline 1997 First steps in Wh-movement. Ph.D. dissertation, Utrecht University. Visser, Fredericus Theodorus 1969 An Historical Syntax of the English Language. Part Three, First Half: Syntactical Units with Two Verbs. Leiden: E.J. Brill.

38

Carson T. Schütze

Warner, Anthony R. 1993 English Auxiliaries: Structure and History. Cambridge: Cambridge University Press. Wijnen, Frank, and Maaike Verrips 1998 The acquisition of Dutch syntax. In The Acquisition of Dutch, Steven Gillis and Annick de Houwer (eds.), 223–299. Amsterdam: John Benjamins. Zuckerman, Shalom 2001 The acquisition of “optional” movement. Ph.D. dissertation, Rijksuniversiteit Groningen. Zuckerman, Shalom, Roelien Bastiaanse, and Ron van Zonneveld 2000 Auxiliary insertion in child Dutch. In WCCFL 19: Proceedings of the 19th West Coast Conference on Formal Linguistics, Roger Billerey and Brook Danielle Lillehaugen (eds.), 631–644. Somerville, MA: Cascadilla Press. Zukowski, Andrea 1996 Auxiliary auxiliaries: Spare parts in early speech. Ms., Boston University.

Dummy auxiliaries in Dutch first language acquisition Shalom Zuckerman

1. Introduction 1.1. Overview This paper presents an investigation of ‘dummy’ auxiliaries in child Dutch. Constructions of the form [subject – dummy-auxiliary – (object) – infinitive], e.g., De jongen doet/gaat een boek lezen ‘The boy does/goes a book read.INF’, are said to be characteristic of preschool children acquiring Dutch as their first language (Blom 2003; Evers and Van Kampen 1995; Hollebrandse and Roeper 1996; Jordens 1990; Roeper 1991; Van Kampen 1997; Zuckerman 2001). These structures are said to be used in a present tense reading, and they are thus identical in meaning to constructions of the form [subject - finite verb - (object)], e.g., De jongen leest een boek ‘The boy reads a book’, in the standard adult grammar. This phenomenon is rather unique in the study of language acquisition as it represents an insertion of an extra element rather than an omission or substitution. Previous studies of this phenomenon propose different hypotheses regarding the function and the origin of dummy auxiliaries, which can be divided into three approaches: an input-based approach according to which dummy auxiliaries are the result of similar structures that appear in children’s input, a structure-based approach according to which dummy elements are inserted to fulfill a structural (syntactic) demand, and a combined input/structure approach according to which both the input and (syntactic) structure determine the use of dummies. In this paper I review the main findings and claims of these approaches and focus on Zuckerman’s (2001) third option which combines both input and structure as the source of dummy auxiliaries in child Dutch. I will review and discuss two experiments from Zuckerman (2001) and then present findings of a new experiment with 51 Dutch preschool children, which aims to support the combined input/structure approach.

40

Shalom Zuckerman

The two experiments from Zuckerman (2001) are a sentence completion (production) experiment in matrix and embedded clauses and a picture selection (comprehension) experiment which examines the interpretation that children assign to the inchoative auxiliary gaan ‘go’. Inchoative gaan is grammatical in adult Dutch (e.g., Ik ga kleuren; I go draw ‘I am going to draw’) but appears in child language as a dummy auxiliary. The new experiment compares children’s performance on these two tasks (production and comprehension). The results of Zuckerman’s (2001) production experiment confirm the existence and the age range of the dummy auxiliary phenomenon and reveal two additional findings: The use of dummy auxiliary is restricted to matrix clauses and the type of the auxiliary that is used is dependent on regional-dialectal factors. The results of Zuckerman’s (2001) comprehension experiment show that young children fail to assign an inchoative interpretation to the auxiliary gaan ‘go’ and instead interpret it as denoting an ongoing event in the present. The new experiment aims to investigate a crucial question: Are children’s production of dummy auxiliaries and their misinterpretation of inchoative gaan-structures related? The assumption is that only children who misinterpret inchoative gaan produce dummy auxiliaries. The following analysis is proposed: Children interpret Aux+INF structures and finite Verb Second (V2) structures as semantically equivalent alternatives. As predicted by the structure-based approach (e.g., Van Kampen 1997), due to economy considerations, dummy auxiliary structures appear in child Dutch in matrix clauses where they are less costly than the finite alternative, but not in embedded clauses in which they are more costly. And as predicted by the input-based approach (Jordens 1990), only when children realize the future denoting interpretation of gaan in their input, they leave the dummy auxiliary stage and fully acquire verb movement and verb inflection. The proposed account is in accordance with (and offers support to) the more general claim that when facing two semantically equivalent derivations, children will apply economy principles to choose between them (Clark and Roberts 1993; Roberts 1998; Williams 1997; Zuckerman 2001). 1.2. The phenomenon It has been observed that early child Dutch includes sentences such as (1) below, which are ungrammatical in the adult standard language (Evers and

Dummy auxiliaries in L1 Dutch

41

Van Kampen 1995; Hollebrandse and Roeper 1996; Jordens 1990; Roeper 1991; Van Kampen 1997): (1)

Ik doe ook verven. I do also paint.INF ‘I also paint.’/‘I’m also painting.’ (Niek 3;10.2, CHILDES, Wijnen Corpus)

In principle, doen ‘do’ is not an auxiliary verb in standard Dutch, except in specific contexts such as V(erb)P(hrase)-topicalization (see Reuland 1983). It is however characteristic of some dialects of Dutch, mainly in the southern regions of the Netherlands (e.g., Cornips 1998; Barbiers, this volume).The standard form of (1) in the adult language is: (2)

Ik verf ook. I paint.FIN also ‘I also paint.’/‘I’m also painting.’

The use of such doen-constructions is said in the literature to constitute a stage in child grammar, roughly characteristic of ages two to four, during which children use both the correct finite form as well as doen with an infinitive verb.1 Several authors point out that children of these ages make use of other auxiliaries (or modals) in the same manner, such as gaan (‘go’) as shown in (3) below. (3)

Ik I

ga kleuren. go color.

(Niek 3;10.2, CHILDES, Wijnen Corpus)

The fact that children produce sentences such as (3) is not surprising, since (3) is grammatical in Dutch under an inchoative reading (‘I’m going to draw’). The claim, however, and this will be the claim in this paper as well, is that children often do not produce (3) in the adultlike aspectual reading but rather as a description of an ongoing event (i.e., ‘I’m drawing’). That is, in child language, gaan- and doen-constructions are considered to differ only with respect to the auxiliary that is used. Most existing accounts of this phenomenon relate it to the process of acquiring verb movement. Before turning to a review of existing analyses and to the current proposal, let us take a look at the properties of verb movement in Dutch that are relevant to this phenomenon.

42

Shalom Zuckerman

1.3. Verb movement in adult and child Dutch Two asymmetries characterize verb movement in Dutch. The first is the matrix/embedded asymmetry, demonstrated in (4) below, and the second is the finite/non-finite asymmetry, demonstrated in (5) below. (4)

The matrix vs. embedded word order asymmetry in Dutch a. De man leesti een boek ti the man reads a book vs. *De man een boek leest. the man a book reads b. *dat de man leesti een boek ti that the man reads a book vs. dat de man een boek leest that the man a book reads ‘that the man reads a book’

(5)

The finite vs. non-finite word order asymmetry in Dutch . a. De man leesti een boek ti The man reads a book vs. *De man een boek leest The man a book reads b. *De man gaat lezen een boek The man goes read.INF a book ‘The man wants to/is going to read a book’ vs. De man gaat een boek lezen.INF The man goes a book read.

The asymmetry in (4) concerns verb movement rather than verb finiteness, since the verb leest (‘reads’) is finite in both matrix and embedded contexts while only in the matrix clause it is moved. The asymmetry in (5) reflects the relationship between verb movement and verb inflection – when the verb in the matrix clause is finite, it moves to the second position (V2) and when it is nonfinite it remains in situ (in the latter case the second position must be occupied by a finite auxiliary or modal verb). Research has shown that Dutch-speaking children master the asymmetry in (5) from the earliest

Dummy auxiliaries in L1 Dutch

43

recorded stages (Wijnen and Verrips 1998). Even in early two-word utterances, children produce virtually no nonfinite verbs in a pre-object position and no finite verb in a post-object position. It might be concluded from this finding that Dutch-speaking children master verb movement from the earliest stages (and that the relevant parameter is either never mis-set or set to the correct value in a stage previous to the onset of speech). Nevertheless, an examination of the two different asymmetries in (4) and (5) above shows that mastery of the movement-finiteness asymmetry does not tell the whole story. It has yet to be established that children make a distinction between the position of the finite verb in matrix clauses in comparison to its position in embedded clauses. Two problems must be considered in order to answer this question. The first is that most spontaneous speech corpora include only a small number of embedded clauses and a proper comparison of them to matrix clauses is not possible. The second problem, which brings us back to the subject of the current study, is that if children use a dummy gaan-structure in order to avoid movement of the lexical verb in matrix clauses, this will only rarely be detected since these structures are identical to the grammatical inchoative structures. One can rarely establish the intention of the child when she utters a structure such as Ik ga kleuren (‘I go draw’). Even if the child uses ga as a dummy auxiliary and actually says “I am drawing”, and not “I am going to draw”, the investigator will most likely miss that intention and judge the sentence as being grammatical. Investigating dummy auxiliary structures is therefore important, not only as a stand-alone phenomenon but also as a crucial piece in the puzzle of the acquisition of verb movement and verb inflection. 1.4. Previous accounts: Three approaches to the dummy auxiliary phenomenon 1.4.1. The input approach Jordens (1990) observed the use of dummy doen and dummy gaan in the spontaneous speech of his 2˗2;6-year old child. He analyzed the use of dummy auxiliary constructions and compared them to the production of lexical finite verbs in second position and to nonfinite verbs in final position. Based on this comparison, Jordens made the observation that the use of a finite auxiliary cannot be taken to indicate the acquisition of V2, but rather indicates the acquisition of finiteness. He showed that increase in the

44

Shalom Zuckerman

production of dummy auxiliary structures was not accompanied by an increase in the use of lexical verbs in second position (which remained stable during the ages of his study), but rather by a decrease in the use of root nonfinite verbs in final position. At 2;6, Jordens’ subject produced 59.4% Modal/Aux+INF structures, 31.2 % lexical verbs in non-final position and less than 10% root infinitives. Based on an analysis of the dummy auxiliary structures, Jordens claimed that there is an increasing tendency to use dummy auxiliary patterns instead of the correct equivalent with verb movement. Jordens made a distinction between the realization of a finiteness feature per se and the realization of it through verb movement. He claimed that children already know that the finiteness feature must be realized, but do not yet understand the associated verb movement. According to Jordens, the Aux+INF phenomenon is thus an intermediate stage between an early stage, which shows no knowledge of finiteness at all (characterized by root infinitive clauses) and a final stage, which shows realization of both finiteness and movement. Jordens further observed that doen- and gaan-structures also occur in the input, doen as part of (some) caregivers’ speech and gaan as a structure with an aspectual meaning of ‘is going to’. This observation is crucial to the current discussion. Lalleman (1986) claimed that caregivers make use of gaan structures to indicate present tense as well (see also: Cornips, this volume). Jordens offered a semantic account to postulate when children abandon the Aux+INF structures and reach the final stage: “The periphrastic use of doet and gaat+INF will decrease in favor of systematic verb fronting as soon as the child acquires a sense of the semantic difference between a particular Aux+INF pattern and its corresponding V-finite alternative” (Jordens 1990: 1437). In more recent versions of this approach (Jordens 2002; Jordens and Dimroth 2006; Jordens this volume) it is proposed that the development of the auxiliary system and specifically the identification of the aux(iliary) + past participle is the trigger for the acquisition of finiteness. In these versions Jordens places his account within a larger information-structural framework that includes not only the acquisition of tense but also other properties such as aspect and negation. In its base, however, it remains a semantic-based approach in which the correct semantic interpretation of auxiliary structures serves as the bootstrap for acquiring syntactic movement.

Dummy auxiliaries in L1 Dutch

45

1.4.2. The structure-based approach Hollebrandse and Roeper (1996) proposed an account for do-insertion in Dutch and English. Their data also included cases of ‘tense doubling’ (see (6) below) but they did not mention cases such as gaan-insertion. (6)

I didn't broke this

(Sarah 3;10:16, Stromswold 1990)

The analysis they proposed to account for these data was that children are unable at this stage to establish the proper syntactic relations that license the inflected verb and therefore they employ a ‘least effort’ strategy, preferring to spell out the tense morpheme rather than the required verb movement operation. Van Kampen (1997) reported similar findings with respect to dummy doen/gaan-structures based on her investigation of the spontaneous speech of two children during the ages of 1;9 to 10;0 and 1;7 to 7;0. The account Van Kampen proposed for the dummy auxiliary phenomenon is a ‘leasteffort’ account, which holds that children initially prefer to avoid verb movement (to the higher C(omplementizer) position) by the insertion of do and modal constants as a ‘least effort’ performance strategy. This account views the dummy auxiliary phenomenon as an intermediate stage that precedes the final stage in which the child realizes the full function of the C position in Dutch. It is when children clearly establish the V2 rule (finite verbs must move to second position) that the use of dummy auxiliary structures will decrease (although they are not completely abandoned). Unlike Jordens (1990), Van Kampen assumed that doen and gaan do not function as carriers of finiteness markings. She described them (following De Haan 1987) as modal constants, which are not yet identified as verbs. This is a crucial difference between the two accounts. According to Jordens (1990), children at this stage have already realized the relationship between the second position and finiteness, while according to Van Kampen, this relationship has yet to be established. In recent work, however, Jordens proposed that the functional stage, in which finiteness is expressed morphosyntactically, is preceded by a lexical stage without any functional categories. The transition from the first to the second stage is assumed to be triggered by auxiliaries (Jordens, 2002; Jordens and Dimroth, 2006, Jordens, this volume). Van Kampen noticed that the ‘least effort’ approach leads to a prediction: Dummy auxiliary structures should not occur in embedded clauses, since movement is not required in embedded clauses in Dutch. She further

46

Shalom Zuckerman

observed that in her spontaneous speech data embedded clauses do not contain Aux+INF structures. This prediction is supported by Zuckerman’s (2001) experimental data below. According to the structure-based approach the Aux+INF phenomenon is not primarily input-related and the decrease in the occurrence of this structure is dependent on syntactic rather than semantic discoveries made by the child. 1.4.3. The combined structure/input approach Zuckerman (2001) proposed a combined approach where both input and economy considerations play a role. According to this approach children ‘mistakenly’ consider gaan-structures that appear in their input to be identical in interpretation to the standard verb movement option, and thus consider these a grammatical option for describing an ongoing event. As mentioned before, this interpretation can be strengthened by such use in childdirected speech as observed by Jordens and Lalleman. As for doenstructures, Zuckerman’s claim was that in this early stage children fail to properly distinguish the dialect from the standard language. In both cases therefore, children are facing the same problem of having two different word orders that yield the same interpretation. Children’s solution to this problem is to reject the possibility of true optionality and prefer the more economical structure, namely the dummy auxiliary structure, which includes no verb movement. Regarding structure, this approach assumes that children are aware of the two available options (using a dummy auxiliary or inflecting the verb) and they prefer the first in matrix clauses and the second in embedded clauses, following economy considerations. Regarding input, this approach assumes that when children understand the obligatory inchoative reading of gaan-structures in the standard language, they stop producing the Aux+INF structure. To conclude, we have seen three attempts to account for the dummy auxiliary phenomenon, all of which consider it to be characteristic of an intermediate stage in child grammar. Jordens’ analysis can be described as input-based and related to the interpretation children assign to the sentences they hear. Van Kampen’s structure-based account is not input-related and considers this stage to be driven by economy. Zuckerman’s combined account adopts both these assumptions: Children produce dummy auxiliary structures because of misinterpretations of the input, but they use them as an alternative only when this is economically preferable.

Dummy auxiliaries in L1 Dutch

47

Zuckerman (2001) presents two experiments to support the combined approach. Because of their importance to the current study, I will review these experiments in detail in Section 2 below. These experiments result in four claims regarding dummy auxiliaries in Dutch first language acquisition and they lead to a prediction that is tested in this paper – the four claims and one prediction are summarized in the next subsection. 1.5. Main claims of the combined structure/input approach – and a prediction Zuckerman (2001) reports two experimental investigations of dummy auxiliary structures resulting in the following four claims: (i) (ii)

(iii) (iv)

Dummy auxiliary structures are characteristic of a stage in early Dutch. Older preschool children produce almost no Aux+INF structures. The type of auxiliaries used by the child in these structures depends on input-related factors. That is, in environments where doen is allowed in the dialect, children use the doen-structure in the dummy auxiliary stage, while in environments providing standard Dutch input, the gaan-structure is used. Dummy auxiliary structures appear in child Dutch only in matrix clauses and not in embedded clauses. Dutch speaking children below four years of age view the gaan+INF structures that appear in their input as ambiguous, having a presentongoing event reading or the standard inchoative reading.

Claims (i) to (iii) above were tested by Zuckerman (2001) with a sentence completion (production) experiment with children of two age groups from the northern and the southern parts of the Netherlands. Claim (iv) was tested by Zuckerman (2001) with a picture selection task (comprehension) experiment. These four claims support the combined structure/input approach. The fact that regional considerations affect children’s performance (claim ii) and the fact that children of the same age range fail to interpret inchoative structures (claim iv) support the crucial role of the input in the account. The fact that dummy auxiliary structures appear indeed only in matrix clauses (claim iii) support the need for a structure-based component in the account.

48

Shalom Zuckerman

Zuckerman’s experiments are presented in detail in the following section as Experiment 1 (production) and Experiment 2 (comprehension). However, there is a missing link between accepting these four claims and supporting Zuckerman’s 2001 combined account. In order to claim that children produce dummy auxiliaries because they misinterpret gaanstructures in their input it needs to be shown that these two phenomena are related. More precisely, I need to show that it is the misinterpretation that leads to dummy auxiliary production and that these two phenomena do not accidentally co-occur in the same age range. A prediction, therefore, of the combined approach that still awaits confirmation is the following: (v)

A correlation exists between children’s tendency to produce dummyauxiliaries and their failure to interpret gaan+INF structures as denoting a future event.

Prediction (v) will be tested with a combined production-comprehension experiment with a single group of children. 2. Zuckerman’s (2001) experimental results In this section I review Zuckerman’s (2001) results that came to support the combined structure/input-based approach and gave rise to claims (i)-(iv) above. The experiments are reviewed in detail since they form the basis of the new experimental investigation presented in this paper (Experiment 3 below). However, for a full review and discussion of the experiments, see Zuckerman (2001). 2.1. Experiment 1: Zuckerman’s (2001) production experiment This production experiment aims to confirm the existence of the dummy auxiliary stage (through a comparison of pre-school and older children) and to reveal both the structure-related environment in which it appears (Dutch matrix vs. embedded clauses) and the input-related environment in which it appears (dialect vs. standard Dutch regions). A note on the Limburg dialect In the Limburg dialect, spoken in the south-east of the Netherlands, doenstructures (as in (7), below) are considered grammatical and are attested in

Dummy auxiliaries in L1 Dutch

49

adult language, although they are usually described as childish and are characteristic of child-directed speech (see also Cornips, this volume). This is also the case in other regional dialects (e.g., Brabants, see Barbiers et al. 2005; Cornips 2011; Giesbers 1983). An observation crucial to our discussion here is that the Limburg dialect also allows doen-structures to occur in embedded clauses as demonstrated by grammaticality judgments of two native speaker of the Limburg dialect in Venlo (where the experiment was conducted), brought in (8) below. (7)

Limburgs: Standard Dutch:

Ik doon speulen. *Ik doe spelen I do.1SG play.INF ‘I play’ or ‘I am playing.’

(8)

Limburgs: Standard Dutch:

As dich duis valle. *Als je doet vallen. If you do.1SG fall-INF. ‘If you fall.’

2.1.1. Subjects 24 Dutch speaking children were divided into two age groups. The younger group included ten children: half of them from Limburg – in the south of the Netherlands and half from Groningen – in the north – (ages 3;0–3;11, mean age: 3;5). The older group included 14 children, all from Limburg (ages 4;8–5).2 The children from Limburg were all exposed to the Limburg dialect at home and in their day care center; they were productive users of both the dialect and standard Dutch. The children from Groningen were exposed only to standard Dutch. It is important to note that the experiment was conducted in standard Dutch for all children. 2.1.2. Materials and procedure A sentence completion test was used, in which the subject had to produce a finite verb and an object. Two pictures were presented to the subject. The experimenter presented the pictures through a coordination structure in which the first conjunct was fully produced by the experimenter and the second conjunct was truncated. The subjects were asked to complete the sentence (that is, to produce the finite verb and the object). The sentences were divided into two conditions: matrix sentences (VO condition, as in

50

Shalom Zuckerman

(9a)) for which the correct answer contained a finite verb and an object, and embedded sentences (OV condition, as in (9b)) for which the correct answer contained an object and a finite verb (i.e., the reversed order). (9)

a.

b.

Matrix condition Experimenter: Deze man snijdt het brood en deze man ... This man cuts the bread and this man..... Subject: snijdt de tomaat. [VO] cuts the tomato [VO] Embedded condition Experimenter: Dit is de man die het brood snijdt en dit is de man die ... This is the man who the bread cuts and this is the man who ... Subject: de tomaat snijdt.[OV] the tomato cuts [OV]

In the VO-condition (9a), the verb has been moved (V2), while in the OVcondition (9b), it is in its base-generated position. 34 picture pairs were presented, half in the VO-condition and half in the OV-condition. 2.1.3. Results The results of the sentence completion experiment are presented in Table 1 below. An analysis of the difference between children’s responses to the VO condition and to the OV condition reveals a significant difference. Table 1. Dummy aux production in matrix and embedded clauses (younger children, i.e., aged between 3;0 and 3;11) Sentence type Matrix

Total responses 145

Correct completion 71 (49%)

Dummy auxiliaries 33 (23%)

Other responses 41 (28%)

Embedded

141

126 (89%)

4 (3%)

11 (8%)

While in the OV-condition children produced 126 correct responses out of 141 items (89%), in the VO-condition they produced only 71 out of 145 items (49%) correctly. This significant difference (t(9)=-5.48, p past errors where the picture representing a past event was chosen. In fact, as can be seen in Figure 2, the future > present error was made more often than any other error. An LSD post-hoc test shows that future > present category significantly differs from each of the other categories (F(7)=18.10, p present) and future interpreted as past (future > past)), we see a significant difference between the two age groups: While the younger children produced more future > present errors than future > past errors, (53% vs. 14%; t(27)=6.591, p present, 11% for future > past; t(18)=.742, p=.47). This suggests that the older children who performed poorly on this structure did not do so because they mistakenly judged gaat-structures to indicate the present tense, contrary to what we have concluded for the younger children. A second difference between the two age groups lies in the responses of the individual subjects. While the younger children did not differ from each other in their responses, the older children did. In the younger group all the children made similar percentages of future > present errors. In the older group, on the other hand, only three of the 19 children are responsible for the majority of these errors in their group. For an elaboration on this point see Zuckerman (2001). The differences between the two age groups are crucial in supporting the validity of this experiment: Although this task is not easy, the older children’s performance shows that the items depicted the different timeaspects correctly and elicited an adultlike performance. 2.2.4. Conclusion The results of Experiment 2 support claim (iv) of the combined approach (Section 1.5 above): Children misinterpret gaan-structures in their input and allow such structures to have a present tense reading as well as a future reading characteristic of the adult grammar. This misinterpretation represents a stage in child Dutch and disappears at later ages. This finding adds support to the input component of the combined approach, but, as argued before, it is not sufficient. Namely, it could be the case that the two stages ˗ the production of dummy auxiliaries and the misinterpretation of gaan-

60

Shalom Zuckerman

structures ˗ only accidentally co-occur between ages three to four years old, but are not really related to each other. The next step, therefore, in supporting the input component of the combined structure/input approach is to show that a correlation exists between children’s tendency to produce dummy auxiliaries and their tendency to misinterpret gaan-structures in their input. A new experiment, presented as Experiment 3 below combines a production and a comprehension task with a single group of children and investigates whether such a correlation exists. 3. Experiment 3: A combined production/comprehension experiment The hypothesis for this experiment is that the scores on a production test (as in Experiment 1 above) negatively correlate with the scores on a comprehension test (as in Experiment 2 above). Specifically, children who score highly on the dummy auxiliary production test will score low on a gaan-structures comprehension test. In order to exclude confounding effects, it is important to show that such a correlation is independent of age. Furthermore, we must remember that although the scores on both the production and the comprehension tasks appear to be gradual in nature – it is the assumption here that they represent two stages. Regarding production, a dichotomy is assumed. A child is or is not in the dummy auxiliary stage. Regarding comprehension she is supposed to be either at what we can call here the ‘misinterpretation of gaan-structures stage’ or not. This point is elaborated on in Section 3.3.3 below, together with its operational implications. The aim of this experiment is not to re-test the claims supported by the first two experiments – this is why it was not necessary to include children of different regions or sentences of various structures. The only purpose here is to test within a single group of children whether a relation exists between the production of dummy auxiliaries and the misinterpretation of gaan-structures. 3.1. Subjects A group of 51 children aged 2;10 to 5;5 (mean age: 4;4) participated in the experiment. All of them were from Utrecht – a province in the middle of

Dummy auxiliaries in L1 Dutch

61

The Netherlands, and none of them spoke a dialect apart from standard Dutch. 3.2. Procedure The experiment included two tasks which were presented to each child in the same order – first a production task followed by a comprehension task. Because the comprehension task included stimuli of the gaan-structure form, it was not possible to change the order of the tasks as this might have biased the child towards a dummy auxiliary response in the production task. The three tasks were presented in one session by two experimenters – one introducing the tasks and the stimuli and the other noting the child’s responses. 3.2.1. Production task A sentence completion task similar to the task used in Experiment 1 above was used. However, in order to improve the present-ongoing aspect of the stimuli, we chose to use movies rather than pictures. In each trial, the child was exposed to two movies presented simultaneously on a computer screen. Both movies showed a different woman performing an action during about four seconds. Both movies were presented in a ‘loop’ creating a continuous action. For example, one film shows a woman eating an apple and the other film shows a woman jumping up and down on a trampoline. The experimenter would prompt a response by saying: (11) Deze vrouw eet een appel en deze vrouw …… This woman eats an apple and this woman …… The child would then answer: springt (op de trampoline). jumps

[a correct response]

gaat (trampoline) springen. goes jump

[a dummy auxiliary response]

or:

62

Shalom Zuckerman

Both the ongoing filmed action as well as the coordination structure elicited a context in which a gaat springen-response is clearly not adultlike. That is, the action loop in the video fragments presents the event as ongoing and the coordination structure in which the verb of the first conjunct is a finite present-tense verb creates a clear bias towards a finite verb for the second conjunct as well. Twelve movie pairs were presented. The child’s response – inflected verb, dummy aux, or other – was noted by the second experimenter leading to a score of zero to twelve for production of dummy auxiliaries. 3.2.2. Comprehension task Based on a pilot study it was decided to use pictures rather than movies for the comprehension task, since in this task three pictures are presented for each trial. As in Experiment 2, one of the three pictures depicted an ongoing action (e.g., A woman is jumping), a second picture depicted the same action as it is about to happen (e.g., Another woman is about to jump). Unlike in Experiment 2, the third picture functioned as a filler depicting another action (e.g., A woman is eating). Experiment 2 above showed that children do not have a problem with interpreting a past tense structure (i.e., they do not have a general preference for ongoing pictures) and therefore this experiment concentrated on future/inchoative interpretations. The experimenter presented the child with a question, for which the child was asked to point to one of the three pictures. (12)

Welke vrouw gaat springen? Which woman goes jump?

[test item]

Welke vrouw springt? Which woman jumps?

[control item]

or: (13)

For the test item, choosing the picture depicting a future event was considered as the correct answer. Nine test items and four control items were presented for every child leading to a score of zero to nine for correctly interpreting gaan-structures as referring to future events.

Dummy auxiliaries in L1 Dutch

63

3.3. Results 3.3.1. Production task The production experiment resulted in 438 correctly inflected responses, 76 dummy gaan-structures and eight doen-structures (all from the same child – we will have to assume that a southern grandparent contributed to this exceptional tendency). Other responses included uninflected lexical verbs (45), aan het-structures (14), zit te-structures (4) and 35 “don’t know” or no responses. All the 76 dummy gaan-structures were produced by only 16 of the 51 children. The mean number of gaan-responses for the 51 children was 1.5 (SD=3.0). 3.3.2. Comprehension task 44 of the 51 children completed the comprehension task choosing correctly three or four control items and without choosing the filler pictures, thus showing an ability to perform the task. For these 44 children, correlations were calculated. In response to the nine gaan+INF questions, children’s correct answers ranged from zero to nine (three children), with a mean of 4.4 correct responses per child (SD=2.2). The individual production and comprehension results for these 44 children are presented in in the appendix. 3.3.3. Correlations A Pearson correlation test between production of gaat+INF and the interpretation of inchoative gaat reveals a moderate negative correlation which approaches statistical significance (n=44, r=-.28, p=.065). The correlation with age is rather strong for both variables (production, r=-.518, p Aux sein/Aux haben > dummy verb and interindividual variation for child L2 acquisition : Faruk: copula/dummy verb/Aux haben > Aux sein Zeren: copula > dummy verb/Aux sein > Aux haben Gül: copula > Aux sein Ne: Me:

copula > dummy verb > Aux haben > Aux sein copula > dummy verb > Aux sein/Aux haben

For the older children, dummy verbs have a pragmatic as well as a structure-building function. The data of Me and Ne reveal that in late child L2 acquisition, dummies are a link to V2, but not to finiteness. The Turkish learners of German produce both dummies with ist and – much less frequently – with machen, and

Dummy verbs in L1 and L2 acquisition of German

237

dummy verbs precede the establishment of auxiliary sentences, and therefore targetlike verbal brackets, in these children. With the dummy verb constructions, Ne and Me create a mould for German sentence structure into which, at a later time, lexical verbs (and targetlike verbal brackets with modals and auxiliaries) can be plugged. Under communicative pressure, they put together what is easily processable, namely automatized formula, e.g., based on the das ist-chunk. This yields a beneficial side-effect, the partitioning of learning tasks, and a stepwise acquisition process. The acquisition of SVA follows the acquisition of V2 in the older children, leading to the impression of a structure-building function of dummy verbs in child L2 acquisition/AO6. In contrast to our findings with respect to AO3 children, dummy verbs work as a lexical linking element in child L2 acquisition (AO6), as has also been reported for adult L2 acquisition (Dimroth 2008; Jordens and Dimroth 2006; Verhagen 2009). Following these assumptions, dummy verbs work as an orientation on the linear structure of German input in child L2 acquisition/AO6 while L1 acquisition and child L2 acquisition/AO3 children seem to rely on an innate domain-specific knowledge of hierarchical relations. Thus, further evidence for an effect of early vs. late AO to the second language in the acquisition of German sentence structure and subject-verb agreement (cf. Chilla 2008; Meisel 2009) may emerge from the study of dummy verb use. Children like Zeren, who still acquires these aspects of second language grammar after four years of age, may profit from the dummy verb option, and hers and Me’s and Ne’s data indicate an increasing role of dummy verbs with slower development and later input in child L2 acquisition. In accordance with this hypothesis, the much slower development of Me due to less favorable input conditions yields an extensive and persistent use of dummy verbs, whereas in the overall rather fast acquisition process of Ne, the dummy verb stage covers only a clear-cut and short time span. To support and strengthen these findings, a comparison of these findings to an extended corpus of children with different first languages and ages of onset as well as elicitation tasks is necessary. Acknowledgments

This research has been supported by the German Science Foundation (DFG) by research grants dedicated to Monika Rothweiler (project E4 within the Research Centre on Multilingualism, SFB 583, 2003-2011). We thank Heide Wegener and Gisela Szagun for the possibility to use the Augsburg corpus and the Oldenburg Corpora. We are grateful to this generous support, and to the helpful comments of the reviewers.

238

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff

Appendix A. Raw scores and percentages of subject-verb agreement -0

-e

-0 3SG

-st

-t

-en

total

forms of sein

L1 acquisition Anna 1 Anna 2

?1/1 100%

?0/1 0%

1/2 50% 1/1 100%

1/1? 100%

Anna 3 Anna 4 Anna 5 Anna 6 Anna 7 Anna 8 Anna 9 Anna 10 Anna 11 Anna 12 Anna 13 Anna 14 Anna 15 Anna 16 Anna 17

0/11 0% 8/15 53,3% 12/15 80% 10/14 71,43% 22/52 42,31% 57/71 80,28% 33/40 82,5% 33/39 84,62% 52/63 82,54% 53/58 91,38% 54/55 98,18% 28/29 96,55% 69/71 97,18%

0/1 0% 0/1 0% 0/2 0%

1/1 100% 2/3 66,67% 1/1 100% 1/1 100% 1/1 100%

3/4 75%

2/2 100%

5/5 100% 18/18 100% 8/8 100% 27/27 100% 24/24 100% 53/53 100% 18/18 100% 22/22 100% 11/11 100% 7/7 100% 9/9 100% 14/14 100%

1/1 100% 2/2 100% 1/1 100% 8/9 88,88% 13/13 100% 17/18 94,44% 19/19 100% 8/8 100% 17/17 100% 21/21 100% 9/9 100% 18/18 100%

1/1 100% 12/13 92,31% 16/26 61,53% 19/26 73,08% 15/16 93,75% 24/25 96% 27/27 100% 36/36 100% 18/18 100% 33/33 100% 20/20 100% 27/27 100% 41/41 100%

0/3 0% 3/4 75% 12/13 92,3% 1/4 25% 2/4 50% 7/9 77,77% 11/12 91,67% 12/12 100% 11/12 91,67% 19/19 100% 14/15 93,33% 11/11 100% 69/69 100%

1/16 6,25% 29/39 74,36% 62/78 86,11% 39/53 73,59% 75/109 68,81% 127/145 87,58% 142/151 94,04% 119/125 95,2% 112/124 90,32% 133/138 96,38% 119/122 97,54% 84/85 98,82% 213/215 99,07%

2/2 100% 6/8 75% 11/13 84,62% 9/9 100% 37/40 92,5% 45/45 100% 22/22 100% 26/26 100% 29/30 96,67% 35/35 100% 31/32 96,88% 62/62 100%

1/1 100% 2/2 100%

9/9 100% 14/15 93,33% 52/55 94,55%

14/16 87,5% 47/54 87,04%

Falko 1-5 Falko 6 Falko 7 Falko 8

6/6 100% 7/8 87,5% 12/15 80%

4/4 100% 16/16 100%

1/1 100%

3/3 100% 1/1 100% 22/22 100%

Dummy verbs in L1 and L2 acquisition of German

Falko 9 Falko 10 Falko 11 Falko 12 Falko 13 Falko 14 Falko 15 Falko 16 Falko 17

-0

-e

-0 3SG

-st

-t

-en

total

forms of sein

13/16 81,25% 13/21 61,90% 10/14 71,43% 37/41 90,24% 29/30 96,67% 45/59 76,27% 75/85 88,24% 95/96 98,96% 133/141 94,33%

1/1 100%

22/22 100% 40/40 100% 24/24 100% 15/15 100% 19/19 100% 41/41 100% 20/20 100% 39/39 100% 29/29 100%

1/1 100% 2/2 100% 9/9 100% 2/2 100% 10/10 100% 25/25 100% 12/12 100% 43/43 100% 32/32 100%

41/42 97,62% 33/33 100% 32/32 100% 24/26 92,31% 24/25 96% 52/52 100% 51/53 96,23% 85/85 100% 72/73 98,63%

2/3 66,67% 15/16 93,75% 26/26 100% 28/30 93,33% 36/36 100% 13/14 92,86% 3/5 60% 67/68 98,53% 50/51 98,04%

80/85 94,12% 104/112 92,86% 101/105 96,18% 109/117 93,16% 121/123 98,37% 183/198 92,42% 164/178 92,13% 335/337 99,41% 321/331 96,98%

27/27 100% 69/74 93,24% 22/24 91,67% 19/21 90,48% 50/51 98,04% 57/57 100% 59/65 90,77% 47/47 100% 163/165 98,79%

3/3 100% 3/3 100% 7/7 100% 3/3 100% 6/6 100% 7/7 100%

Sören 1 Sören 2 1/1 100%

Sören 3

0/1 0%

1/2 50%

1/1 100% 3/5 68% 4/4 100% 10/12 83,33% 28/31 90,32% 27/28 96,43% 45/45 100% 42/44 95,45% 21/21 100%

6/7 85,71% 14/18 77,78% 18/20 90% 68/73 93,15% 205/214 95,79% 279/286 97,55% 228/231 98,7% 317/323 98,14% 205/208 98,56%

Sören 4 Sören 5 Sören 6 Sören 7 Sören 8 Sören 9 Sören 10 Sören 11 Sören 12 Sören 13 Sören 14

239

2/3 66,67% 2/3 66,67% 5/7 71,43% 41/45 91,11% 64/70 91,43% 51/53 96,23% 59/60 98,33% 43/45 95,56%

0/1 0% 0/1 0%

3/3 100%

1/2 50%

5/5 100% 44/44 100% 38/38 100% 82/82 100% 52/52 100% 73/73 100% 88/88 100%

10/11 90,91% 15/15 100% 14/14 100% 33/33 100% 7/7 100%

5/5 100% 7/7 100% 7/8 87,5% 29/30 96,67% 85/86 98,84% 91/91 100% 66/67 98,51% 110/113 97,35% 45/45 100%

5/5 100% 22/22 100% 11/11 100% 7/8 87,5% 6/7 85,71% 15/15 100% 16/16 100% 48/48 100% 78/83 93,98% 132/132 100% 88/89 98,88% 118/122 96,72% 116/119 97,48%

240

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff -0

Sören 15 63/67 94,03% Sören 16 42/47 89,36% Sören 17 35/37 94,59%

-e

-0 3SG

-st

-t

-en

total

forms of sein

3/4 75% 1/1 100% 2/2 100%

52/52 100% 40/40 100% 49/49 100%

21/21 100% 17/18 94,44% 9/9 100%

91/91 100% 49/51 96,08% 36/37 97,3%

60/62 96,77% 26/27 96,3% 18/19 94,74%

290/297 97,64% 175/184 95,11% 149/153 97,39%

102/103 99,03% 48/48 100% 41/42 97,62%

child L2 acquisition/AO 3 Alev 1

17/24 70.8%

4/4 100%

15/16 93.7%

12/12 100%

9/9 100%

6/9 66.7%

63/74 85.1%

31/31 100%

Alev 2

3/5 60% 19/20 95% 12/16 75% 23/26 88.5% 42/42 100% 9/24 37.5% 3/6 50% 7/18 38.9% 6/8 75% 1/1 100% 1/1 100% --

13/13 100% 11/11 100% 12/12 100% 23/24 95.8% 46/46 100% 8/8 100% 1/1 100% 1/1 100% 15/15 100% 2/2 100% 1/1 100% --

2/2 100% 1/2 50% 3/4 75% 1/1 100% --

5/5 100% 8/8 100% 1/1 100% 3/3 100% 3/3 100% -1/1 100% 2/2 100% 8/8 100% --

9/9 100% 7/7 100% 6/6 100% 10/10 100% 11/11 100% 5/5 100% 2/2 100% 20/22 90.1% 21/21 100% --

3/3 100% 7/7 100% 10/10 100% 10/10 100% 13/13 100% 0/6 0% 0/1 0% 4/5 80% 12/16 75% --

--

--

--

--

--

--

2/2 100% --

--

--

--

--

Gül 6

1/1 100% 4/4 100% --

1/1 100% --

--

--

--

1/1 100% --

35/37 94.6% 57/59 96.6% 44/49 89.8% 70/74 94.6% 115/115 100% 27/48 56.3% 10/15 66.7% 34/49 69.4% 65/71 91.5% 4/5 80% 2/2 100% 1/1 100% 3/7 42.9% 5/5 100% --

16/16 100% 13/13 100% 6/6 100% 29/29 100% 18/18 100% 23/24 95.8% 2/3 66.7% 39/39 100% 3/3 100% 8/8 100% 5/5 100% 2/2 100% 38/38 100% 2/2 100% --

Gül 7

--

--

--

Gül 9

--

--

--

1/1 100% --

6/7 85.7% 2/2 100% --

--

--

2/2 100% --

--

Gül 8

4/4 100% 1/1 100% --

Alev 3 Defne 1 Defne 2 Defne 3 Faruk 1 Faruk 2 Faruk 3 Faruk 4 Gül 1 Gül 2 Gül 3 Gül 4 Gül 5

5/5 100% 3/4 75% 0/1 0% 3/3 100% 1/2 50% --

0/4 0% --1/2 50% ---

14/14 100% 18/18 100%

Dummy verbs in L1 and L2 acquisition of German

Gül 10 Gül 11 Zeren 1 Zeren 2 Zeren 3 Zeren 4 Zeren 5

-0

-e

-0 3SG

-st

-t

-en

total

forms of sein

4/4 100% 6/6 100% --

3/3 100% 4/4 100% --

2/2 100% 1/1 100% --

--

2/2 100% 10/10 100% --

0/1 0% 0/2 0% --

11/12 91.7% 21/22 95.5% --

--

0/1 0% 2/3 66% 4/6 66% 7/8 87.5%

1/1 100% 2/2 100% 1/1 100% 1/1 100%

--

--

--

--

1/1 100% --

1/1 100% 9/9 100% 1/1 100% --

2/3 66.7% 14/17 82.4% 8/11 72.7% 9/17 52.9%

1/1 100% 31/31 100% 7/7 100% 26/26 100%

1/5 20% 57/73 78,08% 60/68 88,24% 67/70 95,71% 51/51 100% 201/204 98,53% 154/156 98,72% 228/246 92,68% 269/273 98,53%

1/1 100% 25/28 89,29% 62/75 82,67% 40/47 85,11% 8/10 80% 9/11 81,82% 7/8 87,5% 34/36 94,44% 70/81 86,42% 34/35 97,14% 56/56 100%

1/2 50% 1/1 100%

---

--

0/2 0% 1/1 100% 0/7 0%

--

child L2 acquisition/AO6 Ne 1 (6) Ne 2 (7) Ne 3 (9) Ne 4 (10) Ne 5 (13) Ne 6 (14) Ne 7 (15) Ne 8 (16) Ne 9 (17) Ne 10 (18) Ne 11 (19)

Me 1 (6) Me 2 (7)

241

1/3 33,33% 21/23 91,3% 8/11 72,73% 5/5 100% 4/4 100% 22/26 84,62% 8/8 100% 8/13 61,54% 62/65 95,38%

0/1 0% 6/7 85,71% 5/6 83,33% 6/6 100% 2/2 100% 7/7 100% 1/1 100%

8/8 100%

3/3 100%

0/1 0%

9/9 100% 2/2 100% 29/29 100% 12/12 100%

0/1 0% 0/1 0% 10/11 90,91% 0/1 0%

26/34 76,47% 46/47 97,87% 50/52 96,15% 36/36 100% 149/149 100% 68/69 98,55% 127/135 94,07% 117/117 100%

0/1 0% 1/6 16,67% 1/4 25% 6/7 85,71% 9/9 100% 12/12 100% 75/75 100% 54/58 93,1% 70/70 100%

1/3 33,33% 2/2 100%

242

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff -0

-e

-0 3SG

-st

-t

-en

total

Me 3 (8) Me 4 (9) Me 5 (10) 0/2 0% Me 6 (11) 0/1 0% Me 7 (12)

0/4 0% 1/1 100% 0/1 0%

0/2 0%

2/2 100%

Me 8 (13) Me 9 (14) Me 10 (15) Me 11 (16) Me 12 (17) Me 13 (19)

0/4 0% 1/1 100% 2/3 66,67% 1/3 33,33%

Me 14 (20) Mit Test 0/4 0% Me 15 (21) Me 16 (22)

3/5

Me 17 (24)

2/2 100% 20/20 100%

Me 18 (26)

0/2 0% 2/2 100% 4/4 100% 13/13 100% 0/1 0% 2/4 50% 1/1 100% 13/13 100%

3/3 100%

1/1 100%

0/4 0% 9/10 90% 2/3 66,67% 6/8 1/2 50% 0/1 0%

2/2 100% 22/22 100%

2/2 100%

1/1 100% 14/14 100% 25/25 100% 27/28 74/80 29/44 65,91% 44/60 73,33% 21/24 43/43 100% 38/43 102/125 81,6%

0/1 0% 2/2 100% 1/2 50% 0/11 0% 0/5 0% 0/3 0% 0/6 0% 5/11 45,45% 6/11 0/2 0% 9/9 100% 15/19

4/12 33,33% 2/4 50% 0/12 0% 0/5 0% 1/4 25% 14/26 53,85% 33/39 84,62% 39/50 97/108 89,81% 41/58 70,69% 67/95 70,53% 23/27

forms of sein 1/1 100% 25/26 96,15% 15/16 93,75% 2/2 100%

4/5 80% 4/4 100% 35/37 6/8 15/15 100% 20/21

15/17

74/79

2/2 100% 7/8

3/3 100% 24/26 92,3%

45/50

14/18

173/198 87,37%

94/100 94%

Dummy verbs in L1 and L2 acquisition of German

243

Appendix B. Raw scores of dummy verbs, copular verbs, and auxiliaries subject-verb n dummy agreement verb >90% L1 acquisition

n copula verbs n Aux sein

n Aux haben dummy machen

Anna 1-7

-

0

40

2

5

0

Anna 8

-

3

37

0

1

0

Anna 9

-

0

15

4

8

0

Anna 10

-

1

64

1

9

0

Anna 15

+

1

91

2

12

0

Falko 1-8

-

0

132

3

16

0

Falko 9-15

+

0

476

19

108

0

Falko 16

+

1

71

3

39

0

Falko 17

+

2

206

8

65

0

Sören 1-7

-

0

100

2

3

0

Sören 8-9

+

2

85

3

4

0

Sören 10

+

2

137

5

9

0

Sören 11-17 +

0

896

53

153

0

child L2 acquisition~AO3 Faruk 1

-

2

34

0

3

0

Faruk 2

-

1

15

1

0

1

Faruk 3

-

8

34

0

0

1

Faruk 4

+

3

8

1

2

1

Zeren 1

-

0

0

0

0

0

Zeren 2-3

-

0

22

0

0

0

Zeren 4-5

-

11

19

13

0

2

0

33

0

0

0

child L2 acquisition~AO6 Ne1-2

-

244

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff n copula verbs n Aux sein

n Aux haben dummy machen

Ne3

subject-verb n dummy agreement verb >90% 24

85

0

0

4

Ne4

-

14

56

1

14

3

Ne 5

+

10

12

1

0

5

Ne 6-7

+

2

134

6

64

2

Me 1-3

-

0

4

0

0

0

Me 4-13

-

141

107

10

14

3

Me 14-17

+

2

38

2

5

0

Notes 1. Schwartz and Sprouse (1994) cite a sentence, which, in our judgment, can be qualified as a dummy verb construction without commenting on it. The following utterance was produced by an adult learner with L1 Turkish: (i) Der ist aussteigen. he is get off.INF ‘He is getting off (the bus).’ The same holds for some instances of dummy verbs in the data of Diehl et al. (2000) not analyzed as such, for example: (ii) Ich mache Flöte. I make.1SG flute ‘I play the flute.’ 2. In L2 learners with an OV in their L1, SOV is observed as well, see Keim (1984). 3. The numbers refer to sentences and questions with overt subjects. 4. Me has less contact to German speaking peers than Ne: Her family plans to return to Turkey in the near future and her temperament is, so to speak, less ambitious. 5. There are only few instances of other forms of sein or other dummies (habst, a nonexisting form of haben, and forms of machen).

Dummy verbs in L1 and L2 acquisition of German

245

References Chilla, Solveig 2008 Erstsprache, Zweitsprache, Spezifische Sprachentwicklungsstörung? Eine Untersuchung des Erwerbs der deutschen Hauptsatzstruktur durch sukzessiv-bilinguale Kinder mit türkischer Erstsprache. [First language, second language, specific language impairment? An investigation of the acquisition of the German main clause structure by successive bilingual children with Turkish as mother tongue.] Hamburg: Dr. Kovač. Chilla, Solveig, and Matthias Bonnesen 2011 A crosslinguistic perspective on child SLA: The acquisition of questions in German and French. Linguistische Berichte 228. Clahsen, Harald 1988 Parametrized grammatical theory and language acquisition. A study of the acquisition of verb placement and inflection by children and adults. In Linguistic theory in second language acquisition, Susanne Flynn, and Wayne O’Neil (eds.), 47˗75. Dordrecht: Kluwer. Clahsen, Harald, Jürgen Meisel, and Manfred Pienemann 1983 Deutsch als Zweitsprache. Der Spracherwerb ausländischer Arbeiter. [German as a second language. The language acquisition of foreign workers.] Tübingen: Narr. Clahsen, Harald, and Pieter Muysken 1986 The accessibility of Universal Grammar to adult and child learners. A study of the acquisition of German word order. Second Language Research 2: 93–119. 1989 The UG paradox in L2 acquisition. Second Language Research 5: 1– 29. Clahsen, Harald, and Martina Penke 1992 The acquisition of agreement morphology and its syntactic consequences: New evidence on German child language from the Simonecorpus. In The acquisition of verb placement. Functional categories and V2 phenomena in language acquisition, Jürgen Meisel (ed.), 181–223. Dordrecht: Kluwer. Den Besten, Hans 1977 On the interaction of root transformations and lexical deletive rules. In On the formal syntax of the Westgermania, Werner Abraham (ed.), 47–131, Amsterdam: Benjamins. Diehl, Erika, Helen Christen, Sandra Leuenberger, Isabella Pelvat, and Thérèse Studer 2000 Grammatikunterricht: Alles für der Katz? Untersuchungen zum Zweitspracherwerb Deutsch. [Teaching grammar: Strictly for the

246

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff

birds? Investigations of second language acquisition of German.] Tübingen: Niemeyer. Dimroth, Christine 2008 Age effects on the process of L2 acquisition? Evidence from the acquisition of negation and finiteness L2 German. Language Learning 58: 117–150. Dimroth, Christine, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren 2003 Finiteness in Germanic languages: A stage-model for first and second language development. In Information structure and the dynamics of language acquisition, Christine Dimroth, and Marianne Starren (eds.), 65–94. Amsterdam: Benjamins. Golberg, Heather, Johanne Paradis, and Martha Crago 2008 Lexical acquisition over time in minority first language children learning English as a second language. Applied Psycholinguistics 29: 41˗65. Haberzettl, Stefanie 2003 ‘Tinkering’ with chunks: Form-oriented strategies and idiosyncratic utterance patterns without functional implications in the IL of Turkish speaking children learning German. In Information structure and the dynamics of language acquisition, Christine Dimroth, and Marianne Starren (eds.), 45˗64. Amsterdam: Benjamins. 2005 Der Erwerb der Verbstellungsregeln in der Zweitsprache Deutsch durch Kinder mit russischer und türkischer Muttersprache. [The acquisition of verb placement rules in German as a second language by children with a Russian or Turkish mother tongue.] Tübingen: Niemeyer. 2006 Progression im ungesteuerten Erwerb und in Lehrwerken. In Kinder mit Migrationshintergrund – Spracherwerb und Fördermöglichkeiten. [Progress in untutored and tutored acquisition. In Children with a migration background – language acquisition and options of language training], Bernt Ahrenholz (ed.), 199˗216. Freiburg: Fillibach. Hollebrandse, Bart, and Tom Roeper 1996 The concept of do-insertion and the theory of INFL in acquisition. Proceedings of the Groningen Assembly on Language Acquisition held at the University of Groningen, 7–9 September 1995, Charlotte Koster, and Frank Wijnen (eds.), 269–271. Groningen: Centre for Language and Cognition. Huebner, Thom 1989 Establishing point of view: The development of coding mechanisms in a second language for the expression of cognitive and perceptual organization. Linguistics 27: 111˗143. Jordens, Peter, and Christine Dimroth

Dummy verbs in L1 and L2 acquisition of German 2006

247

Finiteness in children and adults learning Dutch. In The acquisition of verbs and their grammar: The effect of particular languages, Natalia Gagarina, and Insa Gülzow (eds.), 173–200. Dordrecht: Springer.

Keim, Inken 1984 Untersuchungen zum Deutsch türkischer Arbeiter. [Investigations of German spoken by Turkish workers]. Tübingen: Narr. Klein, Wolfgang, and Clive Perdue 1997 The Basic Variety (or: Couldn’t natural languages be much simpler?). Second Language Research 13: 301–347 MacWhinney, Brian 2006 The CHILDES project: Tools for analyzing talk, 3rd edition. Mahwah, NJ: Lawrence Erlbaum Associates. Meisel, Jürgen 1997 The acquisition of the syntax of negation in French and German: Contrasting first and second language development. Second Language Research 13: 227–263. 2008 Child second language acquisition or successive first language acquisition? In Current trends in child second language acquisition: A generative perspective, Belma Haznedar, and Elena Gavruseva (eds.), 55–80. Amsterdam: Benjamins. 2009 Second language acquisition in early childhood. Zeitschrift für Sprachwissenschaft 28: 5–34 Müller, Natascha 1993 Komplexe Sätze. Der Erwerb von COMP und von Wortstellungsmustern bei bilingualen Kindern (Französisch/Deutsch) [Complex sentences. The acquisition of COMP and word order variation by bilingual children (French/German)]. Tübinger Beiträge zur Linguistik Series A: Language Development 16. Tübingen: Narr. Parodi, Teresa 1998 Der Erwerb funktionaler Kategorien im Deutschen. Eine Untersuchung zum bilingualen Erstspracherwerb und zum Zweitspracherwerb. [The acquisition of functional categories in German. A study of bilingual first and second language acquisition.] Tübingen: Narr. 2000 Finiteness and verb placement in second language acquisition. Second Language Research 16: 355˗381. Pienemann, Manfred 1998 Language processing and second language development: Processability Theory. Amsterdam: John Benjamins.

248

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff

Prévost, Philippe 2003 Truncation and missing inflection in initial child L2 German. Studies in Second Language Acquisition 25: 65˗97. Prévost, Philippe, and Lydia White 2000 Missing Surface Inflection or Impairment in second language acquisition? Evidence from tense and agreement. Second Language Research 16: 103˗133. Roeper, Thomas 1991 How a marked parameter is chosen: Adverbs and do-insertion in the IP of child grammar. In Papers in the acquisition of WH: Proceedings of the University of Massachussets Roundtable, May 1990, Thomas L. Maxfield, and Bernadette Plumkett (eds.), 175–202. Amherst, Massachussets: Graduate Linguistic Student Association. Rothweiler, Monika 1993 Der Erwerb von Nebensätzen im Deutschen.[The acquisition of subordinate clauses in German.] Tübingen: Niemeyer. 2006 The acquisition of V2 and subordinate clauses in early successive acquisition of German. In Interfaces in multilingualism: Acquisition, representation and processing, Conxita Lleó (ed.), 91˗113. Amsterdam: Benjamins. Schwartz, Bonnie, and Rex A. Sprouse 1994 Word Order and Nominative Case in Non-Native Language Acquisition. In Language acquisition studies in generative grammar, Teun Hoekstra, and Bonnie D. Schwartz (eds.), 317˗368. Amsterdam/ Philadelphia: Benjamins. Starren, Marianne 2001 The Second Time: The Acquisition of Temporality in Dutch and French as a Second Language. Ph.D. dissertation, Tilburg University. Schulz, Petra, Rosemarie Tracy, Ramona Wenzel 2008 Entwicklung eines Instruments zur Sprachstandsdiagnose von Kindern mit Deutsch als Zweitsprache: Theoretische Grundlagen und erste Ergebnisse. [Development of a language assessment instrument for children with German as a second language: Theoretical basis and first results.] In Kinder und Jugendliche mit Migrationshintergrund – Empirische Befunde und Forschungsdesiderate [Children and adolescents with a migration background – Empirical findings and research wishes], Bernd Ahrenholz (ed.), 17˗41. Freiburg im Breisgau: Fillibach. Szagun, Gisela 2004 German – Szagun. Talk Bank. 2006 Sprachentwicklung beim Kind [Child language development]. Weinheim: Beltz.

Dummy verbs in L1 and L2 acquisition of German

249

Thoma, Dieter, and Rosemarie Tracy 2006 Deutsch als frühe Zweitsprache: zweite Erstsprache. [German as early second language: second first language.] In Kinder mit Migrationshintergrund. Spracherwerb und Fördermöglichkeiten [Children with a migration background: Language acquisition and options of language training], Bernt Ahrenholz (ed.), 58˗79, Freiburg im Breisgau: Fillibach. Tracy, Rosemarie 1991 Sprachliche Strukturentwicklung: Linguistische und kognitionspsychologische Aspekte einer Theorie des Erstspracherwerbs. [Development of language structures: linguistic and cognitivepsychological aspects of a theory of first language acquisiton.] Tübingen: Narr. 2002 Growing (clausal) roots: all children start out (and may remain) multilingual. Linguistics 40-4: 653˗686. Tracy, Rosemarie, and Dieter Thoma forthc. Convergence on finite V2 clauses in L1, bilingual L1 and early L2 acquisition. In Functional categories in learner language, Christine Dimroth, and Peter Jordens (eds.), 1˗44. Berlin: Mouton de Gruyter. Van de Craats, Ineke 2009 The role of ‘is’ in the acquisition of finiteness by adult Turkish learners of Dutch. Studies in Second Language Acquisition, 31: 59˗92. Van Kampen, Jacqueline 1997 First steps in WH-movement. Ph.D. dissertation, University of Utrecht. Vainikka, Martha , and Martha Young-Scholten 1994 Direct access to X’-theory. Evidence from Korean and Turkish adults learning German. In Language acquisition studies in generative grammar, Tom Hoekstra and Bonnie Schwartz (eds.), 265˗316, Amsterdam/Philadelphia: John Benjamins. 1998 The initial state in L2 acquisition of phrase structure. In The generative study of second language acquisition, Suzanne Flynn, Gita Martohardjono, and Wayne O’Neil (eds.), 17˗34. Mahwah, NJ: Erlbaum. Verhagen, Josje 2009 Finiteness in Dutch as a second language. Ph.D. dissertation. Free University Amsterdam and Max-Planck-Institute for Psycholinguistics.

250

Solveig Chilla, Stefanie Haberzettl and Nadja Wulff

Wegener, Heide 1992 Kindlicher Zweitspracherwerb. Untersuchungen zur Morphologie des Deutschen und ihrem Erwerb durch Kinder mit polnischer, russischer und türkischer Erstsprache. Eine Längsschnittuntersuchung. [Child second language acquisition. Studies on German morphology and its acquisition by children with a Polish, Russian and Turkish first language. A long term study.] Habilitationsschrift, Universität Augsburg. Wexler, Ken 1994 Finiteness and head movement in early child grammars. In Verb movement, David Lightfoot, and Norbert Hornstein (eds.), 305˗350. Cambridge: Cambridge University Press. Wode, Henning 1983 Papers on language acquisition, language learning and language teaching. Heidelberg: Groos. Zuckerman, Shalom 2001 The acquisition of ‘optional’ movement. Ph.D. dissertation. University of Groningen. Zuckerman, Shalom, Roelien Bastiaanse, and Ron van Zonneveld 2001 Verb movement in acquisition and aphasia: Same problem, different solutions – evidence from Dutch. Brain and Language 77: 449˗458.

Dummy auxiliaries in children with SLI – a study on Dutch, in monolinguals and bilinguals Jan de Jong, Elma Blom and Antje Orgassa

1. Introduction Specific language impairment (SLI) is a disorder that, while heterogeneous, typically has significant repercussions for a child’s morphosyntactic abilities. While symptoms are most often found in grammatical morphology and syntax, across languages there is variation regarding what is the most problema-tic area. In Germanic languages (more so than in Romance languages) the impairment particularly affects verb inflection (Leonard 1998; 2009). In this chapter we will demonstrate that problems with verb inflection in Dutch SLI may show (i) in errors in choosing the correct inflectional form but also (ii) in ‘avoidance’ of the inflected verb form by using a grammatically correct, but less complex, alternative. That Dutch-acquiring children with SLI have difficulties at using finite verb inflection (i.e., tense and agreement marking) is a well-known fact. Evidence for difficulties with finiteness marking in Dutch SLI was presented in several studies (De Jong 1999; Steenge 2006; Verhoeven, Steenge and Van Balkom 2011; Wexler, Schaeffer, and Bol 2004). In those studies, children with SLI were shown to be significantly less accurate at using the correctly inflected form than the typically developing (TD) control groups. For the subjects in the present study, data on accuracy in agreement marking were reported by Orgassa (2009). The second option, preferred use of a grammatical but less complex form than the one targeted, was not reported on in detail before. In the present study, which deals with verb inflection in the Dutch output of monolingual and bilingual children with SLI (and controls), children’s use of alternative forms is studied. In an early study on lexicon in children with SLI, Leonard et al. (1982) concluded that “the language-impaired children, like the normal children, were more likely to produce words containing sounds already in their repertoires than words whose sounds were absent from their phonologies” (Leonard et al. 1982: 561). In that context they refer to the children’s ‘selection and avoidance strategies’. In a similar

252

Jan de Jong, Elma Blom and Antje Orgassa

vein, Novogrodsky and Friedmann (2006) mention ‘avoidance of relative clauses’ and ‘avoidance of movement’ in their study on relative clauses in SLI. Our rationale in this study is similar: If there are more or less complex candidates to express a meaning, a child with SLI may resort to the less complex form. Preference for a less complex form is consistent with theories of SLI according to which the processing capacities of children with SLI are more limited than those of their TD peers (for an overview of theories, see Leonard 1998; Marinis 2011). These processing limitations have been used to explain what goes wrong in verb inflection. Leonard (1998) describes the processing demands involved in producing grammatical morphemes as follows: “inflected forms (…) require computations in addition to those required for uninflected words (…) these additional operations can cause problems, given the reduced speed of processing in children with SLI” (Leonard 1998: 249). The “extra operations involved (…) increase the likelihood that it (i.e., the inflected form, dJ, B & O) will be reduced to its bare-stem counterpart” (Leonard 1998: 249). We will argue that the ‘additional operations’ required by a morphologically complex form can invoke other strategies to avoid inflecting a verb, besides bare stem use. The particular strategy we focus on in this contribution is children’s use of socalled dummy auxiliaries followed by an infinitival lexical verb, that is, use of a periphrastic verb construction. To summarize, the processing limitations of children with SLI may determine their success at using inflected lexical verbs. The prediction we add is that limited processing capacities cause children with SLI not only to make errors, but also to take recourse to the use of verb constructions (AuxV) that are less taxing in terms of processing demands than inflected lexical verbs. This hypothesis – that dummy auxiliaries are less complex than finite lexical verbs – requires explanation. Therefore, before we discuss the role of auxiliaries in acquisition and in the output of children with SLI, we will first illustrate what is meant by the term dummy auxiliary, focusing in particular on why the label ‘dummy’ is used. 2. Dummy auxiliaries in Dutch Crystal (2008) defines a dummy as “a formal grammatical element introduced into a structure or an analysis to ensure that a grammatical sentence is produced (…) dummy elements have no meaning – they are semantically empty (…) e.g., ‘do’ in question forms is a dummy auxiliary, which carries

Dummy verbs in L1 and L2 acquisition of German

253

the tense-number contrast for the verb phrase” (Crystal 2008: 158˗159). The essential elements of this definition are, first, the grammatical role of dummies, i.e., they make a sentence grammatical, and second, their lack of semantic content. While the first characteristic applies to Dutch auxiliaries (sentences need tense carriers), the second characteristic is dependent on context. In our data, two auxiliaries predominate that can be analyzed as dummies, gaan (‘go’) and doen (‘do’). Use of these two Dutch auxiliaries with and without full semantics will be illustrated in the following paragraphs.1 Gaan The auxiliary gaan (‘go’) has an inchoative aspectual meaning that approximates future tense (Haeseryn et al. 1997). In child utterances (1) and (2), however, both referring to an ongoing action, it is merely a tense carrier without the meaning it has in the adult use of the same verb. If (2) was used while watching a picture sequence to label a picture that ‘precedes’ the action of brushing, the auxiliary would not be a dummy. (1)

Die man gaat ook een zon tekenen. that man go.FIN also a sun draw.INF ‘That man will also draw a sun.’ (boy, SLI, monolingual, 6;3) 2

(2)

(Het meisje) gaat z’n tanden borstelen. the girl go.FIN his teeth brush.INF ‘The girl will brush his teeth.’ (boy, SLI, monolingual, 7;0)

Doen Doen (‘do’) as an auxiliary is only used in certain regiolects of Dutch (Cornips 1998; see also Barbiers, this volume) as well as in child-directed speech (which means it is often part of the input) (Schaerlaekens and Gillis 1991). Haeseryn et al. (1997) describe it as a regional, colloquial form. Although they point out that doen does not add to the meaning of the predicate, they also suggest that – in adult use – it is usually limited to contexts in which the action is habitual. This limitation is not observed in child data.3 An example of dummy use of doen is found in (3). (3)

De jonge doet ook mooi schoen maken. the boy do.FIN also beautiful shoe make.INF ‘The boy also makes a beautiful shoe.’ (boy, TD, bilingual, 5;1)

254

Jan de Jong, Elma Blom and Antje Orgassa

For the present study, non-dummy auxiliaries, i.e., auxiliaries that fit the context grammatically as well as semantically will be excluded. The rationale is that there is no reason to claim that such auxiliaries are solely used to avoid inflection of a lexical verb. A sentence with an auxiliary with full semantics means something else than the same sentence with the lexical verb inflected. This is not true for dummies. In the previous section we have associated the use of ‘less complex forms’ with the impact of processing demands. This is a psycholinguistic approach to complexity, but it is important to note that also in formal, structural terms, dummy auxiliary sentences have been considered to be less complex than equivalents with a finite lexical verb. For instance, the analysis of dummy auxiliaries as a ‘less costly’ alternative to the inflection of a lexical verb derives from Van Kampen (1997). Zuckerman (this volume) calls her position a “structure-based approach according to which dummy elements are inserted to fulfil a structural (syntactic) demand”. This formulation (“to fulfil […]”) echoes Crystal’s “to ensure that […]”. More specifically, Van Kampen claims that in using dummy auxiliaries, children avoid movement to the C(omplementizer) position. Zuckerman’s (2001) account does not refer to (psycholinguistic) processing. Nevertheless, the explanations by van Kampen and Zuckerman support the analysis of dummy auxiliary verbs as less complex than inflected lexical verbs. Regarding SLI, Jakubowicz (2011) has argued that derivational complexity is a factor and that this measure of complexity positively correlates with the appearance of symptoms of SLI. We will revisit this rationale in the next section and discuss its relevance for acquisition. 3. The development of finiteness in Dutch and the role of dummies In a number of studies, dummy auxiliary use has been related to the development of the morphosyntactic reflexes of finiteness in Dutch, that is, moving and inflecting the verb. In this section the typical acquisition of finiteness in Dutch will be outlined in order to explain how dummy auxiliary use fits into the development of finiteness in TD children. This, in turn, will inform our later analysis of dummy use in SLI. The acquisition of Dutch finiteness concerns both verb form and verb placement, because in Dutch finiteness is morphologically expressed through inflection and syntactically via a position in the left periphery of the sentence. More specifically, Dutch is a Verb Second (V2) language, in which the finite verb is produced in second position in the main clause. It

Dummy verbs in L1 and L2 acquisition of German

255

has been argued that V2 is the result of movement of the verb from sentence-final position through application of the V2 rule (Haegeman 1994). Initially, children acquiring Dutch do not apply the V2 rule, but produce infinitival verbs (stem+en) in utterance-final position (as in (4)). Gradually, finite verbs do occur in the V2 position. However, these verbs are not necessarily lexical, and in particular in early stages, children show a strong tendency to restrict their use of finite verbs in V2 position to modals and copulas (see in (5)). In this stage, children’s verbs in finite position often do not carry the finite inflection as found in the regular paradigm, because modals and copulas are irregular, and have either impoverished or suppletive inflectional paradigms. Use of predicates with auxiliaries characterizes a typical stage in the acquisition of finiteness in Dutch, in which auxiliaries are being used as tense carriers instead of tense being expressed on the lexical verb (this stage in the acquisition of inflection has been described by several authors, see: Blom 2003; De Haan 1987; Julien et al. this volume; Van Kampen 1997; Wijnen 1998). (4)

Even buiten kijken. just outside look.INF ‘Look outside for a moment.’ (Tim; 2;01.15/2;02.15; De Haan 1987)

(5)

Moet daarin. must.FIN there-in ‘Has to go in there.’

(6)

Wat doe jij # what do.FIN you ‘What do you make?’

(Tim; 2;01.15/2;02.15; De Haan 1987) maken? make.INF? (Abel; 2;7; Bol 1996)

At the same time, children’s use of finite lexical verbs (i.e., with finite inflection) in V2 position increases (Blom and Wijnen, 2013) and inflected lexical verbs replace not only the early infinitival clauses, as in (4), but also the periphrastic, or analytic, constructions, as exemplified in (6). The ‘end state’ of the acquisition of the grammatical marking of finiteness by means of correct verb placement and verb inflection is characterized by a predominant use of inflected lexical verbs in V2 position.4 Clearly, the use of dummies represents a stage in the acquisition of finiteness. For the purpose of the present study, and thus to outline the role of dummy auxiliaries in typical and atypical development, it is crucial that finite lexical verbs and finite auxiliaries do not necessarily appeal to equiv-

256

Jan de Jong, Elma Blom and Antje Orgassa

alent knowledge or to equivalent processing costs. To explain the occurrence of finite auxiliaries in TD monolingual Dutch toddlers, Blom (2003) suggests that “children overuse (periphrastic verbs) as long as they lack knowledge of the grammatical marking of finiteness by verbal inflection and, consequently, verb movement” (Blom 2003: 168). As a ‘strategic’ alternative for the use of finite lexical verbs, dummy use represents a more economical way of forming a grammatical sentence for the following two reasons; one based on linguistic derivation, the other on psycholinguistic processing demands. First, dummy auxiliaries may lack derivational complexity, assuming as has been argued in previous studies – that dummy auxiliaries are inserted in V2 position and not moved to this position (Blom and De Korte 2011; Van Kampen 1997; Zuckerman 2001). As Barbiers (this volume) puts it: “... insertion of a dummy (...) avoids movement, insertion of a trace and reconstruction and is therefore more economical and transparent”. Having a simple, economical and transparent syntactic derivation may have psycholinguistic repercussions for the language user, since such structures would be expected to be less taxing in terms of processing costs. This approach can be framed in a theoretical explanation. Zuckerman (2001; this volume), following Minimalist Framework (Chomsky 1995) assumptions, analyzed ‘dummy auxiliaries’ such as doen and gaan in Dutch as the outcome of an Economy Principle, in this case as evidence for avoiding movement of the finite lexical verb. Since children using the auxiliaries do not yet know the semantic restrictions for use of auxiliaries, using them is a viable route for them: For mature language users the semantics of the auxiliary verbs would make them inadequate in the contexts in which children use them (Blom and De Korte 2011: 907). To exemplify the children’s different interpretation of the verbs’ semantics, Zuckerman (2001) showed in a comprehension experiment that children “misinterpret the gaan structures in their input as indicating present-ongoing events” (Zuckerman 2001: 138; this volume). The assumption that dummy use is more economical than finite verb use is empirically supported by a difference found between two sentence types in Dutch: Dummy auxiliaries are used more often in main clauses, where the finite lexical verbs need to be moved, than in dependent clauses, where finite lexical verbs can remain in final position. The contrast between main and dependent clauses is found in monolingual TD children (Van Kampen 1997; Zuckerman 2001). The inference from these findings is that movement (which is not needed in dependent clauses in Dutch) is costly. Auxiliaries are not moved, but inserted (in which case the subse-

Dummy verbs in L1 and L2 acquisition of German

257

quent lexical verb remains in final position). Following up on our hypothesis that children with SLI will use more auxiliaries, the rationale, following Zuckerman, is that they do so because they do not select the more costly option which involves movement, but instead select the cheaper dummy auxiliary option. In this respect, it is also relevant to mention findings in other studies on Dutch-acquiring children, both with and without SLI, who were found to be less accurate at using finite verb inflection in main clauses than in dependent clauses (Baker et al. subm.; Blom and Baayen 2012).5 Second, dummy auxiliary use might also reduce processing costs. The additional computations Leonard (1998) alludes to when he mentions the processing costs that could be involved in using inflected verbs are in the well-known Words and Rules model described as operations that unify verbal stems with inflectional suffixes (Pinker and Ullman 2002). This has been described in detail for English regular past tense, but the same principles apply to other regularly inflected forms, and thus to most finite verb inflections in Dutch. Unification operations are not needed for irregular forms, and probably also not for regular high-frequency forms (Prasada and Pinker 1993), because these are stored and can thus be retrieved from memory in fully inflected form. Since the dummy auxiliaries in this study are high-frequency forms (Schaerlaekens, Kohnstamm and Lejaegere 1999), it is likely that language users do not compute these, but instead retrieve them from memory. 6 Direct retrieval does not require ‘additional operations’, in Leonard’s terms. 4. Dummy auxiliaries in SLI Children’s selection of dummy auxiliaries rather than finite lexical verbs gains importance since it could be argued that auxiliary use is more frequent in groups of children with SLI. It has been proposed that the locus of the problem in SLI is a limitation in their processing capacity and speed of processing (Leonard 1998; Marinis 2011). The use of dummy auxiliaries in SLI fits theoretical accounts that suggest that language impairment is due to limited processing resources: Any means of decreasing the processing load might be tapped by children with language difficulties and it can be argued that auxiliary use is such a means, as we have explained above. In this section we will review the data available for the extended use of dummy auxiliaries by children with SLI.

258

Jan de Jong, Elma Blom and Antje Orgassa

De Jong (1999) found that monolingual children with SLI, aged between six and 9;2, in past tense contexts, used auxiliary ging(en) (‘went’) plus infinitive more often than chronological age peers. These data were from narrative retelling of a short animation film. However, the data set was small and the study did not focus on auxiliary use, so there was no design that created a sufficient number of contexts for either auxiliaries or inflected lexical verbs. In addition, the narrative mode is known to encourage persistent use of auxiliary gaan (see the fragments from Frog Story narratives referred to in Roelofs 1988, or examples in older bilingual children’s narratives in Cornips 2000). Bastiaanse et al. (2002) made a comparison between children with SLI, younger TD children, and a group of adult aphasic patients. Each child group included ten children and there were six aphasics. To elicit verbs, Bastiaanse et al. used an experimental task in which they presented their subjects with sentences to be completed by a finite verb. It was found that the children with SLI omitted more verbs and produced more word order errors (the nature of these word order errors is not specified in the study). On the other hand, TD children used more dummy auxiliaries than the children with SLI: “The most frequent error the normally developing children produce are so-called ‘auxiliary dummies’” (Bastiaanse et al. 2002: 126). It is fair to claim that dummy auxiliaries make a sentence grammatical (cf. Crystal 2008). Therefore, we prefer not to interpret them as errors, as Bastiaanse et al. (2002) do. Verhoeven, Steenge and Van Balkom (2011) investigated verb morphology in monolingual and bilingual children, with and without SLI. Their focus was on errors in verb marking. In their study, omission of finite verb inflection was found to be an SLI marker. Auxiliary use, however, does not lead to ungrammaticality. Subsequent on that, in the study by Verhoeven et al. no mention is made of how the use of auxiliaries was analyzed (the same goes for the study by Steenge (2006) on which Verhoeven et al. built). The focus of their study was on error frequency, so auxiliaries were not included. Nevertheless, this omission makes it hard to establish the share that dummy auxiliaries have in the total of correctly inflected forms. As explained earlier in this chapter, processing accounts of SLI support the occurrence of omission errors, but also a preference for structures that are less demanding. If auxiliary use is indeed to be interpreted as such a strategy, then it should be included as a variable when analyzing the realization of finite verb inflection on lexical verbs. In including auxiliaries, our claim is that the impairment not only shows in errors, but also in (lack of)

Dummy verbs in L1 and L2 acquisition of German

259

complexity. 7 This is a different position from the one taken by Bastiaanse et al. (2002), who consider the use of dummies erroneous. A second issue that we will take up in this chapter concerns the question whether extended dummy auxiliary use is exclusive to SLI in bilingual contexts. There are observations that suggest that dummy auxiliaries are also frequent in the language of successive bilingual TD children (Blom and De Korte 2011; Lalleman 1984), and it has been argued that here processing issues might be at stake there as well (Blom and De Korte 2011; Fleta 2003). If that is the case, there may be overlap of behaviours in children with SLI and TD successive bilingual children. The possibility of overlap is relevant for an important issue. According to several studies, surface phenomena are shared between typical bilingual and monolingual SLI, in particular the occurrence of (omission) errors. As Håkansson and Nettelbladt (1996) and Paradis and Crago (2000) and Paradis (2005) have argued, these commonalities lead to diagnostic confusion: Where overlap is found, it becomes more difficult to establish which bilingual children must be labelled SLI. To summarize, it is uncontroversial that dummy auxiliary use is a phenomenon typifying (a certain stage of) the acquisition of finiteness, but whether extended use of dummies is a characteristic of the language production of children with SLI and whether dummy auxiliary use is also a characteristic of SLI in bilingual contexts, we do not yet know. Given the difficulties with verb inflection in SLI and the assumption that use of dummy auxiliaries could be a way of sidestepping verb inflection (i.e., of diminishing the processing load), the prediction that dummies will be more frequent in language production of children with SLI is justified and it would be expected that the effect of SLI holds across monolingual and bilingual populations. Given the fact that dummies are hypothetically associated with SLI and with TD bilinguals, the possible effects of SLI and bilingualism have to be separated. Therefore, a four-group design will be used with monolingual and bilingual children, both with and without SLI. Faced with the challenge of overlap between effects of SLI and bilingualism, it is important to observe what happens when both conditions co-occur, i.e., in bilingual children with SLI. Comparisons of monolingual SLI and bilingual TD children cannot avoid a confound: Both language disorder and language status (bilingual versus monolingual) are involved in the comparison. Using the four groups in the present study, the relative influences of bilingualism and language impairment can be teased apart: A potential outcome is that the effects of bilingualism and SLI can be separated. If an SLI effect

260

Jan de Jong, Elma Blom and Antje Orgassa

can be isolated, the confound can be solved. On the other hand, if both effects are found, the confusion, theoretical as well as diagnostic, subsists. 5. Research questions In this study, we test the hypothesis that extended use of auxiliaries (instead of finite lexical verbs) is a characteristic of the language of children with SLI, and investigate whether it characterizes SLI also in bilingual contexts. The assumption (hence: ‘extended’) is that, within the latter group, auxiliary verbs are used beyond the age range in which they are typically found in (monolingual) TD children. The present study includes monolingual and bilingual children with SLI as well as monolingual and bilingual typically developing children. (The children with SLI are chronologically beyond the age range in which dummy auxiliaries are typically found in monolingual TD children.) The key variable is the share that dummy auxiliaries take in the children’s verb realizations. The specific claim here is that dummy auxiliaries offer children a way to reduce the processing demands that finite lexical verbs create. In testing predictions on dummy use, as we will do here, it is important to specify what this means for methodology, in particular for the data analysis. In previous research, the share that dummy auxiliaries take in the output of children with SLI (how many inflected verbs are lexical, how many are auxiliary verbs?) is often excluded from the analysis, so that it cannot be recovered. Specifically, this means that in reality a significant percentage of inflected verbs may be not lexical but auxiliary verbs. Therefore, testing our prediction that in children with SLI the share of dummy auxiliaries is larger, we will divide responses into (analytic) auxiliary + infinitival lexical verb constructions (Aux+INF) and (synthetic) finite lexical verbs. In this study, we will answer the following questions: 1. Is there a group difference between children with SLI and TD children in the percentage of dummy auxiliaries used? (‘SLI effect’) 2. Is there a group difference between monolingual and bilingual children in the percentage of dummy auxiliaries used? (‘bilingualism effect’) Previous studies have indicated that dummy auxiliary use is subject to much variation, between as well as within children (Blom and De Korte 2011; Hulk and Cornips 2005). For this reason, we also wanted to explore individual strategies in dummy auxiliary use in this study. It may be that

Dummy verbs in L1 and L2 acquisition of German

261

dummy use is not typical of the SLI group as a whole, but of individuals with SLI. 3. Are there individual strategies in the use of auxiliaries? In frequency of dummy auxiliary use: (3a) Do all children use dummy auxiliaries, and do they do so with roughly similar frequency? In contexts of dummy auxiliary use: (3b) Are there children who do not show the expected maindependent clause contrast? In case of dummy auxiliary use with novel verbs: (3c) Are there children who might not have productive morphological rules for using finite lexical verbs? Above, we suggested that use of finite verbs requires more processing resources than use of auxiliaries: Use of finite lexical verbs involves verb movement and additional morphological computation that may not yet be fully automatized. Processing capacity and degree of automatization (Phillips 1995) will differ across children, and hence, if such factors are at play, we would expect individual variation (question 3a). A comparison of main and dependent clauses is informative (question 3b) with respect to distinguishing use of dummy auxiliaries due to derivational complexity and morphological computation. In main clauses, derivational complexity and morphological computation are confounded, because in these sentences use of finite lexical verbs may be costly for both reasons. The two can be distinguished by looking at main versus dependent clauses. By investigating use of dummies in dependent clauses (in which the lexical verb is in base position) we can look at morphological forms in isolation. If a child uses dummy auxiliaries in dependent clauses, morphological computation is relevant, and problems occur either due to a lack of knowledge of inflectional rules or to performance issues related to a lack of processing resources or incomplete automatization of the rule. If a child uses an inflected verb consistently in the dependent clause, it can be ensured that knowledge of how to morphologically express finiteness on verbs by using finite inflection is intact. In order to further establish whether or not dummy auxiliary use could be caused by knowledge gaps (i.e., not knowing the rules for finite verb inflection) or performance factors (i.e., degree to which the use of such rules is an automatic process), we also investigated children’s abilities to use finite verb inflection (question 3c). For that purpose, we will refer to

262

Jan de Jong, Elma Blom and Antje Orgassa

data on nonsense verbs. Eliciting novel forms is a well-established way of testing children’s implicit knowledge. 6. Method 6.1. Subjects In this study, four groups of children participated: monolingual children with SLI, (successive-) bilingual TD children, (successive-)bilingual children with SLI and a monolingual TD control group. Besides the exclusionary criteria common to SLI – the requirement that there is normal hearing, no socio-emotional disorder, no cognitive deficit or frank neurological disorder (Stark and Tallal 1981) – the children were diagnosed with SLI on the basis of their language level (the inclusionary criterion) as established by language tests administered by qualified speech therapists. National guidelines for access to special education demand that children perform at least 1.5 standard deviation (SD) below the norm on tests for at least two language domains or at least 2 SD below the norm on an omnibus test (e.g., Taaltest voor Alle Kinderen (TAK; Verhoeven and Vermeer 2001) or Taaltest voor Kinderen (TvK; Van Bon 1982)). The TAK has standardized norms for native Dutch children and for children belonging to families immigrated from the Mediterranean (including Turkey) and Surinam and the Dutch Caribbean (Verhoeven and Vermeer 2001). All children with SLI fit the national criteria mentioned above. It is important to stress that the criteria as used in the Netherlands are equivalent to criteria used elsewhere to verify a diagnosis of SLI. Consequently, the SLI groups in this study are comparable to children with SLI in the literature. All children had a non-verbal IQ in the normal range (>85). Non-verbal IQ was assessed by the SON-R (Tellegen et al. 1998), a standardized Dutch non-verbal intelligence test. All bilingual children came from families that had migrated from Turkey to the Netherlands. The children were born in the Netherlands and exposed to Turkish in their home environment. Exposure to Dutch mainly involved situations outside the home. All children could be considered successive learners of Dutch: They were exposed to Turkish from birth and exposure to Dutch began between the ages of 1;0 and 4;0. Additional information on the children’s proficiency in their first language, Turkish, was obtained from an adapted version of a parental questionnaire, the Anamne-

Dummy verbs in L1 and L2 acquisition of German

263

se Meertaligheid (‘Questionnaire on Multilingualism’; Blumenthal and Julien 2000). The questionnaire was administered to the caretakers of the bilingual children by a native speaker of Turkish. Questions in the first part focused on the history of speech and language development of the child and its closest family members. The second part covered the current bilingual situation and included questions about the child’s use of Turkish, and its exposure to Dutch. There was variation in age of first exposure, and consequently, in length of exposure to Dutch. None of the children in the bilingual TD group had apparent language or speech problems or had undergone speech therapy. All children lived in (or around) three major cities in the western part of the Netherlands. The bilingual children were tested in both Turkish and Dutch. In this paper we focus on Dutch (for results on Turkish, see: De Jong et al. 2010.) Table 1. Participant information

Number Age in months (M, SD; range) Length of exposure in months

MonoSLI 25 86.8 (7.7) 73-96 86.8 (7.7) 73-96

BiTD 20 88.8 (7.1) 71-99 63 (15.8) 39-101

BiSLI 20 96.3 (7.5) 75-101 61.7 (14.8) 41-91

MonoTD 20 58.2 (7.3) 48-71 58.2 (7.3) 48-71

The children in the monolingual SLI (‘MonoSLI’), bilingual TD (‘BiTD’) and bilingual SLI (‘BiSLI’) group were matched on chronological age (see Table 1). For the monolingual TD (‘MonoTD’) control group, children were recruited who were approximately two years younger, and hence, their language abilities could be assumed to match those of the children in the monolingual SLI group (Paradis 2010). Additionally, although the children in the monolingual TD and bilingual TD group have different ages and therefore differ in cognitive maturity, it is important to note that the younger monolingual TD children provide a better match to the bilingual TD children in terms of length of exposure (LoE) to Dutch than agematched monolingual TD children would be. A one-way ANOVA indicated omnibus effects for Age (F(3,81) = 80.97, p [VP SAID]]] ==> [VP SAID]]] ==> Ø

First, the spell-out of the second instance of ‘have’ as a participle is unexpected under this analysis since the two are part of the same finite chain, and a participle is not a finite form. Moreover, the order derived in (9-iv) is

406

Sjef Barbiers

absent in the relevant Dutch dialects, where we find the orders gezegd heeft gehad (‘said.PPart has had.PPart’) and gezegd gehad heeft (‘said.PPart had.PPart has’), both of which cannot be derived in the way given in (9) (see Koeneman, Lekakou and Barbiers 2011). It is therefore doubtful that ‘have’ and ‘be’ dummies in L1 and in the dialects involve the same derivations. There is another argument supporting the claim that the derivations of dummy ‘have’/‘be’ in L1 and ‘have’/‘be’ doubling in south-eastern Dutch dialects cannot be the same. Koeneman et al. (2011) argue that this construction in the south-eastern dialects involves a combination of a functional and a lexical instance of ‘have’ respectively ‘be’. They show that Standard Dutch has a very similar construction: (10) a. b.

Hij is twee keer getrouwd geweest. he is two times married been ‘He has been married twice.’ Hij heeft het raam de hele dag gesloten gehad. he has the window the whole day closed had ‘He had the window closed all day long.’

The lexical participle in (10a) and (10b) is adjectival, as it cannot follow geweest/gehad (see Hoekstra 1984 for this test that distinguishes between adjectival and verbal participles). Koeneman et al. (2011) argue that the lexical participle in the dummy ‘have’/‘be’ construction in the southeastern dialects is also adjectival. Under the assumption that lexical ‘have’ and ‘be’ select an abstract adjectival head A(djective) (cf. Anagnostopoulou 2003 and references cited there), the difference between Standard Dutch and the south-eastern dialects is that this abstract A only allows for target state participles as its VP-complement in Standard Dutch, while no such restriction holds for the south-eastern dialects. One diagnostic for target state participles is that they can be combined with ‘still’ (nog steeds) (cf. Kratzer 2000). The possibility to use ‘still’ distinguishes between participles like getrouwd (‘married’) and gesloten (‘closed’) which are adjectival in Standard Dutch and verbal participles like gevallen (‘fallen’) which are not. (11) a.

Zij she

is is

nog steeds still

getrouwd. married

Dummy auxiliaries in Dutch dialects, L1 and L2 407

b. c.

De deur the door *De the vase

is is vaas is

nog steeds still is still

gesloten. closed nog steeds gevallen. fallen

In Standard Dutch lexical ‘have’/‘be’ only occur with target state participles, while in the dialects lexical ‘have’/‘be’ can occur with any participle. Since lexical ‘have’/‘be’ can occur in the perfect tense this seems to give rise to ‘have’ / ‘be’ doubling, but we are actually dealing with two different instances of these verbs: auxiliary ‘have’/‘be’ and lexical ‘have’/‘be’. Schematically, this would look as in (12). (12) (i) (ii) (iii)

Structure HAVE doubling [HAVELEX [A [VP]]]] [HAVEAUX Structure BE doubling [BEAUX [BELEX [A [VP]]]] Lexical parametrization: A only selects target state participles. Standard Duch: + South-eastern Dutch: -

If the difference between Standard and south-eastern Dutch is indeed a selectional property of an abstract head A then this syntactic variation is reduced to a lexical property of A in accordance with the Minimalist hypothesis that all syntactic variation can be reduced to the Lexicon or PF. From this analysis we can conclude that the double occurrence of ‘have’ and ‘be’ in perfect constructions in the south-eastern Dutch dialects does not involve dummy auxiliaries. It is not clear whether the same can be concluded for the L1 construction. One way to investigate this would be to look at semantic restrictions on the construction with dummy ‘have’/‘be’ in the perfect tense in L1. In the south-eastern dialects, perfect ‘have’/‘be’ doubling always has a superperfect interpretation: The result of the event denoted by the lexical participle no longer holds. Thus the sentence in (13a) necessarily implies that the vase is no longer in a fallen state, unlike the sentence in (13b). (13) a.

De vaas is gevallen geweest ... the vase is fallen been ‘The vase fell.’ Possible continuation: ... and now it is standing upright again.’

408

Sjef Barbiers

b.

Impossible continuation: ... and it is still lying on the ground. De vaas is gevallen ... the vase is fallen ‘The vase fell / has fallen’ Both continuations of (13a) are possible.

If the interpretational restriction in (13a) also holds for perfect ‘have’/‘be’ doubling in L1 then we are not dealing with dummies in L1 either. It is also possible that ‘have’/‘be’ doubling in L1 is more like the German type, where it occurs in dialects that have lost the preterite and replaces the past perfect (see Koeneman et al. forthc.). We would then expect that Dutch L1 perfect ‘have’/‘be’ doubling occurs in an acquisition stage in which the preterite has not been acquired yet. Both issues require further investigation. 2.2.2. Other ‘be’ dummies The example of the dummy ‘be’ construction in Dutch L2 (see Julien et al. this volume; Verhagen, this volume) presented in Section 1 and repeated here does not occur in any of the Dutch dialects. (14)

Papa is niet komen. daddy is not come.INF ‘Daddy does not come.’

Dutch L2

Very close to it in form is the construction in (15a) that most varieties of Dutch have, but this is clearly not an equivalent of the construction in (14), as (15a) obligatorily has an absentive meaning. It can only mean that daddy is absent because he went out to watch a soccer game. This absentive meaning is lacking in L1 and L2 dummy ‘be’, as Dutch L2 sentences with dummy ‘be’ express ongoing events, just like the German L2 example in (15b) from Haberzettl (2003). Also, only a restricted set of verbs can occur in the absentive construction (see Haslinger 2007) and the verb komen (‘come’) that is used in (14) does not belong to that set (15c). 2 (15) a.

Papa daddy

is is

voetballen soccer

kijken. Standard Dutch watch.INF

Dummy auxiliaries in Dutch dialects, L1 and L2 409

b.

c.

Ein Junge ist die Fussball spielen. a boy is.3SG the football play.INF Target: Ein Jungen spielt Fussball. ‘A boy is playing football.’ Papa is werken / schaatsen / *komen / *gaan. daddy is working / skating / coming / going ‘Daddy went out to work / skate / come / go.’

2.3. ‘Go’ dummies Dummy ‘go’ was not tested in the SAND-project but is wellknown to exist in West-Flemish (Haegeman 1990; Van Riemsdijk 2002). An example is given in (16a). ‘Go’-doubling is restricted to motional ‘go’; it is ungrammatical with future ‘go’ (16b). (16) a. b.

Me gingen atent *(goan) picknikken no den lak. we went always go picknick to the lake ‘We used to go on a picknick to the lake.’ K-goan morgen we (*goan) beter zyn. I-go tomorrow surely GO better be ‘I am going to be better tomorrow.’

Dummy ‘go’ is attested in Dutch L2 (17) but Van de Craats and Van Hout (2010) show that it is a different phenomenon from that in dialects. Unlike in West-Flemish, it occurs with other thematic verbs than ‘go’. Also, it does not have a motional interpretation. Rather, it occupies the finite verb position as a semantically empty dummy, as the translation in (17) shows. (17) Ga kijken naar strand. go.STEM look.INF at beach ‘She is looking at the beach.’ (Van de Craats and Van Hout 2010: 474) 3. Dummy auxiliaries, complexity and economy A central issue in the comparison above between L1, L2 and dialectal dummy auxiliaries was whether their occurrence can be explained as a re-

410

Sjef Barbiers

sult of the tendency to reduce derivational complexity, possibly an instance of the Economy Principle in the Minimalist Program (Chomsky 1995) according to which more economical derivations rule out less economical ones. Inserting a dummy auxiliary in a syntactic position would be more economical than moving the main verb to that position and is therefore preferred in certain developmental stages/dialects. It is not immediately clear, however, that dummies are more economical than movement. Economy can be interpreted in at least the following ways (see also Section 1 above): (18)

Interpretations of economy/complexity (i) Movement (internal merge) is more expensive than insertion (external merge). (ii) The fewer steps a derivation has the more economical it is. (iii) Speaker economy versus hearer/learner economy: Speakers and hearers have different demands with respect to economy: Ease of production versus ease of processing/interpretation/ acquisition. A synthetic construction (no dummy) may be easier to produce but more difficult to process/interpret/acquire than an analytic construction (with a dummy). (iv) The smaller the discrepancy between PF and LF, the easier it is to interpret, and thus to acquire a structure. In a construction with a dummy auxiliary the discrepancy between PF and LF would be smaller than in a movement construction.

The interpretations (iii) and (iv) involve the level of performance, and I will have little to say on these. I will concentrate instead on the interpretations (i) and (ii) that involve the computational system, which is part of the competence. Whether a certain derivation is more economical than another very much depends on the theoretical framework adopted. In the pre-Minimalist Principles and Parameters framework (see Chomsky and Lasnik 1995) movement involved displacement of a constituent and insertion of a trace in the base position of the moved element. This is illustrated with the representation of (19a) in (19b). To be able to interpret the lexical properties of the verb, e.g., its relations with its arguments, it has to be reconstructed into its base position within VP (19c). Indeed, insertion of a dummy in the Iposition (19d) avoids movement, insertion of a trace and reconstruction and is therefore more economical and transparent.

Dummy auxiliaries in Dutch dialects, L1 and L2 411

(19) a. b. c. d.

Ik lees een boek-je. I read a book-DIM ‘I am reading a book.’ [IP ik [I lees [VP tIK [DP een boekje [V tLEES ]]]]] I read a book.DIM [IP ik [I tLEES [VP tIK [DP een boekje [V lees ]]]]] [I DO [VP tIK [DP een boekje [V lezen ]]]]] [IP ik

In Minimalism, movement does no longer involve insertion of a trace. The general idea behind this is the Inclusiveness Condition (Chomsky 1995: 228). A derivation starts with an array of lexical elements (the Numeration). The Inclusiveness condition states that it is impossible to add elements to the structure during the derivation that are not in the Numeration. Insertion of a trace would be a violation of this condition. Movement, now called internal merge, is the result of a three-step process. A constituent that is already present in the structure is copied, then moved and attached to the root and finally, one of the two copies has to be deleted at PF, usually the lower one. The obligatoriness of deletion of one of the copies has been derived by Nunes (2004) from the Antisymmetry theory of correspondence between hierarchy and linear order proposed in Kayne (1994), according to which two identical elements in two distinct positions cannot be linearized. The Minimalist derivation of (19a) would look as in (20a) before PFdeletion and as in (20b) after PF-deletion. (20a) is also the structure that is fed into LF. Notice that although there is a LF – PF discrepancy here, there is no need for reconstruction because the verb lees ‘read’ is interpreted in two positions at LF. (20) a. b.

[IP ik [I lees [VP ik [DP een boekje [V lees ]]]]] [IP ik [I lees [VP ik [DP een boekje [V lees ]]]]]

LF PF

Let us now see if the insertion of a dummy verb reduces derivational complexity. There are at least two ways to analyze dummy verb constructions in the Minimalist Program. Option 1 would be that the dummy verb is present in the Numeration. This would be the minimal difference between the Numerations of the analytical and the synthetic constructions. The derivation would then look as in (19d) (ignoring the trace of the subject which should be a copy on this account). A chain must be formed between ‘do’ and the lexical verb to ensure that agreement and tense features are shared.

412

Sjef Barbiers

This chain formation is also necessary in the case of the synthetic construction, as a precondition for movement. Presumably, the dummy must be deleted when the structure is delivered to LF, as it does not contribute to the semantic interpretation. If we now compare the derivational complexity between the synthetic and the analytical construction in option 1, we see that the synthetic construction involves five (relevant) steps: Merge lexical V, form chain, copy V, merge the copy and delete the lower copy at PF. The analytical construction involves four steps: Merge lexical V, merge the dummy, form chain between dummy and V, delete dummy at LF. Thus, the dummy construction seems to reduce derivational complexity, but this is only true if Copy and Merge Copy are taken to be two steps. If this is one step, the two derivations are equally costly. More importantly, however, according to Minimalist assumptions, there is only competition between derivations that are based on the same Numeration: The most economical of these wins. Since a Numeration with and a Numeration without a dummy auxiliary are not identical, the derivations with and without dummy auxiliary are not in competition and it is irrelevant whether one is more economical than the other. Option 2 for a minimalist derivation of the analytical construction is one in which the dummy is not part of the Numeration but the result of spell-out of a copied feature bundle. In the Distributed Morphology variant of Minimalism (Halle and Marantz 1993), syntactic derivations do not operate on lexical items but on feature bundles. After transfer to PF, the feature bundles are spelled out by Vocabulary Items. The derivation would then look as in (21) (irrelevant features left out): (21) [IP[1SG][I[V.1SG.pres.trans][VP[1SG][DPDP[V .1SG.present.trans]]]]]] PF1 ik lees Ø een boekje Ø PF2 ik doe een boekje lezen In PF1 only the lexical verb has to be inserted and the feature bundle in V must be deleted. In PF2 two vocabulary items have to be inserted, the dummy and the infinitive. Both derivations involve two operations and therefore appear to be equivalent. Notice that there is no difference with respect to reconstruction, as both PF1 and PF2 share the same syntactic input. According to the two Minimalist analyses, dummy auxiliary constructions and movement constructions are either not compared in terms of

Dummy auxiliaries in Dutch dialects, L1 and L2 413

economy or they are equally costly. According to the Principles and Parameters account, the dummy auxiliary construction is more economical than the movement construction. Both Minimalist analyses therefore lead to the expectation that, all things being equal, dummy auxiliary constructions and movement constructions are in free alternation. For the relevant dialects, it is clear that these types of constructions alternate, i.e., dialect speakers produce both, but it is not clear whether they have a preference for one or the other type, as preference data are lacking so far. It is certainly not true, however, that the number of dialects that have movement equals the number of dialects that have dummy auxiliaries: The latter are a small proportion of the former. The Principles and Parameters analysis leads to the expectation that the dummy auxiliary construction is the preferred option. Again, this may be true for the dialects that have both the dummy and the movement construction, but is not true for the Dutch dialect landscape as a whole. Furthermore, the fact that there does not seem to be a stage in Dutch L1 and L2 acquisition that has only dummy auxiliary constructions and no movement of the lexical verb to I also suggests that the Minimalist analyses may be on the right track. 4. Conclusion We have seen in this article that dummy auxiliaries occur both in adult dialects of Dutch and in L1 and L2 Dutch. The auxiliaries involved are ‘do’, ‘have’, ‘be’ and ‘go’. Despite the superficial similarity of dummy auxiliary constructions in dialects, L1 and L2, it was shown that dialectal dummy auxiliaries have a distinct syntactic status. This and the geographical distribution of dummy auxiliaries makes an account in terms of derivational complexity implausible for dummy auxiliaries in dialects, a notion that was shown to be problematic in itself and dependent on the theoretical framework adopted. A Principles and Parameters approach would predict a general preference for dummy auxiliaries, while Minimalist approaches would predict a free alternation between dummy auxiliary and movement constructions. The first prediction is not borne out by the Dutch dialect data because only a minority of the Dutch dialects has dummy auxiliaries. The second prediction should be tested in future research as preference data are lacking so far. Given that the frequency of dummy auxiliaries decreases during L1 and L2 acquisition, both the Principles and Parameters and the Minimalist accounts need to be supplemented with principles applying to

414

Sjef Barbiers

other levels than syntactic computation to fully explain the developmental process. Notes 1. See Section 3 for a critical discussion of the assumption that movement (internal merge) is more complex than insertion (external merge). 2. For a discussion of the absentive construction see Haslinger (2007) who argues that the absentive meaning is an instantiation of obligatory disjoint reference applied to the location component of the deictic center.

References Anagnostopoulou, Elena 2003 Participles and voice. In Perfect explorations, Artemis Alexiadou, Monika Rathert, and Arnim von Stechow (eds.), Berlin: 1˗36. Berlin: Mouton de Gruyter. Barbiers, Sjef 2005 Word order variation in three-verb clusters and the division of labour between generative linguistics and sociolinguistics. In Syntax and variation. Reconciling the biological and the social, Leonie Cornips and Karen Corrigan (eds.), 233˗264. Amsterdam/Philadelphia: John Benjamins. 2009 Locus and limits of syntactic microvariation. Lingua 119: 1607˗1624. Barbiers, Sjef, Johan van der Auwera, Hans Bennis, Eefje Boef, Gunther De Vogelaer, and Margreet van der Ham 2008 Syntactic atlas of the Dutch dialects, Volume II [SAND]. Amsterdam: Amsterdam University Press. Barbiers, Sjef, Olaf Koeneman, and Marika Lekakou 2009 Syntactic doubling and the nature of WH-chains. Journal of Linguistics 46: 1˗46. Blom, Elma 2003 From root infinitive to finite sentence. Ph.D. dissertation, Utrecht University. Blom, Elma, and Siebe de Korte 2011 Dummy auxiliaries in child and adult second language acquisition of Dutch. Lingua 121: 906˗919. Chomsky, Noam 1995 The minimalist program. Cambridge, Mass.: MIT Press.

Dummy auxiliaries in Dutch dialects, L1 and L2 415 Chomsky, Noam, and Howard Lasnik 1995 The theory of Principles and Parameters. In The minimalist program, Noam Chomsky, 13˗128. Cambridge, Mass.: MIT Press. Cornips, Leonie 1994 De hardnekkige vooroordelen over de regionale doen + infinitiefconstructie. [The persistent prejudices about the regional doe + infinitive construction.] Forum der Letteren 35: 282˗294. 1998 Habitual doen in Heerlen Dutch. In Do in English, Dutch and German.History and present-day variation, Ingrid Tieken-Boon van Ostade, Marijke van der Wal, and Arjan van Leuvensteijn (eds.), 83˗101. Amsterdam/Münster: Stichting Neerlandistiek/Nodus Publikationen. Erb, Marie-Christine 2001 Finite auxiliaries in German. Ph.D. Dissertation Tilburg University. Haberzettl, Stefanie 2003 “Tinkering” with chunks: Form-oriented strategies and idiosyncratic utterance patterns without functional implications in the IL of Turkish speaking children learning German. In Information structure and the dynamics of language acquisition, Christine Dimroth and Marianne Starren (eds.), 45˗63. Amsterdam: John Benjamins. Haegeman, Liliane 1990 The syntax of motional goan in West Flemish. In Linguistics in the Netherlands, Reineke Bok-Bennema, and Peter Coopmans (eds.), 81˗90. Dordrecht: Foris. Halle, Morris, and Alec Marantz 1993 Distributed morphology and the pieces of inflection. In The view from Building 20. Essays in honor of Sylvain Bromberger. Kenneth Hale and Samuel Jay Keyser (eds.), 111˗176. Cambridge, Mass.: MIT Press. Haslinger, Irene 2007 The syntactic location of events. Ph.D. dissertation Tilburg University. Hoekstra, Teun 1984 Transitivity. Grammatical relations in Government-Binding theory. Ph.D. dissertation Leiden University. Dordrecht: Foris Publications. Kampen, Jacqueline van 1997 First steps in WH-movement. Ph.D. dissertation UtrechtUniversity. Kayne, Richard 1993 Toward a modular theory of auxiliary selection. Studia Linguistica 47: 3˗31. 1994 The antisymmetry of syntax. LI monograph 25. Cambridge, Mass.: MIT Press.

416

Sjef Barbiers

Klein, Rudolf Martin 1974 Word order: Dutch children and their mothers. Publications of the Institute of General Linguistics. Department of Linguistics, University of Amsterdam. Koeneman, Olaf, Marika Lekakou, and Sjef Barbiers Fortc. Perfect Doubling. Linguistic Variation Yearbook. Amsterdam /Philadelphia: John Benjamins. Kratzer, Angelika 2000 Building statives. In Proceedings of Berkeley Linguistic Society 26, Lisa Conathan (ed.) 385˗399. Berkeley, CA, USA: Berkeley Linguistic Society. Nunes, Jairo 2004 Linearization of chains and sideward movement. LI Monograph 43. Cambridge, Mass.: MIT Press. SAND II see Barbiers et al. 2008. Tieken-Boon van Ostade, Ingrid, Marijke van der Wal en Arjan van Leuvensteijn (eds.) 1998 Do in English, Dutch and German. History and present-day variation. Stichting Neerlandistiek VU Amsterdam. Münster: Nodus Publikationen. Van de Craats, Ineke 2009 The role of is in the acquisition of finiteness by adult Turkish learners of Dutch. Studies in Second Language Acquisition 31: 59˗92. Van de Craats, Ineke, and Roeland van Hout 2010 Dummy auxiliaries in the L2 acquisition of Moroccan learners of Dutch: Form and Function. Second Language Research 26, 473˗500. Van Craenenbroeck, Jeroen 2004 Ellipsis in Dutch dialects. Ph.D. dissertation, Leiden University. Van Riemsdijk, Henk 2002 The unbearable lightness of GOing. Journal of Comparative Germanic Linguistics 5: 143˗196. Zuckerman, Shalom 2001 The acquisition of “optional” movement. Ph.D. dissertation Groningen University.

Index

affirmative declarative, 171˗172, 174˗176, 179˗182, 184, 186˗187, 189, 191, 193˗196, 198˗201 agent, 7, 106, 202, 343˗344, 351˗356, 358˗360, 364, 366˗367, 390 agreement, 31, 104˗105, 133, 171, 173˗174, 177˗178, 181, 185, 187, 192˗193, 199, 202, 212˗213, 217˗ 218, 221˗233, 237, 251, 264, 266, 284, 286, 288˗292, 295˗296, 300, 310˗312, 315˗316, 329˗330, 344, 365, 390, 395˗396, 399, 411 aspect, 6, 41, 44, 52, 56, 60, 62, 78, 80˗81, 84, 88˗89, 91˗94, 97, 101˗102, 104˗106, 110, 131˗133, 142˗143, 146, 158, 191, 192, 202, 215, 253, 282˗285, 292, 299˗300, 301, 312, 320, 346˗347, 353˗355, 358, 369, 372, 374˗376, 385, 387˗388, 398 assertion, 7, 109, 216, 303, 312, 329, 330, 344, 356-358, 360, 362, 366 bilingual, 1-3, 5, 16˗17, 162, 169, 171˗172, 183, 194, 196˗197, 200, 203, 217, 219, 227, 233, 251, 253, 258˗260, 262˗264, 266˗270, 272, 371˗373, 377, 380˗382, 384˗385, 388˗390 blocking, 23, 25˗28, 31˗32, 177˗179, 181, 198, 309 child˗directed speech, 46, 49, 53, 75, 97, 102, 130, 253, 346, 374, 377, 387, 397 codified grammar, 369˗370, 372, 374˗376, 387

comprehension, 2, 4, 6, 40, 47˗48, 54, 57, 61˗65, 68˗69, 82, 97˗98, 256, 293, 300˗301, 315 control, 7, 353˗355, 358, 365˗366 corrective feedback, 191 dependent clause. See embedded clause. derivational complexity, 254, 256, 261, 379˗380, 382, 384, 387˗388, 396, 399˗401, 404, 410˗413 dialect, 2˗3, 7, 19˗21, 40˗41, 46˗49, 51˗53, 62, 66˗67, 75˗76, 102, 130, 132, 142, 162, 172, 174, 176, 178, 181, 191, 199, 201, 253, 346, 370, 373, 376, 390, 396˗401, 403, 405˗ 410, 413 do˗insertion. See do˗support do˗support, 12, 17, 28˗29, 31, 172˗174, 177˗178, 180˗181, 185, 190, 192˗194, 197˗199, 201˗203, 381, 402 doubling, 12, 17, 30, 45, 403, 406˗409 dummy verb, 2, 4˗6, 75˗77, 87˗94, 97˗98, 108, 130, 209˗211, 214˗219, 221˗237, 243˗244, 288, 300˗ 301, 307˗308, 310˗316, 318˗326, 328˗334, 341, 387, 397, 411 economy, 3˗7, 27˗29, 31˗32, 40, 46˗ 47, 52, 54, 67, 76˗77, 91˗94, 109, 133, 159, 180˗181, 197˗199, 201, 256, 307, 383, 396, 409˗410, 412˗ 413 elicitation, 4, 15, 76˗77, 82, 148, 150, 154, 156˗157, 159, 163, 203, 237,

418

Index

264˗265, 274, 282, 314, 316, 321˗ 322, 328 ellipsis, 11, 15˗16, 21, 173˗174, 180, 182, 184, 186˗187, 192, 194, 199, 221, 402 embedded clause, 5, 40, 42˗43, 45˗52, 54, 66˗67, 77, 83, 90, 97˗98, 109, 211˗213, 218, 221, 256˗257, 261, 264˗265, 269˗271, 273, 283, 309 non˗root clause, 371˗372, 377˗384, 387, 389 emphasis, 11, 17, 20, 33, 172, 176, 180, 182, 184, 185˗188, 192, 194, 198˗199, 201 contrastive ~, 22, 173, 175, 184, 186, 188, 196, 200 emotive ~, 22, 175 finiteness, 42˗45, 51, 54, 90, 103˗104, 107˗111, 124, 128, 130˗133, 136, 174, 192, 203, 209, 212˗213, 216, 218, 236, 251, 254˗256, 259, 261, 282˗286, 288˗293, 295˗296, 299˗301, 307˗308, 312, 315, 322, 329˗331, 342, 344˗345, 347, 351, 358, 360, 362˗363, 365, 405 first language acquisition, 1˗4, 7, 39, 47, 66˗67, 144, 147, 163, 194, 395 form˗function mapping, 6, 215˗216, 308, 328˗330 free variation, 5, 20, 172, 176, 181, 189, 199 functional, 5, 20, 24, 45, 104, 177˗178, 180˗181, 215, 312, 341˗342, 344˗351, 356˗359, 361, 363˗366, 369, 395, 399, 406 ~ account, 108˗110, 312 ~ category, 7, 45, 178, 192, 308, 341˗342, 344˗345, 350˗351, 358, 365

~ head, 24, 30, 177, 344, 346˗347, 353, 360˗362, 365, 405 GAP verb. See general all˗purpose verb general all˗purpose verb, 2, 143, 147˗148, 157, 209 inchoative, 40˗41, 43, 46˗47, 53˗54, 59˗60, 63˗67, 69, 81, 102, 132, 253, 369, 374, 385 input, 3, 5, 7, 17, 19, 32, 39˗40, 43˗44, 46˗48, 53˗54, 61, 66˗67, 92, 108, 130, 133, 136, 152, 158, 163, 171, 182˗183, 191, 199˗200, 214, 219, 237, 253˗256, 343, 345, 353, 360, 369, 371˗372, 378˗380, 388, 399, 412 inversion, 11, 25, 67, 174, 182, 185˗ 186, 200˗201, 211˗212, 309, 351, 358, 360˗362, 364, 371, 377˗384, 386˗387, 390, 396, 399 item˗based, 182 Language Acquisition Device, 29 Last Resort, 172˗173, 177, 179˗180, 182, 198˗199 learnability, 92, 172, 194˗195, 197 lexical verb, 4˗5, 29, 43˗44, 64, 66, 70, 75˗77, 81˗82, 88, 90˗91, 93, 101˗105, 107˗110, 112˗113, 115, 119, 122, 124, 126, 132˗33, 135, 141, 144, 158, 171, 200, 209˗210, 215˗216, 234˗237, 252, 254˗258, 260˗261, 265˗268, 270˗272, 294, 296, 299, 301˗302, 307˗308, 310˗ 320, 322, 324˗334, 342, 344, 361˗ 363, 365˗366, 372, 383˗384, 405, 411˗413 contentful verb, 26, 141, 143˗150, 152, 154˗159, 162˗163 thematic verb, 2, 19, 114, 116˗117, 120˗124, 128˗129, 133, 135˗136, 172˗173, 176, 179˗180, 182, 185˗187, 189, 192˗193, 195˗196, 198˗199,

Index 419 202˗203, 211, 216, 221˗222, 225, 228, 281˗283, 285, 288˗290, 292, 295, 300, 303, 409 light verb, 2˗5, 97, 112˗114, 125, 128, 136, 141˗159, 161˗163, 180, 286, 288, 307, 310˗314, 316˗317, 319, 320˗321, 323˗331 Logical Form, 26, 76˗77, 90, 178, 180, 201, 203, 396, 410˗412 main clause, 40, 42˗43, 46, 48, 50˗52, 54, 66˗67, 75, 77, 90, 104, 109, 212, 214, 216, 222, 225, 254, 256, 257, 261, 264˗267, 269˗270, 273, 283˗285, 308˗309, 396, 399˗400 root clause, 371˗372, 377˗380, 383˗384, 386˗387, 390 matrix clause. See main clause. modal, 14, 17, 21, 23˗26, 31˗32, 41˗42, 44˗45, 51˗52, 92, 102˗103, 105˗107, 110˗114, 124˗126, 128˗130, 132, 134, 136, 174˗175, 177, 181, 184, 186, 194, 202, 210, 214˗216, 218, 221˗222, 225, 228, 234, 237, 255, 274, 294, 296˗299, 303, 308, 310, 319˗320, 334, 352˗ 356, 360 monolingual, 3, 5, 13, 103, 109, 111, 128, 135, 172, 193˗195, 197˗198, 200, 202, 216˗217, 236, 253, 256, 258˗260, 262˗270, 272, 370˗373, 377˗379, 381˗382, 387˗389 mood, 5, 12, 24, 31, 104, 180 declarative ~, 12, 19, 25, 30˗31, 152, 178˗179, 185, 187, 189, 195˗196, 203, 212, 284˗285, 375, 378, 397, 399 imperative ~, 24˗25, 30˗32, 114, 185, 221, 378, 386, 397, 400 indicative ~, 5, 7, 25, 31, 181, 203, 267, 283 movement, 4, 12, 29, 43˗45, 67, 106, 109, 133˗134, 145, 153, 162,

177˗178, 180, 252, 254, 256˗257, 273, 283, 390, 399, 401˗402, 405, 410˗413 head ~, 28, 104, 178, 342, 344˗ 345, 351, 358, 362˗363, 365˗366 overt ~, 51 verb ~, 40˗46, 51˗52, 54, 67, 76˗77, 89˗91, 93, 103, 108˗109, 133, 136, 178, 216, 255˗256, 261, 273, 282˗283, 371, 378˗ 379, 383˗384, 395, 399˗400 negation, 11, 17, 25, 30, 32, 44, 67, 172˗180, 182, 184˗186, 193˗194, 197˗199, 201˗202, 213, 285, 287˗288, 309, 313, 315˗316, 321˗323, 331˗333, 337, 358 nonemphatic, 12, 22, 25, 30, 171˗172, 174, 176, 178˗179, 181˗182, 191, 194˗195, 197˗199, 201 nonthematic verb, 112˗114, 116, 122˗125, 128˗129, 133, 136˗137, 218 novel verb, 261, 264, 270˗271 numeration, 26˗27, 30, 32, 411˗412 optional infinitive. See root infinitive. picture selection task, 40, 47, 54, 56, 293˗297, 300 pitch, 22, 175, 186, 188, 194, 201 placeholder, 2, 103, 209˗210, 214, 217˗219, 221, 235, 331, 341 presentational, 6, 310˗311, 317, 319˗ 320, 329 processing, 27, 93, 181, 252, 254, 256˗261, 270˗272, 395, 399, 410 productive, 13, 49, 71, 77, 91, 102, 107, 109˗111, 116, 119˗120, 123˗125, 129, 131, 136, 185, 193, 197˗198, 222, 261, 271, 322, 345, 349, 350 region. See dialect.

420

Index

root infinitive (RI), 44, 70, 107, 109, 113˗124, 126, 128, 132, 173, 192˗ 193, 211, 228˗229, 352˗353 second language acquisition, 1˗3, 6˗7, 67, 147, 210, 345, 395 semantic, 1˗2, 4, 6, 21˗22, 40, 44, 46, 77, 82, 91˗94, 102, 107˗110, 133, 135, 141˗144, 146˗147, 150, 158˗159, 172˗174, 181, 191˗192, 209, 214, 216, 252˗254, 256, 265, 281, 283, 286, 307, 310, 329, 342˗344, 347, 354, 358, 362, 365, 390, 395, 400, 407, 409, 412 Specific Language Impairment (SLI), 5, 97, 216, 251˗254, 257˗273 structural account, 108˗109 subordinate clause. See embedded clause substandard, 176, 346 telicity, 84, 146, 191, 374˗375 tense, 23, 28, 30, 44˗45, 104˗105, 108, 132, 171, 173˗175, 177˗180, 182, 187, 192˗193, 198˗199, 201, 215, 255, 320, 344, 390, 396, 399 future ~, 4, 40, 48, 52, 56˗60, 63, 75˗76, 78˗90, 92˗96, 98, 102, 105, 108, 135, 253, 369, 374, 385, 409

past ~, 4, 17, 19, 23, 35, 44, 56, 59, 60, 63, 71, 76, 78, 80˗86, 88˗92, 175, 187, 192˗193, 196˗197, 203, 257˗258, 288, 317, 402 present ~, 4, 6, 23, 25, 39, 40, 44, 47, 52, 54, 56˗57, 59˗60, 62˗ 63, 75˗76, 78˗93, 95˗96, 104, 175, 192˗193, 196, 284, 294, 296, 298˗302, 310˗311, 315, 318˗319, 369˗371, 374, 376, 385, 387˗388 topicalization, 11, 17, 19, 41, 97, 102, 182, 200, 344˗345, 351, 356, 358˗ 360, 364 typological differences, 284 unstressed, 13, 171, 174˗176, 178˗179, 187, 195, 197, 199, 201, 314, 366 verb raising, 6, 285˗286, 288˗ 292, 295˗296, 300, 307˗313, 315˗316, 320˗326, 328, 330˗333, 342 Verb Second (V2), 40, 42˗43, 45, 50, 52, 54, 67, 75˗76, 77, 90, 93, 97, 104˗105, 107, 108, 114, 128, 132˗ 135, 211˗213, 216˗218, 221˗237, 254˗256, 266, 273, 284, 309, 328, 342, 344, 371˗372, 377˗388