Language Acquisition and the Functional Category System 9783110216219, 9783110216202

Citation preview

Peter Jordens Language Acquisition and the Functional Category System

Studies on Language Acquisition

Editor Peter Jordens

Volume 39

Peter Jordens

Language Acquisition and the Functional Category System

ISBN 978-3-11-021620-2 e-ISBN 978-3-11-021621-9 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. © 2012 Walter de Gruyter GmbH, Berlin/Boston Typesetting: PTP-Berlin Protago-TEX-Production GmbH, Berlin Printing and binding: Hubert & Co. GmbH & Co. KG, Göttingen ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

To someone dear to me

Acknowledgements The present study was supported by a grant from the Netherlands Organization for Scientific Research (NWO) entitled The development of finiteness. From a lexical to a functional category, nr. 365-70-019. With this grant I was able to start this project while I was chair of the Department of Applied Linguistics at Amsterdam, VU University. I wish to express my thanks to Wolfgang Klein who made it possible for me to carry out my research at the Max Planck Institute for Psycholinguistics in Nijmegen. I am also thankful for the suggestions and ideas as well as for the comments and criticisms of my colleagues who participated in the Max Planck research project “The acquisition of finiteness” and the European research group “The Structure of Learner Varieties”. In particular I am grateful to Christine Dimroth and Clive Perdue†. Furthermore, I would like to mention Dagmar Bittner, Theo Bongaerts, Alex Dukers, Tilman Harpe, Eric Kellerman, Leah Roberts, Sarah Schimke, Ad Verbunt, Josje Verhagen and Steffi Winkler who contributed to this book in many different ways. Last but not least, I would like to thank Cynthia, Jasmijn and Andrea who unknowingly participated as my native speaker informants in this study. Nijmegen, 2012

Peter Jordens

Contents Acknowledgements  1 1.1 1.2

 vii

 1 Introduction  Language acquisition from a functional perspective  Overview of the book   3

2 Lexical vs. functional elements   9 2.1 Structure in language   9 2.2 Language structure in production   15 2.2.1 Phrase structure   15 2.2.1.1 Lexical selection   15 2.2.1.2 Phrasal coherence   15 2.2.2 Utterance structure   16 2.2.2.1 Perspective taking   16 2.2.2.2 Contextual embedding   19 2.2.2.3 Lexical categories   21 2.2.2.4 Functional categories   21 2.2.3 Word formation   22 2.2.3.1 Derivation, inherent and contextual inflection   24 2.2.3.2 Lexical and functional categories of word formation  2.3 Learner systems   30 2.3.1 Utterance structure at the initial stage   31 2.3.2 Word formation at the initial stage   33 2.4 First and second language acquisition   34 2.5 Research questions   37 2.6 Data sources   38 3 The Target System   41 3.1 Introduction   41 3.2 Utterance structure   42 3.2.1 Basic word order   42 3.2.2 Word order variation   46 3.2.3 Information structure and lexical meaning  3.2.3.1 VP structure   52 3.2.3.2 FP structure   52 3.2.3.3 The function of F   53 3.2.3.4 Movement   55 3.2.4 Summary   56

 50

 1

 26

x 

 Contents

3.3 3.3.1 3.3.1.1 3.3.1.2 3.3.1.3 3.3.2 3.3.2.1 3.3.2.1.1 3.3.2.2 3.3.2.2.1 3.3.2.2.2 3.3.2.3 3.3.3 3.4

Utterance production   56 Conceptualization, lexicalization and predication  Conceptualization   57  58 Lexicalization  Predication   62 Expressing finiteness and contextual embedding  The semantic function of finiteness   66 Truth value   66 Anchoring   70 Temporal anchoring   71 Spatial anchoring   72 The function of SpecFP   74 Summary   77 Hypotheses on language development   78

4 The initial state   79 4.1 Finiteness at the initial state   79 4.2 Theoretical accounts   80 4.2.1 Morpho-syntactic systems   82 4.2.1.1 Poeppel and Wexler (1993), Wexler (1998)  4.2.1.2 Discussion   84 4.2.2 Semantic systems   87 4.2.2.1 Clahsen (1986)   87 4.2.2.2 Discussion   88 4.2.2.3 Ingram and Thompson (1996)   90 4.2.2.4 Discussion   92 4.2.3 Summary   95 4.3 The alternative   99 4.4 Summary   109

 57

 66

 82

5 The lexical stage   111 5.1 Basic languages, research questions   111 5.2 The analysis of early learner data   115 5.3 Utterance structure at the lexical stage   124 5.3.1 Lexical projections   124 5.3.2 Testing the model: ergatives and particle verbs  5.3.2.1 Ergatives   130 5.3.2.2 Particle verbs   131

 129

Contents 

5.3.3 5.3.3.1 5.3.3.2 5.3.3.3 5.3.4 5.4 5.5 5.6

Word order   134 Subject first   135 Topic first   136  138 Head-initial  Summary   141 Conflicting constraints   145 Conclusion   150 From the lexical stage to the functional stage 

 xi

 153

6 The functional stage   155 6.1 The acquisition of the projection of F   155 6.1.1 Utterance structure at the lexical stage   155 6.1.2 Towards a functional topic position   157 6.1.3 Towards a grammatical subject position   162 6.2 Evidence of the projection of F   164 6.2.1 The functional topic position   164 6.2.2 The functional category F   168 6.2.2.1 Analysis of the modal expressions ‘ulle’, ‘mag-ikke’ and ‘nee’  6.2.2.2 Analysis of the unanalysed modal expressions   169 6.2.2.3 Epistemic modals   170 6.2.2.4 The auxiliary verbs ‘heb, heeft’ and ‘ben, is’   172 6.2.2.5 The auxiliary verbs ‘doe, doet’ and ‘ga, gaat’   178 6.2.3 Topicalization   179 6.3 The topic position   182 6.3.1 The affix in F as a topicalization device   182 6.3.2 Yes/no- and wh-questions   184 6.3.3 Verb-third in adult L2   187 6.4 Finiteness   191 6.4.1 Auxiliary verbs   191 6.4.1.1 Verb placement   191 6.4.1.2 ‘Light verbs’   193 6.4.1.3 Morphological properties of finiteness in L2 Dutch   197 6.4.2 Grammatical aspect   199 6.4.3 Tense   208 6.4.4 Agreement   208 6.5 Summary   209

 168

xii 

 Contents

7 Finiteness in language acquisition research   211 7.1 Finiteness in first language research   211 7.1.1 L1 Dutch: Gillis (2003)   211  215 7.1.2 L1 German: Bittner (2003)  7.1.3 Summary   222 7.2 Finiteness in second language research   223 7.2.1 Verb placement and inflectional morphology   223 7.2.1.1 The contingency between verb placement and morphology   223 7.2.1.2 Methodological questions   224 7.2.1.3 Verb placement and inflectional morphology in second language research   226 7.2.2 Restrictions on the contingency between verb placement and inflection   226 7.2.2.1 Parodi (1998, 2000)   226 7.2.3 The dissociation of verb placement and inflection: The Impaired Representation Hypothesis   229 7.2.3.1 Meisel (1997)   229 7.2.4 The separation of verb placement and inflection: The Missing Surface Inflection Hypothesis   231 7.2.4.1 Prévost and White (2000)   231 7.2.5 The acquisition of auxiliaries as a prerequisite for verb movement   234 7.2.6 The stagewise acquisition of the functional category system   237 7.2.6.1 Rule and Marsden (2006)   237 7.2.6.2 Vainikka and Young-Scholten (1996, 2011): The Organic Grammar approach   239 7.2.6.3 Becker (2005)   242 7.2.7 Summary   248 7.2.7.1 The contingency between verb placement and morphology   249 7.2.7.2 The acquisition of the auxiliary as a carrier of finiteness   250 8 8.1 8.2 8.3 8.4

Conclusion   253 Basic languages   253 The lexical stage   253 The functional stage   262 Driving forces   267

References   271 Index   277

1 Introduction 1.1 Language acquisition from a functional perspective Language acquisition is a developmental process. Research on spontaneous language acquisition both in children learning their mother tongue and in adults learning a second language has shown that language development proceeds in a stagewise manner. Hence, language development is usually studied on the basis of learner utterances that are accounted for in terms of so-called ‘learner languages’. In studies on second language acquisition, learner languages are also referred to with the term ‘Basic Variety’ (Klein and Perdue 1992, 1997). Learner languages or Basic Varieties are language systems that are grammatically rather simple. In fact, they are lexical systems. A typical constraint of the lexical system of learners of Dutch is the principle which holds that “if there is an agent, it occurs in initial position”. This semantic principle of utterance structure implies that a theme can only occur in initial position, if there is no agent to be expressed. Thus, at the lexical stage, both children learning Dutch as their mother tongue (L1) and adults learning Dutch as a second language (L2) may typically produce the utterances as in (1) and (2). (1) child L1 Dutch kannie bal pakke. can-not ball get (2) child L1 Dutch popje valt bijna. doll falls nearly

adult L2 Dutch die kanniet praten nederlands. that can-not talk dutch adult L2 Dutch ik woont in casablanca. I lives in casablanca

At some point in acquisition, the lexical-semantic system develops into a targetlike system. With this targetlike system, learners have reached a stage at which their language system has the morpho-syntactic features to express the functional properties of finiteness and topicality. Evidence of this is word order variation and the use of linguistic elements such as auxiliaries, tense and agreement markers and determiners. Examples are given in (3) and (4).

2 

 Introduction

(3) child L1 Dutch ik heef óók appel gete. I have too apple eaten (4) child L1 Dutch die heb ik wel geplakt. that have I indeed glued

adult L2 Dutch ik heb beetje geld sparen. I have some money saved adult L2 Dutch die heb ik allemaal vergeten. that have I all forgotten

In the present monograph, I will investigate the process of language acquisition from a functional point of view. Within this functional perspective on language acquisition, questions such as the following arise. What is the driving force behind the process that causes learners to give up a simple lexical-semantic system in favour of a functional-pragmatic one? What is the added value of linguistic features such as the morpho-syntactic properties of inflection, word order variation and definiteness? Why is it that in cases of specific language impairment, it is mainly morpho-syntactic properties of the target language that are affected? These are the leading questions of the present volume. They show the relevance of studying learner language as a language system in its own right. In the following chapters, I will show how learner varieties develop both in child L1 and in adult L2 Dutch. At the initial stage, as pointed out before, the functional properties of the target language are absent. This means that utterance structure is determined by the lexical projection of a predicate-argument structure. A typical feature at the lexical stage is the fact that topicalization, i.e. reference to the situation that the utterance applies to, cannot be expressed with the functional means which play a role in the target system. Nevertheless, at the lexical stage, topicalization is expressed within the constraints of the relevant system with a lexical structure that is non-targetlike. Examples from child L1 learners and adult L2 learners are given in (5). (5) child L1 Dutch disse hoeniet meeneme. this-one must-not with-take da kanwel opzitte. there can-indeed on-sit

adult L2 Dutch die wijn magwel drinken. that wine may-indeed drink dan moet daar helemaal opruimen. then must there all up-clean

The type of utterance in (5) has either the object or an adverbial in initial, topic position, while the agent is expressed only implicitly with the semantics of the modal head.

Overview of the book 

 3

I will argue that, at the lexical stage, topicalization is the driving force which leads to the acquisition of a functional projection FP. F is the head of FP. It provides a position for elements carrying the functional properties of finiteness, i.e. for elements such as epistemic modals, auxiliary verbs and later in the acquisition process for (finite) lexical verbs, too. These verbal elements serve as carriers for the expression of finiteness with the morpho-syntactic properties of agreement and tense. SpecFP is the specifier of FP. The position of SpecFP is available either as a topic or as a focus position. Topicalization is expressed with the placement of a constituent in initial position in interaction with the morpho-syntax of definiteness. Focalization can be expressed with a wh-word in SpecFP or in case of yes/no-questions with SpecFP left empty.

1.2 Overview of the book Chapter 2 provides an outline of the principles of linguistic structure building that language perception and production are based on. It discusses the basic syntactic properties of utterance structure and phrase structure, the morphological properties of word formation and the pragmatic properties of information structure. It shows how the underlying knowledge systems of linguistic structure are used in processes of language production, i.e. in lexical selection and perspective taking and in the embedding of an utterance into its situational context. Finally, it argues that the distinction between lexical and functional category systems is relevant for the process of language development. This distinction applies across the levels of utterance structure, phrase structure and word formation. It establishes the focus of the present research. Furthermore, assuming that early learner systems are lexical systems, it is stated in Chapter 2, first, that language acquisition is a process of structure building that is the result of the acquisition, integration and restructuring of lexical and functional linguistic knowledge. Second, it is claimed that it is the acquisition of the linguistic means for the embedding of an utterance into the situational context that serves as the ‘driving force’ for learners to give up a relatively simple lexical system in favour of a complex functional one. Finally, it is argued that the differences between child L1 and adult L2 learners are to be explained relative to the fact that children experience no influence from either advanced levels of cognitive and linguistic development nor from the principles of another language system. The main characteristics of utterance structure in Dutch will be discussed in Chapter 3. First, I will present the arguments for why basic word order in Dutch is considered to be OV. On the basis of OV, word order may vary. Principles underlying word order variation determine the actual form in which utterances occur. In

4 

 Introduction

order to account for the function of variation in utterance structure, I will discuss a proposal that is based on the functional principles underlying the formal properties of utterance structure. From a functional perspective, utterance structure in Dutch integrates two projections of syntactic structure: a functional projection (FP) and a lexical projection (VP). The finite verb in F serves as the head of FP. In declarative main clauses F is the second constituent position. Verbal elements in the position of ‘verb-second’ are carriers of the functional properties of finiteness. It is the function of finiteness to express the pragmatic function of assertion. As an assertion, a finite utterance has to be anchored with respect to both time and space. Temporal anchoring occurs morphologically with the finite verb. Spatial anchoring occurs with elements in the position of SpecFP. In the default case, the functional properties of F are carried by the auxiliary verb. The lexical verb in V is the head of VP. In absence of an auxiliary verb, it is the lexical verb that is used to express the functional properties of finiteness. Hence, variation with respect to the position of the lexical verb serves a functional purpose. Similarly, nominal constituents with an argument function or adverbials may occur in the position of SpecFP. In this position, these constituents serve as carriers of the pragmatic function of contextual embedding. The distribution of verb forms at the initial stage of child L1 Dutch has been accounted for with different theoretical proposals. In Chapter 4, it is argued that neither the Full Competence Hypothesis as proposed in Poeppel and Wexler (1993), nor the Modal Hypothesis of Ingram and Thompson (1996), nor Clahsen’s (1986) model based on the notion of ‘semantic transitivity’ provides an adequate account of the language system in child L1 learners of Dutch or German. That is, the language system at the initial stage is neither innately given nor the reflection of the target input. Rather, it is the result of a process of creative construction. The claim is that the utterance structure at the initial stage is based solely on lexical categories of predicate-argument structure. Absence of the functional category system explains why the morpho-syntactic properties of finiteness and verbsecond are not instantiated. At the relevant stage, variation in utterance structure is accounted for by the opposition between two types of predicate-argument structure: (a) an agentive type of structure as in (1) with the predicate referring to a (causal) action or an agentive motion and an agent as the external argument and (b) a non-agentive type of structure as in (2) with the predicate referring to a state or a change of state and a theme as the external argument. Learner language at the initial state is described in Chapter 5. Here, the spontaneous production data show that initially the learner languages of child L1 and adult L2 Dutch are indeed lexical. The utterance structure is the instantiation of a lexical projection that is used to express a ‘hold-for’ relation between the

Overview of the book 

 5

predicate and the subject. As claimed before, it is either agentive (type A) or nonagentive (type B). (6) Type A:

(7) Type B:

agent

Ctrl

action

Jaja J

mag may gaatie goes-he

dop opdoen. cap on-do [sl]ape. sleep

theme

state

deze this-one

magwel. may-indeed

theme

change of state

poppie doll

valt hier. falls here

As shown in (6) and (7), the difference between the two types of utterance structure lies in the presence or absence of a head position for a modal or aspectual element that is used to express ‘control’. The function of control is exerted by the agent that carries out a causal action or an agentive motion. Absence of control is exerted by the theme that either occurs in a state or undergoes a change of state. Evidence of the lexical stage is the absence of the functional category system of the target language. At the relevant stage, due to the absence of the functional properties of the target language, learners do not have the linguistic means of the target system to express the pragmatic function of an utterance or the embedding of an utterance into its situational context. Thus, grammatically, the learner system at the lexical stage is as simple as can be. The utterance structure serves the default way to express an assertion. Hence, the structural properties of whand yes/no-questions are absent, as is the case with the structural means of the target system to express the pragmatic function of topicalization. Given that the predicate-argument structure is also used to express properties of information structure, the subject is the constituent with topic function. It establishes the relation between the utterance and the situation that it applies to. Furthermore, the predicate is the constituent that is in focus. It is used to express the information that holds for the topic. Chapter 6 discusses the question of the ‘driving forces’. Why is it that learners will give up a simple learner system in favour of a more complex targetlike

6 

 Introduction

system? How does this process of language development evolve? It will be argued that with the instantiation of the functional projection of F, the learner system provides a syntactic position F for the expression of finiteness and a syntactic position SpecFP for elements to express the topic function. With the projection of F, therefore, the learner system develops the linguistic means for the embedding of an utterance into a situational context. This process allows the acquisition of wh- and yes/no-questions as well as the structural means of the target system to express topicalization. Furthermore, it causes modal predicates to serve as the head of a functional projection, while it establishes a position for the acquisition of aspectual auxiliary verbs such as doet (does) and gaat (goes) in the context of an infinitive, and of heb, heeft (have, has) and ben, is (am, is) in the context of a past participle. Finally, it provides the prerequisite for the acquisition of head movement, tense and agreement. Research on the spontaneous production of utterances in child L1 Dutch by Gillis (2003) and child L1 German by Bittner (2003) is discussed in Section 7.1 of Chapter 7. These studies focus on the acquisition of inflectional morphology. In their study of the emergence of so-called ‘mini-paradigms’, both Gillis and Bittner distinguish between a premorphological and a protomorphological stage. They show that, initially, at the premorphological stage, “most verbs are attested in only one morphological form” (Bittner 2003: 60) and that with respect to the placement of verb forms there is a correlation between form and position. These observations are precisely as they appear from the data at the lexical stage as presented in Chapter 5. At the relevant stage, morphology does not yet play a role. Hence, verb forms are unanalysed. At the functional stage, as shown in Chapter 6, the utterance structure has both a functional position for the verb to express finiteness and a lexical position for the main verb. These verb positions are the prerequisite for head movement and with head movement the first form contrasts appear. This process is precisely what Gillis and Bittner account for with the term ‘mini-paradigms’. The emergence of these mini-paradigms is evidence of the instantiation of the morphological expression of tense and agreement. Section 7.2 of Chapter 7 deals with the acquisition of the inflectional morphology of verb forms in L2 acquisition research. Two alternative hypotheses have been proposed: the Impaired Representation Hypothesis (IRH) and the Missing Surface Inflection Hypothesis (MSIH). The IRH suggests that, at the initial stage, L2 learners may not have access to morphology as a feature property of inflection, while they do have access to a position that makes verb movement possible. This hypothesis should account for the observation of random placement. With the MSIH, it is claimed that, at a more advanced stage, L2 learners may have access to morphology as a feature property of inflection. It is this property of inflection that should establish verb movement. However, the MSIH also claims that, at the

Overview of the book 

 7

relevant stage, the morphology of the raised verb may remain underrepresented, i.e. it may not necessarily be realised at the surface-structure level. Thus, it is the assumption of both the IRH at the initial stage and the MSIH at a more advanced stage that inflectional morphology plays a crucial role in the acquisition of verb movement. However, in the discussion of these proposals, I will show that it does not. What is relevant is the functional position for the expression of finiteness. With the acquisition of this position, learners are able to establish the relation between the functional, head-initial position of the verb and its lexical, headfinal position. This relation is the prerequisite for the acquisition of the syntactic relation referred to as verb movement. With its use in functional head position, the lexical verb serves as a carrier of the semantic properties of finiteness, i.e. finiteness as a category of information structure. As the lexical verb occurs in functional, head-initial position, learners are given the necessary condition to acquire the morphological properties of verbal inflection. Hence, it is the acquisition of finiteness as a category of information structure that leads to the acquisition of a functional position serving as the prerequisite for verb movement, whereas verb movement for its part serves as the prerequisite for the acquisition of inflectional morphology with the lexical verb.

2 Lexical vs. functional elements 2.1 Structure in language Communication among human beings occurs by means of spoken or written language, sign language or pictures, gesture or body language. Communication is mutual understanding. It can only be achieved if the means of communication within a community are the same for all its members. The means of communication within a linguistic community are the utterances of a particular language. Both the production of an utterance and its perception are processes of creative construction that are based on a shared knowledge system of linguistic competence. It is this knowledge system of linguistic competence that enables speakers of a particular language to understand and create utterances that have neither been heard nor produced before. The knowledge system that the processes of language perception and production are based on is a system of entities and rules which serves the purpose of linguistic structure building. Linguistic structure building leads to an utterance structure that is hierarchically organized. That is, utterances are organized in terms of phrasal constituents (phrases), phrasal constituents are organized in terms of word forms, word forms are organized in terms of meaningful entities (morphemes), and, finally, meaningful entities are organized in terms of units of sound (phonemes) that are used to produce meaningful contrasts. For example, an utterance such as de koek is op (the cake is all gone) is a linguistic entity that consists of two phrases de koek (the cake) and is op (is all gone). Both phrases are related by the fact that one constituent (is op) qualifies the other (de koek). At this highest level of syntactic structure, utterances can be analysed in terms of constituents with either of these functions. The constituent that serves the function of a qualification is termed the ‘predicate’, the constituent that is qualified is termed the ‘external argument’ or ‘the subject’. Thus, the predicate qualifies or ‘holds for’ the subject. This hold-for relation between the subject and the predicate, termed ‘predication’, is formally expressed with what is called ‘agreement’. That is, given that the subject is singular as in de koek (the cake) the predicate appears as is op (is all gone), whereas in case the subject is plural as in de koeken (the cakes) the predicate has to appear as zijn op (are all gone). Agreement is thus the formal, i.e. morpho-syntactic, expression of the functional relation between the subject and the predicate. A phrasal constituent (XP) is a linguistic entity with a verbal, a nominal, an attributive or a prepositional element which determines the syntactic function of the constituent as a whole. These verbal, nominal, attributive or prepositional elements

10 

 Lexical vs. functional elements

serve the function of the ‘head’ of a constituent. A verbal element (V) is the head of a verb phrase (VP), a noun (N) is the head of a nominal phrase (NP), an adjective or an adverb (A) is the head of an attribute phrase (AP) and a preposition (P) is the head of a prepositional phrase (PP). Syntactically, a VP can be used as the predicate of an utterance, NPs can be used as the subject of an utterance or the object of a predicate, APs and PPs can be used as the attribute of an NP, a VP or a predication. The head of a phrasal constituent XP determines the internal structure of the phrase. Examples of N as the head of an NP structure in Dutch are given in (1). (1) N as the head of NP N: tuin (garden)

N: hek (gate)

de / een tuin the / a garden de mooie tuin the beautiful garden een mooie tuin a beautiful garden

het / een hek the / a gate het mooie hek the beautiful gate een mooi hek a beautiful gate

An NP with the lexical head tuin (garden) occurs with the element de as the definite article, as in de tuin (the garden), while an NP with the lexical head hek (gate) has the element het as its definite article, as in het hek (the gate). Both tuin and hek have the article een as the indefinite article. Furthermore, if an NP with the lexical head tuin is used with an adjective such as mooi (beautiful), it occurs as mooie, as in de mooie tuin (the beautiful garden) and een mooie tuin (a beautiful garden). If an NP with the lexical head hek is used with the adjective mooi, it occurs either as mooie, as in het mooie hek (the beautiful gate), or as mooi, as in een mooi hek (a beautiful gate). Furthermore, NPs may also occur with a possessive pronoun such as mijn as in mijn mooie tuin (my beautiful garden) or with a genitive such as Jans as in Jans mooie tuin (John’s beautiful garden). As with NPs, the internal structure of the VP is also determined by its head. Examples of V as the head of a VP structure in Dutch are given in (2). (2) V as the head of VP V: leest (reads), schrijft (writes), helpt (helps), klimt (climbs), woont (lives) Jan leest een boek. John reads a book Jan helpt zijn vriend. John helps his friend

Structure in language  

 11

Jan schrijft zijn vriend een brief. John writes his friend a letter Jan klimt in de boom / uit het dal. John climbs into the tree / out of the valley Jan woont hier / in het dorp. John lives here / in the village The examples in (2) show different types of VP structure: the VP with the lexical head leest (reads) occurs with the nominal phrase een boek (a book) as the object of the action of reading; the VP with the lexical head helpt (helps) occurs with the nominal phrase zijn vriend (his friend) as the receiver of the action of helping; the VP with the lexical head schrijft (writes) occurs both with the nominal phrase een brief (a letter) as the result of the action of writing and the nominal phrase zijn vriend (his friend) as receiver of the letter; the VP with the lexical head klimt (climbs) occurs with the prepositional phrase in de boom (into the tree) or uit het dal (out of the valley) as the expression of the goal or the source of the action of climbing; and, finally, the VP with the lexical head woont (lives) occurs with the adverbial hier (here) or with the prepositional phrase in het dorp (in the village) to express the location of someone’s state of living. In specific discourse situations the VP can be used as a constituent of its own. With no syntactic context, it exemplifies the form in which it is available as a lexical entity in what is termed the ‘mental lexicon’. VP structures in the examples in (3) appear with the lexical head V (schrijven, helpen, klimmen, eten) in final position. They show that in Dutch the VP is lexically stored with head-final structure. (3) Wat ga je doen? [ Haar een brief schrijven ]VP . what go you do? [ her a letter write ]VP [ Haar helpen ]VP , dat kan ik niet. [ her help ]VP , that can I not [ In de boom klimmen ]VP doe ik niet. [ into the tree climb ]VP do I not [ Fruit eten ]VP moet. [ fruit eat ]VP must Words are the building blocks of both NP and VP structures. The structure of a noun or a verb is the result of a creative process of word formation, i.e. of either inflection, derivation or compounding. Inflection is a rule-based morphological process which modifies the form of a word to express grammatical categories such as voice, tense, aspect, number, person and case. For example, morphological rules of inflection in Dutch determine whether plural nouns may end in

12 

 Lexical vs. functional elements

either -en or -s. That is, they account for the fact that the plural form of boom (tree) is bomen, while the plural form of oom (uncle) is ooms. They also account for the fact that the plural form of bal (ball), meaning ‘a solid or hollow sphere’, is ballen, while the plural form of bal (ball) meaning ‘a gathering for dancing’ is bals. They make clear why some people talk about landelijke barren (country bars) and twee kitkatten (two kitkats), while others would rather say landelijke bars and twee kitkats. With respect to verb forms, morphological rules of inflection discriminate between, for example, a singular form leest (reads) and a plural form lezen (read). They signify whether a verb is used with a present tense form as in leest (reads) or with a past tense form as in las (read), whether a verb is used with perfective aspect as in leest (reads) or with perfect aspect as in heeft gelezen (has read), whether it is used with active voice as in leest (reads) or with passive voice as in wordt gelezen (is being read). Derivation is the morphological process that adds a formal element (affix) to a word stem to create a new word form and, hence, a new concept. For example, the affix -loos may turn the noun kans (chance) into the adjective kansloos (chanceless), while the affix -heid may turn the adjective kansloos into the noun kansloosheid (chancelessness). As with the phrasal structures NP and VP, the element functioning as the head determines both the constituents that it can be combined with and the syntactic function of the lexical structure as a whole. Derivations in Dutch are head-final. This explains why bloem (flower) occurs with the definite article de as in de paardebloem (the dandelion), while bloemetje or bloempje (flower-little) occurs with the definite article het as in het bloemetje and het bloempje. Compounding is the morphological process that creates a new word form out of two or more word stems (lexemes). For example, kurkentrekker (corkscrew), boekenplank (bookshelf), gehaktbal (meatball), wijsneus (wiseacre) and schoonmaken (clean-make), vriesdrogen (freeze-dry), wegpoetsen (away-clean) and koekhappen (cook-bite). These compounds are usually head-final.¹ It explains the difference in meaning between washandje (washcloth) and handwasje (handwash), tuinkabouter (gardendwarf) and kaboutertuin (dwarfgarden), vraagprijs (price asked) and prijsvraag (prize contest), schoolvak (school subject) and vakschool (vocational school), hobbelpad (bumpy road) and padhobbel (speed bump). The structure of a morphological element can be the result of different morpho-phonological processes. For example, in Dutch lexical elements with a stem

1 Exceptions are word forms and phrases such as bal gehakt (ball [of] meat), leraar Engels (teacher [of] English), Hansworst (Jack Pudding).

Structure in language  

 13

form ending in /b/, /d/, /g/ and /v/ /z/ /g/ will become voiceless, if they occur in the final position of a word. For example, /b/ in hebben (have) becomes /p/ in ik heb (I have) and /z/ in grazen (graze) becomes /s/ in de koe graast (the cow grazes). This phenomenon is termed ‘final devoicing’. Another example of a morpho-phonological process in Dutch is the formation of diminutives. The phonological shape of a diminutive ending may differ depending on the nominal stem to which it is attached. Thus, hek (gate) and plant (plant) occur with -je as in hekje and plantje; boom (tree) and raam (window) occur with -pje as in boompje and raampje; zoen (kiss) and maan (moon) occur with -tje as in zoentje and maantje; koning (king) and woning (home) occur with -kje as in koninkje and woninkje; and finally, gang (corridor) and zon (sun) use -etje as in gangetje and zonnetje. In sum, each level of syntactic, morphological and phonological structure constitutes its own domain of linguistic categories and rules. These categories and rules constrain the underlying processes of creative construction. Thus, for example, they account for the fact that is op (is all gone) serves as a predicate that holds for de koek (the cake) and not for de koeken (the cakes). Hence, de koek is op is correct, while *de koeken is op² is not. They account for the fact that duim (thumb) occurs with de in de duim (the thumb) and that duimpje (little thumb) occurs with het as in het duimpje. Finally, as a last example, they account for the fact that een schattig hondje (a sweet dog) is correct, while *een schattige hondje is not. The internal structure of both phrasal constituents and complex word forms in Dutch shows that the head constituent regularly occurs in final position.³ This property of word order is characteristic for Dutch, as it is for German. In Romance languages as in French, however, word order is head-initial. Examples of the relevant types of structure both in Dutch and in French are given in (4). (4) The structure of phrases and complex word forms: head-final vs. head-initial Dutch: head-final

French: head-initial

VP de tafel dekken wegrennen [het] eens zijn

couvrir la table partir en courant etre d’accord

lay the table run away agree

2 The use of * indicates that the relevant example is grammatically incorrect. 3 PPs are different. They occur with P in initial position.

14 

 Lexical vs. functional elements

NP de eetkamer de briefkaart de appelsap de kurkentrekker een dik boek Jans boek

la salle a manger la carte postale le jus de pomme le tire bouchon un livre épais le livre de Jean

the dining room the postcard the apple juice the corkscrew a thick book John’s book

AP nog niet helemaal niet

pas encore pas du tout

not yet absolutely not

In Dutch, the head-final structure is limited to phrasal constituents and lexical elements. At the level of utterance structure, i.e. in assertions, questions and imperatives, word order is head-initial. In French, word order in both cases is head-initial. Compare, for example, in (5) word order variation in Dutch with the fixed word order in French. (5) Word order variation in Dutch and French Dutch head-final

head-initial

French head-initial

head-initial

de tafel dekken wegrennen het eens zijn

hij dekt de tafel hij rent weg wij zijn het eens

couvrir la table partir en courant etre d’accord

il couvre la table il part en courant nous sommes d’accord

The use of head-final vs. head-initial word order in Dutch as in (5) shows that the Dutch language system consists of two major components: on the one hand, a lexical component governing word order at the level of phrase structure and complex word formation and, on the other hand, a syntactic component governing word order at the level of utterance structure.

Language structure in production  

 15

2.2 Language structure in production 2.2.1 Phrase structure 2.2.1.1 Lexical selection In language production, it is the aim of the speaker to convey a particular message to the listener. For the expression of the content of this message, he⁴ must access the lexical knowledge that he has acquired and select the lexical phrases that seem appropriate. The selection of these lexical phrases is based on a process of conceptualization. Conceptualization is the process by which the speaker decides upon what he wants to express. Thus, if a speaker wants to refer to an object that people use to live in, he must be in the position to choose with a greater or lesser degree of specificity from lexical alternatives such as huis (house), woning (residence), rijtjeshuis (terrased house), herenhuis (mansion), villa (villa), helft van twee onder één kap (semi-detached) etc. or from lexical items with a particular connotation such as krot (hovel), hut (hut), paleis (palace), optrekje (cottage). Hence, the options for lexical selection are constrained by what the speaker observes and how he perceives the real world. This is why, for example, in my child data one of the children said poes lacht (kitty laughs) at the moment when the cat got angry and started hissing.

2.2.1.2 Phrasal coherence Lexical selection is the selection of a linguistic structure dominated by a lexical head, i.e. a verb, a noun, an attribute or a preposition. The lexical head determines the lexical structure as a whole. Thus, if the speaker uses the verb vallen (fall) as the lexical head of a VP, he has to use is as the auxiliary verb, as in is gevallen (is fallen), and not heeft, as in *heeft gevallen (has fallen). Furthermore, since vallen is a change-of-state verb, he cannot use it with an element that is the object of an action. Thus, hij valt (he falls), but not *hij valt de bal (he falls the ball). On the other hand, if he uses the expression laten vallen (let fall), he has to use heeft as the auxiliary, as in heeft laten vallen, and not is, as in *is laten vallen. Finally, since laten vallen is a verbal expression referring to a causal action, it has to occur with an element that is the object of this action. Thus, hij laat de bal vallen (he lets the ball fall) and not *hij laat vallen (he lets fall).

4 In the following, the masculine pronouns he, him and his are also used generically.

16 

 Lexical vs. functional elements

A similar situation holds for expressions in which the lexical head is a noun. If the speaker selects the noun huis (house) as the head of an NP, he has to use het as the definite article, as in het huis. If he chooses the noun woning (home), the definite article has to be de. On the other hand, if he decides to use the diminutive form, he has to use het in both cases: het huisje, het woninkje not *de huisje or *de woninkje. Finally, it is the head that is semantically specified in compounds as in gokhuis (gamblinghouse), pakhuis (warehouse), seinhuis (signal cabin), spookhuis (haunted house), kaartenhuis (house of cards), poppenhuis (doll’s house), slakkenhuis (snail’s shell) and warenhuis (department store).

2.2.2 Utterance structure Utterances are used to express not only the content of a predication, but also the functional properties of information structure, such as the pragmatic function of perspective taking and contextual embedding. Perspective taking is an option. For example, it is up to the speaker to describe a situation from the perspective of the agent or its patient. Contextual embedding is a constraint. For example, it depends on what the speaker knows about the knowledge state of the listener, if he may start a conversation with de auto staat voor de deur (the car is outside) or with er staat een auto voor de deur (there is a car outside).

2.2.2.1 Perspective taking In a given situation a speaker could say, for example, (6a) or (6b). (6)

a.

De agent heeft de dief op heterdaad betrapt. the officer has the thief in the act caught

b.

De dief is op heterdaad betrapt. the thief is in the act caught

Choosing one utterance rather than the other is a matter of perspective. Thus, as shown in (6a) and (6b), the speaker may take the perspective of either the officer or the thief by selecting one or the other as the subject of the utterance.⁵ As a con-

5 See, for example, Ertel (1977). Furthermore, as argued by Dik: “We shall say that the selection of any one argument as a point of departure for describing the state of affairs is brought about by assigning the syntactic function Subj to that argument. Thus, Subj assignment determines the perspective from which the state of affairs is described” (1978: 71).

Language structure in production  

 17

sequence of this, the utterance is either an active or a passive sentence. Subject selection as the result of perspective taking also plays a role in (7a) and (7b). (7)

a.

Mies droogt de truien op de verwarming. Mies dries the sweaters on the heating

b.

De truien drogen op de verwarming. the sweaters dry on the heating

Here too, it is up to the speaker, to take the perspective of either the agent as in (7a) or the object as in (7b). However, (7b) is possible only under the condition that the agent is left unexpressed. This explains why in the relevant situation either a transitive predicate as in (7a) or an intransitive predicate as in (7b) is used. The morphological properties of verb agreement in Dutch serve to express the relation between the predicate and the subject. However, for the identification of the subject, verb morphology seems to play a minor role. Evidence for this are the utterances in (8) and (9), in which the 3Sg-form of the verb matches with either of the two NPs. (8)

Man bijt hond. man bites dog

(9)

Boer zoekt vrouw. farmer seeks wife

Word order interacting with the intonation contour will guide the interpretation of the initial noun as either subject or object. In the default case, the NPs man (man) in (8) and boer (farmer) in (9) will be interpreted as the subject. If, on the other hand, these NPs are meant to be interpreted as the object, the speaker must use a marked pattern of intonation. Verb agreement in (10a) shows that with the verb passen-3Pl (fit) the NP de kleren (the clothes) is the subject and the NP de man (the man) the indirect object. In the same way in (10b), the NP deze kleren (these clothes) is the subject, while the pronoun hem (him) is the indirect object. (10)

a.

Deze man passen deze kleren niet. this man fit-3Pl these clothes not

b.

Hem passen deze kleren niet. him fit-3Pl these clothes not

18 

 Lexical vs. functional elements

a′.

Deze man past deze kleren niet. this man fits-3Sg these clothes not

b′.

Hij past deze kleren niet. he fits-3Sg these clothes not

However, instead of (10a) native speakers often produce (10a)′, in which the verb past-3Sg (fits) shows verb agreement with the NP deze man (this man). In the same way, instead of (10b) native speakers often produce (10b)′ in which the verb past-3Sg (fits) shows verb agreement with the subject pronoun hij (he). This shows that for native speakers the initial position is the default subject position. It may overrule the indirect object function of an NP. As shown in (6) and (7), subject selection is due to the perspective taken by the speaker. Perspective taking explains why there is a correlation between subject selection and placement in initial position. This correlation is the default case. It accounts for the interpretation of (8) and (9) and the production of (10a)′ and (10b)′. Further evidence is provided by the fact that, if the initial position is not a subject position, utterance structure is marked. An illustration is given in (11). (11)

De man bijt-ie niet. the man bites-he not

Here, as in (8) and (9), agreement does not play a role. However, it is the use of the subject clitic -ie which is evidence for the listener that the element in initial position cannot be the subject. In fact, the use of -ie serves as a kind of topicalization device. Finally, consider the examples in (12a) and (12b). (12)

a.

Wie denk je dat Karel gaat benoemen. who think you that Charles will appoint

b.

Wie denk je dat hij/hem gaat benoemen. who think you that he/him will appoint

It has been shown that in experimental conditions native speakers unanimously interpret the noun Karel in (12a) as the subject and that given the option in (12b) they opt for the subject-pronoun hij (Jordens 1991). The reason for the selection of Karel as the subject and hij instead of hem is due to the fact that with the use of a proper noun or a pronoun the person referred to is presented as identifiable. In (12a) and (12b) this determines the perspective that the speaker will take and,

Language structure in production  

 19

hence, the process of subject selection. As a consequence, the person referred to with wie (who) is interpreted as the object.

2.2.2.2 Contextual embedding Utterances are produced within a particular context. Contextual embedding, i.e. the relation between an utterance and the relevant context, plays a role in the use of the linguistic expression of finiteness, topicalization, definiteness, tense and pronominalization. See the examples in (13). (13)

a.

Karel heeft een brief geschreven. Charles has a letter written

b.

Hij heeft de brief meteen op de post gedaan. he has the letter immediately in the mail done

c.

De brief heeft-ie meteen op de post gedaan. the letter has-he immediately in the mail done

d.

Die heeft-ie meteen op de post gedaan. that-one has-he immediately in the mail done

b′.

Hij heeft een brief meteen op de post gedaan. he has a letter immediately in the mail done

c′.

Een brief heeft hij meteen op de post gedaan. a letter has he immediately in the mail done

In (13), it is the function of the verb form heeft, i.e. the verb form in the position of verb second, to express the functional properties of finiteness. The functional properties of finiteness indicate that the utterance holds true for a given situation. Hence, finiteness is the linguistic expression of the discourse function of assertion. With the use of the definite article (de or het) the speaker expresses the identifiability of the entity referred to. This is the case in (13b) and (13c), because here de brief (the letter) is interpreted with reference to the same entity as introduced in (13a). Within the context of (13a), the use of een brief in (13b)′ and (13c)′ is inappropriate. This is due to the fact that in (13b)′ and (13c)′ with een brief a new referent is introduced, while at the same time with een brief in (13a) this referent has been made identifiable already within the present context. Hence, it can only be referred to as definite as in (13b) and (13c).

20 

 Lexical vs. functional elements

In (14) and (15) de deur (the door) and de zon (the sun) are also referred to as identifiable. However, here the identifiability is inferred from the non-linguistic situational context. (14)

Jan kwam thuis. De deur was dicht. John came home. the door was closed

(15)

De zon schijnt vandaag niet. the sun shines today not

In (14), the identifiability of the referent of de deur (the door) is established on the basis of the linguistic context in which the word thuis (home) is used to refer to someone’s house. This house is the situational context in which a particular door may serve as an identifiable object for the speaker to refer to. In (15), the frame of reference is the real world in which the identifiability of the sun is a given. Contextual embedding is also achieved with pronouns. Both deictic and anaphoric pronouns as means of contextual embedding are used to refer to elements playing a role in the discourse. Usually, but not necessarily, they are part of the current situation. It explains the use of hij (he) in (13b) and die (that-one) in (13d). Furthermore, tense marking establishes the relation between the topic time (TT) as the time that the utterance holds for and the time of utterance (TU) as the time at which the utterance is actually produced. It takes care of the temporal embedding of the utterance as a whole. A past tense form, as for example in De trein kwam op tijd aan (the train arrived in time), indicates that the TT, i.e. the time span of the event that the speaker focusses on, occurs before the TU, i.e. the time at which the utterance occurs (Klein 1994: 36ff.). Finally, the topic situation is the situation that an utterance applies to (Klein 2008: 289). It is the function of the element in topic position to establish a relation with this topic situation. Topicalization refers to the process in which an element that is part of the verb-argument structure is placed sentence-initially, i.e. in functional, topic position. Topicalization of the patient occurs both in (13c) and (13d). Utterances, as pointed out before, are used to express both the content of a predication and the functional properties of information structure. Linguistic elements used to express the content of a predication are categorized in terms of lexical categories. Linguistic elements and devices used to express the functional properties of information structure are categorized as functional categories.

Language structure in production  

 21

2.2.2.3 Lexical categories Lexical elements are the linguistic representation of ‘referential meaning’. They are used to refer to situations (e.g. actions, states, changes of state), entities (e.g. individuals, objects, sensations, notions, situations) and attributes (e.g. properties, locations, directions). They establish the linguistic representation of the outside world in terms of syntactic categories such as the verb phrase (VP), the nominal phrase (NP), the attribute phrase (AP) and the prepositional phrase (PP). Lexical elements as the linguistic expression of referential meaning are semantically independent, both of each other and of the linguistic context in which they occur. The categories they belong to are therefore open class categories. Hence, new words may come into use (e.g. in Dutch beurtbalkje⁶, homohoefjes⁷), while once common words may become obsolete (e.g. in Dutch nozem⁸, smiecht⁹). Although there is a great amount of overlap in lexical knowledge among the members of a linguistic community, the mental lexicon of one speaker may differ considerably from that of other speakers. This is the reason for why there are dictionaries.

2.2.2.4 Functional categories Functional elements are the linguistic representation of ‘grammatical meaning’, i.e. they are the linguistic expression of relations of cohesion at different levels of linguistic structure.¹⁰ Categories of functional elements relevant for the syntactic structure of an utterance are gender, agreement and case marking. The gender of a noun in Dutch, i.e. the use of a noun as a de-word as de boom (the tree) or a hetword as het bos (the forest), is relevant for the morphology of the nominal phrase as a whole. Hence, een grote boom (a big tree) vs. een groot bos (a big forest). Case marking (nominative, accusative, dative etc.) plays a role at the level of the predicate and predicate-argument structure. It specifies the syntactic function of a nominal constituent. Agreement in person and number is a function of the hold-for relation between the predicate and the subject. Categories of functional elements relevant for the informational function of contextual embedding are finiteness, topicalization, definiteness, tense and pronominalization. Finiteness

6 A rod used to separate one‘s sale items on cash register tape. 7 Heeled boots. 8 Greaser. 9 Crook, rogue. 10 It is for this reason that, in the linguistic literature, functional elements are often characterized as the ‘cement’ that constitutes the connection between lexical elements. However, ‘cement’ is a metaphor. Linguistically it is meaningless.

22 

 Lexical vs. functional elements

is the linguistic expression of the discourse function of assertion. Topicalization establishes the relation between the utterance and the topic situation, i.e. the situation that the utterance applies to. As with finiteness and topicalization, it is the function of definiteness, tense and pronominalization to connect the utterance with its linguistic or extra-linguistic context. Definiteness indicates the identifiability of the entity referred to. Tense establishes the relation between the topic time (TT) as the time that the utterance holds for and the time of utterance (TU) as the time at which the utterance is actually produced. Pronominalization is used to refer to an entity playing a role in the linguistic context. Functional elements are the morpho-syntactic expression of a relation of cohesion. Members of a functional category are used in opposition to each other. This is because the expression of a relation only makes sense if there are alternatives to choose from. Given the fact that the use of a functional element depends on the relation to the other members of its category, individuals within a linguistic community are not free to add a functional element or get rid of one. If they did, the functional meaning of each of the other elements would change and as a consequence communication would become impossible. This explains why functional elements are members of a closed class. As members of a closed class they are part of the linguistic knowledge shared by all native speakers of a language community. The lexical and functional categories that are relevant in Dutch are summarized in Table 1.

2.2.3 Word formation Lexical elements in Dutch may occur as elements of complex lexical entities. In addition to, for example, huis there is huizen (houses), huisje (house-little), huiselijk (home-loving), verhuizen (move-house), uithuizig (out-of-home), behuisd (housed), huiswerk (homework), huisbaas (home-boss); in addition to, for example, woon there is woont (lives), wonen (to live), gewoond (lived-Pp), woonde (lived-Past), bewonen (inhabit), bewoner (inhabitant), verwonen (pay-for-living), woonachtig (living), bewoonbaar (livable), woonboot (living-boat), woonerf (living-ground). These examples are evidence of some of the possibilities of word formation in Dutch. They show that a lexical form can be combined with morphological elements such as -t, -en, -de, ge …(e)d, be-, ver-, -lijk, -achtig, -baar or with another lexical form (lexeme) as in huiswerk, huisbaas, woonboot or woonerf. Processes of word formation with morphological elements are traditionally categorized as either derivation or inflection.

Language structure in production  

Lexical

Functional

verb phrase (VP)i nominal phrase (NP)ii attribute phrase (AP)iii prepositional phrase (PP)iv

gender (de vs. het) case (nominative, accusative, dative etc.) agreement (person, number) finiteness (verb-second)v topicalization definiteness tense pronominalization

 23

i

Subcategories are simple lexical verbs, complex lexical verbs with a verbal particle (Prt) such as in (in), op (on), aan (on), uit (out), deontic modals such as wil as in ik wil het niet (I want this not) or kan as in ik kan toveren (I can do-magic) or the copula as in de koek is op (the cake is all-gone). ii Subcategories of NPs are nouns with their attributes, proper names and other deictic elements. iii Subcategories are APs with adjectives or adverbials. Adverbials include modal particles such as toch (still), maar (just), misschien (maybe), best wel (really) and scope particles such as ook (too), zelfs (even), weer (again), niet (not). iv Prepositional phrases are in fact attributes, cf. op de stapel / bovenop (on the pile / on top). v The morphological properties of finiteness are carried by verbal elements in second position. The position of verb-second is taken by auxiliary verbs, i.e. both the aspectual auxiliaries heb, heeft (have, has) and ben, is (am, is) as well as the epistemic modals such as kan as in de bagage kan eruit vallen (the luggage can fall out) or mag as in dat mag niet gebeuren (that is not allowed to happen). In the absence of an auxiliary verb, it is the turn of the lexical verb to take up the position of verb-second. Table 1: Lexical and functional categories in Dutch

Morphological processes of derivation are geared to create new lexemes, and hence, new concepts. Through the addition of elements such as be-, ver-, -ig, -lijk, -tje the new lexeme often becomes a member of a different word class. This phenomenon of transposition occurs with, for example, dik – dikte – dikkerd (big – bigness – biggie), lopen – loper (walk – walker), storm – stormen (storm – to storm), huis – verhuizen – huiselijk (house – move-house – home-loving), drinken – de drank (to drink – the drink), mooi – een mooie (beautiful – a beautiful-one), groot – groots (great – grand), stad – stads (town – towny), maken – maakbaar – maakbaarheid (make – makeable- makeableness), groot – vergroten (large – enlarge). In sum, derivation can be characterized as a process that ‘feeds’ word formation. Morphological processes of inflection create different word forms of the same lexeme by adding elements such as -0, -en, -t, -er, -est. For example, the verb woon may occur as woont (lives), wonen (to live), gewoond (lived-Pp); the noun tuin (garden) may occur as tuinen (gardens); the adjective groot (big) may occur

24 

 Lexical vs. functional elements

as groter (bigger) or grootst (biggest). Traditionally, derivation and inflection are seen as morphological processes that are grammatically different. This position is known as ‘split morphology’.

2.2.3.1 Derivation, inherent and contextual inflection Inflection is the use of different word forms of the same lexeme. This holds, for example, for both verb agreement as in maak (make-1Sg), maakt (make-2,3Sg), maken (make-1,2,3Pl) and aspect marking as in maken (making-Progressive) vs. gemaakt (made-Perfect). However, verb agreement depends on its syntactic context, i.e. the person and number of the subject, while aspect marking is syntactically independent. The same distinction holds for adjective agreement as in een groot bos (a big forest) and een grote boom (a big tree) compared to number marking as in de boom (the tree) and de bomen (the trees). Therefore, Booij distinguishes between verb agreement as a case of contextual inflection, i.e. “inflection that is dictated by syntax”, and aspect-marking as a case of inherent inflection, i.e. “inflection that is not required by the syntactic context” (Booij 1996: 2; 2000: 365). Contextual and inherent inflection are also categorizations of a semantic difference. While word forms with contextual inflection are semantically transparent, word forms with inherent inflection may develop idiosyncratic meaning. See, for example, gemaakt (made-Pp) in zijn lachen is gemaakt (his laughing is unnatural) or boeken (books-Pl) in de accountant controleert de boeken (the accountant examines the business accounts). Finally, word forms with inherent inflection may feed word formation, while word forms with contextual inflection may not. Thus, een gemaakt lachje (a forced laugh-little), etens-tijd (dinner-time), meisjes-achtig (girl-like), huizen-markt (housing market) are examples of inherent inflection feeding word formation, while *maakt-werk (makes-work) and *zwarteboek (black-book) are examples of contextual inflection with which feeding is not possible. Hence, given that inflection is the use of different word forms with the same lexeme, word forms with contextual inflection are (a) syntactically dependent, (b) they are semantically transparent, and (c) do not feed word formation or compounding. Word forms with inherent inflection are (a) syntactically independent, (b) they can be used with idiosyncratic meaning, and (c) may feed derivation and compounding. Therefore, as illustrated in Table 2, instead of an account in terms of split morphology (derivation vs. inflection), Booij proposes a three-way distinction in terms of contextual inflection, inherent inflection and derivation. Booij notes that this categorization is in fact a scale whereby “contextual inflection is peripheral with respect to inherent inflection, and inherent inflection is peripheral with respect to derivation” (2000: 366).

Language structure in production  

Derivation new lexeme / transposition

Inflection

syntactically semantically word formation

independent idiosyncratic feeding

same lexeme / no transposition Inherent Contextual independent dependent idiosyncratic transparent feeding no feeding

verbal

e.g.: ver-, ont-, be-.

aspect

vermaken (remodel)

maken; gemaakt (make-Inf; made-Pp)

e.g.: -er -ing, -sel, -heid, -schap, -dom. maker; maaksel (maker; make-N) e.g.: -baar, -lijk, -ig, -achtig, -loos. maakbaar; vermakelijk (makeable; amusing)

number

examples:

nominal examples: adjectival examples:

 25

maker; makers (maker-Sg; makers-Pl) comparative; superlative maakbaarder; vermakelijkst (more makeable; most amusing)

verb agreement; tense¹¹ maakt vs.maakte vs.maken (makes-3Sg vs. made1.2.3Sg vs. make1,2,3Pl) case marking¹²

adjective agreement maakbaar vs. maakbare (makeable vs. makeable)

Table 2: Three-way categorization of word formation

It seems true that there is no principled distinction to be made between the morphological processes underlying inherent inflection and derivation. In both cases word forms are subject to some degree of lexicalization, i.e. the word form as a whole may appear with idiosyncratic meaning which allows it to serve as input for processes of word formation. Thus, on the basis of their morphology, plural nouns, infinitives and participles are usually categorized as inflectional forms of nouns and verbs. On the basis of their semantics however, they can also be categorized as examples of derivation. This phenomenon that Booij refers to as ‘lexical split’, accounts for the fact that “the semantic relation between two formally related words is no longer transparent” and, therefore, that inflection has

11 In Booij (1996: 2; 2000: 365) 'tense' is categorized as inherent inflection, because 'tense' seems syntactically independent. 12 In Dutch, case marking plays a role only in pronouns.

26 

 Lexical vs. functional elements

“not purely a compositional function of the meaning of its morphological constituents” (2000: 364). Lexicalization explains why plural nouns may refer to entities with the semantics of collectivity if they are unanalysed, i.e. not in opposition to a singular form. Examples are buren (neighbours), kosten (costs), troepen (troops), zaken (affairs), haren (hair-Pl), gelden (money-Pl). It also explains why unanalysed plural nouns are used with the function of an attribute in compounds as in boekenbeurs (bookfair), meisjesachtig (girl-like) and kinderstoel (child-chair). Similarly, infinitives may refer to an event as an entity, if they are not used in opposition to finite verb forms. It explains why they are used as nouns in, for example, het eten staat op tafel (the food is on the table), etenslucht (smell of cooking), het is een komen en gaan (it is a coming and going). And finally, participles may also be unanalysed. It explains why they are used as attributes as, for example, afgelopen (finished), gelikt (polished), gewaagd (risky), beleefd (polite), bedroefd (sad), besloten (private). In sum, unanalysed plurals, infinitives and participles may each behave syntactically differently from the inflectional form. This, however, is precisely what derivation is about, i.e. the creation of a new lexeme for a new concept. Hence, there seems to be no reason why there should be a categorical distinction between inherent inflection and derivation.

2.2.3.2 Lexical and functional categories of word formation Finite verb forms and adjectives occur with what Booij refers to as ‘contextual inflection’. Word forms with contextual inflection are semantically transparent. That is, the meaning of the word form as a whole is predictable on the basis of its morphological components. For word forms with contextual inflection categorical reanalysis as an unanalysed word form with an idiomatic meaning is impossible, nor is it possible for these word forms to feed word formation. This explains why, for example, the finite verb form maakt cannot occur in compounds such as *maaktwerk or *maaktheid, while the use of maak in maakwerk (frame-up) and gemaaktheid (artificial) is correct. Similarly, it explains why, for example, the adjective form goede cannot occur in *goedekoop or *goedeheid, while the use of goed in goedkoop (cheap) and goedheid (goodness) is correct. Tensed verb forms are semantically transparent, too. However, in Booij (1996: 2; 2000: 365) tense is categorized as syntactically independent and therefore morphologically as an instance of his category of ‘inherent inflection’. The reason for this is possibly the fact that tense as opposed to verb agreement is not required by its syntactic context. However, as is the case with finite verb forms, tensed verb forms cannot be recategorized, nor can they feed word formation. Hence, the verb form bezorgde (delivered) in het bezorgde pakje (the delivered package)

Language structure in production  

 27

comes from the past participle bezorgd, not from the past tense form bezorgde. As a past participle form bezorgd can be recategorized as an adjective, while as a past tense form it cannot. Furthermore, incorrect word forms such as *zorgdelijk and *zorgde-instelling as opposed to the correct alternatives zorgelijk (precarious) and zorginstelling (institution to provide care) show that the tensed verb form zorgde (delivered) does not feed either derivation or compounding. Thus, as is the case with the inflectional marking of finiteness, verb forms marked for tense are semantically compositional. They will never be used with idiosyncratic meaning and are therefore not to be categorized like, for example, verb forms that are marked for aspect. The semantic transparency of finite and tensed verb forms is due to the fact that the semantic contribution of both finiteness and tense does not apply to the lexical element to which they are attached. While the verb as it occurs in second position serves as a carrier of finiteness and tense, these functional properties are properties of the utterance as a whole. Finiteness as a functional property of the utterance serves to express the pragmatic function of an assertion. As an assertion the utterance applies to a particular topic situation. Tense establishes the relation between the time span that the utterance applies to and the time of the utterance. Thus, both finiteness and tense are relevant with respect to the situational context, not with respect to the meaning of the verb.¹³ Hence, it is the functional properties of finiteness and tense due to which the semantic transparency of finite and tensed verb forms is warranted. In sum, there is a categorical distinction to be made between derivation as the morphological process to create a new lexeme and inflection as the morphological process to create a different word form of the same lexeme. The relevant distinction is based on the semantic difference between idiosyncracy vs. transparency.¹⁴ If a complex word form such as a plural noun, an infinitive or a participle is used with idiosyncratic meaning, it functions as a morphological entity. This lexical process of word formation establishes a new lexeme. This new lexeme may undergo transposition to another lexical category and, furthermore, it may serve to feed word formation, too. On the other hand, complex word forms such as finite and tensed verbs are semantically transparent. They may not feed word forma-

13 In a similar vein, Booij states that “inflectional markers often do not pertain to the meaning of the complex word itself, but express the relation of a word to situation and context.” He also notes that “[t]ense (…) expresses the time relation between the event or situation expressed by the verb and the moment of speaking” (Booij 2000: 366). 14 Thus, it seems that the relevant distinction is not a matter of degree as claimed in Booij: „[C]ontextual inflection is peripheral with respect to inherent inflection, and inherent inflection is peripheral with respect to derivation” (Booij 2000: 366).

28 

 Lexical vs. functional elements

tion. Word forms that are semantically transparent are the result of a functional process of word formation in the sense that they are sensitive to the situational or grammatical context. While the situational context plays a role with verb forms that are marked for finiteness and tense, the grammatical context plays a role in case marking and adjective agreement. In both cases, inflection is the linguistic representation of grammatical meaning, i.e. it is the linguistic expression of some form of cohesion. As a functional process, cohesion, thus, occurs either at the level of the discourse in the use of finite and tensed verb forms or at the level of utterance structure in the use of case marking and adjective agreement. In both cases, cohesion is a means of anchoring linguistic structure into a larger whole. The categorical difference between lexical and functional processes of word formation becomes evident in the order in which the relevant morphemes appear. Morphemes serving lexical processes of word formation occur before morphemes that are used to express a functional meaning. This explains why in blauw-achtig-e (blue-like) the adjectival morpheme -e occurs after the derivational morpheme -achtig, why in macht-ig-de (authorize) the past tense morpheme -de occurs after the derivational morpheme -ig, and why in German as in, for example, Büch-er-n (books-Dat) the case morpheme -n occurs after the plural ending -er. Thus, as far as the morphological structure of word forms is concerned, there is a semantic opposition between idiosyncracy and transparency. This opposition, which has traditionally been accounted for in terms of derivation vs. inflection, is due to a categorical difference between lexical and functional processes of word formation. This two-way categorization of the relevant types of word formation is represented in Table 3.

syntactically semantically word formation verbal derivation nominal derivation adjectival derivation

Lexical

Functional

new lexeme / transposition independent idiosyncratic / lexical feeding

same lexeme / no transposition dependent transparent / functional no feeding

aspect verb agreement; tense e.g.: ver-, ont-, be-. number case marking e.g.: -er -ing, -sel, -heid, -schap, -dom. comparative; superlative adjective agreement e.g.: -baar, -lijk, -ig, -achtig, -loos.

Table 3: Two-way categorization of word formation

Language structure in production  

 29

It seems interesting to note that the distinction between lexical-idiosyncratic and functional-transparent word forms also plays a role in the agrammatic language of patients with Broca’s aphasia. While these patients are able to use verb forms with idiosyncratic meaning such as derivations, infinitives and participles, they are unable to produce verbs forms that are semantically transparent such as finite and tensed verb forms (de Bleser and Bayer 1988). In sum, within the framework proposed above a dichotomous distinction can be made between lexical and functional processes of word formation. Lexical processes of word formation occur with complex word forms, if they are going to develop idiosyncratic meaning. As a result of this, transposition may occur, i.e. as a new lexical entity, a complex word form may become a member of another word class. Lexical processes of word formation are involved in the acquisition of lexical knowledge as an ongoing process in which new words come into use, while old words become obsolete. Functional processes of word formation occur with word forms in which the internal structure of the lexical entity remains semantically transparent. With functional processes of word formation the development of idiosyncratic meaning is impossible, because the inflectional element does not contribute to the meaning of the word. It serves the function of embedding within a larger context. Functional elements of word formation are, therefore, “visible on the top node of its word, because otherwise it is invisible for the syntactic context” (Booij 1996: 8). This explains why the past tense of complex verb forms such as roerbakken and vriesdrogen cannot be *roerdebak or *vroordroog, but occur as roerbakte and vriesdroogde. It also explains why the 3Sg-Pres form of roerbakken and vriesdrogen is not *roertbak and *vriestdroog, but roerbakt and vriesdroogt. Finally, it explains why regardless of the use of plural morphology, a complex Dutch noun as geitenwollensokkenangsthaas (goat-woollen-socks-coward) is a singular form. A similar observation can be made with respect to the use of the article in German compounds. Regardless of the use of male and female nouns, the article is the neuter form das for a complex German noun as Terrorbekämpfungergänzungsgesetz (terror-fighting-completion-law). In sum, lexical processes of word formation such as aspect marking with verbs and number marking with nouns serve the expansion of conceptual knowledge. Functional processes of word formation such as finiteness and tense with verb forms and case marking and adjective agreement with nouns are the linguistic expression of anchoring and cohesion, either at the level of the discourse or at the level of utterance structure.

30 

 Lexical vs. functional elements

2.3 Learner systems Language development is a stepwise process both in children learning their mother tongue and in adults learning a second language. It is stepwise in the sense that some structural properties of the target language are acquired as prerequisites to other properties. In children, language development is remarkably similar across individuals. Within a limited period of time, every child acquires the adult language system. Second language learners, however, show a lot of variation both with respect to the time they need and the ultimate level they can obtain. Nevertheless, for adult learners, too, the developmental route as such appears to be rather similar. Evidence for this has been found in studies on the acquisition of German as a second language by native speakers of Italian and Spanish within the framework of the Heidelberg Project (Klein and Dittmar 1979) and by native speakers of Italian, Portuguese and Spanish within the ZISA study (Clahsen, Meisel and Pienemann 1983).¹⁵ Both these projects were carried out in situations in which the learners acquired the target language spontaneously, that is, without the help of instruction. As in child first language acquisition, these learners were only exposed to target language input as it occurred in situations of daily use. Following up on these early studies, a longitudinal project on Second Language Acquisition by Adult Immigrants funded by the European Science Foundation (ESF) was carried out among learners with different language backgrounds learning French, German, English, Dutch and Swedish as a second language. In this project, it was found that “after some time, all 40 learners investigated developed a relatively stable system to express themselves”. Furthermore, Klein and Perdue, who coordinated this study, noted that this learner system (a) “seemed to be determined by the interaction of a small number of organizational principles”, (b) “was largely (though not totally) independent of the specifics of source and target language organization”, (c) “was simple, versatile and highly efficient for most communicative purposes” (1997: 303). They also found that at some point learners appear to be comfortable with a typical learner system which they refer to as ‘the Basic Variety’ (BV). As Klein and Perdue argue, “[f]or about one-third of the learners investigated, acquisition ended on this structural level; some minor variation aside, they only increased their lexical repertoire and learnt to make more fluent use of the BV” (1997: 303). These observations led Klein and Perdue to claim that their notion of a Basic Variety is fundamental for understanding why the human language system functions the way it does.

15 ZISA = Zweitspracherwerb italienischer, portugiesischer und spanischer Arbeiter

Learner systems 

 31

“We believe that the BV not only plays a particular role in the process of second language acquisition but also that it represents a particularly natural and transparent interplay between function and form in human language. In a way, fully fledged natural languages are but elaborations of this BV. They add some specific devices such as inflectional morphology or focus constructions; they also add some decoration, pleasant to the ear, hard to learn, but faithfully handed down from one generation to the next. But essentially, they build on the same organizational principles” (Klein and Perdue 1997: 303f.).

2.3.1 Utterance structure at the initial stage Language development as a stepwise process is evident in child first and adult second language learners of Dutch. The examples in (16) and (17) are produced by both child and adult learners at the initial stage of their language acquisition process. They are typical for learners with a language system that serves as a Basic Variety. (16) child L1 Dutch Jasmijn (1;10–2;0)

Andrea (2;0–2;2)

die opa koffie. that grandfather coffee pop ook blaad. doll also sheet poesje vinger bijte, magwel. kitty finger bite-Inf, may-indeed poppie haartie wast. doll hair washed-Pp Mijnie tiktak om. M watch on Mijnie kan losmake. M can loose-make-Inf nee thee, lust nie. no tea, like-1Sg not valtie om. falls-3Sg-he over kanwel optille. can-indeed up-lift-Inf zo moettie. this-way must-it

koffietijd nog niet. coffee time not yet deze ook mooi. this also beautiful eve jurk uitdoen. just dress off-do-Inf Cynthia ook bote kijkt? C too boats looked-Pp goene aan. shoes on? Jaja mag dop opdoen. J may cap on-do-Inf bokkies kome aan. goats come-3Pl to gaatie niet? goes-3Sg-it not? disse hoeniet meeneme. this must-not with-take-Inf deze mag ook niet. this may also not

32 

 Lexical vs. functional elements

dit is dop. this is cap

deke is nou? blanket is now?

(17) adult L2 Dutch Osman (L1 Turkish, 1.1–2.7)

Mohammed (L1 Arabic, 1.6–2.1)

vandaag beetje wandelen. today a-little walk-Inf ik niet lassen. I not weld-Inf misschien twaalf uur huis. maybe twelve o’clock home en dan nederlandse les ook. and then dutch lesson too hoe kost dan? how costs-3Sg then? mag niet, niet koed kafe. [it] may not, not good bar en dan krant lezen. and then newspaper read-Inf niet slapen, moet film klaar. not sleep-Inf, must film ready ik beetje boek lezen. I a-little book read-Inf ik moet andere huis kijken. I must other house look-Inf die is nou werken. that-one is now work-Inf ikke komt hier werken. I comes-3Sg here work-Inf nee, kan niet. no, cannot

duiven ook kopen. doves too buy-Inf hij wil kopen een renault. he wants buy-Inf a renault ik lees die boek. I read-1Sg that book ik niet lezen op de huis. I not read-Inf on the house dan hij pakken een mayonaise. then he get-Inf a mayonnaise maar jij pakt mij zoon. but you get-2Sg my son hier pauze hier eten. here break here eat-Inf hij eten die ei. he eat-Inf that egg ik morgen terug naar huis. I tomorrow back to home hij is crossfiets spelen. he is cross-bike play-Inf hij doet niet. he does [= works] not zij ook woont bij hij ouders. she too lives-3Sg with he parents komt daar ook allemaal die boermensen. comes there too all that farm-people

The examples in (16) and (17) show that, at the relevant stage, there is no productive use of functional elements such as auxiliary verbs, tense and agreement markers, determiners, wh-words and prepositions, and also no variation in word order. These child L1 and adult L2 utterances are typical for learners at the initial stage of the acquisition process. At this stage, learners appear to produce their utterances with lexical elements only. The Basic Variety, being the learner system at the initial stage of language acquisition, is thus essentially a lexical system.

Learner systems 

 33

The ultimate goal, i.e. the end state of the acquisition process, is the creation of a language system in which the integration of both lexical and functional properties of the target system has been achieved. As will be argued, the creation of a functional system is a process that occurs abruptly. It is the result of a developmental process in which learners acquire the linguistic means for the embedding of an utterance into the relevant situational context.

2.3.2 Word formation at the initial stage As far as the processes of word formation are concerned, it is claimed that the dichotomy between inflection and derivation is due to the categorical difference between lexical and functional processes of word formation. Word forms based on lexical processes of word formation are semantically idiosyncratic and syntactically independent. The prediction is that they are learned early, initially unanalysed and analysed possibly later. In a similar vein, Snow distinguishes between ‘semantically strongly-based systems’ and ‘semantically weakly-based systems’. Semantically strongly-based systems such as singular vs. plural are “based on obvious distinctions that are important in our understanding of the world” (1976: 151). Semantically weakly-based systems, however, need first to be noticed before their grammatical function and distribution can be acquired. Semantically strongly-based morphology is, therefore, easier to acquire than syntactically weakly-based morphology. Thus, initially, i.e. at the lexical stage, word forms may appear as marked for number, aspect, comparative and superlative. Word forms based on functional processes of word formation such as verb agreement, tense marking, case marking or adjective agreement are learned later. This prediction seems to be supported by data from studies on language acquisition and specific language impairment. Word forms that are semantically idiosyncratic and syntactically independent are acquired earlier than word forms that are semantically transparent and syntactically dependent. Snow (1976), for example, pointed out that the distinction between both types of morphology explains why it is that in the morpheme order studies on L2 English -ing and plural marking appear to be acquired before the 3rd person singular -s. Furthermore, in Schaerlaekens and Gillis (1987), it is argued that plural nouns and infinitives are learned before finite verb forms.¹⁶ Slobin provides evidence that aspectual distinctions

16 Finite verb forms occur from early on. However, as I will show in Chapter 3, at the initial stage, they are unanalysed.

34 

 Lexical vs. functional elements

occur before person marking (1986: 10).¹⁷ Early studies on the acquisition of morphemes both in L1 and L2 English by De Villiers and De Villiers show that word forms with an idiosyncratic meaning, such as the present progressive, the plural, the past irregular and the possessive are learned before word forms that are semantically transparent such as the past regular and the third person singular (1986: 68). Finally, in a similar vein, the distinction between lexical and functional processes of word formation also explains why it is that the Basic Variety, as a potential stage of fossilization, is a system in which “all features are weak” (Klein and Perdue 1997: 337).

2.4 First and second language acquisition Learners of a second language hardly ever reach the so-called ‘end state’, i.e. the target state of a fully-fledged language system. Children, on the other hand, learning their mother tongue, usually achieve native competence within a relatively short period of time. Why is it that adult second language learners cannot do the same? Language learning in children occurs on the basis of the linguistic input that is available for the purpose of processing. However, children are confronted with a particular learning problem, due to the so-called ‘poverty of stimulus’. That

17 It seems that morpho-syntax is not only difficult to acquire but also difficult to lose. Clahsen (1989), for example, has shown that for learners with SLI it is particularly difficult to acquire subject-verb agreement, case and article-noun agreement, while, on the other hand, in Clahsen and Rothweiler (1992) it is found that SLI children do not have problems with the acquisition of past participles. On the other hand, Hutz, who investigated processes of attrition in German immigrants in the United States, concludes that “it is the lexicon which is first and most severely affected by language attrition, whereas morphology and syntactic structures seem to be more resistant to language loss” (2004: 203). Why should this be the case? Morpho-syntax serves the function of contextual embedding, i.e. the function of anchoring an utterance in time and space. In order for learners to acquire the linguistic means of contextual embedding, they have to give up their initial ego-centric perspective on language processing in favour of a context-sensitive view. For children with SLI, this may be precisely the problem they have to face in language acquisition. Thus, it seems that a delay in the cognitive development of SLI children explains why the acquisition of morpho-syntax takes a while. On the other hand, the morpho-syntactic properties of the functional category system are vital for the expression of the informational function of the language system. Given that this functional system plays a role in every situation of language use, attrition would mean loss of the informational function of language. Thus, it seems that its communicative relevance and, therefore, its frequent use explains why morpho-syntax is less prone to attrition than the lexicon.

First and second language acquisition 

 35

is, utterances addressed to children are limited in their repertoire and also often not targetlike, ‘defective’ as it is called. In addition, it has been shown that feedback with respect to the formal correctness of a child’s utterance has no effect on the child’s developing linguistic competence. Nevertheless, children are able to acquire linguistic knowledge of the target language system, i.e. the linguistic knowledge that allows them not only to produce utterances that are grammatically correct, but also to judge the grammatical correctness of any given utterance. In order to account for this, children are assumed to have access to a linguistic capacity that is geared to the acquisition of the formal properties of language. It is this capacity that children are supposed to have access to only for a limited period of time. The time span during which children are supposed to be able to acquire a language in a native-like fashion is called ‘the critical period’ (Lenneberg 1967: 176). This limited period of linguistic sensitivity should explain why adults are unable to learn a language as children do. Not only do adult language learners hardly ever achieve native language competence, they also show a great deal of variation with respect to the level of competence they will eventually achieve. As straightforward as this ‘window-of-opportunity’ approach may seem, on second thought it appears less self-evident. Although adult second language learners usually do not achieve the same level of competence as native speakers, there is no evidence that they should not be able to use the same principles of language acquisition and achieve the same type of linguistic knowledge as children do. For example, Bley-Vroman (1990) and Clahsen and Muysken (1986, 1989), who tried to show that the mechanisms underlying L2 acquisition are fundamentally different from the mechanisms underlying L1 acquisition, have been proven wrong. Rather, as Schwartz, for example, has claimed: “the nature of interlanguage ‘grammars’ is comparable to the nature of native-language grammars” (1996: 227). Hence, it seems fair to assume that the interlanguage systems of both first and second language learners are instantiations of a common set of possible language systems. Given that L2 learners are able to operate on the basis of the same principles of language acquisition as children, why is it, then, that adult second language acquisition appears so different from child first language acquisition? It seems there are two reasons for this. First, adults and children differ with respect to the way in which they process linguistic input, particularly at the initial stage. For example, what counts as a linguistic entity for a child may not count as a linguistic entity for an adult. Second, adult learners are faced with the problem of keeping the linguistic systems of their native language and the second language separate. This is because the linguistic rules and categories of a second language system are likely to be processed and acquired in terms of the linguistic knowledge system acquired first.

36 

 Lexical vs. functional elements

Assuming that the principles of spontaneous, i.e. untutored, language acquisition in children and adults are essentially the same, what does this mean in terms of developmental progress over time? It seems that two main driving forces are going to play a role. On the one hand, i.e. at the initial stages of acquisition, language learning will be geared to achieve maximal communicative potential with a minimal amount of formal complexity. For the purpose of communication, the content of the message is what matters. Hence, knowledge of the lexical category system is indispensable, or as Hatch phrases it: “When our first goal is communication, when we have little of the new language at our command, it is the lexicon that is crucial. If we can but find the words, we know we can take care of our immediate needs. The words (along with all the gestures we have to help us get meaning across) will make basic communication possible. As Krashen has pointed out, learners don’t carry grammar books around in their pockets, they carry dictionaries” (1983: 74).

Thus, at the initial stage, the learner’s system is grammatically simple, i.e. linked to the here and now, while the focus of the learning effort is on lexical expansion. At this lexical stage, the learner system is constrained by the general principles of the Basic Variety as outlined in Klein and Perdue (1997)¹⁸. On the other hand, i.e. at a more advanced stage of acquisition, the learner’s system will develop such that it is able to accommodate for the contextual variability of any given discourse situation. This means that the learner system will become grammatically more complex, i.e. independent of the here and now. At this functional stage, language learners have to find out how to express the pragmatic function of an utterance and how to accommodate an utterance to the situational context that it applies to. That is, they have to acquire the linguistic properties of information structure as they are established with the functional categories of finiteness and topicalization.

18 Klein and Perdue (1997: 313) distinguish between three types of constraints: phrasal constraints, semantic constraints and pragmatic constraints. Phrasal constraints are, for example, NP1-V or NP1-V-NP2. They pose restrictions on possible sentence structures. Semantic constraints are: SEM1: The NP-referent with highest control comes first; SEM2: Controller of source state outweighs controller of target state; SEM3: Theme before relatum in target position. The pragmatic constraint referred to is: PR1: Focus expression last.

Research questions  

 37

2.5 Research questions Language learning is a developmental process. At the initial stage, utterance structure is purely lexical. Lexical categories are learned on the basis of linguistic input. At some point in the acquisition process, the lexical system develops into a targetlike, functional system. At this advanced stage of acquisition, learners appear to be able to acquire a language system in which two knowledge systems interact. First, there is the knowledge system of how ‘things’ are called. Everything that is relevant has a name. This holds for entities and attributes and for the way in which they play a role in actions, states and changes of state. This referential function is executed with the linguistic knowledge that is stored in what is called the mental lexicon. Second, there is the knowledge system of how to use lexical structure for the expression of an utterance, i.e. for a linguistic structure that is interpreted as an assertion, a question or a command. This informational function is executed with the linguistic categories serving the expression of functional knowledge. It seems obvious that from a communicative point of view lexical knowledge is a prerequisite to functional knowledge. In a communicative situation a speaker may confine himself to the expression of lexical structures with only referential function, because he knows that he can rely on the inferential competence of the listener. Thus, from the point of view of communicative relevance, there is an implicational relationship between lexical and functional knowledge. This relationship seems to be reflected in children’s cognitive development. Children are initially ‘self-centered’, i.e. whatever they perceive and whatever they pursue, they assume that it will be understood by the listener. It takes them a while to find out that this assumption is wrong. This means that, in order to be able to communicate successfully, children have to learn how to take into account the point of view of the listener. This process of cognitive development is described by Piaget (1983) as a process in which an initial, egocentric point of view is given up in favour of an adult-like social orientation which is essentially context-sensitive and cooperative. Thus, both from a linguistic-acquisitional and a cognitive-developmental point of view knowledge of the lexical system seems a prerequisite to knowledge of the functional system. The lexical system determines the shape of the early learner varieties. The functional system determines the way in which advanced varieties of learner languages are used. Research on language development as a process in which an early lexical system develops into an advanced functional system poses the following empirical questions: 1. Given that the language system of learners at the lexical stage is structurally different from the language system at the functional stage, what is the nature of the acquisition process? The claim is that language acquisition is

38 

2.

3.

 Lexical vs. functional elements

not simply a process of the accumulation of linguistic knowledge. Rather, it seems that it is a process of structure building that is the result of the acquisition, integration and restructuring of lexical and functional linguistic knowledge. What are the driving forces for learners to give up a relatively simple lexical system in favour of a complex, functional system? The causal factor seems to be the acquisition of the linguistic means for the embedding of an utterance into the situational context. This explains the additive value of functional categories such as finiteness (verb-second), topicalization, agreement, definiteness, tense and pronominalization. Why is it that children normally acquire the adult-like functional system within a limited period of time, whereas adult second language learners vary greatly both in terms of the amount of time they need and the ultimate level they achieve? It seems that children are generally more successful than adult language learners due to the fact that in processing language input they experience no influence from either advanced levels of cognitive and linguistic development nor from the principles of another language system.

2.6 Data sources The data of this study originate from investigations on processes of language acquisition by children learning Dutch as their native language and by adults learning Dutch in an untutored second language environment. The child data are diary data collected by the present author (Jordens 2002). This study is based on the data of two children. The second language data come from the European Science Foundation (ESF) project “Second Language Acquisition by Adult Immigrants” (Perdue 1993a, b). These data consist of spontaneously produced utterances by adult learners of Dutch. The present study is based on the data of two informants: one informant has Turkish as his native language, the other informant Moroccan-Arabic. Both the first and second language data were collected longitudinally, the first-language data on a daily basis over a period as indicated in Table 4, the second language data in three cycles of 10 months each. In the examples below, a reference such as, for example, ‘Jasmijn 1;9’ means that the utterance occurred when Jasmijn was 1 year and 9 months of age, ‘Mohammed 1.2’ means that the utterance by Mohammed occurred at cycle 1 of month 2.

Data sources  

 39

L1 Dutch

diary data

J = Jasmijn (1;7–2;2) A = Andrea (1;8–2;4)

L2 Dutch

ESF audio recordings

O = Osman (L1 Turkish) M = Mohammed (L1 Moroccan-Arabic)

Table 4: Data sources.

3 The Target System 3.1 Introduction Language acquisition is a developmental process that occurs in stagewise progression. At the initial stage of this process, learners create a relatively simple language system in order to be able to communicate on a basic level of proficiency. This holds for children learning their mother tongue as it does for adults learning a second language in a natural context. In research on second language acquisition by Klein and Perdue, this simple learner system is called the Basic Variety (1997). Later on, in the course of the acquisition process, this simple learner system will be given up in favour of a more complex, targetlike version. Adult competence of the target system (the ‘end state’) is the final goal in children learning their native language; in fact, it is the norm. In second language acquisition, however, native speaker competence is an exception. Language development is commonly studied with respect to the ‘end state’ of the acquisition process: the principles and rules of the target language system. This holds particularly for the underlying principles of word order and word order variation. Basic word order in Dutch is taken to be OV (object – verb). As shown in (1) basic OV order becomes manifest in main clauses with an auxiliary verb. (1)

Zij gaat de krantO lezenV. she goes the newspaper read

On the basis of OV, word order may vary. Variation, however, is constrained. Principles underlying word order variation determine the actual form in which utterances occur. Knowledge of these principles accounts not only for the types of utterances that native speakers produce spontaneously, but also for why native speakers can provide judgements as to which forms of an utterance are correct and which are incorrect. In the following, I will provide an overview of the main characteristics of utterance structure in Dutch. First, I will present the arguments for why basic word order in Dutch is considered to be OV. Second, I will discuss a proposal to account for variation in utterance structure that is based on the functional principles underlying the formal properties of finiteness and ‘verb-second’.

42 

 The Target System

3.2 Utterance structure 3.2.1 Basic word order Utterances are used to communicate. Information that the speaker sees as relevant is shared with the listener. However, this is not the only function of an utterance. Every utterance entails a pragmatic function. As shown in (2), an utterance may function as an assertion (2a), (2b), (2c), a question (2d), (2e), (2f), (2g) or an imperative (2h). (2)

a.

Zij leest vandaag de krant. she reads today the newspaper

assertion (SVO)

b.

De krant leest ze vandaag. the newspaper reads she today

assertion (OVS)

c.

Vandaag leest ze de krant. today reads she the newspaper

assertion (Adv-VSO)

d.

Wie leest vandaag de krant? who reads today the newspaper ?

wh-question (SVO)

e.

Wat leest ze vandaag? what reads she today?

wh-question (OVS)

f.

Wanneer leest ze de krant? when reads she the newspaper?

wh-question (Adv-VSO)

g.

Leest zij de krant vandaag? reads she the newspaper today?

yes/no-question (0-VSO)

h.

Lees jij de krant eens vandaag! read (you) the newspaper today!

imperative (VSO)

The pragmatic function of assertions and questions is essentially different from the pragmatic function of imperatives. Assertions and questions are used to express a propositional content that holds or may hold true for the situational context that the utterance applies to. With the use of imperatives, however, there is no truth value involved. Assertions and questions typically occur with a verb form that is referred to as ‘finite’. Finite verb forms are marked for number, person and tense. They always occur in second position. The examples in (2) demonstrate that relative to this verb-second position the initial position may or may not be empty. The examples

Utterance structure 

 43

in (2) furthermore demonstrate that the interpretation of an utterance as an assertion or a question depends on whether there is an element in first position and, if so, whether or not this element is a question word. That is, in the default case as in (2a), (2b) and (2c), in which the first position is taken by the subject (S), an object (O) or an adverbial (Adv), the utterance is an assertion. A wh-word in first position as in (2d), (2e) and (2f) turns an assertion into a wh-question and in case of a yes/no-question, as in (2g), the initial position appears empty. With imperatives there is no initial position. This is because there is no situational context that the utterance applies to. Verb-second is a word order property of main clauses in Dutch. In the examples in (2), the position of verb-second is taken by a lexical verb (leest / lees). This is not the case in (3). Here, it is the auxiliary verb (gaat) that occurs in the position of verb-second, while the lexical verb (lezen) appears in final position. (3)

a.

Zij gaat vandaag de krant lezen. she goes today the newspaper read

assertion (S-Aux-OV)

b.

De krant gaat ze vandaag lezen. the newspaper goes she today read

assertion (O-Aux-S-V)

c.

Vandaag gaat ze de krant lezen. today goes she the newspaper read

assertion (Adv-Aux-S-OV)

d.

Wie gaat vandaag de krant lezen? who goes today the newspaper read?

wh-question (S-Aux-OV)

e.

Wat gaat ze vandaag lezen? what goes she today read?

wh-question (O-Aux-S-V)

f.

Wanneer gaat ze de krant lezen? wh-question (Adv-Aux-S-OV) when goes she the newspaper read?

g.

Gaat zij vandaag de krant lezen? yes/no-question (0-Aux-S-OV) goes she today the newspaper read?

Utterances with an auxiliary verb in the position of the finite verb, as in (3), are used to express the same pragmatic functions as in (2). They show that the pragmatic function of an utterance is linked to the relation between the finite verb and the element (or its absence) in initial position, regardless of whether the finite verb is an auxiliary or a lexical verb. In the position of verb-second, the finite verb serves as a carrier of the pragmatic function of assertion. This “invisible function” is what Lasser refers to as semantic finiteness: S-finiteness. The relevant function is formally expressed with the properties of morphological finiteness. This

44 

 The Target System

“overt form” is what Lasser refers to as morphological finiteness: M-finiteness (1997: 77).¹⁹ In main clauses in Dutch, the auxiliary verb always occurs in second position. In the absence of an auxiliary verb, the lexical verb takes the position of verb-second. Hence, the position of the lexical verb, as shown in (2) and (3) may vary. While the lexical verb may occur in second position when the auxiliary is absent, in the default case it occurs in final position. Here, the lexical verb has its default, non-finite morphology. The default position of the lexical verb accounts for the view that basic word order in Dutch is OV. Further evidence of a basic OV order comes from utterances with particle verbs as in (4). Particle verbs are compositional. They consist of a simple verb and a separable non-verbal element: the particle. Elements used as particles are in (in), op (up), uit (out), aan (on), mee (with), weg (away) etc. (4)

a.

Zij leest vandaag de krant uit. she reads today the newspaper through

assertion (SVO-Prt)

d.

Wie leest vandaag de krant uit? wh-question (SVO-Prt) who reads today the newspaper through?

g.

Leest ze vandaag de krant uit? yes/no-question (0-VSO-Prt) reads she today the newspaper through?

As shown in (4), the verbal component leest (reads) of the particle verb uitlezen (through-read) is morphologically finite. Depending on the pragmatic function of the utterance, this finite element may occur with a constituent in initial position as in (4a) and (4d) or with the initial position empty as in (4g). The particle uit (out) is the non-verbal component of the particle verb. Hence, although it serves as a predicate, it cannot be finite. A non-finite verb form occurs in its default, sentence-final position. Therefore, the position of the particle is always sentencefinal. These observations on the principles underlying word order variation in Dutch can be summarized as follows. In main clauses, basic word order in Dutch is determined by the default, sentence-final position of the lexical verb. The

19 “A terminological distinction is (…) necessary between the overt form that finiteness takes and the invisible function that the finite form serves. I will use the term M-finiteness (for morphological finiteness) and S-finiteness (for semantic finiteness) for these two different notions. M-finiteness concerns any overt (“audible”) marking of (some aspects of) S-finiteness” (Lasser 1977: 77).

Utterance structure 

 45

lexical verb appears in second position only if there is no auxiliary verb present. With the lexical verb in final position, basic word order in Dutch is OV. Final evidence for the basic OV order in Dutch comes from utterances with an embedded clause structure. Word order in the embedded clause does not contribute to the pragmatic function of the utterance as a whole. Hence, in the embedded clause there is no verb-second position and, thus, verb forms, occur in their basic word order position, either with non-finite morphology (lezen) as in (5a) and (5b), or with finite morphology (leest) as in (5c) and (5d). (5)

a.

[ Vandaag de krant lezen ] is niet zijn plan. today the newspaper read is not his plan

assertion

b.

Hij beweert [ vandaag de krant te lezen ]. he claims today the newspaper to read

assertion

c.

Ik denk [ dat hij vandaag de krant leest ]. I think that he today the newspaper reads

assertion

d.

Denk jij [ dat hij vandaag de krant leest ]? yes/no-question think you that he today the newspaper reads?

In sum, in main clauses, verb forms in second position serve as carriers of the pragmatic function of the utterance. They are used to express the function of an assertion or a question. The position of verb-second is the position of the auxiliary verb. Lexical verbs occur in second position only in the absence of the auxiliary verb. The default position of the lexical verb is sentence-final. Hence, basic word order in Dutch is OV. Utterances are structured in terms of the relations between constituents. The precise nature of the constituent structure of an utterance can be determined by the procedure of substitution. As illustrated in (6), this procedure is based on the fact that constituents are members of syntactic categories that are relevant at particular levels of categorization. Sets of constituents at a lower level of categorization can be substituted by one constituent of a syntactic category at a higher level of categorization. (6)

a.

Vandaag gaat ze [ de krant lezen ]. today goes she the newspaper read

b.

Vandaag gaat [ ze slapen ]. today goes she sleep

46 

 The Target System

c.

Het [ gaat regenen ]. it gets rain

d.

[ Het regent ]. it rains

Thus, in Vandaag gaat ze de krant lezen in (6), both the elements de krant and lezen can be substituted by one element such as, for example, slapen in (6b). It shows that at another, higher level of analysis de krant and lezen function as one constituent. The same holds for ze and slapen in (6b), gaat and regenen in (6c) and, finally, het and regent in (6d). This procedure of analysis through substitution demonstrates that utterance structure is hierarchical in nature. A representation of the hierarchical structure of (6) in terms of a tree structure as in Figure 1 shows that the utterance structure in Dutch is right branching.

Vandaag gaat ze today goes she

de krant lezen the newspaper read

Figure 1: The hierarchical structure of the main clause in Dutch

3.2.2 Word order variation Word order in Dutch is variable. As shown in the examples in (2) and (3) repeated here in (7) and (8), word order variation in Dutch is linked to the position of the finite verb. (7)

a.

Ze she

gaat goes

vandaag today

b.

De krant gaat the newspaper goes

ze vandaag she today

c.

Vandaag today

ze she

gaat goes

de krant the newspaper

lezen. read lezen. read

de krant the newspaper

lezen. read

Utterance structure 

g.

(8)

Gaat ze vandaag goes she today

de krant the newspaper

leest reads

de krant. the newspaper

a.

Ze she

b.

De krant leest ze vandaag the newspaper reads she today

c.

Vandaag today

g.

vandaag today

leest ze reads she

de krant. the newspaper

Leest ze vandaag reads she today

de krant? the newspaper?

 47

lezen? read?

As pointed out before, examples as in (7) and (8) show that, while the finite verb always occurs in second position, the initial position may be taken by constituents such as an NP-subject (7a) and (8a), an NP-object (7b) and (8b), an adverbial  (7c) and (8c), or that it may be left empty (7g) and (8g). Furthermore, the finite verb can be either an auxiliary verb as in (7a), (7b), (7c), (7g) or a lexical verb as in (8a), (8b), (8c), (8g), and if the finite verb is an auxiliary verb, the lexical verb occurs with non-finite morphology in final position. As shown in (7) and (8), the constituents in initial position are used with the syntactic function of the subject (7a, 8a), the object (7b, 8b) or an adverbial (7c, 8c). The syntactic function of the constituents in initial position (XP) is determined by the lexical verb (V). The word order underlying both (7) and (8) is represented in Figure 2. XP

F

S | ze she

(Adv) | vandaag today

  [ O | de krant the newspaper

V ]Pred | lezen read

Figure 2: Underlying word order in Dutch. F = the position of the finite verb; V = the position of the lexical verb; XP is a position that may also be empty.

It is a lexical property of the verb lezen (read) that it establishes a lexical structure for the syntactic positions of the subject ze (she) and the object de krant (the newspaper). Syntactically, the object occurs immediately before the lexical verb. O and V are elements of a constituent that serves as the predicate of a predicate-

48 

 The Target System

argument structure. Syntactically, it is preceded by S. Semantically, S is the external argument that the predicate holds for. The interpretation of ze in (7a) and (8a) as the subject is due to the fact that in (7a) and (8a) there is no element occurring in the subject position (S) before OV. The interpretation of de krant in (7b) and (8b) as the object is due to the fact that in (7b) and (8b) there is no element occurring in the object position (O) before the verb (V). This relation between the element occurring in initial position and its position within the syntactic representation of its lexical structure is referred to as ‘movement’. It is represented by the specification of an empty position ei as in both (7a′) and (7b′) and (8a′) and (8b′). (7)′

(8)′

a.

Zei she

b.

De krant i gaat the newspaper goes

a.

Zei she

b.

De krant i leestj ze vandaag ei the newspaper reads she today

gaat goes

ei

vandaag de krant today the newspaper

ze vandaag ei she today

leestj ei reads

lezen. read lezen. read

vandaag de krant today the newspaper

ej. ej.

Thus, from the fact that in (7a′) and (8a′) there is no element in ei, i.e. the position of the subject, it is possible for the speaker to conclude that ze in initial position has subject function. Furthermore, from the fact that in (7b′) and (8b′) there is no element in ei, i.e. the position of the object, it is possible for the speaker to conclude that de krant in initial position has object function. A similar situation holds for the relation between the lexical verb and the position formally indicated with ej in (8a′) and (8b′). Within the syntactic representation of a lexical structure, the lexical verb occurs in final position. It occurs in this position in utterances with an auxiliary as in (7a′) and (7b′). However, it must occur in verb-second position as in (8a′) and (8b′), when this position is not taken by an auxiliary. For utterances as in (7a) and (8a), it is assumed that there is an underlying empty subject position represented as in (7a)′ and (8a)′. Evidence for this is provided by the use of scope particles such as ook (also) and niet (not). Examples are shown in (9) and (10). (9)

a.

Vandaag today

gaat goes

ook MaRIE also Mary

de krant lezen. the newspaper read

Utterance structure 

(10)

b.

[Ook MaRIE]i gaat also Mary goes

c.

Mariei Mary

 49

ei

vandaag de krant lezen. today the newspaper read

OOK ei also

vandaag de krant lezen. today the newspaper read

a.

Straks gaat niet JAN in a moment goes not John

de krant lezen. the newspaper read

b.

[Niet JAN]i not John

gaat ei goes

straks de krant lezen. in a moment the newspaper read

c.

Jani John

gaat NIET ei goes not

straks de krant lezen. in a moment the newspaper read

gaat goes

In (9a) and (10a) the scope particles ook and niet are used with narrow scope, i.e. only one constituent – here the subject NP – is within their scope. Movement into sentence-initial position as in (9b) and (10b) shows that subject and scope particle appear as one constituent. In (9c) and (10c), however, movement only applies to the subject. This leaves ook and niet with no element in scope. Nevertheless, there is no difference in meaning between (9a) and (9c) and (10a) and (10c). In both cases ook and niet are still interpreted with narrow scope over the subject NP. Therefore, there is reason to believe that in (9b) and (10b) the underlying subject position is empty. Figure 3 provides a graphic representation of the possibilities of word order variation in Dutch in terms of ‘movement’. It shows that constituents with the syntactic function of the subject, the object or the predicate may occur in initial or second position of the utterance. If this is the case the original position is left empty.

(Vandaag) (today)

heeft has

ze she

de krant the newspaper

leest reads

Figure 3: Variation of word order in Dutch

gelezen. read

50 

 The Target System

In sum, word order may vary as a function of the informational properties of the utterance. In main clauses, as shown in 3.2.1, the finite verb always occurs in second position. Absence or presence of a particular constituent in initial position plays a role in the use of an utterance as an assertion or a question. Nonfinite verb forms occur in default, i.e. sentence-final, position. This default position accounts for the fact that basic word order in target Dutch is OV. Structural variation in utterances as shown in 3.2.2 is represented in terms of ‘movement’. However, structural variation described in terms of movement does not acknowledge its functional relevance. That is, it leaves untouched the question of what this kind of variation in utterance structure is good for. In the following, both utterance structure and structural variation in assertions will be discussed from this functional point of view. It will be shown how utterance structure and structural variation serve the expression of the linguistic properties of information structure.

3.2.3 Information structure and lexical meaning The examples in (7′) and (8′) show that main clauses in Dutch consist of two major components. A component which is dominated by the finite verb (F) and a component which has the lexical verb (V) as its main constituent. As argued before, the two initial positions that are dominated by the finite verb are linked to the pragmatic function of the utterance. They determine the functioning of the utterance as an assertion, a wh-question or a yes/no-question. The component that has the lexical verb as its main constituent, projects a set of syntactic positions that are each linked to a specific lexical-semantic function. The relation between the component dominated by the finite verb in F and the component dominated by the lexical verb in V is a hierarchical one. This is illustrated in Figure 4. Both F gaat (goes) and V lezen (read) are the ‘head’ constituent of the structure they project, i.e. they determine its semantic and syntactic properties. The head may require a particular constituent as its ‘complement’. Thus, the VP ze de krant lezen (she the newspaper read) is the complement of F and the NP de krant (the newspaper) is the complement of V. Both head and complement entertain a ‘hold-for’ relation with respect to the constituent that is termed the ‘specifier’ (Spec). In Figure 4, the adverb vandaag (today) serves as the specifier of FP, while the NP ze (she) is the specifier of VP. The component dominated by F is a functional projection (FP). FP accounts for the functional properties of information structure of an utterance. With the expression of finiteness and tense in F the speaker expresses both the pragmatic

Utterance structure 

SpecFP Vandaag today

FP | F′ | F (Head) gaat goes SpecVP ze she

 51

Complement ‖ VP | V′

Complement de krant the newspaper

V (Head) lezen read

Figure 4: Main clause structure in Dutch

function and the spatio-temporal anchoring of the utterance. SpecFP establishes its contextual embedding. SpecFP is a non-argument position. This position is available for both arguments and adverbials as constituents that F′ holds for. Hence, SpecFP is the position for constituents with the pragmatic function of topic or focus. The component dominated by V is a lexical projection (VP). VP serves as the complement of F. VP accounts for the lexical-semantic properties of predicate-argument structure. It serves to express that a predicate (with or without an internal argument) holds for a particular external argument (the subject). The relation between the lexical projection of VP and the functional projection of FP is such that VP serves as the complement of F. The lexical projection VP thereby establishes the propositional content of an utterance, which is linked to a particular contextual situation by the information structural properties of its functional projection. Although the projections in FP and VP serve different functions, they involve the same structural relations between their constituents. The structure as given in Figure 5 represents the general form of phrase structure referred to as X-bar syntax. In Cook and Newson, the general properties of X-bar syntax are described as follows: “X-bar theory claims that all types of phrase need two internal levels of structure. It proposes that all phrases in all languages share a simple cell-like structure with two levels to each phrase: one (X″) [here XP] consists of the head and possible specifiers; the other (X′) consists of the head (X) and possible complements. Note that specifier and complement are

52 

 The Target System

not themselves syntactic categories but functional labels for positions in the structure that may be filled by actual syntactic categories such as NPs and VPs” (1996: 144).

SpecXP

XP | X′ | X (Head)

Complement

Figure 5: Hierarchical structure of the projection of XP

3.2.3.1 VP structure The VP constituent is the representation of the lexical-semantic properties of a predication which expresses a hold-for relation between V′, i.e. the verbal head (V) with its complement, and the specifier in SpecVP. The specifier in SpecVP is the external argument – or the subject – of V′. The lexical-semantic structure of the predicate V′ is stored as lexical knowledge in the mental lexicon. This also holds for the syntactic configuration of its head-complement structure as either head-initial or head-final. That is, for native speakers of Dutch a verbal head such as lezen (read) is stored as part of a lexical entity which projects a semantic category object, such as krant (newspaper), as its complement and a syntactic configuration in which the verb occurs in head-final position as in krant lezen (newspaper read). In English, the semantic structure of the verbal head read is the same as lezen in Dutch. However, its syntactic configuration is different. The verbal head read appears in head-initial position. That is, unlike krant lezen in Dutch, read and newspaper are lexically stored as read newspaper.

3.2.3.2 FP structure As represented in Figure 6, it is the function of FP to express that the predication (VP) has to be interpreted as an assertion with respect to the topic situation (TS). The topic situation (TS) is the situation that the assertion applies to. While F as the head of FP is used to express the pragmatic function of the assertion, SpecFP establishes the relation between the assertion and the topic situation (TS). It provides the condition for the assertion to express that the speaker holds the predication to be true with respect to a particular TS (Klein 1998: 227, 2008: 287, 290).

Utterance structure 

TS ← SpecFP

FP | F′ | F (Head)

 53

VP

Figure 6: Hierarchical structure of the projection of FP

3.2.3.3 The function of F The position of F in Dutch is used to express the pragmatic function of assertion. Auxiliary verbs may serve as carriers of this function. More specifically, they are used to express the assertion of aspectual meaning. Examples are given in (11). (11)

a.

Ze heeft vandaag de krant gelezen. she has today the newspaper read

b.

De krant is nog steeds verdwenen. the newspaper is still disappeared

c.

Ze gaat vandaag de krant lezen. she goes today the newspaper read

d.

Ze is net de krant aan het lezen. she is just the newspaper on the read

e.

Ze doet vandaag de krant lezen. she does today the newspaper read

In (11a) and (11b), the position of F is taken by the auxiliary verbs heeft (has) and is (is). These auxiliary verbs occur with a past participle form of the lexical verb. They are used to express perfect aspect. In (11c) and (11d), F has the auxiliary verbs gaat (goes) and is (is). They occur in the context of an infinitival form of the lexical verb in V. They are used to express perfective or progressive aspect. Finally in (11e), F is represented by the auxiliary doet (does). This type of auxiliary verb is register- or regionally bound. It also occurs with an infinitival form of the lexical verb in V. As in (11c) and (11d), it is used to express perfective or progressive aspect. The examples in (11) show that while the properties of F are carried by the auxiliary verb, the lexical verb occurs in final position. However, the lexical verb occurs in F in case the auxiliary verb is absent. As pointed out before, this phenomenon is known as ‘verb-second’. In Dutch, the formal properties of F are

54 

 The Target System

expressed by inflectional morphology. This explains why the lexical verb as the carrier of F occurs with finite morphology, whereas in final position it is used in a non-finite form, i.e. either as an infinitive or as a past participle. The pragmatic function of the assertion in Dutch may also be carried by a modal verb. Modal verbs in the position of F are used to express the assertion of a modality. Examples are given in (12). (12)

a.

Ze wil vandaag de krant lezen. she wants today the newspaper read

b.

Ze kan vandaag de krant lezen. she can today the newspaper read

c.

Ze mag vandaag de krant lezen. she may today the newspaper read

d.

Ze moet vandaag de krant lezen. she must today the newspaper read

With the utterances in (12), the speaker may assert that he has the wish (12a), the ability (12b), the permission (12c) or the obligation (12d) to perform some kind of action. This is not possible in examples as given in (13). (13)

a.

Deze dingen kunnen gebeuren. these things can happen

b.

Het kopje mag niet breken. the cup may not break

c.

Het werk moet vandaag af zijn . the work must to-day finished be

In (13), the speaker asserts that it is a possibility (13a), an option (13b) or a necessity (13c) that a particular situation occurs. Thus, modal verbs in Dutch can be used in different ways. In (12), they are used to express the assertion of a modality, while in (13) they are used to express the modality of an assertion. In sum, from a functional perspective, utterance structure in Dutch integrates two projections of syntactic structure: a functional projection (FP) and a lexical projection (VP). Both consist of a head, a complement and a specifier constituent. Table 1 lists the syntactic and semantic functions of each of these constituents.

Utterance structure 

 55

projection

syntactic

semantic

functional: FP

F-head Complement of F SpecFP

assertion predication (VP) reference to TS

lexical: VP

V-head Complement of V SpecVP

predicate object (= internal argument) subject (= external argument)

Table 1: The syntactic and semantic function of the constituents of FP and VP

3.2.3.4 Movement In Dutch, the structural positions within an utterance are fixed. However, the head of the lexical projection VP may occur in the head position of the functional projection FP, as it serves as a carrier of its functional properties. Thus, as shown in (14b), the lexical verb leest (reads) may occur in the position of F, if there is no auxiliary available for the expression of the pragmatic function of assertion. (14)

a. b. c. d.

Zei she Zei she De kranti the newspaper Vandaagi today

heeft has leestj reads heeft has heeft has

ei

vandaag today ei vandaag today ze vandaag she today ze ei she

de krant the newspaper de krant the newspaper ei

gelezen. read ej.

gelezen. read de krant gelezen. the newspaper read

Hence, in the absence of an auxiliary verb, variation with respect to the position of the lexical verb serves a functional purpose. Similarly, as shown in (14c), nominal constituents with argument function may occur in SpecFP position. In this position, as is the case with the adverbial in (14d), these nominal constituents are used to refer to the contextual situation that the utterance applies to. The representation in Figure 4 serves as an abstract formal means to describe the possibilities of variation in utterance structure. It accounts for the fact that variation in utterance structure is constrained. That is, it explains the fact that native speakers generally agree on the kind of language structures that they judge as either correct, as in (14), or incorrect (*), as in (15). (15)

*a.

De krant j zei heeft the newspaper she has

ei

vandaag today

ej

gelezen. read

56 

 The Target System

*b.

Vandaagj today

*c.

Vandaagi today

*d.

Vandaagi today

zei heeft she has heeft has lezenj read

ei

ej de krant gelezen. the newspaper read

de krant j the newspaper

ze she

ei

ej gelezen. read

ze (3P.Sg) ei de krant ej . she the newspaper

3.2.4 Summary In Figure 4, main clause structure is represented as the expression of the pragmatic function (FP) of a lexical predicate-argument structure (VP). The predicateargument structure VP (ze – de krant lezen) is dominated by the lexical verb (V) lezen. This lexical head projects a complement position for the internal argument, i.e. the object (de krant). The lexical head and its complement serve as the predicate V′ that holds for SpecVP (ze). Thus, the constituent in SpecVP is the external argument, i.e. the subject. The predicate-argument structure VP (ze – de krant lezen) is the complement of F (gaat). F′, i.e. F and VP, represents the expression of an utterance (gaat – ze de krant lezen), while SpecFP provides the position for the constituent (vandaag) that is used to refer to the contextual situation that the utterance applies to.

3.3 Utterance production The principles of utterance structure as described in Section 3.2 determine both the grammatical and functional competence of adult native speakers of Dutch. That is, they account for these speakers’ intuitions not only on the types of utterance structure they judge as either grammatically correct or incorrect, but also on the communicative function with which different types of utterances are used. Utterance structure is the result of processes of production. Principles of language production interact with the current state of structural linguistic knowledge. Therefore, it seems relevant to investigate from the point of view of the production system not only the knowledge system underlying utterance structure but also the way in which utterance structure is used for the expression of communicative functions. Language learners are typically in the process of discovering how to use linguistic knowledge as a means to communicate. Given the interaction between processes of production and the principles that language

Utterance production  

 57

systems are based on, (early) processes of language production may account for the particular properties of form and functioning of early learner systems. Language production is a matter of content selection and information structuring. Content selection and information structuring determine the eventual shape of the utterance structure, not only with native speakers of the fully-fledged system of the target language, but also with L1 and L2 learners using the simple language system of their Basic Variety. Content selection is the result of the interaction of processes of conceptualization and the knowledge systems of lexicalization and predication. Its relevance for the initial stages of language acquisition will be discussed in 3.3.1. Information structuring is the result of the interaction between processes underlying the embedding of an utterance into its situational context and the knowledge systems that serve the expression of finiteness and topicalization. Its relevance for the advanced stages of language acquisition will be discussed in 3.3.2.

3.3.1 Conceptualization, lexicalization and predication As argued above, content selection and the use of basic utterance structure are achieved by the interaction of the processes of conceptualization, lexicalization and predication.

3.3.1.1 Conceptualization Prerequisite to both the processes of lexicalization and predication is the process of conceptualization. See, for example a scene as in Figure 7, in which someone is breaking a stick.

Figure 7: The man is breaking a stick © Tilman Harpe 2012

58 

 The Target System

EVENT(e′) | PERSON(x) CAUSE PHYS.OBJ(y)

EVENT(e) | BECOME

STATE(s)

Figure 8: Conceptual analysis of the situation of someone breaking a stick

The picture in Figure 7 shows a person performing an action with an object. There are several perspectives from which a speaker may inform the listener about what is happening. One possibility is to describe the situation from the perspective of the person who is carrying out the action. Another option would be to describe the situation from the perspective of the physical object. Whatever the speaker may decide, the outcome is the result of a process of conceptualization. It requires a conceptual analysis of the relevant situation as represented in Figure 8. The conceptual analysis given in Figure 8 entails a relation between two structures: x causes e and y becomes s. If the speaker decides to describe the situation in Figure 7 from the perspective of the person who is carrying out the action, the situation is conceptualized as x causes e; if he decides to describe the situation from the perspective of the physical object, the situation is conceptualized as y becomes s. The situation as shown in Figure 7 can, thus, be conceptualized at least in two ways, i.e. either as a causative action or as a change of state.

3.3.1.2 Lexicalization The process that leads to the formal expression of a conceptual structure is not at all straightforward. This is because, as Bowerman argued, “not all the cognitive distinctions [the speaker] may be aware of while [he] produces an utterance are linguistically relevant and those that are not should not appear in the linguistic representation of the utterance” (1974: 155). Hence, the encoding of a conceptual structure is constrained by the types of lexical structure that a particular language system has available. Lexicalization is the process that matches possible conceptualizations with the linguistic means that the language provides for their expression. Given a conceptual analysis as in Figure 8, the outcome of the lexicalization process in Dutch is the option between either of two lexical projections. That is, for each of the conceptual structures x causes e and y becomes s there is a lexical projection available in Dutch. The relation between these conceptual structures and the relevant lexical projections is represented in (16).

Utterance production  

 59

(16) Lexicalization: ‘Someone breaking a stick’ a.

PERSON(x) ↕ Agent(x) | iemand someone

+

CAUSE + EVENT(e) conceptual structure ↕ ↕ + [ Object(y) + Pred ] Pred′ lexical projection | stok breken stick break

b.

PHYS.OBJ(y) ↕ Theme(y) | stok stick

+ BECOME + STATE(s) ↕ + Pred | breken break

conceptual structure ↕ lexical projection

Lexical projections are the linguistic means for the representation of a situation in terms of (a) predicates (Pred) and (b) arguments. Predicates serve the expression of semantic categories such as actions, states and changes of state, i.e. they refer to different types of situation. Arguments serve the expression of elements with the semantic function of an agent, an object or a theme, i.e. they refer to participants playing a particular role in a particular situation. As shown in (16), there are two lexical projections matching the conceptual structure in Figure 8. One option, as shown in (16a), is the lexical projection of an action in which two arguments are involved, each with its own semantic function: the person as the agent of the action and the stick as the object of this particular action. On the other hand, as shown in (16b), the same conceptual structure can also be represented as the lexical projection of a change of state with only the stick involved. Here, the stick is the element that undergoes the process as it takes place. In the absence of an agent, its semantic function is referred to as the theme. Thus, the interaction between the conceptual structure as in Figure 8 and the possibilities of lexicalization as in (16) provides the speaker with two alternative ways to describe the scene in Figure 7. That is, either as a causal action or as a change of state. In Dutch, the relation between the lexicalizations in (16a) and (16b) is systematic. Examples are given in (17a) and (17b). (17a) causal action

(17b) change of state

Hij breekt de stok. he breaks the stick

De stok breekt. the stick breaks

60 

 The Target System

Ze plakt de poster tegen het raam. she sticks the poster on the window Hij lost de poeder op in het water. he solves the powder in the water De kok smelt de boter in de pan. the cook melts the butter in the pan De wind beweegt het gordijn. the wind moves the curtain

De poster plakt tegen het raam. the poster sticks on the window De poeder lost op in het water. the powder solves in the water De boter smelt in de pan. the butter melts in the pan Het gordijn beweegt in de wind. the curtain moves in the wind

The examples in (17a) and (17b) show that in addition to an agentive predicate with both an agent and an object, there is a non-agentive equivalent in which the object has the function of the theme. However, this type of relation does not hold across the board. Thus, a situation in which someone throws a stone can be referred to with the agentive predicate gooien (throw) as in (18a). However, as shown in (18b), gooien (throw) cannot be used to express the relevant event as a change of state. The same holds, for example, for stukmaken (kaput-make) and many other agentive verbs. (18a) causal action

(18b) change of state

Hij gooit een steen tegen de ruit. he throws a stone against the window Hij maakt het speelgoed stuk. he makes the toy kaput

* De steen gooit tegen de ruit. the stone throws against the window * Het speelgoed maakt stuk. the toy makes kaput

There are alternatives to (18b). As shown in (18c), the expression of a change of state can be achieved with the auxiliaries worden (is) or gaan (go). (18a) causal action

(18c) change of state

Hij gooit een steen tegen de ruit. he throws a stone against the window Hij maakt het speelgoed stuk. he makes the toy kaput

Een steen wordt tegen de ruit gegooid. a stone is thrown against the window Het speelgoed gaat stuk. the toy goes kaput

Similarly, as shown in (19a) and (19b) state and change-of-state verbs such as trillen (tremble), vallen (fall) or schrikken (get-frightened) cannot be used to express a causal action.

Utterance production  

 61

(19a) change of state

(19b) causal action

De grond trilt. the soil trembles De steen valt. the stone falls Het kind schrikt. the child gets-frightened

* Iets trilt de grond. something trembles the soil * Hij valt de steen. he falls the stone * Dit schrikt het kind. something gets-frightened the child

However, here too, there are alternatives. As shown in (19c) the expression of a causal action can be achieved with the auxiliary verbs laten (let), doen (make) or maken (make). (19a) change of state

(19c) causal action

De grond trilt. the soil trembles De steen valt. the stone falls Het kind schrikt. the child gets-frightened

Iets doet de grond trillen. something does the soil tremble Hij laat de steen vallen. someone lets the stone fall Dit maakt het kind aan het schrikken. something makes the child to get-frightened

Processes of lexicalization are language specific. Hence, cross-linguistically, languages may differ with respect to the meaning with which lexical verbs are used. Thus, for example, in English it is possible to use sink both in The enemy sinks the ship and The ship sinks, while the Dutch equivalent zinken is only possible as in Het schip zinkt (the ship sinks). The same holds for drop, slide, increase, shrink etc. The Dutch equivalents vallen (drop), glijden (slide), toenemen (increase), krimpen (shrink) etc. can only be used with a theme. Furthermore, in English it is possible to use catch both in The cat catches a mouse and The bolt does not catch, while the Dutch equivalent vangen is only possible in De kat vangt een muis (the cat catches a mouse). The same holds for divide, heighten, combine etc. The Dutch equivalents verdelen (divide), verhogen (heighten), samenstellen (combine) etc. can only be used with an agent. Lexicalization is the process that matches possible conceptualizations with a set of linguistic structures provided by the present language system. As argued before, the conceptual structure y becomes s, is a possible conceptualization of a situation as in Figure 7. Lexicalization of y becomes s as a change-of-state predicate is shown in (16b) and (17b). However, lexicalization of y becomes s as a change-ofstate predicate is not the only possibility. As shown in (20b) y becomes s can also be lexicalized as a result-state predicate as, for example, with dicht zijn (closed be).

62 

 The Target System

(20) Lexicalization: Someone closing the door a.

PERSON(x) ↕ Agent(x) | iemand someone

+

CAUSE + EVENT(e) ↕ + [ Object(y) + Pred ] Pred′ | deur dichtdoen door closed-do

b.

PHYS.OBJ(y) + BECOME + STATE(s) ↕ ↕ Theme(y) + Pred | | deur dicht zijn door closed be

conceptual structure ↕ lexical projection

conceptual structure ↕ lexical projection

A result-state predicate may serve the lexicalization of y becomes s, because it has the implication of a change of state. The relation between lexical projections as in (20a) and (20b) is systematic and productive. As shown in (21a) and (21b), it occurs with complex action predicates, i.e. particle verbs such as uitdoen (off-do), opeten (up-eat), aanzetten (on-turn), weggooien (away-throw), stukmaken (kaputmake). While the head of these predicates serves to express a causal action, the relevant particles are used to express the result state that the object is in. (21a) causal action

(21b) result state

Het meisje eet de soep op. the girl eats the soup gone De man gooit het papier weg. the man throws the paper away Het kind maakt het horloge stuk. the child makes the watch kaput De vrouw doet het licht uit. the woman does the light off De jongen zet de tv aan. the boy puts the tv on

De soep is op. the soup is gone De krant is weg. the paper is away Het horloge is stuk. the watch is kaput Het licht is uit. the light is off De tv is aan. the tv is on

3.3.1.3 Predication Semantically, lexical projections are the linguistic means to express a situation in terms of a predicate-argument structure. The predicate as the head of the lexical

Utterance production  

 63

projection specifies the semantic functioning of its arguments as, for example, agent, object or theme. Syntactically, lexical projections are the expression of a predication, i.e. the relation between a predicate – if applicable with its internal argument(s) – and its external argument, i.e. the subject. The subject is the constituent that the predicate holds for. Assuming that with a conceptualization as in Figure 8 the mental lexicon has at its disposal both a lexicalization of x causes e and y becomes s, it is up to the speaker to decide which of these will be selected for production. Whether it will be the causal action or the change of state is a matter of subject selection. If the speaker decides to select the agent as the subject, the predication will be of type (22a). If he decides to select the theme as the subject, the predication will be of type (22b). (22) Lexicalization and predication a.

Agent(x) + [ Object(y) + PredACTION ] Pred′ ↕ Subject(x) + [ Object(y) + PredACTION ] Pred′

lexical projection ↕ predication

b.

Theme(y) + ↕ Subject(y) +

lexical projection ↕ predication

PredSTATE / CHANGE OF STATE PredSTATE / CHANGE OF STATE

In the relevant case, the selection of either an agentive (22a) or a non-agentive (22b) type of utterance structure is, thus, the outcome of a process of subject selection. Subject selection serves the function of information structuring. The subject refers to the entity that the predicate holds for or is about. It explains why in the default case the subject occurs in sentence-initial position. Subject selection is, thus, a crucial step towards the eventual word order with which an utterance is produced in Dutch. The representations in (22a) and (22b) differ with respect to the semantic structure of the lexical projection. Syntactically, however, they are both the expression of a predication in which the predicate holds for the subject. Elements with the semantic function of an agent or a theme may serve as entities that the predicate is about. This does not hold for the object. The object is part of the complex predicate (Pred′). Evidence for this is the way in which a native speaker might respond to a question like Wat ga je doen? (what will you do?). A natural answer would be, for example, de krant lezen (the newspaper read), het nieuws kijken (the news watch) or een boek pakken (a book take), but not simply lezen (read), kijken (watch) or pakken (take). Thus, a predicate with a verb and an object as in (22a) serves as an entity similarly as in lexical compounds such as tanden-

64 

 The Target System

poetsen (teeth-brush), handenwassen (hands-wash), koffiedrinken (coffee-drink), boevenvangen (villain-catch) or de tafel dekken (the table set). As with lexical compounds, verb and object are stored as a lexical entity in the mental lexicon. Placement of the object before the verb as in (22a), is thus in fact a lexical feature. While the relation between a lexical verb and its object serves the function of a complex predicate, the relation between the subject and the predicate is different. Subject and predicate do not function as a lexical entity. This is shown in the examples in (23a) and (23b). (23a)

Jan gooit de stenen op de grond. John throws the stones on the ground

(23b)

De stenen vallen op de grond. the stones fall on the ground

While in (23a) de stenen is the object of gooien, in (23b) it is the subject of vallen. Due to its function as the complement of a lexical verb in (23a), stenen may occur as part of a nominalization with gooien, as in (23a′). (23a)′

Het stenengooien. the stones-throwing

This is not possible for stenen and vallen in (23b′). (23b)′

*Het stenenvallen. the stones-falling

A similar situation holds for the use of passen in (24a) and (24b). (24a)

Jan past kleren in de kleedkamer. John fits on clothes in the dressing-room

(24b)

De stukjes passen in de puzzel. the pieces fit into the puzzle

In (24a) kleren is the complement of passen, while in (24b) stukjes is not. Hence, the nominalization in (24a)′ is correct, while in (24b)′ it is not. (24a)′

Het klerenpassen. the clothes-fitting

Utterance production  

(24b)′

 65

*Het stukjespassen. the pieces-fitting

Both gooit de stenen (throws the stones) and past kleren (fits-on clothes) in (23a) and (24a) and vallen (fall) and passen (fit) in (23b) and (24b) are predicates. For the production of an utterance the speaker needs to identify the entity that the predicate holds for. This hold-for relation as it occurs between the predicate and the subject also occurs in examples as in (23a)″, (23b)″, (24a)″ and (24b)″. (23a)″

Het stenengooien van Jan. The stone-throwing of John

(23b)″

Het vallen van de stenen. the falling of the stones

(24a)″

Het klerenpassen van Jan. the clothes-fitting of John

(24b)″

Het passen van de stukjes. the fitting of the pieces

The relation between the nominalized infinitives stenengooien, klerenpassen, vallen and passen and the van-phrase is a relation of attribution. This explains why the van-phrase serves as a semantic paraphrase of the hold-for relation between the predicate and its subject. In sum, the process of conceptualization is prerequisite to the processes of lexicalization and predication. Whether a causal event becomes lexicalized as an agentive or a non-agentive utterance is due to the process of subject selection. If the agent is going to be the subject, the predicate that holds for the agent is an action. If the theme will be the subject, the predicate that holds for the theme is a state or a change of state. The subject serves the function of information structuring. As the constituent that the predicate holds for, the subject occurs in initial position. Complex predicates with an object and a lexical verb are stored as lexical entities in the mental lexicon. Hence, the OV structure that is seen as a characteristic feature underlying the syntactic structure of utterances in Dutch, is in fact a structural property of the lexicon.

66 

 The Target System

3.3.2 Expressing finiteness and contextual embedding 3.3.2.1 The semantic function of finiteness 3.3.2.1.1 Truth value As shown in Figure 4 in Section 3.2.3, main clauses in Dutch consist of two major constituents: a structure FP which is the maximal projection of F and a structure VP which is the maximal projection of V. FP is a functional projection. Within FP, both the position of SpecFP and the position of F are linked to the pragmatic properties of information structure. As the head of FP, F is the position of verb forms used to express the semantic function of finiteness. The semantic function of finiteness can be demonstrated with a dialogue as in (25). (25)

A:

Het lijkt of het huis bewoond is. ‘It seems as though the house is inhabited.’

B:

Het huis IS bewoond. ‘The house IS inhabited.’

In (25), A claims that a particular situation seems to exist, while B responds by saying that this is indeed the case. In his answer, B’s use of the verb is plays a crucial role. The verb is is a copula which occurs in the position of F. As a copula, is is relatively meaningless. Therefore, it is well suited to show the functional meaning of the category F. As argued in Klein (1998: 226f.), the meaning of F appears most prominently in a context in which is is used with contrastive stress. In B’s natural response to A, contrastive stress with is evokes a contrast in meaning which applies to the truth value of the predication. That is, B asserts that a situation is INDEED the case as opposed to what seems to be implied in what is said by A, i.e. that the relevant situation may NOT be the case. Hence, as stated by Klein (1998: 225), “being the carrier of assertion is the main function of finiteness”. That is why Lasser (1997: 77) refers to this as S-finiteness, i.e. semantic finiteness. S-finiteness indicates “the invisible function that finiteness serves”. It constitutes a property of the utterance as a whole. As shown in Figure 4 in Section 3.2.3, here repeated as Figure 9, the semantic content of the utterance is represented by the complement of F, i.e. the predication in VP. The predication in VP expresses a hold-for relation between the predicate in V′ and the constituent in SpecVP. The constituent in SpecVP is the external argument, i.e. the subject, of the predication. Semantically, the external argument is attributed a theta role which it receives from the predicate. Thus, with

Utterance production  

SpecFP vandaag today

FP | F′ | F (Head) gaat goes SpecVP ze she

 67

Complement ‖ VP V′ Complement de krant the newspaper

V (Head) lezen read

Figure 9: Main clause structure in Dutch

an agentive predicate such as kopen (buy) in, for example, Hij koopt een boek (he buys a book), the external argument hij (he) serves as an agent. A participant has the function of an agent, if it carries out a particular action. On the other hand, with a non-agentive predicate such as vallen (fall) in, for example, De pen valt op de grond (the pencil falls onto the ground), the external argument serves as the theme. A participant has the function of a theme, if it occurs in a particular state or if it undergoes a change of state. In utterances as in (26) and (27), the position of F is taken by a modal verb. The modal verbs in (26) are used with what is called ‘deontic modality’. That is, the utterances in (26) express the assertion of the modality of a particular state of affairs. (26)

deontic modality

a.

De minister wil duidelijkheid verkrijgen. the minister wants clarity to obtain

b.

Hij kan er geen antwoord op geven. he can it no answer on give

c.

We mogen achter de voordeur kijken. we may behind the front door look

d.

Je moet heel voorzichtig te werk gaan. you must very careful proceed.

68 

 The Target System

In utterances with deontic modality, the external argument receives its theta role from the modal predicate. Thus, wil (want) in (26a) refers to the wish of de minister (the minister), kan (can) in (26b) refers to the ability hij (he) has, mag (may) in (26c) refers the permission we (we) have, moet (must) in (26d) refers to the obligation of je (you). Deontic modal verbs are used to express the willingness, the ability, the permission or the obligation of the external argument to perform an action. Hence, the external argument of a deontic modal predicate has the theta role of an agent. Modal verbs can also be used with epistemic modality²⁰. Examples are utterances as in (27), which are used to express the modality of a particular assertion. (27)

epistemic modality

a1.

De bagage kan eruit vallen. the luggage can it-out fall

a2.

Je kunt van de fiets vallen. you can off the bicycle fall

b1.

Dat mag niet meer gebeuren. that may not anymore happen

b2 .

Hij mag niet langer blijven. he may no longer stay

c1.

Het ergste moet nog komen. the worst has-to yet come

c2 .

Je moet heel voorzichtig zijn. you must very careful be

The utterances in (27) are the expression of a possibility (27a) het kan (it can), an option (27b) het mag (it may) or a necessity (27c) het moet (it must). They hold for the event referred to with the projection of lexical verbs such as vallen (fall), gebeuren (happen), blijven (stay), komen (come) and zijn (be). The modal verbs in (27) are used differently from the way in which they are used in (26). While the external argument of the utterances in (27) serves as the external argument of the modal verb, it receives its theta role from the lexical verb. The lexical verb refers to a state as in blijven (stay) and zijn (be) or to a change of state as in vallen (fall), gebeuren (happen) and komen (come). Therefore, in utterances as in (27) the external argument has the theta role of a theme.

20 Barbiers (2002: 58) also uses the term ‘indirectly deontic modality’.

Utterance production  

 69

Modal verbs, whether used with deontic or epistemic meaning, occur in the position of the head of FP. Hence, they determine the pragmatic properties of an assertion as a whole. In utterances as in (26) they assert the willingness, the ability, the permission or the obligation of the agent to perform an action. In utterances as in (27) they assert the possibility, the option or the necessity that the theme occurs in a state or undergoes a change of state. In the absence of a modal predicate, the element in F is used to express the default function of an utterance, the expression of an assertion. That is, it simply expresses the truth value of an utterance: the claim that a particular situation holds true. As the formal expression of the default function of assertion, the position of F can be taken by an auxiliary verb such as ga, gaat (go, goes), doe, doet (do, does), heb, heeft (have, has), or ben, is (am, is). Examples are utterances as in (28). In case the predicate has no auxiliary verb, the default function of assertion is expressed with the finite form of the lexical verb (Vfin) as in (29). (28)

auxiliary verbs in F

a.

Ik ga niet achter de voordeur kijken. I go not behind the front-door look

b.

Hij doet daar geen antwoord op geven. he does there no answer to give

c.

De minister heeft nu duidelijkheid verkregen. the minister has now clarity got

d.

De fiets is op de grond gevallen. the bicycle is on the ground fallen

e.

Ik ben te laat gekomen. I am too late come

(29)

lexical verbs in F

a.

Ik kijk niet achter de voordeur. I look not behind the front-door

b.

Hij geeft daar geen antwoord op. he gives there no answer to

c.

De minister verkrijgt nu duidelijkheid. the minister gets now clarity

70 

 The Target System

d.

De fiets valt op de grond. the bicycle falls on the ground

e.

Ik kom te laat. I come too late

In sum, modal verbs are carriers of the assertion of a modality (deontic) or they are used to express the modality of an assertion (epistemic). In the absence of a modal verb, either an auxiliary verb (AUX) or a lexical verb (Vfin) may serve as the carrier of the functional properties of finiteness. In the position of F, they are used as a means to express the default function of assertion. The semantic function of finiteness as the expression of an assertion is summarized in Figure 10.

SpecFP

F

→

FP | F′ | F

Complement

MOD deontic: volition (wil), ability (kan), permission (mag),obligation (moet); MOD epistemic: possibility (kan),optionality (mag), necessity (moet); AUX (ga, gaat; doe, doet; heb, heeft; ben, is); Vfin (kijken, geeft, verkrijgt, valt, kom etc.).

Figure 10: The semantic function of finiteness

3.3.2.2 Anchoring With the expression of an assertion the speaker claims that a particular state of affairs holds true. However, in order for an utterance to be interpretable as true, it has to be accommodated such that it applies to a specific situation. This situation must be identifiable. Identifiability means that the listener can establish the time interval and the spatial setting that the utterance applies to. Spatio-temporal anchoring of an utterance is what finiteness is about. Or as Klein phrases it: “In an utterance, a sentence base and a situation are brought together, and this is what happens when the sentence is made finite” (2008: 288).

Utterance production  

 71

3.3.2.2.1 Temporal anchoring As with finiteness, temporal anchoring in Dutch is achieved in F. This is illustrated in (30). In this dialogue, as in (25), the utterance of B is a natural response to A. (30)

A:

Hij was een goede collega. ‘He was a good colleague.’

B:

Hij IS een goede collega. ‘He IS a good colleague.’

In (30), A claims that a particular situation was true in the past, while B responds by saying that it is STILL true. As in (25), the verb is in (30) is a copula. As a copula is is relatively meaningless. Here, contrastive stress with is evokes a contrast with respect to tense. That is, B asserts that a situation is the case at the moment of speaking and NOT at some point earlier in time. Identification of an event (i.e. an action, a state or a change of state) on the dimension of time is conceptualized in terms of the relation between the time at which the utterance occurs (TU) and what Klein refers to as the topic time (TT), i.e. the particular time span of an event that the speaker focuses on. The interaction between the topic time and the time of utterance is expressed with the formal properties of tense (Klein 1994: 120ff.). “Starren (2001) uses the metaphor of a video camera to explain TT – it is the time the camera is “shooting”. Imagine you are a witness in court, and the judge asks you, “What did you see when you entered the room?” The crucial time span corresponds to your entering the room, and just this time span is filmed by the camera. You answer, “A man was trying to open the safe. He looked Japanese.” The time span occupied by ‘man try to open safe’, and indeed the time span occupied by ‘man look Japanese’ – the “situation times” – are considerably longer than it took you to enter the room. It would indeed be surprising if the man did not still look Japanese as you speak. But this was not what you were asked. The TT is your entering the room, and your and the judge’s use of past tense puts this TT (but not necessarily TSit) before the time of utterance. The time of action of trying to open the safe, TSit, encompasses the TT: TT ≥ TSit. This aspectual relation is imperfective and explains the use of the past progressive aspect in your answer. (Imperfective aspect contrasts with perfective aspect, where TSit is within, or coincides with, TT.)” (Perdue, Benazzo and Giuliano 2002: 854).

Tense provides the formal means to establish temporal anchoring with respect to the time of utterance. As illustrated in Figure 11, the time of utterance (TU) can be either (a) simultaneous with TT, (b) after TT, or (c) before TT.

72 

 The Target System

(a) TU incl TT

(b) TU after TT

(c) TU before TT

+++ [ TT ] +++ TU

+++ [ TT ] +++ TU

+++++[ TT ] + TU

De trein komt op tijd aan

De trein kwam op tijd aan

De trein zal op tijd aankomen

(the train arrives in time)

(the train arrived in time)

De trein komt op tijd aan (the train will arrive in time)

Figure 11: Tense as the expression of the relation between topic time (TT) and time of utterance (TU). The ‘time of situation’ (TSit) is represented as ++++++

If the time of utterance co-occurs with the topic time as in (a), this is expressed on the finite verb with present tense morphology, if the topic time occurs before the time of utterance as in (b), the finite verb occurs with past tense morphology, and if the topic time occurs after the time of utterance as in (c), the finite verb is either a form of the auxiliary zullen (will, shall) or it has present tense morphology.

3.3.2.2.2 Spatial anchoring An assertion not only holds for a particular topic time, but also for a particular spatial setting. Identification of an event on the dimension of space is conceptualized in terms of entities and/or locations that are represented by means of lexical expressions, i.e. directly with nominal and/or adverbial elements, indirectly with pronominal and/or proadverbial elements. These entities and/or locations may serve the purpose of spatial anchoring, if they are part of the listener’s frame of reference. Examples as in (31a) and (31b) show how this works. (31)

a.

Ik lust deze snoepjes niet. I like these candies not

b.

Ik lust geen snoepjes. I like no candies

In (31a), the speaker apparently assumes that the entity snoepjes is part of the listener’s frame of reference, while in (31b), there is no such assumption involved. The difference between (31a) and (31b) seems to be due to the use of the definite NP deze snoepjes vs. the indefinite NP snoepjes. With the definite NP the speaker expresses his assumption that the referent is uniquely identifiable, whereas with the indefinite NP he expresses that it is not. Definiteness thus seems to play a role with respect to the spatial anchoring of an utterance. However, examples

Utterance production  

 73

as in (32) and (33) show that identifiability is not just a matter of definiteness vs. indefiniteness. (32)

(33)

a.

Een fiets staat niet in de stalling. a bake is not in the shed

b.

Er staat geen (= niet een) fiets in de stalling. there is no (= not a) bike in the shed

a.

De politie heeft de weg nog niet vrijgegeven. the police has the road not yet cleared

b.

De politie heeft nog niet de weg vrijgegeven. the police has not yet the road cleared

In both (32a) and (32b), the NP een fiets (a bike) is indefinite. Nevertheless, in (32a) een fiets refers to a particular object with a particular function. This object is assumed to be uniquely identifiable within the listener’s frame of reference. As it refers to a specific entity, it can establish spatial anchoring. In (32b) however, een fiets refers to any object with this particular function. It is not assumed to be uniquely identifiable. As it is unspecific, it cannot establish spatial anchoring. Furthermore, in both (33a) and (33b) the NP de weg is definite. In (33a) de weg refers to an object as uniquely identifiable and, hence, it is assumed to establish spatial anchoring. In (33b) however, de weg has no particular referent. In fact, in (33b) de weg may not refer to an object at all. It seems to function as part of a complex predicate phrase meaning ‘open for traffic’. Although it is definite, identifiability does not play a role. Given that identifiability and, thus, spatial anchoring is not a matter of definiteness, what seems to matter is the position of the NP. As shown in (31), (32) and (33), NPs are used to express identifiability, if they are placed before the position of the negator. The reason for this is the informational function of niet as a sentential negator. Sentential negation is the negative option of finiteness. The default option of finiteness is the expression that a situation is true, its negative option is that a situation is not true. The expression of the semantic function of finiteness, thus, occurs in F either without or with negation. F has scope to the right. The predication within the scope of F is asserted, either as true or untrue, elements not within the scope of F are assumed. They are assumed to be identifiable. Therefore it is that an NP or any other element within the scope of F may NOT serve the functional property of spatial anchoring, while outside the scope of F, i.e. to its left, it does. This explains why in (31), (32) and (33) the NPs that are placed before negative particles such as niet (not), geen (not-a/no), nog niet

74 

 The Target System

(not yet), nooit (never), nergens (nowhere) serve to express identifiability, while the NPs that are placed after these negative particles do not. Spatial anchoring provides a functional account of a syntactic phenomenon, commonly referred to as ‘scrambling’. In sum, it is the function of finiteness to express the pragmatic function of assertion. As an assertion a finite utterance has to be anchored with respect to both time and space. Temporal anchoring occurs morphologically with the finite verb. Spatial anchoring occurs lexically outside the scope of F.²¹

3.3.2.3 The function of SpecFP As the specifier of F, the element in SpecFP is used “to identify the situation to which the speaker’s assertion is confined” (Klein 2008: 290). That is, elements in SpecFP serve as a ‘hook up’ of the assertion to the situational context. Klein refers to this situational context as the ‘topic situation’ (2008: 287). As indicated in Klein, there are two possible ways for the listener to find out what situation the utterance is about, i.e. either through “information sources different from what the sentence itself provides” or through part of “the descriptive context of the sentence” (2008: 293). Klein refers to these processes as “external and internal topic situation identification”. External identification of the topic situation occurs through information that is present either in the situational context as in [ Noise ] The refrigerator has turned on or in the linguistic context as in We arrived around 10. Mary opened the kitchen door. The light was on. Internal identification of the topic situation takes place if the topic situation can be identified through an element that is part of the content of the utterance itself. For example, in John left early identification of the topic situation is possible with John. John is a ‘topic entity’. A topic entity “may be marked in a special way as contributing to IDENTIFY the topic situation” (2008: 290). In the following, a few examples may serve to illustrate how both internal and external identification of the topic situation can be achieved. Consider, for example, a situation in which children are playing a ball game. At some point a ball breaks through the window. In this situation a speaker, who notices that something is happening, may produce either of the following utterances.

21 Spatial anchoring occurs lexically. It is achieved with reference to entities and/or locations. Therefore, as opposed to temporal anchoring, spatial anchoring is not restricted to a set of linguistic elements that are mutually exclusive and, hence, it does not constitute a closed-class category.

Utterance production  

(34)

a.

De bal / Een bal vliegt door de RUÌT. the ball / a ball breaks through the WÌNdow

b.

De BÁL / Een BÁL vliegt door de ruit. the BÁLL / a BÁLL breaks through the window

c.

Er vliegt een BÁL (* de BÁL) door de ruit. there breaks a BÁLL (* the BÁLL) through the window

 75

The utterances in (34) may occur in a situation in which both speaker and listener are aware of the fact that there are children around playing a ball game. In an utterance like (34a), i.e. either De bal vliegt door de RUÌT or Een bal vliegt door de RUÌT, the NP de bal or een bal refers to an entity with which the topic situation can be identified internally. The utterance in (34b) may occur in a similar situation. All of a sudden there is the sound of breaking glass. In that situation the speaker may say (34b), i.e. either De BÁL vliegt door de ruit or Een BÁL vliegt door de ruit. With (34b) the utterance itself does not provide the information to identify the topic situation. The topic situation is identified externally through the sound of the breaking glass and/or the children screaming. The use of the definite or indefinite NP in (34a) and (34b) depends on the assumption of the speaker as to the number of balls the children are playing with. If he assumes that there is just one ball, the NP de bal is definite, because the ball is uniquely identifiable. If he assumes there is more than one ball, the NP een bal is indefinite, because he is not able to tell which particular ball it is. However, the ball is identifiable to the extent that it belongs to a particular set. Hence, the NP een bal in (34a) and (34b) is used with specific reference. The utterance in (34c) may occur in a situation in which the listener must not necessarily be aware of what is going on. The speaker hearing the noise of breaking glass may say Er vliegt een BÁL door het ruit. In this “seemingly ‘topic-less’ sentence” (Klein 2008: 290), it is the function of this utterance itself to establish a topic situation. In such a topic-less situation the ball is not assumed to be identifiable as an element of a given set. Hence, the meaning of the NP een bal is ‘unspecific’. The differences between the examples in (34) are represented schematically in (35).

76 

 The Target System

(35) Topic situation: Children playing with a ball Means to identify TS

Utterance SpecFP

F′

a. [ de / een bal ] De bal / Een bal the ball / a ball

vliegt door de RUÌT. flies through the WÌNdow

b. [ sound ]

De BÁL / Een BÁL the BÁLL / a BÁLL

vliegt door de ruit. flies through the window

c. [ (sound) ]

Er [= placeholder]

vliegt een BÁL (* de BÁL) door de ruit. there flies a BÁLL (* the BÁLL) through the window

Whether a topic situation is established internally (34a) or externally (34b, 34c) depends on the intonation with which the expression in SpecFP is used. In his decision to choose either an internal or an external means, the speaker relies on the current knowledge state of the listener.²² Internal identification of the topic situation is realized in SpecFP. SpecFP is, thus, the position for expressions to represent the topic situation that the utterance applies to. Examples of different types of topic situation are given in (36) (36) Topic situations Topic situation

Utterance SpecFP

F′

a. Event at some place

Op het schip on the ship

is brand ontstaan. has fire broken out

b. Event at some time

Vorige week last week

zag alles er nog zonnig uit. saw everything still sunny

22 As pointed out above, utterances can only be interpreted as assertions, if they are anchored within a situational context. However, reference to a specific situation is not a necessary condition. This is shown in, for example, Er is vast wel iemand die het weet (there is certainly someone who it knows). Here, the position of SpecFP is empty. Hence, there is no element to refer to the topic situation either internally or externally, and, therefore, there is no reference to a situation that the utterance could apply to. Nevertheless, this utterance is an assertion, because with tense marking on the finite verb is it is anchored in time.

Utterance production  

 77

c. Event with some entity

Dit sieraad this jewelry

kan bijna niemand zich veroorloven. can nearly nobody afford

d. Sudden event

Opeens suddenly

vliegt een bal door het raam. flies a ball through the window

e. Unexpected event

Toch yet

is hij er weer in geslaagd. has he again succeeded

f. Causing event

Daarom therefore

heb ik het maar niet gezegd. have I this not said

As shown in (36a, b, c), the expressions in SpecFP may refer to the place or the time of the topic situation or to an entity that plays a role in the topic situation. However, “the topic situation may also be characterized by many other types of information” (Klein 2008: 289). As shown in (36d, e, f), this happens with adverbial expressions, each in a different way. With opeens in (36d) the topic situation is identified as an event that occurs all of a sudden, with toch in (36e) it is identified as opposed to what one would expect and with daarom in (36f) as the reason for why a particular situation takes place. In sum, the situation that the utterance applies to is called the ‘topic situation’. Identification of the topic situation occurs either internally or externally. Internal identification is achieved if the topic situation can be identified through a ‘topic entity’ in SpecFP, i.e. an element that is part of the content of the utterance itself. External identification of the topic situation occurs through information that is present either in the situational or in the linguistic context.

3.3.3 Summary The content of an utterance is established by the processes of lexicalization and predication. Prerequisite to both lexicalization and predication is the conceptualization process. Whatever may be the result of a conceptualization process, the expression of the content is constrained by the possibilities of lexicalization in a particular language. Whether an event is lexicalized as a causal action, a state or a change of state depends on the process of subject selection. The subject is the external argument of a predication. It is the element that the predicate holds for. Subject selection serves the function of information structuring. In the default case the subject occurs in sentence-initial position. The predicate of a predication is a complex lexical entity. It consists of a simple predicate with its internal argument, if there is one. This complex predicate is stored in the mental lexicon as a

78 

 The Target System

lexical entity. Placement of the object before the verb in Dutch, i.e. OV structure, is a lexical property. The default function of an utterance is the expression of an assertion. With the expression of an assertion the speaker claims that the situation as described by the utterance is true. However, for an utterance to be true, it has to be anchored. In Dutch, this is achieved by means of both temporal and spatial anchoring. Temporal anchoring occurs with tense morphology on the verb in the position of F. Spatial anchoring occurs lexically, with constituents outside the scope of F. Spatio-temporal anchoring of an utterance applies to a particular situation, i.e. the topic situation. Identification of the topic situation occurs by means of either internal or external identification. The topic situation is identified internally by the referent of a constituent in SpecFP, externally by the inference of a referent that plays a role in the context of the relevant situation. SpecFP, thus, specifies the means by which an assertion is embedded in a particular situational context.

3.4 Hypotheses on language development The linguistic knowledge as represented in Figure 4 [repeated in Figure 9] seems rather complex. One may wonder how it is possible that language learners can derive this abstract knowledge system from the input they receive. One solution to account for this so-called learnability problem is to assume that relevant properties of linguistic structure are innate. However, I have chosen to pursue a different approach. That means that I will focus on the acquisition of structural knowledge as a function of the linguistic properties of information structure. My claim is that at the initial stages of language acquisition utterance structure is based on the lexical-semantic properties of the target language. At the relevant stage, this lexical-semantic knowledge is simultaneously used for the purpose of information structuring. However, this learner state of linguistic knowledge is unlike the linguistic knowledge state of the adult native speaker. In the native language system of the adult, properties of information structure are expressed by means of functional elements. Therefore, I will argue that it is the principles of information structure that are the driving force, causing learners to develop the basic lexical variety they use at the initial stages of language acquisition into the functional variety of the target system at stages that are more advanced.

4 The initial state 4.1 Finiteness at the initial state. Language production and perception are based on the internal linguistic knowledge system of the individual. This holds not only for adults speaking their native language, but also for children learning their mother tongue and adults learning a second language. Relevant data from spontaneous language use provide information on the structure of the underlying language system, whether it is the target system of a native speaker or the intermediate system of a language learner. Learner languages or learner varieties are often seen as deficient compared to the system of the target language. However, from a linguistic perspective, there is no reason why this should be the case. Klein even argues for the contrary. “Learner varieties are a genuine manifestation of the human language faculty, and the careful and systematic investigation of how they are internally structured and how they develop over time is a genuine contribution to the understanding of the human language faculty. In fact, I believe that learner varieties are the core manifestation of the human language faculty, and “real languages”- or a speaker’s perfect knowledge of a real language – are borderline cases. They are particularly interesting for social and cultural reasons, they are also interesting because they often exploit the structural potential of the human language faculty to a particular high extent. But to the linguist, they should be no more privileged than is the noble lion over the humble drosophila melanogaster to the biologist” (Klein 1997: 5).

One possible way to investigate the principles of a linguistic system is through analysis of (spontaneous) production data. A data base of production data is commonly used to establish form and function of learner grammars in both child L1 and adult L2 learners at the initial stages of acquisition. However, the use of spontaneous production data as the basis of empirical research poses a methodological problem. That is, to investigate a learner’s linguistic competence, it has to be established which categories of analysis are appropriate. It is too simple to assume that elements occurring in a learner’s language play a role in the same way as they do in the target language. The functioning of linguistic categories at any level of acquisition is determined by the properties of the system at the relevant stage. Therefore, an adequate account of learner data can only be provided from the perspective of the learner system. However, for the study of a particular learner variety these categories and rules are also the object of investigation. For this vicious circle there is no simple way out. Statements on the form and functioning of a learner system have to be tested with respect to their internal consistency and the type of data that is claimed to be accounted for. Furthermore, it has to be acknowledged that learner languages are evidence of a process of

80 

 The initial state

development. Hence, an adequate representation of the learner system should also account for the fact that specific aspects of the target language are not (yet) accommodated for or only in an interesting precursory way.

4.2 Theoretical accounts In research on the acquisition of child L1 and adult L2 Dutch, the use of verb morphology plays an important role. Particularly the observation on the distribution of ‘finite’ vs. ‘infinite’ verb forms. At the initial stages of acquisition, it seems as though verb morphology is already used in a targetlike way. As in the target language, it correlates with the position of the verb. Examples of verb forms as they occur most frequently in my L1 and L2 data are given in (1) and (2). (1) L1 Dutch. Utterances with ‘finite’ (a) and ‘infinite’ (b) verb forms a.

M valt. (J 1;10) M falls poes komt niet. (J 1;11) kitty comes not poes ligt. (J 1;11) kitty lies is dat? (A 2;1) is that? J heeft au. (A 2;1) J has ow J vindt vies, deze. (A 2;1) J finds yuk, this

b.

C koek happe. (J 1;10) C cookie bite M ook heppele. (J 1;10) M too help eve melluk pakke. (J 1;10) just milk get da ook kusse doen. (A 2;0) there too pillow do poppie niet J help. (A 2;0) doll not J help papa eve make. (A 2;1) daddy just make

(2) L2 Dutch. Utterances with ‘finite’ (a) and ‘infinite’ (b) verb forms a.

ik woon zwarte zee. (Os/T 1.8) I live black see jij blijft thuis. (Os/T 2.3) you stay home hij vindt leuk. (Os/T 2.5) he finds nice ik werk vijf maand. (Mo/A 1.3) I work five months hij loopt met straat. (Mo/A 1.5) he walks with street

b.

en dan ik ook slapen. (Os/T 1.2) and then I also go-sleep die lamp maken. (Os/T 1.5) that lamp make zoiets ik niet eten. (Os/T 1.8) such-thing I not eat hij arabisch lezen. (Mo/A 1.2) he arabic read hij maken die fiets. (Mo/A 1.3) he make that bicycle

Theoretical accounts  

gisteren ik slaap bij mij oom. (Mo/A 2.1) yesterday I sleep at my uncle

 81

hij staan voor de weg. (Mo/A 1.6) he go-stand before the road

In German, verb morphology is used in the same way as in Dutch. Hence, it should come as no surprise that the observations on L1 and L2 Dutch as presented in (1) and (2) are similar to what is found in research on L1 and L2 German. In L2 learners the correlation between verb morphology and verb placement seems less strict compared to children learning their L1. A few examples of the L2 Dutch of Osman/T and Mohammed/A are presented in (3). (3) L2 Dutch. Infinite verb forms in head-initial position. die man kopen een auto. (Os/T 1.9) that man buy a car dan hij vallen achter. (Os/T 2.9) then he fall behind hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault hij pakken mayonaise. (Mo/A 1.6) he get mayonnaise The use of infinite verb forms in head-initial position in (3) seems to be due to the fact that in L2 learners the infinitive may function as the default verb form. For L2 Dutch, this observation has been documented in Jansen et al. (1981), for L2 German in Meisel (1997) and Prévost and White (2000). Overall, however, in both the L1 and L2 learner varieties, verb morphology seems to correlate with verb placement such that ‘finite’ verb forms occur in initial position while ‘infinite’ verb forms occur in final position. Given this observation, how should this distributional difference be accounted for? What are the principles that it is based on? Given the fact that there are two verb positions, which of these might represent the basic or underlying word order of the relevant learner system? For the acquisition of verb placement in children learning Dutch and German as their native language, two types of hypotheses have been put forward. The relevant data as given in (2) suggest that these hypotheses might hold for adult L2 learners in the same way. According to the assumption of the Full Competence Hypothesis, verb placement in (1) and (2) should be accounted for in terms of the morpho-syntactic principles as they occur in the system of the target language. That is, the relation between the placement of finite and infinite lexical verbs is based on ‘movement’. Furthermore, as in the system of the target language, basic word order is

82 

 The initial state

represented with the infinite lexical verb occurring in final position. This hypothesis has been stated for L1 German in Poeppel and Wexler (1993), Wexler (1994), Wexler (1998), Ionin and Wexler (2002) and similarly for L1 Dutch in Verrips (1996), Wijnen (1997), Van Kampen and Wijnen (2000). Alternatively, there are a few hypotheses which state that verb placement in (1) and (2) should not be accounted for in terms of movement. Rather, they claim that the distributional difference has to be explained semantically: different sets of lexical verbs are used in complementary distribution. However, researchers do not agree as to the semantic criteria that are involved. Clahsen (1986) has argued that transitivity is the relevant feature: infinite verbs are transitive, they occur in final position; finite verbs are intransitive, they occur in initial position. Ingram and Thompson (1996) adhere to their Modal Hypothesis: infinite verbs in final position are used with a modal meaning; finite verbs in initial position are used with a non-modal meaning. In sum, under the assumption of the Full Competence Hypothesis verb placement at the initial stage of language learning is based on a morpho-syntactic system which entails an operation of movement. Under the assumption of a semantic hypothesis, verb placement is determined by the semantics of the lexical verb. Both types of hypotheses will be discussed below.

4.2.1 Morpho-syntactic systems 4.2.1.1 Poeppel and Wexler (1993), Wexler (1998) Among theories of L1 acquisition the Full Competence Hypothesis has attracted a great deal of attention. In their investigation of German L1 acquisition, Poeppel and Wexler argue that “young children’s grammars have functional categories and the principles which govern them” (1993: 2). Evidence for the presence of functional categories in early child German are ‘finiteness’, ‘agreement’, ‘head movement’ and ‘nonsubjects in first position’. Poeppel and Wexler’s investigation is based on a corpus of 282 utterances from Andreas (2;1). One of the claims put forward is that, at the relevant stage, there is “a contingency between the position of the verb and its inflectional status” (1993: 5). That is, “[+finite] verbs systematically appear in second clausal position, whereas [-finite] verbs systematically remain in final position” (1993: 5). Hence, it is claimed that “the finiteness distinction is made correctly at the very earliest stages of grammatical development” (1993: 6f.). The relevant morphosyntactic distribution is represented in Table 1 and illustrated with the examples in (4).

Theoretical accounts  

morphology

placement

finite root infinitive (-en)

initial final

 83

Table 1: Poeppel and Wexler (1993), Wexler (1998): Verb morphology and placement

(4) Andreas (2;1). Finite verb forms and (root) infinitives Thorsten Caesar haben. Thorsten has [the doll] Caesar Ich hab tein Bürse. I have [a] small brush The situation in L1 German is similar to the situation in L1 Dutch. In Jordens (1990), it has been claimed that the occurrence of finite verb forms in initial and infinitives in final position is due to the absence or presence of modal verbs in the input. Therefore, finite verbs are used in utterances with a non-modal meaning, while non-finite verbs are used with a modal meaning. For German, Poeppel and Wexler, however, conclude that the relevant distribution does not correlate with a semantic difference. “Our figures indicate that 20 of the 37 nonfinite verbs are realized, nonmodal uses. The other 17 are unclear; they may very well be descriptive …”. Hence, “[t]he infinitives are often clearly used in the same way that an adult would use a finite verb…” (1993: 16). Furthermore, Poeppel and Wexler argue that in early child German the agreement system is basically used correctly. Evidence from L1 German shows that, given -t inflection, the subject is a third person singular. Since “the child almost never uses -t in an incorrect syntactic context” (1993: 9), they state “[t]he singular agreement system is thus in place quite early …” (1993: 10). Evidence for the presence of head movement comes from the distribution of finite and infinite lexical verbs. Contrary to the results of studies by De Haan (1987) and Jordens (1990) on the acquisition of L1 Dutch, Poeppel and Wexler argue that in early child German finite verbs in V2 position and infinite verbs in final position belong to overlapping sets. Hence, they claim that “head movement as a morphosyntactic process is in place in the early grammar” (1993: 11). However, as opposed to adults, “[t]he one feature that seems to distinguish Andreas’ knowledge of German from an adult’s knowledge of German is that Andreas optionally allows infinitives as matrix verbs” (1993: 29). Apparently, for children, utterances with these root infinitives are grammatical at this age. Therefore, Wexler claims “the discovery (…) of the Optional Infinitive stage in grammatical development” (1998: 27).

84 

 The initial state

Finally, Poeppel and Wexler observe that non-subjects frequently occur in first position followed by a finite verb. From the fact that in 28% of the sentences overt subjects occur in non-canonical word order, they conclude that “the V2-phenomenon, which is attributed to the existence of a CP system, is in place” (1993: 15). These observations on the early use of finiteness, agreement, head movement and non-subjects in first position are all tied to properties of the functional category system of the target language: “the morphosyntactic processes associated with finiteness and attributable to the availability of functional categories are in place” (1993: 2). “Children at a remarkably young age know some very abstract properties of grammar, including head movement, the properties of inflection that require head movement – in particular from V to I to C – and properties forcing a constituent to move into Spec, C” (1993: 30; see also Wexler 1998: 28). Therefore, the relevant data are interpreted as evidence in favour of the Full Competence Hypothesis. Since the Full Competence Hypothesis is a theory of the initial state of the language faculty, it provides a solution to the learnability problem. However, given Poeppel and Wexler’s (1993) claim that for early child German “the best model of the data is the standard analysis of adult German” (1993: 2), one may wonder what is left for children to learn.

4.2.1.2 Discussion Poeppel and Wexler take the use of hab vs. haben in (4) as an example of the kind of data they see as evidence that finiteness is a property of the child’s grammar. This is because the verb ‘have’ occurs morphologically both in a finite (hab) and an infinite (haben) form. Learners who know about finiteness are able to use a particular verb differently, morpho-syntactically, i.e. they use an infinite verb form in final position and a finite verb form in initial position. However, even if learners use verb forms as in (1), is this evidence of finiteness as a morphosyntactic category? That is, how do we know that for the learner both forms are manifestations of the same verb? Second, how do we decide that both forms are instantiations of an opposition that is morphologically systematic? What functions as a morphological opposition in the target language could in fact be an irrelevant feature of the learner’s language. Poeppel and Wexler, however, tacitly assume that verbs and verb forms at the initial stage of language acquisition can be analysed in terms of categories of the adult system. Hence, the assumption of innateness seems to be an artefact of the analysis of child utterances in terms of the adult model. In order to find out if morpho-syntactic finiteness is a relevant category in the learner’s language, there must be evidence that learners know that particular

Theoretical accounts  

 85

morphological properties are linguistically relevant. However, it seems questionable that children at the relevant stage can discriminate between verb forms on the basis of morphological variation. Poeppel and Wexler refer to the procedure as proposed in Clahsen (1986) to argue that in early L1 acquisition the singular agreement system is in place. However, a targetlike use of agreement morphology can also be due to the fact that, given a particular position, the relevant verb forms are learned as unanalysed items. In the target language, the agreement system is based on a morphological rule. If learners know about agreement, their production should be rule-based. Evidence of a rule-based acquisition of morphology are errors of generalization. With respect to agreement morphology, evidence of rule-based acquisition in L1 Dutch is obtained if children produce 3Sg t-endings in contexts in which the 3Sg has a 0-ending. This might occur with modal verbs such as kan (can) or mag (may). However, at the relevant stage, there are no examples of 3Sg *kant or *magt. Poeppel and Wexler’s assumption underlying their Optional Infinitive Hypothesis is the claim that the relevant child data show random head movement. As is the case in German, head movement also occurs in Dutch. However, contrary to Poeppel and Wexler’s claim for German, I have argued in Jordens (1990) that, in early Dutch, finite and infinite verbs are used in complementary distribution. Given “[t]he small amount of overlap between verbs used in first/ second position and final position” (1990: 1431), it seems highly unlikely that children have discovered verb movement. Moreover, the distribution of particle verbs in early child Dutch provides a case of counterevidence that is even more telling. If a particle verb in Dutch is finite, the verbal part occurs in verb-second position, while the particle remains in final position. Most of the particle verbs used by children, and probably in the adult input too, are causal action verbs. In child Dutch, they are always used as one entity in final position. That is they always occur infinite, i.e. as instances of ‘root infinitives’ as in opeten (up-eat) and stukmaken (kaput-make). On the other hand, children also use a few particle verbs that are change-of-state verbs. These are always finite as in komt (er)aan (comes at-it) or vliegt weg (flies away). Hence, the use of particle verbs in child Dutch indicates that, with respect to the use of finite vs. infinite morphology, semantic properties are involved. At any rate, the relevant observation establishes a clear case against random head movement. Head movement is possible if the language system has two verbal positions. This holds for German as well as for Dutch. The position of the finite verb is in fact the position of the auxiliary verb. Hence, head movement may occur, if the language system has such a position. However, at the relevant stage a prominent feature of learner languages is the absence of auxiliary verbs. The point is, how

86 

 The initial state

can a lexical verb occur in the position of the auxiliary verb if the language system has no auxiliary verbs? And if there is no position for auxiliary verbs, how can there be head movement? Furthermore, since auxiliaries are typically absent in early child German and Dutch, how can there be non-subjects occurring in first position? Finally, at the relevant stage there are no utterances with wh-question words in initial position, nor is there evidence of tense marking. If one assumes the presence of a functional position for the finite verb, why is it that tense is absent and that there is no functional position for wh-movement? Poeppel and Wexler also note that “children do not use overt complementizers” at the initial stage (1993: 19). They claim that this is because presumably for reasons of a limited processing capacity “young children rarely use subordinate clauses”. Given that “absence of evidence for some category does not constitute evidence for its absence”, they claim that “[t]he lack of overt complementizers (…) does not conclusively prove the absence (or unavailability) of the linguistic category in question” (1993: 20). However, contrary to Poeppel and Wexler, the examples in (5) show that Dutch children produce simple forms of embedded clauses already quite early. (5) L1 Dutch. Early forms of embedding. Finite verb form in final position ulle afdoet. (J 1;10) want, (he it) off-does ulle pit indoet. (J 1;10) want, (he) pit in-does papa v(r)age toetje is. (J 1;10) daddy ask, desert is kijk es poppie mooi is. (J 1;11) look, doll nice is mama kijk es Ernie doet. (J 1;11) mommy look, Ernie does wip wap is? (A 2;0) see saw is? kijke pakband is. (A 2;1) look, tape is kijk es papa Jaja hand heeft. (A 2;1) look daddy, J hand has papa kijk es mooi is. (A 2;2) daddy look, nice is

Theoretical accounts  

 87

These embeddings are systematically used with no complementizer.²³ They are introduced with particular verbal elements such as kijk es (look how), ulle (I want that), v(r)age (I want to ask). This indicates that they are used as structurally unanalysed elements and, hence, they serve as evidence that, at the relevant stage, the functional category in question is absent. To sum up, in early child German and Dutch evidence for the presence of a functional category system is lacking. Morphological endings in early child language that look like properties of a functional projection appear to be part of lexical elements that are used as elements that are morphologically unanalysed.

4.2.2 Semantic systems 4.2.2.1 Clahsen (1986) Clahsen (1986) analysed the use of verb forms in early L1 German in terms of a correlation between morphology and semantic transitivity. He argues that his notion of semantic transitivity should be interpreted in the sense of Hopper and Thompson (1980: 252). That is, ‘transitive’ means actively involved or intentional, while ‘intransitive’ means less actively involved or non-intentional. According to Clahsen, verb forms with a t-ending are used intransitively, while verb forms without a t-ending are used transitively. Hence, he claims that verb forms are semantically distributed as in Table 2.

morphology

semantics

t-ending no t-ending

intransitive transitive

Table 2: Clahsen (1986): Verb morphology and transitivity

The functioning of t-morphology in early L1 German is illustrated in (6) with an example from Clahsen (1986: 102). It shows the difference in use between dreht (turns) and drehen (turn).

23 Similar data have been observed in L1 French, L1 German and L1 Swiss German (Müller and Penner 1996).

88 

 The initial state

(6) dreht (intransitive) vs. drehen (transitive) a.

dreht immer (M 2;5) turns always ‘M. points to a carousel which is turning’

b.

diese drehen (M 2;5) this-one turn ‘M. is turning the lens of a camera’

In (6) the infinitive form drehen (turn) is used with its transitive meaning of ‘someone turns something around’, while the finite form dreht (turns) is used with its intransitive meaning of ‘something or someone turns around’. That is, in its transitive use the verb drehen refers to an action carried out by an agent, in its intransitive use the verb dreht refers to a state that applies to a theme.

4.2.2.2 Discussion The semantic notion of transitivity as it is used in Clahsen (1986) differs from the syntactic notion of transitivity. This can be demonstrated with some child utterances in L1 German in which the verb forms hat (has-3Sg) and haben (have-Inf) are used. The examples in (7) are taken from Poeppel and Wexler (1993; 5f., 11) and Behrens (1993: 141). (7) hab, hat (intransitive) vs. haben (transitive) a.

ich hab tein bürse. (Andreas 2;1) I have [a] small brush hat noch schinken drauf. (Sim 2;1,18) has still ham on it

b.

Thorsten Caesar haben. (Andreas 2;1) Thorsten [the doll] Caesar have [= get] du das haben. (Andreas 2;1) you that have [= get] saft habe. (Sim 1;10,20) juice have [= get]

In these examples, both hat and haben are used syntactically transitively, i.e. the lexical verb is used with an object. Semantically however, hat is used less transitively compared to haben. The same semantic difference in use occurs in child L1 Dutch not only with heeft (has-3Sg) vs. hebben (have-Inf), but also, as

Theoretical accounts  

 89

shown in (8), with doe(t) (it works-3Sg) vs. doen (do-Inf), prikt (it pricks-3Sg) vs. prikken (prick-Inf), gaat (goes-3Sg) vs. gaan (go-Inf), and zit (is-3Sg) vs. zitten (sitdown-Inf). (8) Semantic transitivity: ‘intransitive’ (a) vs. ‘transitive’ (b) a.

Tompoes heeft koffie. (J1;11) b. T has coffee Jaja heeft au. (A 2;1) J has ow hij doe(t) niet. (J 1;11) he does not [ ‘it does not work’ ] hij doet. (J 1;11) it works ja doet wel. (A 2;2) yes, works indeed papa pikt niet, opa pikt hel. (A 2;1) daddy pricks not, grandfather pricks indeed deze gaat niet. (A 2;1) this-one works not grote auto. gaatie inne garage. (J 2;2) big car. goes-he into garage beetje melk zit in. (J 1;11) little milk is in

poes bal hebbe. (J 1;11) kitty ball have papa vorrek hebbe? (A 2;1) daddy fork have? Mijnie zelf doen. (J 1;11) M self do ik doen. (J 1;11) I do eve koud doen. (A 2;2) just cold do vokkik pikke. (A 2;2) (with) fork prick óok gaan, paarde. (A 2;1) [me] tóo go, horses mama, school toe gaan. (J 2;1) mommy, school to go ikke da zitte. (J 1;11) I there sit-down

Contrary to Clahsen however, it seems that the use or non-use of t-endings is not a matter of semantic transitivity or degree of activity. Rather, I would claim that it is categorical, i.e. that it is based on the semantic opposition between, on the one hand, state and change-of-state verbs and, on the other hand, action verbs. It explains the evidence in (8), i.e. for example, the use of the state verb heeft (has3Sg) vs. the action verb hebben (have-Inf), the change-of-state verb gaat (goes3Sg) vs. the agentive motion verb gaan (go-Inf) and the state verb zit (sits-3Sg) vs. the agentive motion verb zitten (sit-down-Inf). Further evidence is the use of state verbs such as lust (likes-3Sg), weet (knows-3Sg), vindt lekker (finds nice-3Sg) and change-of-state verbs such as lukt (succeeds-3Sg), past (fits-3Sg). They typically occur with t-morphology. Action verbs however, such as geven (give-Inf), lezen (read-Inf), losmaken (loose-makeInf), opeten (up-eat-Inf) and agentive motion verbs such as klimmen (climb-Inf), zwemmen (swim-Inf), glijden (slide-Inf) typically occur with en-morphology.

90 

 The initial state

In sum, Clahsen’s notion of semantic transitivity or degree of activity is in fact a categorical one. It accounts for the difference between two types of lexical verb categories, i.e. state and change-of-state verbs as in (7a) and (8a) and agentive verbs as in (7b) and (8b). State and change-of-state verbs occur with t-morphology, agentive verbs occur with en-morphology. This categorization however, leaves unanswered the question why it is that the former category occurs with t-morphology, whereas the latter occurs with en-morphology.

4.2.2.3 Ingram and Thompson (1996) Ingram and Thompson analysed the L1 German data from four children: Dorothy (1;11–2;5), Nicole (1;8), Katrin (1;5) and Andreas (2;1). Andreas is the child that has also been studied in Poeppel and Wexler. However, unlike Poeppel and Wexler, Ingram and Thompson come to the conclusion that finite and infinite verb forms belong to different sets of lexical verbs: “Inflected verbs are initially single morphemes unrelated to their nonfinite counterparts” (1996: 101). Furthermore, they argue that “[t]he use of inflections, in and of itself, is not sufficient evidence to claim that they are acquired”. This is because “the large majority of verbs occur in only one inflected variant form” (1996: 111). Hence, they conclude that, initially, inflected verbs are learnt holistically. These observations lead to what Ingram and Thompson regard as their “basic assumption” (1996: 101) on how to account for their child L1 data. Differences in the use of verb morphology are simply a matter of “what they hear is what you get” (1996: 97). “That is, children will by the nature of their conservative learning strategies show surface forms that look like the input language to which they are exposed. In and of themselves, therefore, such forms cannot be taken as evidence of the adult grammar.” (1996: 101). Ingram and Thompson, furthermore, argue that this does not mean that the use of verb morphology is meaningless. Rather, they claim that, as opposed to the adult grammar, in early child German verb morphology correlates with a semantic distinction which can be accounted for in terms of what they call the Lexical-Semantic Hypothesis. This hypothesis is “(…) lexical in the sense that early inflected forms are first acquired as lexical items, not as roots plus affixes. (…) It is semantic in the sense that children will assign distinct semantic functions to grammatical forms” (1996: 101). “Finite forms are heard in contexts involving ongoing events, and nonfinite forms are heard in modal contexts, with overt modals in the adult German input”. (1996: 109). Therefore, “(…) the infinitives are semantically associated with modality” (1996: 102). Examples of child utterances with infinite forms in a modal context are given in (9). These child utterances are used as the answer to a preceding question by an adult.

Theoretical accounts  

 91

(9) Child utterances. Infinitives in a modal context (Ingram and Thompson 1996: 106f.) Mother:

Was möchtest du haben? what want you have?

Katrin:

Stift haben? crayon have?

Mother:

Ach, du möchtest einen Stift haben. yes, you want a crayon have

Grandfather:

Ich glaub(e) der will mit dem Auto fahren wohl? I think he wants with the car drive, right?

Andreas:

Auto fahren [fahrn] Nikolaus. car drive Nikolaus

However, examples as in (9) are exceptions. Ingram and Thompson state that, in all four children, “root infinitives were rarely produced as answers to immediately preceding questions containing modals. Rather, children produce them in a variety of contexts, whenever they need to express a desire, request an object, state a wish, and so on” (1996: 114). Hence, the use of infinitives is not the result of dropping elements from structures as they were used in the preceding utterance of the adult. Infinitives are claimed to be used with a modal meaning by themselves. This observation is accounted for with what Ingram and Thompson call the Modal Hypothesis. It entails that “German children in their early stage of acquisition use infinitives as main verbs in sentences that contain a modal interpretation, i.e. that some activity will, can or should occur” (1996: 102). „The Modal Hypothesis (…) claims only that a correlation exists between modality and finiteness” (1996: 113). The relevant distribution is given in Table 3.

morphology

semantics

‘finite’ root-infinitive (-en)

non-modal modal

Table 3: Ingram and Thompson (1996): Verb morphology and modality

Comparing the Lexical-Semantic Hypothesis of Ingram and Thompson with the Optional Infinitive Hypothesis put forward by Poeppel and Wexler, it appears that both hypotheses make rather different claims. Given the utterances of Andreas

92 

 The initial state

in (4), for example, Ingram and Thompson will argue that they are used with a difference in meaning. Contrary to Poeppel and Wexler, they assume that haben has a modal meaning (want to have), while hab (have-1Sg) has not. The figures in Table 4 summarize the data from Ingram and Thompson (1996: 108, Table 3). They show that root infinitives are used with a modal meaning, while the meaning with which ‘finite’ verb forms are used is non-modal. These figures are therefore interpreted as evidence in favour of the Modal Hypothesis.

Dorothy

Nicole

Katrin

Andreas

Mean

Infinitives Modal meaning

139  74 (.53)

93 93 (1.00)

103  91 (.88)

162 121 (.75)

124 (.79)

Finite verbs Modal meaning

126   3 (.02)

33 13 (.39)

103   4 (.04)

202  10 (.05)

125 (.13)

Table 4: Ingram and Thompson (1996): The use of infinitives and finite forms in modal contexts

In sum, Ingram and Thompson argue that in early L1 German children show no evidence of grammatical knowledge that can be accounted for in terms of verb movement. Inflected verb forms and infinitives constitute two separate verb categories. According to the Lexical-Semantic Hypothesis these verb categories are used to express different semantic properties. They are claimed to be due to properties of the input, in which infinitives occur in “modal contexts, with overt modals …” (1996: 109). It explains why in children modality becomes a semantic property of infinitives. With this modal meaning infinitives come to be used as a means to express the pragmatic function of a wish, a desire, a request, etc.

4.2.2.4 Discussion According to Ingram and Thompson, en-morphology is a formal property of verb forms that are used to express some modal meaning. However, evidence in Table 4 shows that there are also cases in which en-morphology does not correlate with a modal use. Thus, it is not the case that all ‘root infinitives’ are modal. Furthermore, in Ingram and Thompson the modal meaning of root infinitives is accounted for by what they phrase as “what they hear is what you get”. Thus, utterances with root infinitives in child German are modelled after adult expressions with an infinitive. These infinitives are used with a modal meaning both in expressions with and without a modal verb as in (10) and (11).

Theoretical accounts  

 93

(10) Adult input: Modal-Infinitive Mother:

Was möchtest du haben? what want you have?

Mother:

Ach, du möchtest einen Stift haben. yes, you want a crayon have

Grandfather:

Ich glaub(e) der will mit dem Auto fahren wohl I think he wants with the car drive indeed

Mother:

Ja, dann müssen wir es hier in den Mülleimer tun, ne? yes, then must we it here into the wastebasket do, right?

Father:

Willst du wieder malen? want you again paint?

Father:

Was sollen wir jetzt malen? what shall we now paint?

(11) Adult input: Infinitive Grandfather:

Mit (de)m Auto fahren? with the car drive?

Father:

Was machen? what do?

Father:

Papa malen? Papa paint?

Furthermore, it should be noted that in the adult language infinitives may occur in modal contexts with an object or an adverb in initial position. In fact, this type of utterance seems most frequent. Examples are some of the adult utterances in (10). They are repeated here in (12). (12) Adult input: Object-Modal-Infinitive; Adverb-Modal-Infinitive Mother:

Was möchtest du haben? what want you have?

Mother:

Ja, dann müssen wir es hier in den Mülleimer tun, ne? yes, then must we it here into the wastebasket do, right?

Father:

Was sollen wir jetzt malen? what shall we now paint?

94 

 The initial state

However, despite the type of input as in (12), utterance structure in child language has the subject in initial position. Hence, children produce either (13) or (14). (13) Child utterances: Subject-Object-Infinitive Nicole:

Nicole Wurst haben. Nicole sausage wants to have

Dorothy:

Bebi haye machen. Baby wants sleep make

Andreas:

Ich aber haben. I but want to have

Katrin:

Katrin machen. Katrin wants to do

(14) Child utterances: Object-Infinitive Katrin:

Stift haben. crayon have

Andreas:

Auto fahren [fahrn] Nikolaus. car drive Nikolaus

Katrin:

Da daraus [daus] machen [maren]. das there-out make

Why would that be the case? That is, why is it that they produce either (13) or (14) but never utterances like Was du haben? (what you have?), Dann wir es in den Mülleimer tun (then we in the wastebasket do), Was wir jetzt malen (what we now draw) (Poeppel and Wexler 1993: 16f.; Weissenborn 1992). If utterances as in (12) occur in the input so frequently, why is it that children do not use them as a model? In sum, the Lexical-Semantic Hypothesis provides an explanation for the relation between verb morphology in early child language and the presence or absence of a modal meaning. However, it does not provide an explanation for the fact that not all infinitives are modal. Furthermore, the claim that the use of root infinitives by children should be accounted for in terms of the adult input does not explain why it is that utterances with a modal infinitive as they occur in the input as in (12) are not produced by children.

Theoretical accounts  

 95

4.2.3 Summary The investigations by Poeppel and Wexler (1993), Clahsen (1986) and Ingram and Thompson (1996) revolve around the use of verb morphology at the initial state of German child language. The hypotheses put forward concern the question of the grammatical status of the morphological difference between finite and infinite verb forms at the relevant stage. In Poeppel and Wexler (1993) and Wexler (1998) the morphological opposition ‘finite’ vs. ‘infinite’ is claimed to be relevant only with respect to the positioning of the verb. Morphology correlates with positioning, i.e. the same verb can be used in either of two positions each with its appropriate morphology. The discussion with other proposals, as presented in De Haan (1987), Jordens (1990) and Ingram and Thompson (1996), focuses on ‘the amount of overlap’, e.g. the number of verbs that occur in both categories simultaneously. Poeppel and Wexler in fact argue that an overlap of 8 out of 51 verb types can be taken as evidence for a correlation between verb morphology and verb placement, i.e. as evidence of head movement. However, in Poeppel and Wexler two observations have not been taken into account. First, contrary to Poeppel and Wexler, it seems that finite verb forms and infinitives are distributed semantically differently even, as observed in Ingram and Thompson, across children. Thus, brauchen (need), können (can), möchten (want), wollen (want), müssen (must), lassen (let), sein (be), werden (become), hängen (hang), heissen (to be called), fliegen (fly), gehen (go), kommen (come), finden (find), kriegen (get), nehmen (take), wissen (know) typically occur in their finite form, while putzen (clean), spielen (play), werfen (throw), zeigen (show) typically occur in their infinite form. The finite types of verb are modal verbs and mostly state or change-of-state verbs. The infinite types of verb are (causal) action verbs. Interestingly, the same holds for particle verbs. Particle verbs mostly refer to (causal) actions. As observed in Poeppel and Wexler, they occur as unseparated entities and with infinite morphology in final position. Examples are: sauber machen (clean-make), abtrocknen (dry-make), anziehen (on-dress), aufblasen (up-blow), dareintun (in-put), fertigmachen (ready-make), hineintun (in-do), hineinmachen (in-do), Tor machen (goal make). There are a few exceptions of particle verbs that refer to a state or a change of state. It seems no coincidence that it is precisely these types of particle verbs that occur in separated form and with finite morphology in initial position. Examples are: kippt um (falls-3Sg over) and fährt ab (drives-3Sg away). Second, contrary to Poeppel and Wexler, I would argue that acquisition data as presented in (4), (6), (7) and (8) should not be analysed as evidence of head movement. The finite and infinite verb forms are used with different meaning.

96 

 The initial state

They are instances of polysemy. A clear example of polysemy is the use of dreht (turns) vs. drehen (make turn) in (6). A similar example from L1 Dutch children is doet (works) vs. doen (do) in (8). In these examples the finite verb is used to express a state or a change of state that applies to a theme, while the non-finite verb is used to express an action performed by an agent. In sum, at the initial stage of acquisition, both in German and Dutch, finite verb forms and infinitives are used in complementary distribution. This means that in early Dutch and German head movement does not exist. Hence, there is no functional projection to account for finiteness as a functional category either. According to both Clahsen (1986) and Ingram and Thompson (1996), morphology correlates with the meaning of the verb. In Clahsen, the opposition between the presence and absence of t-morphology is explained in terms of transitivity. Transitivity is used not as a syntactic notion but as a semantic one. It refers to the degree of agentivity. In Ingram and Thompson, the opposition between the presence and absence of en-morphology is interpreted as evidence of the presence or absence of the semantic feature of modality. This distributional difference is a reflection of the input. It accounts for the modal meaning with which root infinitives are used and the non-modal meaning of unanalysed finite verb forms. However, it seems that the following observations have been overlooked. First, the data from Ingram and Thompson show that root infinitives are not necessarily modal. Second, given the assumption that child utterances with the order SubjectObject-Infinitive are produced on the basis of the adult model with a modal verb, there is no explanation for why this adult model does not serve as a model for children to produce utterances with the order Object-Subject-Infinitive as in Was du haben? (what you have?) or Adverb-Subject-(Object)-Infinitive as in Dann wir es in den Mülleimer tun (then we in the wastebasket do) (see also Chapter 5, fn. 39). On the other hand, according to Poeppel and Wexler children do have utterances with the order Object-Verb-Subject and Adverb-Verb-Subject. However, the relevant data seem to be produced as unanalysed phrasal entities. They are not productive and, therefore, they can not be taken as evidence for a functional projection. In the accounts given by Clahsen, Wexler et al. and Ingram and Thompson, focus is on the formal feature of morphological variation²⁴. They entail the

24 Rizzi (1993) assumes that, at the early stage, non-finite sentences in child language are evidence of a grammar that has truncated structures with no CP node. Rizzi’s claim accounts for the fact that initially, while finiteness is lacking, non-finite sentences are used abundantly. However, it does not account for those verbs that seem to appear with finite morphology. Or, how in the absence of finiteness the observation of the distributional properties of verb forms can be accounted for.

Theoretical accounts  

 97

assumption that there is a systematic relation between finite and infinite verb forms occurring in different positions. This distributional opposition is interpreted by Clahsen as +/− transitive, by Ingram and Thompson as +/− modal and by Poeppel and Wexler in terms of head movement. It also entails the assumption that one position represents basic word order. As in the adult grammar, this is determined on the basis of the position of the root infinitive. However, the question is, whether such a morphological categorization provides an adequate account of the language system at the relevant stage. Do children really discriminate between verb forms on the basis of their morphological endings? That is, is it the case, as claimed by Poeppel and Wexler with their Optional Infinitive Hypothesis, that for learners a particular morphological feature is one of two alternatives that are equally possible? Or is it the case, as argued by Clahsen and Ingram and Thompson, that a particular morphological feature correlates with a particular meaning? Although it is generally assumed that, at the initial stage of L1 German and Dutch, morphology is relevant as a grammatical feature of the learner language, the following observations on the acquisition of L1 Dutch suggest that this assumption does not provide a fair account. First, a comparison of the typetoken ratios (number of different verbs vs. total number of verbs) of finite and non-finite verbs shows that, at the initial stages of acquisition, Dutch children use relatively many different types of infinite verb forms relatively infrequently. Finite verbs on the other hand, constitute a comparatively small class of different verb types, each however occurring relatively frequently. This observation may be interpreted as in Ingram and Thompson (1996), i.e. as evidence that finite verbs are used as unanalysed entities. It is confirmed by the fact that there is no evidence that agreement rules are applied productively. Rather, as illustrated in (15), children may even use verb forms that include the clitic as part of the morphology of the verb. (15)

da hootie oma. (J 1;11) there lives-it grandmother hootie da Cynthia. (J 1;11) lives-it there C da valtie Oscar. (J 1;11) there falls-it O mammie, valtie / vallik. (J 1;11) mommy, falls-he / fall-I

98 

 The initial state

Hence, in early child grammar of Dutch there is no evidence of a system of inflectional morphology that entails a distributional opposition between finite and infinite verb morphology. Utterances characterized as ‘root infinitives’ are used most frequently. The infinitives are the head of the predicate. However, at the relevant stage, as shown in (16), other types of predicate are used with the same function and the same distribution. (16)

ulle poesje toe. (J 1;10) want kitty to mama kanniet kusje. (J 1;11) mommy cannot kiss hoeniet plak op. (A 2;0) must-not glue on niet oppe tegels. (J 1;11) not on-the tiles Mijnie tiktak om. (J 1;10) M watch on Jaja dop op. (A 2;0) J cap on Erre huis toe. (A 2;0) E home to

The types of predicate in (16) are evidence that, for children, the relevant category of linguistic analysis is not the infinitive but the predicate. In other words, at the initial stage of L1 Dutch a morphological analysis of verb forms seems inadequate. In language acquisition research in German and Dutch, the morphological variation of verb forms has become relevant due to the fact that the child data have been studied from the perspective of the adult grammar. From the perspective of the child, however, morphological finiteness as a grammatical property seems irrelevant. For children, instead of the verb, it is the predicate that counts. Furthermore, predicates are used in complementary distribution either with head-initial or with head-final structure. Therefore, neither one nor the other may establish basic word order. From the arguments given above, it seems evident that, at the initial stage of language acquisition, morphological ‘finiteness’ only reflects the form in which a verb may be used in the adult input. Placement of the verb is due to the internal structure of the predicate as either head-initial as in (17a) or head-final as in (17b).

The alternative  

 99

(17) a.

M valt. (J 1;10) b. M falls poes komt niet. (J 1;11) kitty comes not ich hab tein bürse. (Andreas 2;1) I have [a] small brush dreht immer. (M 2;5) turns always

C koek happe. (J 1;10) C cookie bite papa eve make. (A 2;1) daddy just make Thorsten Caesar haben. (Andreas 2;1) Thorsten [the doll] Caesar have [= get] diese drehen. (M 2;5) this-one turn

As indicated in (6), (7) and (8), both in L1 Dutch and L1 German, predicates that are head-final are semantically different from predicates that are head-initial. Hence, a semantic categorization seems adequate to account for the two types of utterance structure that children in early Dutch and German apparently use. In the following, I will demonstrate that the complementary distribution of predicates, not verbs, is based solely on semantic properties. More specifically, I will show that there is a correlation between the semantic category of the predicate and placement of the predicate in either head-initial or head-final position.

4.3 The alternative The investigations presented above have shown that verb forms in early L1 German are used in complementary distribution. Poeppel and Wexler (1993) argue in favour of a distribution which is purely morphologically determined. That is, morphological properties are accounted for in terms of head movement. For Poeppel and Wexler, head movement is evidence in favour of the Full Competence Hypothesis. However, as noted in Ingram and Thompson, the distribution of verbs in two sets is not random, not only within but also across subjects. Hence, there must be “some factor other than a syntactic rule of movement that determines finiteness” (1996: 112). As pointed out, Clahsen (1986) made a similar observation. As Ingram and Thompson, he claims that this distributional difference should be accounted for in terms of a semantic distribution. The following observations are based on the data provided by Poeppel and Wexler, Ingram and Thompson, and Clahsen. These data lead to the conclusion that finite and infinite verb forms seem to belong to non-overlapping sets of verbs. First, one set of verbs typically occurs with finite form in initial position. Examples from Poeppel and Wexler are: kommt (comes), weiß (knows), geht (goes), findet (finds), fliegt (flies), heißt (is called), kippt um (falls over), fährt ab

100 

 The initial state

(drives away) and modals such as braucht (has to), kann (can), möchte (wants), muß (must), will (wants) (1993: 12f.). Examples from Ingram and Thompson are: kommt (comes), frißt (eats), geht (goes), gehört (belongs), guckt (looks), reitet (drives), sticht (pricks), tut (does), gibt (gives), liegt (lies), weiß (knows) and modals such as kann (can), möchte (wants), muß (must), will (wants) (1996: 15ff.). Clahsen has the examples: fährt (drives), passt (fits), fehlt (lacks), weint (cries), hält (stops) (1986: 97). Second, another set of verbs typically occurs with infinite form in final position. Examples from Poeppel and Wexler are: putzen (clean), werfen (throw), zeigen (show); abtrocknen (off-dry), anziehen (on-dress), hineintun (in-put), sauber machen (clean make) (1993: 12f.). Examples from Ingram and Thompson are: essen (eat), fahren (drive), hauen (hit), holen (take), lassen (leave), malen (paint), putzen (clean), raten (guess), schlafen (sleep), schreiben (write), spielen (play), suchen (search), werfen (throw), zeigen (show); anfassen (on-take), angucken (at-look), aufblasen (up-blow), mitfahren (with-drive), vorlesen (to-read) (1996: 117f.). Clahsen (1986: 112) has the examples: rausräum (out-take), drauftun (on-do). Third, a few verb forms may occur both in initial position with finite morphology and in final position with infinitival morphology. However, these verb forms are used with a difference in meaning that correlates with their position. Examples are dreht (turns) and drehen (make turn) in Clahsen (1986: 102) and hat (has) and haben (want to have) in Poeppel and Wexler (1993: 11). The situation as it holds for German can also be observed in Dutch, not only in early L1 Dutch, but to some extent in early L2 Dutch, too. As pointed out before, in early L1 Dutch, a small set of verbs is used in initial position with ‘finite’ morphology, while a large set is used in final position with ‘infinite’ morphology. A few examples were given in (1). More examples are given in (18). They provide a representative sample of the types of verb that were used most frequently in L1 Dutch. (18) L1 Dutch. The distribution of verb forms in head-initial (a) and head-final (b) position Jasmijn a.

M valt. (J 1;10) M falls poes ligt. (J 1;11) kitty lies poes komt niet. (J 1;11) kitty comes not

b.

C koek happe. (J 1;10) C cookie bite M ook heppele. (J 1;10) M too help eve melluk pakke. (J 1;10) just milk get

The alternative  

poes lus niet. (J 1;11) kitty likes not magwel, dat hebbe. (J 1;11) may-indeed, that have hoortie daarin. (J 1;11) belongs-it therein past. (J 1;11) fits heettie? (J 1;11) is-called-he?

 101

ulle pop hebbe. (J 1;10) want doll have poesje vinger bijte. (J 1;10) kitty finger bite water indoen. (J 1;11) water indo mama kanniet kusje. (J 1;11) mommy cannot kiss niet omgooie. (J 1;11) not overthrow

Andrea a.

gaatie niet? (A 2;0) works-it not? moet hier. (A 2;0) must here is dat? (A 2;1) is that? J heeft au. (A 2;1) J has ow J vindt vies, deze. (A 2;1) J finds yuk, this hier kan wel. (A 2;1) here can indeed gaap kom niet. (A 2;1) sheep come not

b.

da ook kusse doen. (A 2;0) there too pillow do poppie niet J help. (A 2;0) doll not J help niet au. aaie. (A 2;0) not ow. pet eve kijke. (A 2;0) just look papa uitdoen. (A 2;0) daddy off-do papa eve make. (A 2;1) daddy just make goene aandoen. (A 2;1) shoes on-do

As with the examples dreht (turns-3Sg) vs. drehen (make-turn-Inf) and hat (has3Sg) vs. haben (want-to-have-Inf) in L1 German, there are also examples in L1 Dutch, showing that the same verb may occur in both head-initial and head-final position. As in German, these verb forms are used with a difference in meaning which correlates with their position. Examples were given in (8). In (8) doet (works-3Sg) refers to a state, while doen (do-Inf) refers to a causal action; zit (is-3Sg) is used as a state verb, while zitten (sit-down-Inf) refers to an agentive motion; p(r)ikt (pricks-3Sg) refers to an experience, while prikken (prick-Inf) refers to a causal action. Thus, in early child Dutch the morpho-syntactic opposition of verb forms correlates with a lexical distribution of different types of verb. Cases of overlap are in fact examples of polysemy, i.e. they are used semantically differently.

102 

 The initial state

A similar situation seems to occur with adult learners of L2 Dutch. However, with these learners, the distribution is less clear-cut. In L2 learners of Dutch, finite verb forms occur in initial position, while infinite verb forms may occur both in final and initial position. This phenomenon has been accounted for in terms of the Missing Surface Inflection Hypothesis (Prévost and White 2000: 103). It states that the infinitive as a representation of the default form may be used in initial position, too. Further differences between child L1 and adult L2 learners of Dutch are clearly due to the native language system of the L2 learner. Examples from the early L2 Dutch of native speakers of Turkish and Moroccan-Arabic were given in (2). More examples are given in (19). They provide a representative sample of the types of verb that these learners use most frequently. (19) L2 Dutch. The distribution of verb forms in head-initial (a) and head-final (b) position Osman/T a. dan ik zitten alles samen. (Os/T 1.6) then I sit all together die man gaat naar ander stad. (Os/T 1.6) that man goes to other town ik woon zwarte zee. (Os/T 1.8) I live black sea hij zit thuis. (Os/T 1.9) he sits home die man kopen een auto. (Os/T 1.9) that man buy a car jij blijft thuis. (Os/T 2.3) you stay home hij vindt leuk. (Os/T 2.5) he finds nice hij vindt leuk en … (Os/T 2.5) he finds nice and … dan hij vallen achter. (Os/T 2.9) then he fall behind

b. en dan ik ook slapen. (Os/T 1.2) and then I also go-sleep hier koekfabriek pakket maken. (Os/T 1.3) here cookiefactory parcel make die lamp maken. (Os/T 1.5) that lamp make ik moet huis kopen. (Os/T 1.6) I must house buy en dan ook krant pakken. (Os/T 1.6) and then too newspaper get die man kroket eten. (Os/T 1.6) that man croquette eat zoiets ik niet eten. (Os/T 1.8) such-thing I not eat jij moet maken. (Os/T 1.9) you must make …. dan kopen. (Os/T 2.5) …. then buy

The alternative  

 103

Mohammed/A a. hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault ik zit. (Mo/A 1.3) I sit ik werk vijf maand. (Mo/A 1.3) I work five months ik heb diploma maar … I have diploma but … ik woont in casablanca. (Mo/A 1.4) I lives in casablanca hij blijft met mij. (Mo/A 1.5) he stays with me hij loopt met straat. (Mo/A 1.5) he walks with street hij pakt een kroket. (Mo/A 1.6) he gets a croquet hij pakken mayonaise. (Mo/A 1.6) he get mayonnaise gisteren ik slaap bij mij oom. (Mo/A 2.1) yesterday I sleep at my uncle

b. hij arabisch lezen. (Mo/A 1.2) he arabic read een auto kopen. (Mo/A 1.2) a car buy hij maken, die fiets. (Mo/A 1.3) he make that bicycle … kan niet goed werken. (Mo/A 1.3) … can not good work hier pauze eten. (Mo/A 1.3) here break eat hij wil staan. (Mo/A 1.3) he wants stand-up morgen hij crossfiets spelen. (Mo/A 1.3) tomorrow he cross-bike play hij staan, voor de weg. (Mo/A 1.6) he go-stand in-front-of the road hij iets pakken. (Mo/A 1.6) he something get ja ik taart maken. (Mo/A 1.9) yes I cake make

In sum, the complementary distribution of types of verbs as illustrated in (18) and (19) is evidence against head movement. With the absence of head movement, it seems warranted to conclude that ‘finite’ and ‘infinite’ verb forms do not establish a morphological opposition. Hence, at the relevant stage, morphological finiteness is not a grammatical property of the learner grammar. Furthermore, the concept of head movement entails the notion of two positions. It is the default position which determines basic word order. For early L1 and L2 Dutch, as for L1 and L2 German, basic word order is usually assumed to be head-final, as is the case in the target language. However, given that the early Dutch learner system has no head movement yet, there is no reason why the assumption of a basic, head-final word order should hold. Absence of evidence is not evidence of absence. Given that there are two semantically distinct sets of verbs one could argue as follows. Verbs belonging to one semantic set have to move to a position where, as is the case in the input, they receive finite morphology. Verbs belonging to the other semantic set have to remain in a position where, as is the case in the input, they appear as infinitives.

104 

 The initial state

However, the problem with this idea is that it provides no explanation for why learners would discriminate between two semantic sets of verb in the first place. Furthermore, I will provide evidence for an alternative proposal. I will argue not only that both the child L1 and the adult L2 data can adequately be accounted for on the basis of their semantic properties, but I will also explain why this is so. In the following, I will argue that utterances at the initial stage are the expression of a predicate-argument structure. Placement of the predicate in either initial or final position is due to the particular type of predicate-argument structure that the predicate belongs to. Furthermore, I will claim that at the relevant stage modal predicates are used as lexical heads with an initial, specifier position for the agent. Action predicates may occur as the complement of this modal head in final position. State and change-of-state predicates are non-agentive. They cannot be used as the complement of a modal head. Hence, they occur in head position with an initial, specifier position for the theme. Both types of predicateargument structure are distributed as given in Table 5 and illustrated with the examples in (20) and (21).

semantics

position

state / change of state causal action / agentive motion

initial final

Table 5: Semantics and placement

(20) L1 Dutch. The distribution of predicate-argument structure state / change of state

action

poes komt niet. (J 1;11) kitty comes not hoortie daarin. (J 1;11) belongs-it therein J heeft au. (A 2;1) J has ow

ulle poesje toe. (J 1;10) want kitty to mama kanniet kusje. (J 1;11) mommy cannot kiss papa eve make. (A 2;1) daddy just make

(21) L2 Dutch. The distribution of predicate-argument structure state / change of state

action

die man gaat naar ander stad. (Os/T 1.6) that man goes to other town

ik moet huis kopen. (Os/T 1.6) I must house buy

The alternative  

hij vindt leuk. (Os/T 2.5) he finds nice Ik heb diploma maar … I have diploma but …

 105

zoiets ik niet eten. (Os/T 1.8) such-thing I not eat … kan niet goed werken. (Mo/A 1.3) … can not good work

In sum, it is the possibility or impossibility to occur with a modal head which explains the distributional difference between agentive predicates in final position and state or change-of-state predicates in initial position. Furthermore, the input is relevant for the position and the form of a predicate. Agentive predicates are used as the complement of a modal head in final position. As in the target language, predicates in this position are infinitives. State and change-of-state predicates do not occur as the complement of a modal head. They are expressed as verbal heads in initial position. As in the target language, predicates in this position are finite. Thus, at the relevant stage, placement and morphology of the predicate appear as they occur in the input. Assuming that at the initial stage inflectional morphology does not play a role, root infinitives as such do not constitute a grammatical category either. This explains why the complement of a modal head does not necessarily has to be an ‘infinitive’. This also holds for utterances for which I assume that the position of the modal head is empty. The utterances in (22a) vs. (22b) and in (23a) vs. (23b) show that all kinds of predicate – not just infinitives – can be used as the complement of a modal head. The utterances in (22) vs. (23) show that this may occur with or without the use of a modal element. The assumption of an empty modal head in (23) accounts for the modal meaning with which most of these predicates seem to be used. (22) L1 Dutch. Verbal (a) and non-verbal (b) predicates as the complement of a modal head a.

ulle pop hebbe. (J 1;10) want doll have poppie nee ape. (A 2;0) doll no go-sleep Mijnie kan losmake. (J 1;10) M can loose make mag-ikke ijssie hebbe? (A 2;1) may-I ice-cream have? Jaja mag dop opdoen. (A 2;0) J may cap on-do

b.

ulle poesje toe. (J 1;10) want kitty to Mijnie nee daahee. (J 1;9) M no there-to mama kanniet kusje. (J 1;11) mommy cannot kiss mag-ikke ijssie? (A 2;1) may-I ice-cream? hoeniet plak op. (A 2;0) must-not glue on

106 

 The initial state

kusse hebbe magwel. (J 1;11) pillow have may-indeed magwel pek hebbe. (J 1;11) may-indeed candy have moet inzitte. (J 1;11) must in-sit hoef aaie? (A 2;0) must pet? handigniet kusje geve (A 2;1) handy-not kiss give kanniet pinkusse pakke. (A 2;1) cannot jump-mattress get

koppie thee magwel. (J 1;11) cup of tea may-indeed mag-ikke buite toe? (A 2;0) may-I outside to? minne hoene uit. (J 1;7) want shoes off unne die ope. (J 1;7) want that open ulle uit. (J 1;9) want out kanniet bij. (A 2;1) cannot at

(23) L1 Dutch. Verbal (a) and non-verbal (b) predicates as the complement of an empty modal head a.

Cynthia koek happe. (J 1;10) C cake bite Mijnie zelf doen. (J 1;10) M self do Mijnie ook heppele. (J 1;10) M too help eve melluk pakke. (J 1;10) just milk take Jaja dop hebbe? (A 2;0) J cap have? poppie niet Jaja help. (A 2;0) doll not J help papa eve make. (A 2;1) daddy just make

b.

niet oppe tegels. (J 1;11) not on-the tiles pek op. (J 1;10) candy on Mijnie tiktak om. (J 1;10) M watch on Jaja dop op. (A 2;0) J cap on Erre huis toe. (A 2;0) E home to Mijnie die deruit. (A 1;9) M that-one it-off deke over. (A 1;11) pillow over

Thus, the examples in (22) and (23) show that instead of the verbal category ‘root infinitive’ the relevant generalization is any verbal or non-verbal element that can function as the predicate of a modal head. Furthermore, the categorization of verbal elements as morphologically finite makes sense only if these elements function in opposition to verbal elements that are non-finite. In early child language, there is no such opposition. Rather, the opposition is in terms of predicates not in terms of verbs. Hence, finiteness as a morphological category is not a grammatical property of the learner grammar at the relevant stage.

The alternative  

 107

At the relevant stage, the modal head is claimed to project a lexical structure with an agent in initial, specifier position and an agentive predicate as its complement in final position. State and change-of-state predicates do not occur with a modal element. Therefore, they are the head of a lexical projection with a theme in initial, specifier position. In both cases, basic word order is, thus, head-initial. In utterances with agentive predicates it is the modal predicate that functions as the head, while with non-agentive, i.e. state and change-of-state predicates the head is the predicate itself. Hence, the basic word order at the initial stage is head-initial, and therefore, a discussion about whether or not the distributional properties of verb forms may show evidence of head movement is vacuous. In sum, the distribution of verb forms at the initial stage has been accounted for from different theoretical points of view. Three variables have been claimed to play a role: morphology, placement and semantics. These variables may be correlated in three different ways. Figure 1 shows that each possible correlation has been (or is) claimed to apply to the grammatical system of the learner language: morphology and placement by Wexler et al., morphology and semantics by both Clahsen and Ingram and Thompson, semantics and placement by the present author.

head movement (Wexler et al., 1993; Wexler, 1998) morphology

placement

transitivity (Clahsen, 1986)

agentivity (Jordens, this volume)

modality (Ingram and Thompson, 1996) semantics Figure 1: Morphology, placement and semantics: the distribution of verb forms at the initial stage

The German data studied by Poeppel and Wexler are similar to the Dutch data that I have analysed. Why is it that Poeppel and Wexler come to the conclusion that children have access to what they refer to as Full Competence? Figure 1 provides a representation of the structure of the target grammar in German as given in Poeppel and Wexler.

108 

 The initial state

(a) V-finite CP kommt, geht, Spec fährt, fliegt, liegt, weiß, C findet, kippt um, will, kann, möchte, muss

(b) V-infinite putzen, holen, malen, essen, werfen, suchen; anziehen, drauftun, aufblasen, vorlesen, abtrocknen, sauber machen

C′ IP Spec

I′ VP

Spec

I V′

NP

V

Figure 2: Poeppel and Wexler: The structure of adult German grammar

As pointed out before, spontaneous L1 production data show that, at the initial stage, verb forms in Figure 2 occur either as in (a) V-finite or as in (b) V-infinite. According to Poeppel and Wexler the verb forms in (a) are evidence of a C position, while the verb forms in (b) are evidence of a V position. Given this account of the data, Poeppel and Wexler note that “the best model of the data is the standard analysis of adult German” (1993: 2). Hence, the conclusion is that nothing needs to be learned: the Full Competence Hypothesis. However, with this interpretation of the data there are two problems. First, there is an empirical problem in the sense that it makes the wrong predictions. That is, at the initial stage, on the basis of the structure in Figure 1, children should be able to produce utterances with auxiliary verbs and utterances with a topicalized element in SpecCP. However, at the relevant stage this does not occur. Children produce neither utterances with auxiliary verbs nor utterances which show evidence of topicalization. Secondly, there is a methodological problem in the sense that the hypothesis of Full Competence involves a way of reasoning which is typically circular. This is because the production data are analysed in terms of categories of the target language system, i.e. in terms of finite vs. infinite morphology. No wonder that the best account is the adult model. As argued before, at the initial stage of L1 acquisition, basic word order in Dutch is assumed to be head-initial. Other acquisition studies however, usually assume that basic word order is head-final. The reason for this is the word order as it occurs, for example, in ‘root infinitives’. Root infinitives are assumed to represent the underlying, default word order of the verb phrase. In normal discourse these root infinitives may occur as the answer (B) to a question (A) as in (24):

Summary  

(24)

A:

Wat ga je doen? what go you do?

B:

Een ijsje halen an ice-cream get.

 109

The root infinitive halen (get) is a transitive predicate that is subcategorized for an object. The position of the object een ijsje (an ice-cream) before the verb halen (get) shows that in root infinitives the complement occurs before the verb.²⁵ The same situation holds for structures with an object and a particle as in licht uit (licht out), jas aan (coat on), dop op (cap on). Here, it is the particle functioning as a transitive predicate that is subcategorized for an object. Complement-head structure in Dutch also appears in particle verbs like kapotmaken (kaput-make), dichtdoen (closed-do), thuiskomen (home-come), weggaan (away-go) and deverbal compounds such as schroevendraaier (screwdriver), schoenmaker (shoemaker), kurkentrekker (corkscrew) etc.²⁶ Thus, root infinitives are predicates that are subcategorized for their complement. They function not as single words, but together with the information on the linguistic categories that they subcategorize for. In other words, they are lexical entities like particle verbs and deverbal compounds. Now, fact is that lexical structure and utterance structure are different ball games. That is, complement-head structure that holds for the internal structure of lexical entities may not hold for the syntax of utterance structure. This explains why, at the initial stage of L1 and L2 Dutch, root infinitives as lexical entities with complement-head structure may function as the complement of a modal head that syntactically serves as the head of a head-complement structure.

4.4 Summary Neither the Full Competence approach as proposed by Poeppel and Wexler, nor the input approach of Ingram and Thompson, nor Clahsen’s morpho-semantic model provides an adequate account of the language system of L1 and L2 learners

25 In similar cases in English and French, predicates that are subcategorized for an object occur with head-complement structure. Hence, for example: What are you going to do? Having an ice cream. 26 In French, but not in Dutch and English, nominal compounds usually occur with headcomplement structure as in tire bouchon (kurketrekker, corkscrew), salle de bain (badkamer, bathroom), tarte aux pommes (appeltaart, apple cake), fonds d’urgence (noodfonds, emergency fund), prévisions météo (weersverwachting, weather forecast).

110 

 The initial state

of German and Dutch. This language system is neither innately given nor is it the reflection of the target input. On the basis of general principles of language learning and the adult input, the linguistic system of learner language is the result of a process of creative construction. As I argued above, at the initial stage this is a grammatical system which is based solely on lexical categories of predicateargument structure. Furthermore, I have argued that in L1 and L2 learners of Dutch this grammatical system is initially based on lexical projections with a head-initial word order. Variation in utterance structure can be accounted for in terms of an opposition between two types of predicate-argument structure: predicates referring to a state or a change of state occur with a theme as the external argument; predicates referring to a (causal) action or an agentive motion occur with the agent as the external argument. State and change-of-state verbs cannot occur with a modal head, agentive verbs may or may not. However, even if utterances with an agentive predicate have no modal head, the modal head position is still present. Elements in modal head position are used to express the volition, the ability, the permission or the obligation of an agent to perform an action. In the default case, i.e. if the position of the modal head is empty, the speaker may either leave the modal meaning to be inferred from the context, or he uses this empty position simply to express the default pragmatic function of assertion. In the absence of a modal head position, utterances with a state or a change-of-state predicate can only be used to express the pragmatic function of assertion by default. In the following, I will investigate the structure of early learner varieties of Dutch. The assumption is that the initial learner grammar is the expression of predicate-argument structure. The functional projection FP is claimed to be absent and, therefore, the syntactic properties of finiteness or verb-second are not established yet. Absence of these syntactic properties has its repercussions for the present learner system. This particularly holds for the availability of the linguistic means of information structuring that are linked to the availability of the functional category system. These linguistic means are used to link the content of an utterance to the contextual situation that it applies to. Platzack (1996), therefore, seems right in stating: “I consider finiteness to express anchoring of the clause in time and space, hence lacking finiteness, the nonfinite sentences are not supposed to be as closely bound to the speech act as the finite ones. Speculating about the deeper reasons for the abundant use of nonfinite sentences in early child language, and noticing that the option seems to fade away around the age of three, it is tempting to relate it to the egocentric standard observed by Piaget, which is replaced by a contextsensitive standard at age four (see e.g. Piaget [1983])” (388).

5 The lexical stage 5.1 Basic languages, research questions Pit Corder was one of the first to study learner utterances at the beginning of the language acquisition process both in children learning their mother tongue and in adults learning a second language. He discovered that in both situations the structure of these learner utterances was surprisingly similar. Comparing these utterances with the target language model, he noted that early in the acquisition process some linguistic features are systematically missing, grammatical features in particular. Hence, he described the learner grammar with the following characteristics: “a simple or virtually non-existent morphological system, a more-or-less fixed word order, a simple personal pronoun system, a small number of grammatical function words and grammatical categories, little or no use of the copula, absence of the article system (less often the absence of deictic words). The semantic functions (…) of tense and aspect are typically performed, when at all by lexical means, e.g. adverbs (…). The basic syntactic relations are expressed by word order” (Corder 1978: 2).

The early language system of the language learner, i.e. of language learners in general, can therefore be best described as a kind of basic language. Perdue has stressed the point that basic languages, or Basic Varieties as they are called in Klein and Perdue (1997), are language systems in their own right. That is, they are: “self-standing, independent systems showing their own regularities (….) on the structural, semantic and informational level” (Perdue 2006: 859).

Wolfgang Klein has taken this line of thought even a step further in arguing that: “Learner varieties are a genuine manifestation of the human language faculty … In fact, I believe that learner varieties are the core manifestation of the human language faculty and real languages … are the borderline cases” (Klein 1997: 5).

Basic languages are thus ‘simple systems’ that are used not only by children learning their mother tongue, but also by learners of a second language in situations of spontaneous language acquisition. In the following, I will show that in Dutch this simple language system in fact consists of a limited set of lexical structures. I will show that these lexical structures are characteristic of the spontaneous language

112 

 The lexical stage

use of both children learning Dutch as their mother tongue and adults learning Dutch as a second language. A small sample of utterances from two children learning Dutch as their mother tongue and from two adults learning Dutch as a second language is given in Table 1. These utterances were produced spontaneously and are characteristic of the early stages of language acquisition. They are representative of the types of utterance of a basic language system.

child L1 Dutch

adult L2 Dutch

ik chocomel hebbe. I chocolate milk get jij opemake. you open-made ikke glijbaan maakt. I slide made kannie bal pakke. can-not ball get popje valt bijna. doll falls nearly gaap [= schaap] kom niet. sheep comes not Jaja vindt vies deze. J finds dirty this kanniet zellef. can-not self goene aan. shoes on pop da in. doll there in hoefniet plak op. must-not glue on

ik alles doen. I everything do jij mag niet drinken. you may-not drink ik toen die auto gekocht. I then that car bought die kan niet praten nederlands. that can-not talk dutch ik woont in casablanca. I lives in casablanca die man gaat naar ander stad. that man goes to other city hij vindt niet leuk. he finds not nice dan ik moet zelf huis. then I must self home ik heb fiets kwijt. I have bicycle gone dan zij kwaad op mijn. then she mad at mine moet jij koptelefoon uit. must you headset off

Table 1: Lexical structures in L1 and L2 Dutch

A first analysis of the data in Table 1 shows that the relevant basic languages are mainly lexical. Functional features systematically do not occur. Hence, there is virtually no inflectional morphology, there are no function words and there is no variation in word order. The functional categories which play a role in target Dutch are summarized in Table 2.

Basic languages, research questions  

 113

Functional systems

Categories

Examples

morphology

finiteness

kom (come-1SgPres), komt (come-2,3SgPres), komen (come-1,2,3PlPres) vs. komen (Inf). maken (make-Progr) vs. is gemaakt (is madePerf). schoen (shoe-Sg) vs. schoenen (shoes-Pl); auto vs. auto’s (cars-Pl). hapt (bite-3SgPres) vs. hapte (bit-3SgPast).

aspect number tense function words

auxiliary verbs determiners anaphoric pronouns prepositions

word order variation

heb (have), heeft (has); ben (am), is (is). de (the), het (the) vs. een (a, an). hij (he), zij (she), hem (him), haar (her), het (it), er (there). in (in), op (on), aan (on), uit (out), met (with), van (of).

verb-second

We gaan straks een glaasje drinken (we go later a glass drink) vs. Straks gaan we een glaasje drinken (later go we a glass drink). topicalization Dat geloof ik niet (that believe I not) vs. Ik geloof dat niet (I believe that not). questioning Wie heeft dat gedaan? (who has that done?) vs. Heeft hij dat gedaan? (has he that done?). scrambling Hij heeft een antwoord nooit gekregen (he has an answer never received) vs. Hij heeft nooit een antwoord gekregen (he has never an answer received). main- vs. sub. clause We gaan straks een glaasje drinken (we go later a glass drink) vs. Als we straks een glaasje gaan drinken (when we later a glass go drink).

Table 2: Functional categories

The functional categories listed in Table 2 show that the functional category system in Dutch consists of the morphological categories of finiteness, aspect, number and tense marking, of the function word categories of auxiliary verbs, determiners, anaphoric pronouns²⁷ and prepositions and, finally, of word order variation as established in verb-second (or: head movement), topicalization, whand yes/no-questions, scrambling and clause type.

27 At the relevant stage, children and adults differ with respect to the use of pronouns. While children may only use deictic pronouns such as ik, ikke (I) and jij (you), adult second language learners use both deictic and anaphoric pronouns. The reason is that children are more closely bound to the here and now than adults.

114 

 The lexical stage

As shown in Table 1, learner utterances at the initial stage are typically lexical. They are used to refer to situations with persons and objects each playing a particular role. Grammatically, learner utterances are the expression of a predicate-argument structure in which constituents with verbs, nouns, adjectives and adverbs occur with different semantic functions. A comparison of the learner utterances at the lexical stage as they are shown in Table 1 with the functional categories of the target language listed in Table 2 shows that, at the initial stage, not all morphological categories are missing. At the relevant stage some morphological features such as tense marking do not occur, while others such as finiteness, number and aspect marking do. As pointed out in Chapter 2, it should be noted however, that number and aspect marking are subject to processes of lexicalization. Therefore, nouns that are marked for number and verbs that are marked for aspect may initially be used as unanalysed wholes. The morphological properties of finiteness marking are initially also unanalysed. They are the result of learning by rote. Morphological finiteness is, thus, initially not a productive feature of the learner system, either.²⁸ In sum, at the initial stages of language development, child L1 and adult L2 learners of Dutch appear to create a simple, basic language variety which is essentially the same across individuals. The utterances in Table 1 are representative of the underlying linguistic knowledge system, which consists of a coherent set of principles that are derived from the target language input. Basic languages are language varieties with lexical structures and no functional elements. Nevertheless, they serve the basic communicative needs that language learners may have. In the following, I will investigate the acquisition of Dutch focussing on three questions. First, what are the principles of a basic language system? Second, how do learners progress from their basic learner variety to a more advanced system? Finally, what does insight into the acquisition process tell us about the faculty of human language acquisition.

28 With respect to the use of the term ‘finite’, a distinction is made between ‘finite forms’ (morphological finiteness) and ‘finiteness’ (semantic finiteness) as a concept of information structure (Klein 1998). My claim is that at the lexical stage of the acquisition process finite forms do appear. However, they occur as the reflection of the input and not as the representation of a functional category. Finiteness as a functional category is the result of a process of language acquisition.

The analysis of early learner data 

 115

5.2 The analysis of early learner data In the acquisition of Dutch, there is a prominent role for the use of infinitives, particularly at the early stages of the language learning process. Utterances in which the verbal part of the predicate appears as an infinitive are in the literature known as ‘root infinitives’. At the initial stage, these utterances occur most frequently. Examples from child L1 Dutch and adult L2 Dutch are given in type A1. Type A1. Infinite verb form (‘root infinitive’) child L1 Dutch

adult L2 Dutch²⁹

ik chocomel hebbe. (J 2;10) I chocolate-milk get Mijnie zelf doen. (J 1;11) M self do jíj gijbaan make. (A 2;1) you slide make deze slagroom ete. (J 1;10) this-one cream eat gaag boekje leze. (A 2;0) like booklet read poes bal pakke. (J 1;11) kitty ball get onne deke slape. (J 1;11) under blanket sleep papa nieuwe kope. (A 2;1) daddy new-one buy dit losmake. (J 1;11) this loose-make Mijnie uitklimme. (J 1;10) M out-climb

veel geld hebben. (Os/T 1.7) much money get ik alles doen. (Mo/A 2.2) I everything do hier koekfabriek he pakket maken. (Os/T 1.3) here cookie-factory right package make die man kroket eten. (Os/T 1.6) that man croquette eat eerste jaar arabisch lezen. (Mo/A 1.2) first year arabic read en dan ook krant pakken. (Os/T 1.6) and then too newspaper get ikke deze avond hier slapen. (Os/T 1.3) I this night here sleep als jij man auto kopen. (Mo/A 2.2) if you man car buy en dan krant helemaal dichtmaken. (Os/T 2.3) and then newspaper fully closed-make misschien jij weglopen huis. (Os/T 2.3) maybe you away-run home

As noted in Section 2.6, the data of the present study originate from investigations on children learning Dutch as their native language and adults learning Dutch in an untutored second language learning environment. Both the L1 and

29 As shown in Jordens and Dimroth (2006: 182f.), the same data base has similar examples also from other adult L2 learners of Dutch. For example, dan auto bijna vallen (then car nearly fall-down. Mahmut/T 2.9), hij even zoeken (he just search. Abdullah/T 2.9).

116 

 The lexical stage

L2 data come from longitudinal studies of utterances produced spontaneously. The L1 data originate from a corpus of diary data collected from two children of the present author: Jasmijn (J) and Andrea (A). In the examples below a reference such as, for example, ‘J 1;9’ means that the utterance occurred when Jasmijn was 1 year and 9 months of age. The L2 data stem from the European project “Second Language Acquisition by Adult Immigrants”, funded by the European Science Foundation. In this project, data collection was organized in three cycles of 10 months each, which means that similar types of elicitation techniques (e.g. film retellings and picture descriptions) were repeated three times during the process of data collection (see Perdue 1993). In the present study, I will investigate data from two L2 learners: Osman who has Turkish (T) as his native language and Mohammed who speaks Moroccan-Arabic (A). In the examples ‘Os/T 1.2’ means that the utterance by Osman whose L1 is Turkish, occurred at month 2 of cycle 1.³⁰ At the initial stage, as pointed out before, ‘root infinitives’ of type A1 are produced most frequently. As is evident in the examples given, the infinitive always appears in final position. Simultaneously with the occurrence of root infinitives there is also a type of utterance with a finite verb form. Although this type of utterance is produced less frequently, it appears in both the child and the adult data. Examples from child L1 Dutch and adult L2 Dutch are given in type B1. Type B1. Finite verb form child L1 Dutch

adult L2 Dutch³¹

poesje, heb jij? (J 1;10) kitty, have you? da zit mama. (J 1;11)

ik heb veel vriendin. (Mo/A 1.3) I have many girl-friend hij zit thuis. (Os/T 1.9)

30 Does this mean that N equals 2? That is correct. A language system is a knowledge system that is shared by a particular set of language users. Therefore, if I want to know if a particular structure occurs in such a system, one informant would in fact suffice. If this informant judges an utterance as ‘correct’ or ‘incorrect’, other users of the same language system should be expected to come up with the same judgement. This is as is the case with a bowl with marbles. If I know that in a bowl with marbles all the marbles are of the same colour, one trial is enough to decide what colour all the other marbles have. 31 As shown in Jordens and Dimroth (2006: 182f.), the same data base has similar examples also from other adult L2 learners of Dutch. For example, die is zuster (that-one is sister. Mahmut/T 1.5), ik heb niet geld (I have not money. Mahmut/T 3.5), ik hoefniet ziektewet (I mustnot health-insurance. Mahmut/T 2.7), eenduizend kanwel sparen per jaar (one-thousand canindeed save per year. Mahmut/T 2.2), dan magniet rijden (then may-not drive. Abdullah/T 2.7), vrouw moet keuken (wife must kitchen. Mahmut/T 2.2).

The analysis of early learner data 

there sits mommy da poes blijf hier. (J 1;11) the kitty stay here da woont Gaga. (A 2;1) there lives G ís niet pot. (J 2;0) is not kaput gote paard is hier. (A 2;1) big horse is here dit is dop. (J 1;11) this is cap is hanne nou? (A 2;1) is hands now? Jaja heef koud. (A 2;1) J has cold Jaja vindt vies deze. (A 2;1) J finds awful this popje valt bijna. (J 1;10) doll falls nearly uil zo komt. (J 1;10) owl so comes popje gáát niet mee. (A 2;1) doll goes not with deze kanniet ope. (A 2;1) this can-not open da kan ook. (A 2;1) that can too poes magniet oppe aanrecht. (J 2;1) kitty may-not on-the sink kijk, zo moettie. (J 1;11) look, so must-it deze moet hier. (A 2;1) this-one must here Ja dees ook slaapt. (J 1;10) yes, this too sleeps kusse hebbe, magwel. (J 1;10) pillow get, may-indeed deze magwel. (J 1;10) this-one may-indeed

 117

he sits at-home hij blijft met mij. (Mo/A 1.5) he stays with me ik woont in casablanca. (Mo/A 1.4) I lives in casablanca nou is beetje moeilijk, hè? (Os/T 1.8) now is little difficult, ok? [krant] is in de zak. (Mo/A 1.3) newspaper is in the bag ja ik niet weten wat is ja dat. (Mo/A 2.6) yes I not know what is yes that mijn werk is vlak bij. (Mo/A 2.8) my work is close by ik heb groot probleem. (Os/T 1.6) I have big problem hij vindt leuk en dan kopen. (Os/T 2.5) he finds nice and then buy dan hij vallen achter. (Os/T 2.9) then he fall behind ik komt hier. (Os/T 1.6) I comes here die man gaat naar ander stad. (Os/T 1.6) that man goes to other town kanniet weg. (Os/T 1.3) can-not all-gone die kan ook. (Os/T 2.6) that can too pilske magwel hè, een twee per dag. (Mo/A 2.8) beer may-indeed okay, one two per day moet terug. (Os/T 2.2) must back dat is tweede straat, moet linksaf. (Os/T 1.9) that is second street, must left gisteren he jij slaapt bij un vriend. (Mo/A 2.1) yesterday right you sleep with a friend sigaretten roken, magniet. (Os/T 1.8) cigarettes smoke, may-not die magniet, daar veertig kilometer rijen. (Os/T 1.9) that may-not, there forty kilometers drive

118 

 The lexical stage

As shown in type B1, finite verb forms occur systematically in second and, in only a few cases, in third constituent position. Only second language learners use infinitive forms in these positions as, for example, in dan hij vallen achter (then he fall behind).³² The similarity between children and adults with respect to the distribution of infinitives and finite verb forms is striking. Particular types of verb systematically occur as infinitives in final position, while other types of verb usually occur as finite verb forms in head-initial, that is in second (or third) position. In sum, utterances of type A1 and type B1 are characteristic for learners of Dutch. They occur at the initial stage of the spontaneous process of language acquisition. While verb forms may occur in either of two positions, there is a correlation between position and form: infinite verb forms occur in final position; finite forms occur in head-initial position. Given these observations, it seems obvious why utterance structure at the lexical stage is usually analysed in terms of a complementary distribution that is based on the morphological opposition finite vs. infinite. It also explains why notions such as ‘root infinitives’, ‘finiteness’ as a morphological property, ‘agreement’ and ‘movement’ play a central role in describing the shape of the learner grammar at the initial stage (see Clahsen 1986, Poeppel and Wexler 1993, and Ingram and Thompson 1996 as discussed in Chapter 4). However, the conclusion of a complementary distribution based on morphology becomes less obvious, if we consider the fact that this distribution is actually based not on morphology but on semantics. That is, verb forms that occur in one position do not occur in the other and vice versa. This observation runs contrary to the overlap hypothesis put forward in, for example, De Haan (1987) and Poeppel and Wexler (1993). Furthermore, although nothing is wrong with the data of type A1 and type B1, they are not representative at all. They are a selection from the perspective of the target language system, more specifically from the perspective of the verb-second phenomenon. For a complete picture of the language capacity of learners at the relevant stage, there are other data that should be taken into account as well. Fact is that, simultaneously with type-A1 utterances, there is also a variant shown in the utterances of type A2.

32 Infinite forms in second position only occur in second language learners. A possible explanation for this is given in the Missing Surface Inflection Hypothesis (Prévost and White 1999, 2000). It will be discussed in Chapter 7.

The analysis of early learner data 

 119

Type A2. Modal / Aspectual element + infinitive child L1 Dutch³³

adult L2 Dutch

ulle ijsje hebbe. (J 1;8) want ice-cream get ulle glijbaan make. (J 2;0) want slide make nee bad zitte. (A 2;1) no bath sit ik wil mellek pakke. (J 1;10) I want milk get kanwel papa zitte. (A 2;1) can-indeed daddy sit-down

ik wil gaan naar moskee. (Os/T 1.8) I want go to mosque ik wil uh honderd dirham geef uh … (Mo/A 1.3) I want em hundred dirham give em .. als jij jullie niet wilt geef een huis. (Mo/A 1.6) if you not want give a house hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault en dan kanniet lezen, kan begrijpen niet. (Os/T 1.7) and then can-not read can understand not die ander kanniet lezen op school. (Mo/A 1.4) that other-one can-not read in school die magniet daar veertig kilometer rijen. (Os/T 1.9) that-one may-not there forty kilometers drive dan mag-ik even vragen? (Os/T 2.3) then may-I just ask? en dan wij moet zo doen. (Os/T 2.2) and then we must so do die is nou werken. (Os/T 1.3) that-one is now work hij gaat snel lopen. (Mo/A 1.6) he goes fast run hij komt snel lopen direct. (Mo/A 1.6) he comes fast run immediately ikke komt hier werken. (Os/T 1.3) I comes here work

kannie bal pakke. (J 1;9) can-not ball get Jaja mag dop opdoen. (A 2;0) J may cap on-do mag-ikke ijssie hebbe? (A 2;1) may-I ice-cream get? Peter moet zitte. (J 1;11) P must sit doettie alles opete. (J 1;11) does-it everything up-eat gaatie [sl]ape. (A 2;1) goes-he sleep komma hare kamme. (A 2;1) come hair comb doemaa dit drinke. (J 1;11) do this drink

33 As shown in Jolink (2009, 2010), child L1 learners also produce examples with is instead of a modal verb. For example, die eisje is tieke (the girl is draw), paadje is alle biele opete (horsie is all wheels up-eat), toen is e vogel da vliege (then is a bird there fly). Similar examples occur in adult L2 Dutch. For example, daar is mensen niks doen (there is people nothing do. Abdullah/T 2.7).

120 

 The lexical stage

Utterances as in type A2 occur with a modal or an aspectual predicate in head-initial position. The subject in first position refers to the speaker or another person. It is not necessary that it is explicitly expressed. Utterances as in type A2 are used to express that the subject wants, can, may, must or is going to do something or is currently involved in doing something. This is represented with the modal elements wil (want), kan (can), mag (may) or moet (must), while aspectual elements such as gaat (goes), komt (comes) and doet (does), indicate that the subject is actively involved. The opposite meaning is expressed with unanalysed negative modal elements such as wilniet (want-not), kanniet (can-not), magniet (may-not) and hoefniet (must-not). What the subject actually does or does not, is expressed with the complement of the modal or aspectual head. This complement functions as a lexical entity, i.e. its structure is as it is stored lexically. This explains the infinite form of structures such as ijsje hebbe (ice-cream get), bal pakke (ball get), kopen een renault (buy a renault), niks anders doen (nothing else do), etc. Utterances of type A1 (the ‘root infinitives’) are, regardless of the frequency with which they occur, a special case of type A2. They are a special case in the sense that in root infinitives of type A1, the position of the modal head is empty. Nevertheless, root infinitives are a structural feature of the present language system, because in a normal speech situation the meaning of a modal element in this position can easily be inferred from the relevant context. The examples in Table 3 are evidence of the variable use of utterances of type A1 and type A2. They

Jasmijn (2;0) a. Vinf glijbaan, aanmake. slide, on-make glijbaan vastmake. slide up-tie Herrie vinde. H. find

b. Mod/Aux + Vinf doemaar aanmake. do on-make ulle glijbaan make. want slide make ik kanniet Herrie vinde. I can-not H. find

Andrea (2;3) a. Vinf mellek indoen. milk in-do papa jij pakke? daddy you get? papa eve diepe kuil make. daddy just deep hole make

b. Mod/Aux + Vinf doemaa mellek indoen. do milk in-do opa magniet mij pakke. grandpa may-not me get doemaa diepe kuil make. do deep hole make

Table 3: Utterances with Vinf and Mod + Vinf in L1 Dutch

The analysis of early learner data 

 121

show that Jasmijn (2;0) and Andrea (2;3) vary between the use of infinitives (Vnf) and modals + infinitive (Mod + Vinf) even with the same lexical verbs. Comparable with the data of type A2 are those of type A3. In type A3, however, the complement is a non-verbal predicate. Type A3. Modal + non-verbal predicate child L1 Dutch

adult L2 Dutch

kannie zellef. (J 1;11) can-not myself magniet oppe dak. (J 1;11) may-not on-the roof mama kanniet kusje. (J 1;11) mommy can-not kiss kanniet beker. (J 2;0) can-not cup mag-ikke ook gijbaan? (A 2;0) may-I too slide? kanniet goed niet. (J 1;10) can-not good not minne [= ik wil] hoene uit. (J 1;7) want shoes off moettie hier? (A 2;0) must-it here? poppie kan. (A 2;1) doll can [sth]

moet zelf, hè? (Os/T 3.2) must myself, right? magniet binnen. (Os/T 2.6) may-not inside dan ik kanwel bromfiets of motor. (Os/T 2.5) then I can-indeed moped or motorcycle dan mag-jij bord. (Os/T 2.6) then may-you plate ik magwel alles. (Os/T 2.6) I may-indeed everything moet tempo, hè? (Os/T 2.8) must fast, right? moet-jij koptelefoon uit. (Os/T 2.8) must-you headphone off moet linksaf. (Os/T 1.9) must left-turn jij magwel. (Os/T 2.6)³⁴ you may-indeed [sth]

In research on spontaneous processes of language learning, utterances such as those of type A3 are usually not taken into consideration. However, the only difference with the examples of type A2 is the fact that a complement such as zelf (myself) or oppe dak (on-the roof) is not a verbal constituent. Semantically, it has the same function as in cases in which it would occur with an infinitive. Thus, kannie zellef (J 1;11) means ik kan ‘t niet zelf doen (I can it not myself do) and magniet oppe dak (J 1;11) means hij mag niet op het dak klimmen (he may not on the roof climb). These non-verbal complements are used with the function

34 At the relevant stage of acquisition, predicate forms such as mag-ikke (may-I), minne (I-want), moettie (must-it), mag-jij (may-you) and moet-jij (must-you) occur only as unanalysed wholes.

122 

 The lexical stage

of a predicate. As a non-verbal predicate it is non-finite, as is the case with the infinitive. The utterances of type A1, A2 and A3 are all variants of the same type of utterance structure. They occur with a modal element in head-initial position or with a structural position available for it.³⁵ In this respect, type-A utterances differ from utterances of type B. The difference between type A and type B is, thus, not a question of morphology – infinite vs. finite – but of the presence or absence of a structural modal head position. Typical child utterances are utterances with in (in), op (on), aan (on), uit (off), om (on). Initially, they occur as the predicate of one-word utterances, later also as the predicate of two- and three-word utterances. Examples of two-word utterances are toti in (pencil in. A 1;8), buggy in (buggy in. J 1;8), dop op (cap on. A 1;8), deze, paartie op (this, horse on. A 1;11), jas aan (coat on. A 1;10), [s]tekker uit (plug out. J 1;11) en tiktak om (watch on. J 1;7). Predicates like in, op, aan, uit, om are called particles. Utterances with particle predicates can also be categorized as variants of either type A or type B. In the following examples, the elements in brackets [ ] are added on the basis of contextual information. Type A4. Particle

variant: Particle + infinitive

Mijnie in, tiktak. (J 1;8) M in, watch Mijnie bril op. (J 1;8) M glasses on dop op. (A 1;8) cap on goene aan. (A 1;11) shoes on zak uit. (A 1;11) bag off tittat om. (J 1;7) watch on

da [melk] ingenke. (J 1;10) there milk in-pour [blaadje] oplegge. (J 1;10) piece-of-paper on-put dop opdoen. (A 1;11) cap on-do goene aandoen. (A2;0) shoes on-do dit uithale. (J 1;11) this off-get [tiktak] omdoen. (J 1;11) watch on-do

variant: Modal + particle

unne pleister op. (J 1;10) want plaster on hoeniet plak op. (A 1;11) must-not glue on

minne hoene uit. (J 1;17) want shoes off minne tittat om. (J 1;7) want watch on

35 This is a problem for Ingram and Thompson. Their Modal Hypothesis claims that it is the morphology which serves to carry modality as part of the lexical meaning of the infinitives: “(…) the claim is that the infinitives are semantically associated with modality as part of their lexical information” (1996: 102).

The analysis of early learner data 

 123

Utterances of type A4 with in, op, aan, uit, om are variants of type A1 and A2. Evidence is the fact that the particles of type A4 are used with the same function as the infinitives of type A1. This is shown in variants of type A4 with particle verbs such as ingenke (in-pour), oplegge (on-put), aandoen (on-do), uithale (off-get), omdoen (on-do). See, for example, the child utterances goene aan (shoes-on) next to goene aandoen (shoes on-do) and tittat om (watch on) next to [tiktak] omdoen (watch on-do). The claim that particle utterances of type A4 are in fact variants of type A1 and A2 is, furthermore, confirmed by the fact that, as with type A2 and A3, these particles may also occur with a modal element. Examples are unne pleister op (want plaster on), hoeniet plak op (must-not glue on), minne hoene uit (want shoes off), minne tittat om (want watch on). The particles in (in), op (on), uit (off) may also occur as predicates in utterances of type B4. Type B4. Particle pop da in. (J 1;10) doll there in fiets in. (J 1;8) bicycle in Mijnie buggy in. (J 1;8) M buggy in mauw op [plaatje]. (A 1;9) kitty on picture Jaja paartie op. (A 1;9) J horsie on die deruit. (A 1;9) that it-out [popje] daaruit. (J 1;11) doll there-out

variant: Finite form + particle [popje] sittie da-in. (J 1;10) doll sit-it there-in kom es inne huisje. (J 2;0) come in-the house passie (= pastie) in. (A 2;1) fits-it in [paard] zittie oppe dak. (J 1;11) horse sits-it on-the roof

kommie da-uit. (A 2;0) comes-it it-out da valtie uit. (J 1;7) there falls-it off

Utterances of type B4 are variants of type B1. Evidence is the fact that the type-B4 particles in, op, uit are used with the same function as particles occurring in utterances with a finite verb form as, for example, in zit in (sits in), kom in (come in), past in (fits in), zit op (sits on), komt uit (comes out), valt uit (falls off). Examples are child utterances such as pop da in (doll there in) next to [popje] sittie da in (doll sits-it there in), [boekje] op [schoot] (booklet on lap) next to [paard] zittie oppe dak (horse sits-it on the roof) and die deruit (that it-out) next to kommie da uit (comes-it it-out). The claim that particle utterances of type B4 are in fact

124 

 The lexical stage

utterances of type B1 is confirmed by the fact that, as with utterances of type B1, variants with a modal element do not occur. In sum, some utterances with particle predicates are adequately analysed in terms of variants of type A, while other particle utterances are adequately analysed as variants of type B. However, on what grounds exactly might this be done? Neither the position nor the morphology of the particle provides a clue. Particles always occur in final position without verbal morphology. The only reason for the relevant distinction is the fact that particle utterances of type A4 may occur with a modal element, while those of type B4 may not. To conclude, the language system of beginning language learners is a lexical system. At the lexical stage of language learning the presence or absence of a structural position for a modal head determines the function and the position of the lexical predicate. The form of the lexical predicate is determined by the form in which it occurs the input. Therefore, a lexical predicate with the function of the complement as in type A occurs in final position and appears as an infinitive. A lexical predicate with the function of the head as in type B occurs in initial position and appears as morphologically finite. Finally, comparing utterances of type A with utterances of type B, it seems that some predicates typically occur in final position, while others preferably occur in initial position. In the following, I will pursue the question why this might be the case.

5.3 Utterance structure at the lexical stage 5.3.1 Lexical projections At the initial stage, the utterances of type A and type B differ not only as far as the structural position of the lexical predicate is concerned. They also differ with respect to the lexical-syntactic structure of the predication as a whole. Figure 1 is the representation of the opposition between types of predication that are agentive (type A) vs. non-agentive (type B) . In Figure 1A, the presence of a modal head or an element with an aspectual default function expresses the fact that there is an agent serving as the controller of an action, i.e. either a (causal) action or an agentive motion as, for example, in dop opdoen (cap on-do), bad zitte (bath sit-down), tiktak om (watch on-do). The modal elements wil (want), kan (can), mag (may), moet (must) indicate that this agent has the willingness, the ability, the permission or the obligation to perform the action. The negative modal equivalents wilniet (want-not), kanniet (can-not), magniet (may-not) and hoefniet (must-not) are, of course, used to express the opposite.

Utterance structure at the lexical stage  

A: agentive

 125

B: non-agentive

Ctrl″

Pred″

agent Ctrl′ (carries out)

theme Pred′ (occurs, undergoes)

Ctrl complement | action (causation, motion): e.g. dop opdoen (cap on-do), tiktak om (watch on-do), bad zitte (bath sit-down).

Pred (complement) | state: e.g. zit (sits), lust (likes), is op (is it-on), deze moet hier (this-one must here). | change of state: e.g. valt (falls), komt (comes).

modal: e.g. wil (want), kan (can), mag (may), moet (must). | aspectual: e.g. gaat (goes), komt (comes). | default: e.g. doet (does), 0 = to be inferred. Figure 1: The lexical-syntactic structure of type A and type B

At the relevant stage, modal elements in head position are in fact lexicalizations of a variation in control. This lexical use of modals is shown in (1). (1) L1 Dutch. Modality as the expression of control Mijnie kannie. (J 1;10) M can-not [do] kannie zellef. (J 1;11) can-not [do] myself ik kanwel dat. (J 1;10) I can-indeed [do] that mag-ikke buite toe? (A 2;1) may-I outside [go] to? hoefnie meer. (J 1;11) [I] must-not [do] more

126 

 The lexical stage

unnie niet. (A 2;0) I want-not [do] this wil naar poesje toe. (J 1;11) [I] want kitty [go] to In (1), the complement of the modal head which is used to refer to the action that is under control, is left more or less implicit. In the English equivalent the relevant action is represented with the verbal proform do (causal action) or go (agentive motion). In Figure 1B, as opposed to Figure 1A, there is no structural position for a modal element with the meaning of ‘control’. Utterances are used to express that a person or an object occurs in a state as, for example, in zit (sits), lust (likes), is op (is it-on), deze moet hier (this-one must here)³⁶ or undergoes a change of state as, for example, in valt (falls), komt (comes). The entity that a state or change of state applies to is the theme. Note that in Figure 1B modal elements such as kan (can), mag (may) or moet (must) do not express the meaning of ‘control’. Rather, they refer to a situation that is represented as a possibility, an optionality or a necessity.³⁷ In situations of spontaneous language production, learners do not necessarily have to express every constituent of an utterance. Nevertheless, given the underlying structure of a predication, both the position and function of its constituents is available. The so-called ‘root-infinitives’ are typical examples in

36 Past-participle forms such as valle (fallen. J 1;10), afpakt (away-taken. J 1;10), krege (got. J1;10), had (had. A 2;0), potmaakt (kaput-made. A 2;0), kamd ([has] combed. A 2;0) are typically used to express a result state. At the relevant stage, there is no grammatical opposition between infinitival predicates expressing the intention of an agent to carry out a particular action and past-participle predicates expressing the result state of an action. Hence, infinitival and past participle forms are frequently used in lexical oppositions such as hebben (get) vs. gekregen (got), zoeken (look for) vs. gevonden (found), weglopen (run away) vs. verstopt (hidden), maken (make) vs. kapotgemaakt (kaput-made), eten (eat) vs. opgegeten (eaten up), glijden (slide) vs. gevallen (fallen). See Jordens (1990: 1415ff.) 37 In the target language, state and change-of-state predicates are commonly used with modal verbs. Possible utterances in the target language are de boom kan omvallen (the tree can down-fall), je kunt hier wel zitten (you can here indeed sit), dit voorstel mag in de prullenbak (this proposal may in the waste-paper basket). These modal verbs are not the head of a lexical projection as in Figure 1A. They are neither used to express control, nor do they refer to a state or a change of state of the subject. They are the head of a functional projection and, therefore, they function as auxiliary verbs. At the lexical stage, functional projections do not occur. Hence, the utterances mentioned before should not occur, either. This is precisely what can be observed.

Utterance structure at the lexical stage  

 127

case. Root infinitives are utterances in which the position of the element used to express control is left empty. The structures in Figure 1A and 1B serve as representations of the learner system in terms of two types of predication. Notions such as ‘action’, ‘state’ and ‘change of state’ refer to a particular type of situation. Notions such as ‘agent’ and ‘theme’ refer to the role of individuals and objects therein. Utterances are used to express both the content of a predication and the pragmatic function with which they are used. Thus, in agentive utterances with the structure as in Figure 1A, modal predicates are used not only to express control, i.e. the willingness, the ability, the permission or the obligation on the part of an agent. They are also used to express the pragmatic function of assertion. This becomes evident in the form in which these modal elements may appear. Examples are given in Table 4.

volition ability permission obligation assertion

affirmative

negative

ulle, mag-ikke [I-]want, may-I kan(wel) can-indeed mag(wel) may-indeed moet must 0 + (wel) (indeed)

nee [I-]want-not kanniet can-not magniet may-not hoe(f)nie must-not 0 + niet not

Table 4: Modal predicates and the pragmatic function of assertion ³⁸

In structures such as kanwel (can-indeed) vs. kanniet (can-not) etc., wel and niet are merged with the modal predicate. As an integrated part of the predicate, wel (indeed) and niet (not) are used to express the pragmatic function of assertion. This explains why it is that at the relevant stage object scrambling does not occur. That is, why learners systematically produce non-targetlike structures as ik

38 The elements presented in Table 4 seem to suggest that the set of modal predicates might constitute a ‘closed class’ and, hence, establish a functional category. However, although the number of modal predicates is, indeed, relatively restricted, this is due to the input. The possibility for the child to create a modal predicate such as handigniet in handigniet kusje geve (handy-not kiss give. A 2;1) is evidence that at the relevant stage modal predicates constitute an ‘open class’, i.e. they are members of a lexical category.

128 

 The lexical stage

kanniet die pakken (I cannot that-one get) and are unable to produce a targetlike utterance like ik kan die niet pakken (I can that not get). Absence of the modal element in structures of type A establishes the function of assertion as the pragmatic default. Given this to be the case, the utterance is left with the optional element wel (indeed) vs. the negative element niet (not) to serve the affirmative or negative value of an assertion. The use of wel (indeed) vs. niet (not) to express the value of assertion also occurs with non-agentive utterances with the structure represented as in Figure 1B.

affirmative

negative

Peter luswel. (J 1;10) P likes-indeed Mijnie lust. (J 1;11) M likes Herrie, zo komtwel. Herrie zooo komt. (J 1;11) H, just comes-indeed. H just comes hij doet, hij doetwel. (J 1;11) he works, he works-indeed Peter luswel, Peter lus. (J 1;11) P likes-indeed, P likes ikke doetwel, ikke doet. (J 1;11) I do-indeed, I do dà kanwel. (A 2;1) that can-indeed gaathel. (A 2;1) goes-indeed opa hel pikt, opa pikthel. (A 2;1) granddad indeed pricks, granddad pricks-indeed

nee tee, lustnie. (J 1;10) no tea, like-not lustniet. (J 1;11) [I] like-not

hij doetniet. (J 1;11) he works-not Mijnie lusniet. (J 1;11) M likes-not

hier kanniet. (J 2;1) here can-not gaatniet, goed. (J 2;1) goes-not, good papa piktniet. (J 2;1) daddy pricks-not

Table 5: State and change-of-state predicates and the expression of assertion

Modal predicates can be used as the lexical head in utterances of both type A and type B. With an agent as the external argument, they serve the expression of control, i.e. they serve the willingness, the ability, the permission or the obligation of the agent to perform. At the relevant stage, the agent often refers to the speaker. In utterances in which the agent refers to the speaker the modal predicate expresses the pragmatic function of a wish, a promise, a request or a demand. Modal predicates with a theme as the external argument are used to express the possibility, the option or the necessity that a situation occurs. In sum, at the initial stage of language acquisition, utterance structure is the instantiation of a predication. Any lexical head may serve the function of a predi-

Utterance structure at the lexical stage  

 129

cate, no matter whether in the target system it occurs as a verb form, a particle or any other non-verbal element. Learner languages at the relevant stage have only two types of utterance structure. The difference between the two types of utterance structure lies in the presence or absence of a head position for a modal or aspectual element that is used to express control. The function of ‘control’ is exerted by the agent, absence of control is exerted by the theme. Hence, one type of utterance structure (type A) has an agent and a modal or aspectual element expressing control over actions and agentive motions, while another type of utterance structure (type B) has a theme occurring in a state or undergoing a change of state. Both types of utterance structure are used to express the pragmatic function of assertion. The value of an assertion can be expressed affirmatively with the optional use of wel (indeed) or negatively with the use of niet (not). As stated above, a morpho-syntactic categorization of the data as, for example, in Poeppel and Wexler (1993) and in Ingram and Thompson (1996), poses a restriction on the kind of data to be accounted for. If at the relevant stage of acquisition utterances are categorized in terms of verbs, or verb categories such as finite verb, infinitive and past participle, it is tacitly assumed that these categories are relevant as they are in the target language. It may lead to the decision to regard ‘verbless’ utterances of less or no relevance compared to utterances with verbs. As I have argued earlier, this explains why Poeppel and Wexler come to the conclusion that for L1 German “the best model of the data is the standard analysis of adult German” (1993, 2). The analysis of learner utterances in terms of categories of lexical-syntactic structure shows that the linguistic categories that are relevant at the initial stages of language acquisition may differ from those that are relevant at later stages in the acquisition process. For example, the verb may not be relevant as a category of the basic language system, while the predicate is. The same is particularly true for the morphological properties of the verb. Morphological properties of the target language system seem to be irrelevant as a feature of the learner language at the initial stage. Hence, if learner data are analysed in terms of morpho-syntactic categories of the target language system, learners are attributed a level of linguistic knowledge for which there is no evidence. Moreover, if the system of the target language determines the point of view from which the learner data are analysed, there is a large amount of data that are not taken into account.

5.3.2 Testing the model: ergatives and particle verbs The analysis of the spontaneous production data as given in Section 5.1 can be tested empirically with the distribution of ergative- and particle verbs.

130 

 The lexical stage

5.3.2.1 Ergatives In the target language, ergative verbs are verb forms that can be used with both an agentive- or a non-agentive meaning. Examples are breken (break), koken (cook), smelten (melt) etc. in utterances as in (2a) and (2b). (2) Ergative verbs in Dutch a. agentive

b. non-agentive

het kind breekt het kopje. the child breaks the cup moeder kookt het water. mother cooks the water de kok smelt de boter in de pan. the cook melts the butter in the pan

het kopje breekt. the cup breaks het water kookt. the water cooks de boter smelt in de pan. the butter melts in the pan

The verb forms in (2a) are used with an agentive meaning. They refer to an action in which the agent performs a causal action on the object. The verb forms in (2b) are used with a non-agentive meaning. They refer to situations in which the theme occurs in a state or undergoes a change of state. The question is, how would learners use an ergative verb at the lexical stage when the basic language system is constrained by the two types of utterance structure as presented in Figure 1A and 1B? On the basis of the relevant system, it can be expected that ergative verbs occur in a different position and with a different form depending on whether the subject is an agent or a theme. With an agent as the subject, these verbs will appear as in type A, i.e. in final position with the form of an infinitive. With a theme as the subject, these verbs will appear as in type B, i.e. in initial position with the form of a finite verb. The relevant data show that, initially, ergative verbs with non-agentive meaning as in (2b) are not used very frequently. However, whenever they occur, they are distributed differently from their agentive counterparts. Ergatives Type A1. agentive

Type B1. non-agentive

child L1 Dutch Mijnie zelf doen. (J 1;11) M self do poes bal hebbe. (J 1;11) kitty ball get

hij doe(t) niet. (J 1;11) he does (= works) not Tompoes heeft koffie. (J 1;11) T has coffee

Utterance structure at the lexical stage  

papa vorrek hebbe? (A 2;1) daddy fork get? mama school toegaan. (J 2;1) mommy school towards-go óók gaan, paarde. (A 2;1) too go, horses vokkik pikke. (A 2;2) fork prick

 131

Jaja heeft au. (A 2;1) J has ow gaatie inne garage. (J 2;2) goes-it into garage deze gaat niet. (A 2;1) this goes (= works) not papa pikt niet, opa pikt hel. (A 2;1) daddy pricks not, grandfather pricks indeed

adult L2 Dutch en dan nou drie weken stage doen. (Os/T 1.6) and then now three weeks internship do ik alles doen. (Mo/A 2.3) I everything do en terug daar iets doen. (Mo/A 3.7) and back there something do toen hij staan, hij kijk achter hem. (Mo/A 2.6) when he stand-up, he look behind him allemaal nou zitten, in auto. (Mo/A 2.9) all now sit-down, in car

en nou motor doet niet. (Os/T 1.9) and now motor does (= works) not doet die motor zo. (Mo/A 2.7) does (= works) that motor this-way mij auto doet niet. (Mo/A 3.7) my car does (= works) not die naam hier staat niet. (Mo/A 2.8) that name here stands not hij [= sjaal] zit op hem gezicht. (Mo/A 2.9) he [= scarf] sits on him face

As shown in the examples above, the alternatives occur in complementary distribution exactly as predicted. An ergative verb with an agentive meaning appears in utterances of type A1. An ergative verb with a non-agentive meaning appears in utterances of type B1.

5.3.2.2 Particle verbs Further evidence of a basic language system as represented with Figure 1A and 1B (Section 5.3.1) is the distribution of particle verbs. Particle verbs are verb forms such as opdoen (on-do) and omvallen (down-fall). They consist of separable elements such that opdoen may also occur as doet op and omvallen may also occur as valt om. There are two types of particle verbs: particle verbs with an agent as the subject like opdoen (on-do), losmaken (loose-make) and vasthouden (tight-hold) and particle verbs with a theme as the subject like omvallen (down-fall), uitkomen (out-come) and ingaan (in-go). At the lexical stage, both types of verb are expected to be distributed differently. The agentive type of particle verb should occur in final position and therefore, as in the target language, with an infinite

132 

 The lexical stage

form, i.e. as opdoen, losmaken and vasthouden. The non-agentive type of particle verb should occur in initial position and, therefore, as in the target language, with a finite form, i.e. as valt om, komt uit and gaat in. In the spontaneous language use of children, particle verbs occur relatively frequently. In production data from adult second language learners, however, particle verbs are less common. For this reason, the following examples are a selection from my child first language data. Particle verbs. Child L1 Dutch Type A. agentive

Type B. non-agentive

Jaja dop opdoen. (A 2;0) J cap on-do papa indoen. (J 1;11) daddy in-do vlokke opdoen. (A 2;0) chips on-do goene aandoen. (A 2;1) shoes on-do dit losmake. (J 1;11) this loose-make viesmake. (A 2;1) dirty-make omslaan? (J 1;11) over-turn? openslaan? (J 1;11) open-make? punne aanslijpe. (J 1;11) points on-scharpen afdroge. (J 1;11) off-dry mondje afvege. (J 1;11) mouth off-wipe opdrinke? (J 1;11) up-drink? Cynthia weglope. (J 1;11) C away-run

valtie om. (J 1;10) falls-it down Mijnie valt om. (J 1;11) M falls down Tompoes komt aan. (J 2;0) T comes at kommaa mee. (A 2;1) come with Peter komt thuis. (J 1;11) P comes home gaap kom niet Jaja toe. (A 2;1) sheep comes not J to kommie da uit. (A 2;1) comes-it there out gaatie af. (A 2;1) goes-it off gaattie inne garage. (J 2;2) goes-it into garage poppie gáát niet mee. (A 2;1) doll goes not with gaanne huis toe. (A 2;1) go-we house to ben zo trug. (J 1;11) am right back vliegtie weg. (A 2;3) flies-it away

These examples show precisely what could be expected. The agentive type of particle verb occurs in utterances that are instantiations of type A. That is, it occurs

Utterance structure at the lexical stage  

 133

in final position and appears with infinite morphology. The non-agentive type of particle verb occurs in utterances that are instantiations of type B. That is, it occurs in initial position and appears with finite morphology. In sum, utterances are either agentive or non-agentive. Agentivity determines the presence or absence of a modal element and as a consequence the position of the predicate. With a structural position for a modal element, the predicate occurs in final position. In the absence of such a position, the predicate occurs in initial position. It is the position of the predicate that determines the form with which the predicate is learned and not vice versa. The basic language system as represented in Figure 1 A and B is a simple lexical-syntactic system. It is the reflection of situations that can be categorized in terms of the presence or absence of control. In hindsight this does not really come as a surprise. The notion of control is essential for a human being to make an assessment of the actual situation, because it matters if someone has the possibility to influence a situation or whether he is just exposed to it. The representation in A and B of Figure 1 indicates the semantic difference between the two types of utterance that learners use at the lexical stage. Structurally, however, the two representations are identical. For language learners, it might therefore not be too difficult to discover that these types of utterance are variants of the same underlying structure. The result of this is represented in C of Figure 2. The hierarchical structure of Figure 2C consists of two different relations. First, there is the relation between the lexical head (V) and its complement. The head-complement relation is characterized by the fact that the grammatical category of the complement is determined by the head. In utterances of type A, the head (V) is a modal element. It expresses the function of control that is exerted by the agent. The complement that is under control of the head is an action predicate. In utterances of type B there is no function of control. The head (V) is a non-agentive predicate. It serves to express a state or a change of state. Besides the head-complement relation, there is also the relation between both head and complement (V′) and the element that V′ holds for, i.e. the subject. In type-A utterances the subject is the agent and in type-B utterances the subject is the theme. The structure represented in Figure 2C is well-known in linguistic theory. It accounts for the linguistic knowledge that is the result of a creative process of language learning. The categorization in terms of an opposition between two types of lexical-syntactic categories seems to be founded on a basic cognitive distinction between situations that are under control and situations that are not.

134 

 The lexical stage

A: agentive

B: non-agentive

Ctrlʺ agent Ctrl

Predʺ Ctrlʹ

theme

complement

Predʹ

Pred (complement)

VPʺ

C: subject

Vʹ V (head)

Ctrl mag may

complement

Type A:

agent Jaja J

action dop opdoen. (A 2;0) cap on-do

Type B:

theme state deze magwel. (J 1;10) this-one may-indeed theme change of state poppie valt hier. (A 2;0) doll falls here

Figure 2: Utterance structure at the lexical stage

5.3.3 Word order The analysis of the Dutch learner data in Section 5.3.1 shows that at the initial stage of language acquisition utterance structure is rather simple. This holds for the structural relations between constituents as given in Figure 2. Now, what about the rules of word order?

Utterance structure at the lexical stage  

 135

5.3.3.1 Subject first At the relevant stage, the word order rules are also very simple. The data as presented in Section 5.2 show that in agentive utterances the agent occurs in initial position, while in non-agentive utterances it is the theme. Both agent and theme have subject function. Thus, word order at the lexical stage appears to be determined by the principle ‘Subject first’.³⁹ Compare, for example, the child L1 data in (3). (3)

Subject (agent or theme)

Lexical stage a.

Jaja mag dop opdoen. (A 2;0) J may cap on-do

b.

poppie valt hier. valtie hier? (A 2;0) doll falls here. falls-it here?

Functional stage c.

broodje mag Cynthia opete. (A 2;4) bun may C up-eat

d.

da kanne kindere inzitte. (A 2;4) there can children insit

e.

dà heef Jaja maakt. (A 2;4) that has J made

f.

da ben ikke ook wees. (A 2;3) there am I too been

The examples in (3a, b) are typically produced at the lexical stage. At this stage, utterances with a subject not in initial position as in (3c, d, e, f) are absent. They only appear with the acquisition of the functional projection of F. Evidence that F is not yet available at this stage, is the absence of auxiliary verbs such as heeft

39 In my child data at the lexical stage, there are only 4 examples with either Object-SubjectVerb or Adverb-Subject-Verb: dit Mijnie vasthoue (this M tight-hold. J 1;10), hier poes krabd (here kitty scratched. J 2;0), ditte ik kleurd (this I coloured. J 2;0) and da Goover viege (there G fly. A2;0). It seems that these utterances are produced as a type of left-dislocation similar to examples such as deze ook toetie heg zette (this too, stool away put. A 2;0) and da ook appeltaart bakt (there too, apple-cake baked. A 2;2).

136 

 The lexical stage

(has) and ben (am) as they occur in (3e, f). Auxiliaries appear at some later point in the acquisition process when, with the acquisition of the functional projection of F, there is also the possibility for the subject to occur in non-initial position. The acquisition of both auxiliaries and non-initial subjects is evidence that learners have reached the functional stage. In sum, at the lexical stage, word order is based on the principle ‘Subject first’. It is due to the fact that the learner grammar has not yet established the functional projection of F that would make word order variation possible.

5.3.3.2 Topic first In the system of the target language, the initial position is the topic position. The topic constituent serves to establish the relation between the utterance and the topic situation (TS). As illustrated in Figure 3, TS is the situation that the utterance applies to.

TS

topic | VP

Figure 3: The utterance applies to a topic situation TS

The predication of VP that serves to express the hold-for relation between the predicate and the subject provides the information that the speaker wants to communicate. Hence, as Klein argues, “the descriptive content of the sentence is somehow split between parts which help to identify TS [the topic situation], and parts which do not serve this function.” (…) “This split, I believe, is what underlies the fundamental dichotomy which is traditionally covered by oppositions such as ‘psychological subject vs psychological predicate’, ‘topic vs comment’, ‘presupposition vs focus’, ‘background vs focus’ and similar ones” (Klein 2008: 293). At the lexical stage, in the absence of F, there is no topic position, yet. Furthermore, as shown above, absence of F accounts for the fact that the subject at the relevant stage can only occur in initial position. Hence, the subject at the lexical stage serves as a carrier of the topic function at the same time. Compare, for example, the child L1 data in (4).

Utterance structure at the lexical stage  

(4)

 137

Topic (agent or theme)⁴⁰

Lexical stage TS

a.

Jaja mag dop opdoen. (A 2;0) J may cap on-do

TS

b.

poppie valt hier. valtie hier? (A 2;0) doll falls here. falls-it here?

Functional stage TS

c.

wie is dat? kijke is? (A 2;3) who is that? see is?

TS

d.

wat is dit nou van kleur? (A 2;3) what is this now of colour?

TS

e.

waar is v[l]iegtuig nou? (A 2;3) where is airplane now?

At the lexical stage, the examples in (4c, d, e) do not occur. They only occur at the functional stage when the projection of F has been established. This is due to the fact that with the projection of F there is an initial position which serves both as topic and focus position. In the absence of F, there is no such position. It explains why both the utterances in (3c, d, e, f) and in (4c, d, e) may appear only later in the acquisition process, i.e. when learners have reached the functional stage. In sum, learner grammar at the relevant stage is as simple as can be. While the subject is used to express the topic function, the predicate refers to information that is in focus. In other words, at the lexical stage there is a 1:1-correspondence between the syntactic structure of an utterance and its properties of information structuring. A representation of this is given in Figure 4.

40 Spontaneous child utterances usually apply to topic situations that are linked to the here and now of the moment of speaking. This explains why in child language particularly deictic elements, i.e. proper names, e.g. Jaja in (a), and pronouns such as ik, ikke (I) and deze (thisone) are used in first position.

138 

 The lexical stage

VPʺ subject

Vʹ (predicate) V (head)

complement

Type A:

agent

Ctrl

Type B:

theme

state change of state ↕ focus

TS

↕ topic

action

Figure 4: Utterance structure at the lexical stage: 1:1-correspondence between syntax and information structure

5.3.3.3 Head-initial The word order within the predicate V′ is determined by the position of the head relative to the complement. The following examples show that in utterances of type A, V′ is head-initial. Type A child L1 Dutch

adult L2 Dutch

ulle ijsje hebbe. (J 1;8) want ice-cream get nee au doen. (A 2;1) want-not ow do kannie bal pakke. (J 1;9) can-not ball get Jaja mag dop opdoen. (A 2;0) J may cap on-do doettie alles opete. (J 1;11) does-it everything up-eat gaatie [sl]ape. (A 2;1) goes-he sleep

hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault ik wil gaan naar moskee. (Os/T 1.8) I want go to mosque kannik niks anders doen. (Os/T 3.6) can-I nothing else do moettik eerst mijn eten maken. (Mo/A 2.4) must-I first my dinner make hij doet krant helemaal opendoen. (Os/T 3.3) he does newspaper fully open-do hij gaat snel lopen. (Mo/A 1.6) he goes fast run

Utterance structure at the lexical stage  

 139

In utterances of type A, the head of V′ is either a modal element like ulle [= wil] (want), wil (want), nee [= wil niet] (want-not), kan (can), kannie[t] (can-not), mag (may), moet (must) or its aspectual default doet (does) or gaat (goes). It serves to express control. The complement ijsje hebbe (ice-cream get), au doen (ow do), bal pakke (ball get), dop opdoen (cap on-do), alles opete (everything up-eat), slape (go asleep) and kopen een renault (buy a renault), gaan naar moskee (go to mosque), niks anders doen (nothing else do), eten maken (food make), krant opendoen (newspaper open-do), snel lopen (fast run) is the object of control. In utterances of type B, V′ is head-initial, too. Type B child L1 Dutch

adult L2 Dutch

Mijnie valt om. (J 1;10) M falls down vindt vies. (A 2;1) finds dirty gaanne huis toe. (A 2;1) go-we home to hij doet niet. (J 1;11) he does not

ik woont in casablanca. (Mo/A 1.4) I lives in casablanca hij vindt leuk … (Os/T 2.5) he finds nice hij blijft met mij. (Mo/A 1.5) he stays with me ik komt hier. (Os/T 1.6) I comes here

In utterances of type B, the head of V′ is the lexical verb valt (falls), vindt (finds), gaan (go), doet (does), woont (lives), blijft (stays), komt (comes). The complement is a particle om (down), vies (dirty), niet (not) or an adverbial in casablanca (in Casablanca), met mij (with me), hier (here). The conclusion from this is that word order in utterances of type A is the same as in type B. Hence, at the lexical stage, word order of the basic language system is head-initial. Contrary to this conclusion, it is generally claimed that at the intial stage the underlying word order in child L1 Dutch is head-final, whereas in English, for example, it is head-initial. Evidence of this should show up in early child language with utterances as in (5a) in Dutch and (5b) in English⁴¹. (5) Head-complement structure in child L1 Dutch and child L1 English a.

poes bal pakke. (J 1;11) kitty ball get

41 The examples in (5b) are from Bowerman (1990: 1267 ff.). E = Eve.

140 

 The lexical stage

gaag boekje leze. (A 2;0) want booklet read b.

have cookie. (E 1;8) read book me. (E 1;9) make ring me. (E 1;9) give candy for me. (E 1;10)

However, in child utterances as in (5a) and (5b), the predicates bal pakken (ball get), boekje lezen (booklet read) and have cookie, read book me etc. are in fact lexical entities. That is, they are structurally similar to compounds such as handenwassen (hands-clean) and tandenpoetsen (teeth-brush) in Dutch. Hence, the idea that the underlying word order in Dutch is head-final, is based on the internal structure of the predicate in utterances of type A. As shown in type A2, it does not apply to type-A utterances as such. Nor does it apply to utterances of type B. The examples in (6a) and (6b) show that in adult L2 Dutch there is some variation with respect to the internal, head-complement structure of the predicate. (6) Head-complement structure of the predicate in adult L2 Dutch a.

ik wil gaan naar moskee. (Os/T 1.8) I want go to mosque als jij jullie niet wilt geef een huis. (Mo/A 1.6) if you not want give a house hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault hij 0 maken die fiets alles. (Mo/A 1.3) he fix that bike everything

b.

ik moet huis kopen. (Os/T 1.6) I must house buy ik wil uh honderd dirham geef uh … (Mo/A 1.3) I want em hundred dirham give em .. die mag niet daar veertig kilometer rijen. (Os/T 1.9) that-one may not there forty kilometers drive en dan ook krant pakken. (Os/T 1.6) and then also newspaper take

Both in (6a) and (6b) the position of the modal verb or its default shows that syntactically the structure of the utterance is head-initial. Lexically, however, i.e. as far as the internal structure of the complement is concerned, the ordering varies.

Utterance structure at the lexical stage  

 141

In (6a), entities such as gaan naar moskee (go to mosque), geef een huis (give a house), kopen een renault (buy a renault), and maken die fiets (fix that bike) are head-initial, while in (6b) the entities huis kopen (house buy), honderd dirham geef (hundred dirham give), veertig kilometer rijen (forty kilometers drive) and krant pakken (newspaper get) are head-final. This kind of variation may be due to the interaction in the Dutch learner language between native and target language word order. That is, basic word order in Turkish is head-final, in Arabic it is headinitial. In the target language, however, word order varies. For example, in utterances with a finite lexical verb such as eet (eats) in Jan eet een koekje (John eats a cookie), the lexical verb occurs before the object, while in utterances with an infinite lexical verb such as eten (eat) in Jan gaat een koekje eten (John goes a cookie eat), the lexical verb is placed behind it. The interaction between basic word order in the native language and both options as they appear in the target language input may explain why it is typical for Arabic learners of Dutch to produce utterances with head-initial word order as in (6a), as opposed to Turkish learners who typically produce utterances with head-final word order as in (6b). This interaction of native and target language word order in L2 Dutch learner language has been accounted for by Jansen et al. (1981) in terms of what they refer to as the Alternation Hypothesis. “Assume that in target language A there is an alternation between two surface structures, and that in source language B only one of the two surface structures occurs. Then speakers of source language B acquiring language A will overgeneralize in their interlanguage grammar the structure which corresponds most closely to the structure in their own language” (Jansen et al. 1981: 315f.).

5.3.4 Summary In the acquisition of child L1 and adult L2 Dutch, there is an initial stage at which utterance structure is based on two types of lexical projections. That is, on the one hand, lexical projections with agentive predicates such as bal pakke (ball get), Jaja mag dop opdoen (J may cap on-do), klimme (climb), ik wil paard opklimme (I want horse on-climb) and deze nee afdoen (this-one no off-do) and, on the other hand, lexical projections with non-agentive predicates such as Mijnie valt (M falls), valt om (falls down), poes ligt (kitty lies), ik heb niet die (I have not thatone), dit lus niet (this like not), is kapot (is kaput), valt (falls), komt (comes), vliegt weg (flies away), kommie da uit (comes there out). Both types of utterances, referred to as type A (agentive) and type B (nonagentive), are lexical structures that specify a hold-for relation between a predi-

142 

 The lexical stage

cate and a constituent that may function either as an agent or a theme. The constituent that the predicate holds for is the subject. The internal structure of the predicate is determined by the position of the head constituent. Both types of utterance structure are head-initial. In utterances of type A, this head-initial position is taken by a modal predicate or its aspectual default. It expresses the semantics of control (Ctrl). Its complement is a predicate that refers to a causal action or an agentive motion. In utterances of type B, the initial head position is taken by a state or a change-of-state predicate. Given that both these agentive and non-agentive types of utterance are produced as instantiations of a relation between a predicate and the constituent that it holds for, learners are in the position to infer the generalization of a structural configuration as in Figure 4. The subject is the constituent that the predicate holds for. In the absence of the functional projection of F, the subject occurs in initial, topic position. In this position, it establishes a relation with the situation that it applies to, i.e. the topic situation (TS). Due to the absence of F, word order is fixed. Hence, the structure as presented in Figure 4 accounts for a 1:1-correspondence between the syntactic structure of the lexical projection and the way in which it is used to express the properties of information structure. Syntactically, utterance structure in both types of utterance is head-initial (see Hoekstra and Jordens 1994). Lexical entities serving as the complement of a modal or aspectual head appear in final position and with non-finite morphology. In child L1 Dutch, these lexical structures are head-final as, for example, in Jaja mag dop opdoen (J may cap on-do). In adult L2 Dutch, the internal structure of the complement can be either head-initial as in hij wil kopen een renault (he wants buy a renault) or head-final as in ik moet huis kopen (I must house buy). The relevant variation in utterance structure seems to be due to the interaction between the head-complement structure of the L1 and the input of the target language (see Jansen et al. 1981). In the absence of the functional projection of F, there is no SpecFP position and, hence, the learner system has no movement. That is, no topicalization, no wh-questions and no yes/no-questions. Furthermore, in the absence of the functional projection of F, utterance structure has no position to express morphological finiteness. Thus, inflection as it seems to occur in state and change-of-state predicates as in Mijnie valt (M falls), poes ligt (kitty lies), komt (comes), vliegt weg (flies away), is not a morphological feature of the learner system. At the relevant stage, verb morphology simply reflects the form in which the predicate occurs in the input. A similar observation has been made with respect to data of child L1 German. In absence of evidence of productivity, verb forms showing features

Utterance structure at the lexical stage  

 143

of agreement are interpreted as instances of “holistic learning” (Ingram and Thompson (1996: 111). Auxiliary verbs are functional elements. In the target language, auxiliary verbs such as heb, heeft (have, has) and ben, is (am, is) are used in utterances with past participle forms to express perfect aspect. Furthermore, modal auxiliary verbs are used in utterances like de boom kan omvallen (the tree can downfall), je kunt hier wel zitten (you can here indeed sit), deze brief moet vandaag nog weg (this letter must today still away), dit voorstel mag in de prullenbak (this proposal may in the waste-paper basket). At the relevant stage, in the absence of the functional projection of F, utterances with auxiliary verbs (modal or aspectual) do not occur. At the lexical stage, modal elements seem to constitute a ‘closed class’, and hence, they seem to constitute a functional category. However, the fact that the number of modal elements is relatively restricted is due to the input. At the relevant stage, modal elements serve as the head of a lexical projection. Examples are nee (want-not) vs. ulle (want), mag-ikke (may I) or g(r)aag (want); kanwel (canindeed) vs. kanniet (can-not); magwel (may-indeed) vs. magniet (may-not) and even handigniet (handy-not). These modal predicates are used as unanalysed elements. Evidence for this is the fact that at the relevant stage the target language phenomenon referred to as ‘object scrambling’, as in, for example, ik kan die niet hebbe (I can that-one not get), does not occur. Learner grammar at the relevant stage only allows for structures like ik kanniet die hebbe (I cannot that-one get). It shows that modal elements at the lexical stage are not used as elements of the functional category F. At the lexical stage, the modal head of an agentive predicate is used to express the willingness, the ability, the permission or the obligation that holds for the agent. If the agent refers to the speaker, modal elements are used to express the pragmatic function of a wish, a promise, a request, a demand etc. In the absence of an agent, utterances with a predicate referring to a state or a change of state can only be used to express the default pragmatic function of assertion. To conclude, at the initial stage of child L1 and adult L2 Dutch acquisition, utterance structure is determined lexically. Despite the fact that the input provides ample evidence of the use of functional elements, functional categories are not part of the grammatical system, yet. The grammatical system of learner language is therefore relatively simple. In the absence of both F and SpecFP, there are no functional positions for the expression of the target language properties of semantic finiteness (S-finiteness) and topicalization. That is, there are no functional means geared towards the use of an utterance as a contribution to a larger discourse. The properties of information structure are expressed with the syntactic categories of the subject and the predicate. The subject is used with topic

144 

 The lexical stage

function, the predicate serves to express focus information. The embedding of the utterance into the situational context is achieved with lexical means such as deictics and adverbials. The consequences of the absence of the functional projection of F with respect to the structure of the basic language system are summarized in (7). (7)

no F no SpecFP

→ no auxiliaries (no scrambling) → no inflection (no agreement, no tense) → no topicalization → no determiners → no wh-questions → no yes/no-questions

In the basic language system, the pragmatic function of assertion (S-finiteness) is represented lexically. With agentive predicates, it is carried by a modal predicate, an aspectual element (doet, gaat, is) or its position is left empty; with non-agentive predicates, it is carried by the lexical predicate itself. Finally, usually learner data are represented selectively, for example on the basis of particular morphological features. However, the problem with this approach is the lack of evidence that at the relevant stage, these properties are productively used as elements of learner language. In fact, the criteria for data selection and categorization are usually taken from the target language. As a consequence, a host of data that are characteristic of the shape of learner language are left unanalysed. Additionally, and more seriously, a host of data are categorized incorrectly. For example, the classification of verb forms in terms of finite verb forms and root infinitives turns out to be inadequate with respect to the learner variety that it is claimed to hold for. The representation as given above establishes the claim that without exception all learner utterances at the initial stage of language acquisition can be accounted for in terms of two types of lexical projection, i.e. lexical projections with an agentive or a non-agentive predicate. These projections are the reflection of a distinction at the conceptual level between situations that are categorizable in terms of situations that are subject to control and situations that simply happen to occur.

5.4 Conflicting constraints The basic language system provides a relatively simple means of communication. With the complex target system in place, learners have available more efficient means to communicate effectively. Perdue phrases it as follows:

Conflicting constraints 

 145

“The early varieties are less efficient communicative systems than the later ones. Their progressive complexification can then be attributed to the learner’s attempts to overcome these communicative limitations” (Perdue 2006: 859).

Efficient communication is achieved with the discourse-functional means provided by the functional category system. As pointed out above, it is the initial position of the utterance that serves the function of contextual embedding. Elements in initial position are used with topic function, i.e. they refer to the situation that the utterance applies to. At the lexical stage, the initial position is both the subject and the topic position. Furthermore, the subject is either the agent or the theme. This explains why at the relevant stage learners produce utterances as in (8a) and (8b), whereas examples with an object or an adverbial in first position as in (9) only occur later in the acquisition process at the functional stage of language acquisition. (8) Subject in topic position

TS

a.

Jaja mag dop opdoen. (A 2;0) J may cap on-do

TS

b.

poppie valt hier. valtie hier? (A 2;0) doll falls here. falls-it here?

(9) Object or adverbial in topic position broodje mag Cynthia wel opete. (A 2;4) bun may C indeed up-eat dan moet Cynthia weer make. (J 2;2) then must C again make Thus, in the absence of the functional position SpecFP, the agent and the theme always occur in initial, topic position. However, as far as the topic function is concerned, there is no functional reason why the object of a causal action or an adverbial could not be used with topic function and hence, why these constituents could not occur in initial position as in (9). Hence, at the lexical stage, structural properties of the basic language system and principles of information structure are in conflict with each other, a conflict that needs to be resolved during the course of the language learning process. Given that children usually talk about the here and now, they do not have to worry about the linguistic means to achieve temporal or local embedding. Hence,

146 

 The lexical stage

it is not necessary for children to be able to refer to a particular time or place. L2 learners, on the other hand, do make use of lexical means of contextual embedding. Temporal or local embedding is achieved by an adverbial or a nominal element in XP position, sentence-initially. At the lexical stage, L2 learners, thus, typically produce utterances with a structure as in (10). (10) Finite verbs in V3 position in L2 Dutch⁴² dan politie wilt charlie slaan. (Mo/A 2.6) then police wants Charlie beat toen die oma wilt ook auto rijen. (Mo/A 3.9) then that grandmother wants also car drive toen die meisje gaat snel lopen. (Mo/A 3.9) then that girl goes fast walk However, utterances as in (10) are not targetlike. Rather than being a solution of the conflict, placement of the element in initial position is a means to avoid this conflict. It works with adverbials, because an adverbial may serve as an adjunct. It works with nominal constituents, because a nominal constituent may occur in the position of left-dislocation. A solution of the conflict between lexical-syntactic and functional-pragmatic principles of word order is initially achieved with a kind of vanishing act. Agentive utterances may occur with the object or an adverbial in initial, topic position if the agent is not expressed. At the relevant stage, this explains the use of nontargetlike utterances as in (11).

42 The well-known ZISA study (e.g. Pienemann 1986) provides evidence of the same phenomenon in the L2 German of adult speakers of a Romance L1. These L2 learners of German initially produce utterances with SVO structure. When ADV occurs in initial position, it has no effect on the position of S. This explains learner utterances as in (i). (i)

dann ich schreiben (then I write) da kinder spielen (there children play) un einmal die war in garderobe (and once that-one was in wardrobe) sonntag autopolizei sun da (sunday car-police are there) In: Clahsen, Meisel and Pienemann (1982).

Conflicting constraints 

 147

(11) Agentive utterances with an object or an adverbial in topic position⁴³ child L1 Dutch

adult L2 Dutch

nee losmake. (J 1;10) [this] wantAG-not loose-make dit nee afdoen. (J 1;10) this wantAG-not off-do die [mag]niet afpakke! (J 1;11) that mayAG-not away-take! da kanwel opzitte. (J 2;0) there canAG-indeed on-sit da moet op drukke. (J 2;1) there mustAG on press papa, moet make. (A 2;0) daddy, [this] mustAG make disse hoeniet meeneme. (A 2;1) this mustAG-not with-take die maa hier doen. (A 2;1) that justAG here do da kanniet zitte. (A 2;1) there canAG-not sit

dan kanniet lezen, kan begrijpen niet. (Os/T 1.7) then canAG-not read, canAG understand not hier schoenen uitdoen moet. (Os/T 1.8) here shoes off-do mustAG en dan moet teruggeven. (Os/T 2.3) and then [this] mustAG back-give die wijn magwel drinken. (Os/T 2.3) that wine mayAG-indeed drink naam moet schrijven. (Os/T 2.3) name mustAG write die magwel kopen. (Os/T 2.5) that mayAG-indeed buy dan moet daar helemaal opruimen. (Os 2.5) then mustAG there all up-clean dan moet woordenboek kijken. (Os/T 2.7) then mustAG dictionary look die stad magniet met 40 km hard rijden. (Os/T 2.9) that city mayAG-not with 40 km fast drive Charlie zegt: nou magwel weglopen. (Os/T 2.9) C says: now mayAG-indeed away-run of kan andere sturen naar turkije. (Os/T 3.3) maybe canAG other send turkey

da kanniet pakke, visie. (A 2;1) that canAG-not get, tv deze magniet teke. (A 2;2) this mayAG-not draw hier moet poesje eve kamme. (A 2;3) here mustAG kitty just comb

The data in (11) show that the object or an adverbial can be placed in topic position, because the relevant structure does not provide a position for an agent. It is a creative solution of the internal conflict within the constraints of the language system at the lexical stage.⁴⁴ However, the fact that this type of utterance has no structural position for the agent does not mean that the agent does not play a role.

43 AG indicates that the agent is used implicitly with the modal predicate. 44 Evidence of a similar solution in child L1 German are examples such as des auchnoch rausmach (that too out-make) and da daman aufmachen (there may-one open-make) in Dimroth et al. (2003: 80, 82).

148 

 The lexical stage

On the contrary, whenever there is an action, there is also an agent. And precisely because this is the case, it might not be necessary to express the agent explicitly. The spontaneous learner data in (11) show that, at the relevant stage of the language acquisition process, this type of utterance is used productively, regardless of the fact that it does not occur in the input. It serves the specific function of a topicalization device and it is the best kind of evidence that learner utterances at the early stages of language acquisition are creatively produced on the basis of an abstract system of linguistic principles.⁴⁵ Utterances with an object in topic position as in (11) have also been noticed in earlier research on child L1 Dutch. In Verrips (1996) and Van Kampen and Wijnen (2000) this type of utterance is analysed either as a learner version of a passive sentence or as an overgeneralization of an ergative structure. As a learner version of a passive sentence the utterance die magwel kopen (that-one may-indeed buy) is analysed as the precursor or, to use a more interesting term, the underspecification of an utterance such as die mag wel gekocht worden (that-one may indeed bought be). As the overgeneralization of an ergative structure, the utterance die magwel kopen may be produced as the analogy of a target utterance such as die kanwel breken (that-one can-indeed break). However, for both these interpretations there is not much proof. Passives appear much later in the acquisition process. Furthermore, at the relevant stage ergatives are rare. Hence, it seems unlikely that they could prompt a structural case of overgeneralization. Moreover, overgeneralization raises the problem known as Baker’s paradox (1979). Baker argues that language acquisition is driven by positive evidence. With Baker, it is therefore generally assumed that negative evidence by means of correction has no effect on the process of acquisition. The question then is, how would learners be able to constrain an incorrect generalization, if counter-evidence does not play a role. Van Kampen and Wijnen (2000) seem to be aware of this when they note: “It is yet unclear how this learning process takes place”⁴⁶ (268). Finally, both these accounts do not recognize the fact that, at the relevant stage of acquisition, the utterances with an object in topic position always occur with a modal element. Thus, it does not seem to be a coincidence that the title of one of the publications on this matter reads “Potatoes must peel” (Verrips 1996) and not “Potatoes peel”. At the relevant stage, modal elements serve as the head of an action predicate. Hence, it is these modal predicates that allow the agent not to be expressed.

45 The spontaneous production of this type of utterance can be used as an argument in the discussion on the „usage based account of development“ in „item-based grammars“ (Bannard, Lieven and Tomasello 2009: 17284f.). 46 In Dutch: “Hoe dat leerproces in zijn werk gaat is nog geheel onduidelijk”

Conflicting constraints 

 149

The solution of the basic language system that allows the object or the adverb to appear in topic position is possible because the agent is not expressed. Nevertheless, as argued above, in utterances as in (11) the agent does play a role. Evidence of this is the fact that at the relevant stage the system appears to develop the means of an affix to express the implicit agent function. Examples are given in (12). (12) The affix as a means to express the agent child L1 Dutch

adult L2 Dutch

doetie alles opete. (J 1;11) does-he everything up-eat nou gaatie weer naar huis toe gaan. (J 2;2) now goes-he again to home to go

dan magg’k even vragen? (Os/T 2.3) then may-I just ask? dan hier moette kaartje maken. (Os/T 2.4) then here must-we card make moetje deze meisje trouwen niet andere. (Os/T 2.6) must-you this girl marry not other-one dan moette daar kijken. (Os/T 2.8) then must-we there look kan’k proberen, kan’k niks anders doen. (Os/T 3.6) can-I try, can-I nothing else do

doense same zitte. (A 2;2) do-they together sit-down moete nieuwe make, vokke. (A 2;2) must-we new-ones make, flakes zo moettet rije. (A 2;2) this-way must-it drive

Verb forms such as omdraaien (around-turn), opeten (up-eat), vragen (ask) and maken (make), but also naar huis toe gaan (to-home to go) and zitten (sit) meaning ‘sit-down’⁴⁷ refer to actions and as actions they imply an agent. In (12) the agent is expressed with an affix -ie, -e, -et, -se, -’k, -je. Although these affixes may serve as the pronouns hij (he), zij (she), het (it), zij (they), ik (I), jij (you) they occur with no particular syntactic position. The question now is: is it possible for the language system to become reorganized such that the object may occur in initial, topic position, while the agent is attributed a position according to the syntactic principle ‘Subject first’? “Contexts of conflicting constraints are very fertile for observing language development” (Perdue 2006: 862). This observation by Perdue seems to apply to the conflict between ‘Subject first’ and ‘Topic first’. In Chapter 6, it will be shown that this

47 Language learners use the verb zitten both as zit (sit) with the meaning of a state and as zitten (sit-down) with the meaning of an agentive motion.

150 

 The lexical stage

conflict serves as the driving force in a process in which the early lexical system develops into a targetlike system with both lexical and functional categories.

5.5 Conclusion Evidence from child L1 and adult L2 learners of Dutch shows that the relation between language input and the shape of the learner variety cannot simply be described as “what they hear is what you get” (Ingram and Thompson 1996: 97). Given a particular amount of target language input, learners appear to be able to create a basic language system which has no functional projections and, hence, no functional categories. At the initial stage, utterance structure is claimed to be the instantiation of a lexical projection that has a configuration as in Figure 2C, here repeated as Figure 5. VP subject

V′

V (head) Type A: agent Jaja J

Ctrl mag may

complement action dop opdoen. (A 2;0) cap on-do

Type B: theme state deze magwel. (J 1;10) this-one may-indeed theme poppie doll

change of state valt hier. (A 2;0) falls here

Figure 5: Utterance structure of the Dutch basic language system

The configuration in Figure 5 accounts for the expression of two types of utterance as in A and B of Figure 1, here repeated as in Figure 6. They are representative of a categorization of the outside world in terms of agentive situations that are under control and non-agentive situations that are not.

Conclusion  

 151

In Figure 6, A accounts for the syntactic structure of lexical projections with agentive predicates (type A) and B accounts for the syntactic structure of lexical projections with non-agentive predicates (type B).

A: agentive

B: non-agentive

Ctrl″

Pred″

agent Ctrl′ (carries out)

theme Pred′ (occurs / undergoes)

Ctrl complement | action (causation, motion): e.g. dop opdoen (cap on-do), tiktak om (watch on), bad zitte (bath sit-down).

Pred (complement) | state: e.g. zit (sits), lust (likes), is op (is it-on), deze moet hier ( this-one must here). | change of state: e.g. valt (falls), komt (comes).

modal: e.g. wil (want), kan (can), mag (may), moet (must). | aspectual: e.g. gaat (goes), komt (comes). | default: e.g. doet (does), 0 = to be inferred. Figure 6: The syntactic structure of type A and type B

Agentive situations are represented syntactically in terms of a predicate-argument structure with a modal or an aspectual predicate as the head, an agent as the specifier and an action predicate as the complement. The action predicate may refer to either a causative action or an agentive motion. Modal elements serve to express that the agent has the willingness, the ability, the permission or the obligation to carry out a particular action. Aspectual elements such as gaat (goes) and komt (comes) are used to express that the agent has the intention to carry out the action, while doet and is indicate the default function of control (Ctrl). Typical examples are Jaja mag dop opdoen (J may cap on-do; A 2;0), gaatie [sl]ape

152 

 The lexical stage

(goes-he sleep; A 2;1), hij komt snel lopen direct (he comes fast run immediately; Mo/A 1.6), doettie alles opete (does-it everything up-eat; J 1;11), daar is mensen niks doen (there is people nothing do; A/T). If the position of the + Ctrl element is empty, as for example in poes bal pakke (kitty ball grab; J 1;11) a modal or aspectual meaning is left to be inferred from context. Events that are non-agentive are represented syntactically in terms of a predicate-argument structure with a state or a change-of-state predicate as the head and a theme as the specifier. It serves to express that the theme occurs in a state or undergoes a change of state. Typical examples are poes ligt (kitty lies; J 1;11), Mijnie valt (M falls; J 1;11), is hanne nou? (is hands now; A 2;1), Jaja heef au (J has ow; A 2;1), dit magwel? (this may-indeed? J1;11). Verb morphology at the relevant stage is due to the position of verb forms in the input. This explains why agentive predicates in complement position appear with an infinitival verb form, while non-agentive predicates in head position appear with a finite verb form. Hence, inflectional morphology is not a formal feature of the learner variety at the lexical stage. It is neither used to express + vs. – modality as claimed by Ingram and Thompson (1996), nor + vs. – head movement as claimed by Poeppel and Wexler (1993), nor + vs. – finiteness as it is used in the target language. Modal predicates appear very early (see, for example, De Haan 1987, Jordens 1990, and Gillis 2003). At the initial stage, they function as lexical heads, not as auxiliary verbs. With agentive predicates, they are used to express the willingness, the ability, the permission or the obligation of the agent. As state or changeof-state predicates, they refer to a possibility, an option or a necessity. Utterance structure at the lexical stage is head-initial (see Hoekstra and Jordens 1994). In L1 Dutch, the complement of a modal or aspectual element occurs with head-final word order as in dop opdoen (cap on-do), and in L2 Dutch with head-final or head-initial word order as in huis kopen (house buy) vs. kopen een renault (buy a renault). In L2 Dutch, transfer of the L1 structure seems to interact with variation in the target language input. Utterance structure at the initial stage of acquisition can be accounted for relatively simply. Utterances are lexical projections with Spec-Head-Complement structure and head-initial word order. Syntactically, these lexical projections are complementarily distributed on the basis of the semantic feature of agentivity. Action predicates occur in the position of the complement of a modal or aspectual head with the semantic feature of control. State and change-of-state predicates occur in the position of the lexical head. In lexical structures in which an agent plays a role, the agent occurs in initial, subject position. In lexical structures in which there is no role for an agent, the theme occurs in initial, subject position.

From the lexical stage to the functional stage 

 153

Thus, a non-agentive argument may only occur in initial position if an agent is not available. At the lexical stage, functional projections are absent. Hence, topicalization does not occur. Nevertheless, learners seem to be able to provide a solution for this. They produce utterances like disse hoeniet meeneme (this-one must-not withtake; A 2;1), en dan moet teruggeven (and then must back-give; Os/T 2.3). These utterances have the object or an adverbial in initial position, while the agent is expressed implicitly with the modal predicate. At the initial stage, the lexical head carries the pragmatic function of an assertion. It validates the relation between the predicate and the subject. In agentive utterances, validation is carried by modal and aspectual predicates in the position of +Ctrl. In non-agentive utterances validation is carried by the predicate itself. The expression of validation may occur with an element attached to the head as in wilniet (want-not), kanniet (can-not), magniet (may-not), hoefniet (must-not) vs. kanwel (can-indeed), magwel (may-indeed) and in lustnie (likenot) vs. luswel (like-indeed) or integrated into the head as in nee (want-not) as opposed to ulle (want). This explains why, at the relevant stage, scrambling, as for example in ik lust pindakaas niet (I like peanut-butter not), does not occur. In sum, at the initial stage of language acquisition a representation of the learners’ grammar which focusses on morpho-syntactic phenomena such as finite vs. infinite verb forms, agreement, movement etc. seems inadequate. It deals with the acquisition data from the point of view of the properties of the target language system. That is, it uses categories of the target system that are not part of the learner’s grammar, yet. In such an approach, the learner system is dealt with not only selectively, but also incorrectly. Learner grammar as the instantiation of lexical projections, however, appears to be able to provide a full and, it seems, adequate account of the types of utterance structure which occur at the early stages of both child L1 and adult L2 Dutch.

5.6 From the lexical stage to the functional stage At the lexical stage, verb-argument structure serves the expression of both lexical-syntactic and information structure. At the functional stage, the linguistic means specifically geared towards the expression of the informational function of anchoring and contextual embedding will become available with the acquisition of the position for the category AUX. At the functional stage, AUX is the head of a functional projection. The position of AUX is referred to as ‘verb-second’. It serves the expression of semantic finiteness (S-finiteness). That is, it is a position for verb forms that are used to

154 

 The lexical stage

validate the hold-for relation between the predicate and the external argument in terms of time and place. Verb forms in the position of AUX are modal or aspectual auxiliary verbs. In the absence of an auxiliary verb, the position of AUX is taken by the lexical verb. In this position of verb-second, the lexical verb serves as the head of the functional projection of F. The lexical verb in F is marked with morphological finiteness (M-finiteness). Placement of the lexical verb in F is accounted for with ‘head movement’. With the projection of F, the SpecFP position serves the embedding of the utterance into a particular discourse situation. The constituent in SpecFP establishes the relation between the utterance and the situation that the utterance applies to. It has either topic or focus function. Placement of elements in the position of SpecFP is accounted for as ‘topicalization’ and ‘focalization’. The functional projection of F establishes a relation between, on the one hand, the functional properties of finiteness and topicalization / focalization and, on the other hand, the syntactic properties of word order variation. It explains why in the process of language development functional elements and the possibilities of word order variation are learned simultaneously.

6 The functional stage 6.1 The acquisition of the projection of F 6.1.1 Utterance structure at the lexical stage As pointed out in Chapter 3, contextual embedding in Dutch is achieved with the structural properties of the projection of F as represented in Figure 1.

SpecFP Vandaag today

FP | F′ | F (head) gaat will SpecVP ze she

complement ‖ VP | V′

complement de krant the newspaper

V (head) lezen read

Figure 1: Utterance structure of the target language system

As shown in Figure 1, evidence of FP is a syntactic position F for the expression of finiteness and a syntactic position SpecFP for elements to express the topic function. Finiteness as a category of information structure is carried by a verbal element, i.e. an auxiliary verb or a lexical verb. At the lexical stage, as shown in Chapter 5, FP is absent. Contextual embedding is achieved with the structural properties of the projection of V as represented in Figure 2. At the lexical stage, V is used to express the pragmatic function of assertion. The initial, SpecVP position is both a subject and a topic position. In agentive utterances, the agent is the subject. Hence, with an agentive predicate only the agent can occur in initial, topic position. Other elements, such as the object or an adverbial cannot. This explains why at the lexical stage, targetlike utterances as in (1), do not occur.

156 

 The functional stage

VP SpecVP

V′

V (head) complement | | Type A: agent Ctrl action Jaja mag dop opdoen. (A 2;0) J may cap on-do Type B: theme state deze magwel. (J 1;10) this may-indeed theme change of state poppie valt hier. (A 2;0) doll falls here Figure 2: Utterance structure of the basic language system in Dutch

(1)

da mag papa wel doen. (A 2;3) that may daddy indeed do broodje mag Cynthia wel opete. (A 2;4) bun may C indeed up-eat dan moet Cynthia weer make. (J 2;2) then must C again make hier mag je neus snuite. (J 2;2) here may you nose blow

Furthermore, at the lexical stage, utterances with state and change-of-state predicates occur with the theme as the subject and thus, here, the theme occurs in initial, topic position. This explains why at the relevant stage, utterances as in (2) do not occur either. (2)

koppie thee vindt papa wel lekker. (J 2;2) cup tea finds daddy indeed nice die vin Jaja gekke boom. (A 2;4) that finds J funny tree nou valt ik niet. (J 2;1) now falls I not zo doet Cynthia. (A 2;3) this-way does C

The acquisition of the projection of F 

 157

6.1.2 Towards a functional topic position Although the basic language system does not allow the kind of utterance structure as in (1) and (2), there are communicative reasons for why learners may want to use an object or an adverbial in initial, topic position. That means that the learner system has to face an internal conflict, because at the lexical stage it is either the agent or the theme that can occur in initial position. Given the constraints of the relevant system, a solution of this conflict is achieved in types of utterance as in (3) and (4). (3) Object or adverb in topic position, agent missing⁴⁸ child L1 Dutch⁴⁹

adult L2 Dutch

nee losmake. (J 1;10) dan kanniet lezen, kan begrijpen niet. (Os/T 1.7) [this] wantAGnot loose-make then canAGnot read, canAG understand not da kanwel opzitte. (J 2;0) en dan moet teruggeven. (Os/T 2.3) there canAGindeed on-sit and then [this] mustAG back-give deze magniet teke. (A 2;2) die magwel kopen. (Os/T 2.5) this mayAGnot draw that mayAGindeed buy hier moet poesje eve kamme. (A 2;3) charlie zegt: nou magwel weglopen. (Os/T 2.9) here mustAG kitty just comb C says: now mayAGindeed away-run (4) Object or adverb in topic position, theme missing child L1 Dutch nee tee, lust nie. (J 1;11) no tea, [that] like not. dit lus nie. (J 1;11) this like not. die heb ook. (J 2;1) that have too

48 The same phenomenon can be observed in L1 and L2 learners of German. In L1 German: des auchnoch rausmach (that [mustAG] also out-take. Lisa 2;0) in Dimroth et al. (2003: 80); da auch machen? (that [mustAG] also do? Caroline 2;05.31) in Winkler (2009: 23); hier kannnich raus (here canAGnot out. M. [The child is pointing to three children who are locked in a room and cannot get out]) in Clahsen (1986: 89, 112). In L2 German with L1 Polish: diese paket fahren nach polen (this parcel [mustAG] take to poland); diese schal fahren im polen (this scarf [mustAG] take in poland); das muss sagen (that mustAG say) in Dimroth (p.c.). 49 AG indicates that the agent is used implicitly with the modal predicate.

158 

 The functional stage

Utterances as in (3) and (4) are evidence of the structural possibilities at the lexical stage. That is, the initial, topic position may be taken by an (empty) object or an adverbial only in the absence of the subject. At the relevant stage, both the L1 and L2 learners use this type of utterance rather productively. It serves the communicative function of topicalization. Nevertheless, due to the fact that the basic learner system has to deal with the constraints of the lexical system, its structure is grammatically not targetlike. The data of both the L1 and L2 learners in (3) and (4) pose the question of the driving forces of the acquisition process. Given the possibilities of the basic language system with respect to its lexical and informational properties, why would learners want to give up this simple learner system in favour of a more complex targetlike system? Furthermore, given the fact that at some point they will give it up, how does this process take place? These questions may not only lead to understanding language acquisition, they may also provide insight into the way in which the system of the target language operates. As shown in (3) and (4), it is possible for utterances to occur with the object or an adverb in topic position due to the fact that there is no position for the subject. In (3), it is the use of the agentive predicate that allows the agent to play its role implicitly. Similarly in (4), the non-agentive predicate indicates that the theme is implicitly present. As pointed out in Chapter 5, at the relevant stage L1 and L2 learners of Dutch may also produce agentive utterances with the (empty) object or an adverbial in topic position, while the agent is expressed with the affix -ie (or -e, -et, -se, -’k, -je). This affix is attached to either an aspectual or a modal head. Examples are given in (5). (5) The affix as a means to express the agent⁵⁰ child L1 Dutch

adult L2 Dutch

doetie alles opete. (J 1;11) does-he everything up-eat

dan magg’k even vragen? (Os/T 2.3) then may-I just ask?

50 Similarly, child L1 learners of German produce agentive utterances with the object or an adverbial in topic position, while the agent is expressed with an affix such as -ma(n) (one) or -se (they) attached to a modal head. Examples in Dimroth et al. (2003) are: den damannich essen (that may-one-not eat. Valle 1;11); da daman aufmachen (this may-one open. Valle 1;11); des buch soll’ma / buch anguckn (the book must-one / book on-look. Valle 1;11); da hier mussen’se hin (there here must-they away. Valle 1;11).

The acquisition of the projection of F 

 159

nou gaatie weer naar huis toe gaan. (J 2;2) dan hier moette kaartje maken. now goes-he again to home to go (Os/T 2.4) then here must-we card make doense same zitte. (A 2;2) moetje deze meisje trouwen niet do-they together sit-down andere. (Os/T 2.6) must-you this girl marry not other-one moete nieuwe make, vokke. (A 2;2) dan moette daar kijken. (Os/T 2.8) must-we new-ones make, flakes then must-we there look zo moettet rije. (A 2;2) kan’k proberen, kan’k niks anders this-way must-it drive doen. (Os/T 3.6) can-I try, can-I nothing else do At the lexical stage, utterances as in (5) are possible because with the affix there is no need for a structural position for the agent. Hence, the utterance structure does not violate the word order restrictions at the lexical stage. Given the absence

VP V′ SpecVP

V (head)

complement

topic

Ctrl mod-affix[AG] do-affix[AG] ga-affix[AG]

action

[] 0i

maggik may-I

wel hebbe. (J 2;2) ei indeed get

pop dolli

wildie wants-he

mee spele. (J 2;2) ei with play

[] 0i

doetie does-he

viesmake. (A 2;1) ei dirty-make

[] 0Adv

doetie does-he

alles opete. (J 1;11) everything up-eat

nou now

gaatie will-he

weer naar huis toe gaan. (J 2;2) again to home to go

Figure 3: Topicalization at the lexical stage

160 

 The functional stage

of a position for the agent, the initial position is available as a topic position for the object or an adverbial. At the lexical stage, utterances with an agentive predicate, an affix referring to the agent and a topic position are used as a structural device to express the pragmatic function of topicalization. The relevant structure is given in Figure 3. As illustrated in Figure 3, the agentive predicates are used with lexical elements expressing the semantic function of control. That is, predicates of causal action are used with a modal or a form of doen (do), while predicates of agentive motion are used with a form of gaan (go). Utterances as in (3), (4) and (5) are evidence of a creative solution given the constraints of the learner system at the relevant stage. However, the solution not to provide a position for the agent or the theme is not targetlike. A targetlike solution for the topicalization of objects and adverbials will be reached as soon as the learner system finds a way to reanalyse the affix as a regular pronoun. As shown in Figure 2, learner utterances at the lexical stage can be accounted for in terms of two types of utterance structure: type A and type B. The difference between type-A and type-B structures is based on the opposition between utterances with predicates that are either agentive or non-agentive. The type-A structure has a head position for an element to express control, a complement position for an agentive predicate and a specifier position for the agent. The type-B structure has a head position for a non-agentive predicate and a specifier position for its theme. The utterances with an agentive predicate and the object or an adverbial in topic position as in (3) are not targetlike, because the agent is left implicit. Relevant for the process of language development at this stage is the lexical predicate in the position of the complement. In utterances of type A such as Jaja mag dop opdoen (J may cap on-do. A 2;0), opdoen (on-do) serves as the lexical head of the projection of V, i.e. with both an agent in SpecVP and an object as its complement. Reanalysis of the complement in (3) as the full projection of V leads to the re-establishment of a structural position for the agent. Integration of this full projection of V at the position of the complement causes a process of reinterpretation by which the head-initial modal V is reinterpreted as a modal auxiliary verb, while the initial, specifier position is reinterpreted as a topic position. The developmental process to achieve this process of integration and reinterpretation is represented graphically in Figure 4. The utterance in (a) has an object (deze) in topic position, the modal predicate (kan) as its head and an action predicate (meenemen) as its complement. As pointed out, at the lexical stage, utterances as in (a) entail an internal conflict between principles of syntax and pragmatics. Syntactically, the agent should occur in initial position, for pragmatic reasons however, the object is. Given that

The acquisition of the projection of F 

(a)

object

V (head)

complement

dezei this

kan can

ei meenemen ei with-take (b) agent V (head) ik I

(c)

kan(wel) can-indeed

 161

complement deze meenemen this with-take

topic (d) Aux

(e)

topic

Aux

[ agent V (head)

deze thisi

kan canj

ik I

(wel) (indeed) ej

complement ] ei

meenemen with-take

Figure 4: Integration of the type-A structure

the object occurs in first, i.e. topic position, this conflict is solved by the fact that the agent of the action predicate is left unexpressed. The structure in (b) is the agentive type of utterance structure at the lexical stage. Integration of this structure at the position of the predicate meenemen in (a) provides a position for the missing agent. Due to the fact that the structure in (b) has both a position for an agent (ik) and an object (deze), a reinterpretation of grammatical functions takes place. As a result, the initial position is reanalysed as a topic position as in (c), while the second position becomes a position for the auxiliary as the carrier of the pragmatic function of assertion (d). The result of this process of integration and reinterpretation is the structure in (e). In (e) the conflict between the principles ‘Agent first’ and ‘Topic first’ is solved with the creation of two types of ‘first positions’. A functional, first position sentence-initially for an element with topic function and a semantic, first position for the agent sentence-internally. The way in which at the relevant stage a simple, lexical system develops into a fullyfledged, discourse-functional system is how language development proceeds. As Perdue puts it: “Learner utterances show organizational regularities right from the beginning of the acquisition process, and these early organizational principles do not disappear, but rather interact with new organizational principles as and when they are acquired” (Perdue 2006: 864).

162 

 The functional stage

6.1.3 Towards a grammatical subject position The data in (6), (7) and (8) are evidence of the learner system at the developmental stage at which the utterance structure has both a topic and a subject position. (6) Object topic and agent affix / pronoun child L1 Dutch

adult L2 Dutch

die mag je nog hebbe. (J 2;2) that may you also get die mag ik lekker opete. (J 2;2) that may I nicely up-eat die wilt ik hebbe. (J 2;2) that want-I get da mag je ook mee hebbe. (J 2;2) that may you also with get mag jij lekker opete mette ei. (A 2;2) [this] may you nicely up-eat with-the egg Ruti nome hemmik. (A 2;2) R taken have-I heb ik oppegete. (A 2;2) [this] have I up-eaten die hem ik van Jasmijn kege. (A 2;4) that-one have I from J got

moet jij vragen. (Os/T 2.8) [this] must you ask kan hij helemaal vergeten. (Os/T 3.2) [this] can he totally forget moet jij zelf betalen. (Os/T 3.2) [this] must you yourself pay kan ik proberen. (Os/T 3.6) [this] can I try kan jij kopen. (Mo/A 3.2) [this] can you buy die heb ik hier gedaan. (Mo/A 3.5) that have I here done ‘n klein flat, die wil ik niet (Mo/A 3.6) a small flat, that want I not [get] dat heb ik nooit gezegd. (Mo/A 3.8) that have I never said

As noted in Chapter 2, Section 2.6, the data in (6) are transcriptions of utterances taken from collections of spontaneous oral production. Therefore, it is difficult to decide whether the element referring to the agent is an affix or a pronoun. However, as soon as it is a pronoun, the learner system must have created an argument position for the subject. Evidence showing unambiguously that there is a non-initial, subject position for the agent, is the use of an agent NP, a stressed pronoun, or jij (you) instead of je (you) and ikke (I) instead of ik (I) in utterances as in (7). (7) Object topic and agent pronoun / NP in L1 Dutch heef Cynthia maakt. (J 2;0) [this] has C made mag jij opete. (J 2;1) [this] may you up-eat

The acquisition of the projection of F 

 163

die heef mama maakt. (J 2;1) that-one has mommy made mag jij hebbe. (J 2;2) [this] may you get die mag boze wolf niet potmake, de muts. (J 2;2) that may bad wolf not kaput-make, the cap mag poekie niet meer aankome. (J 2;2) [this] may kitty not anymore touch die heef Cynthia gemaakt. (J 2;2) that has C made mag jij lekke opete mette ei. (A 2;2) [this] may you nicely up-eat with-the egg moete mammie ook kope. (A 2;2) [this] must mommy also buy mag IK doen. (A 2;3) [this] may I do da mag papa wel doen. (A 2;3) that may daddy indeed do broodje mag Cynthia wel opete. (A 2;4) bun may C indeed up-eat Utterances showing that the theme occurs in non-initial, subject position are given in (8) (8) Object topic and theme in L1 Dutch dit heeft Cynthia. (J 2;0) this has C. die hebbe wij. (J 2;1) this have we lust Tompoes die? lust Tompoesje wel. (J 2;1) likes T that-one? [that] likes T indeed weer ponypaard heb ik. (J 2;2) again pony-horse have I koppie thee vindt papa wel lekker. (J 2;2) cup-of tea finds daddy indeed nice die wilt ik. (J 2;2) that-one want I vogel hoort ik. (A 2;2) bird hear I

164 

 The functional stage

hoog ben ik. (A 2;2) high am I die vin Jaja hekker, fokomel. (A 2;3) that finds J nice, chocolate-milk zefde heef Jaja. (A 2;3) same has J da wus ik wel. (A 2;3) that like I indeed die geef Jaja s(tr)aks aan Jasmijn. (A 2;3) that gives J in-a-minute to J In sum, the data in (6), (7) and (8) are evidence of an utterance structure with positions for both a topic and a non-initial subject. As in the target language, they show that the position of the topicalized object may be left empty.

6.2 Evidence of the projection of F 6.2.1 The functional topic position The topic position is a functional position. It is not used as a position for the expression of a particular semantic role. Evidence that at the relevant stage the initial position has become a topic position is the fact that not only objects as in (6), (7) and (8), but also adverbs may occur in this position. Examples are given in (9) and (10). (9) Adverbial in topic and agent in non-topic position child L1 Dutch daa kan ik niet meer lope. (J 2;2) there can I not anymore walk

adult L2 Dutch

dan moet jij helemaal doorrijen. (Os/T 2.8) then must you fully further-drive nou gaatie weer naar huis toe gaan. (J 2;2) dan moet ik huis zoeken. (Os/T 3.6) now will-he again to home to go then must I house look-for hier mag je neus snuite. (J 2;2) dan kan ik niks anders doen. (Os/T here may you nose blow 3.6) then can I nothing else do dan moet Cynthia weer make. (J 2;2) hier kan ik niet lezen, hè? (Mo/A 2.8) then must C again make here can I not read, ok? nou mag Jaja peenie in. (A 2;2) dan moet ik trouwen. (Mo/A 3.6)

Evidence of the projection of F  

now may J pacifier in [do] zo kan Jaja niks zien. (A 2;3) this-way can J nothing see zo kan ikke Jaja wel niks zien. (A 2;3) this-way can I J indeed nothing see da mag papa wel zitte van Jaja. (A 2;3) there may daddy indeed sit-down of J nou mag papa weer teke. (A 2;4) now may daddy again draw hier wilt Jaja ook denkik naa toe. (A 2;4) here wants J also think-I to [go] hier kan ikke op s[t]aan. (A 2;4) here can I on stand-up

 165

then must I marry

(10) Adverbial in topic and theme in non-topic position child L1 Dutch da ben ik weer. (J 2;0) there am I again

adult L2 Dutch

eerste dieping wonen andere mensen. (Os/T 2.5) first floor live other people daar zit Pino. (J 2;0) woon jij in tilburg of? (Os/T 2.6) there sits P live you in tilburg or? da moet schaar. (J 2;1) en daar staat een winkel. (Os/T 2.9) there must scissors and there stands a shop daar zit jij? (J 2;1) op dat … staat mooie asbak. (Os/T 3.3) there sits you? on that … stands nice ashtray waar is snoepdonald? (J 2;2) in tas zit krant. (Os/T 3.3) where is candy-donald? in bag sits newspaper nou bent mama weer terug. (J 2;2) in el jadida woont goeje mensen. (Mo/A 1.8) now am mommy again back in el jadida lives good people da ga ik toe bij papa. (A 2;2) en dan blijft die jongen … (Mo/A 2.3) there go I to with daddy and then stays that boy da wone wij. (A 2;2) waar staat die kindje? (Mo/A 2.6) there live we where stands that child? waars vokkik nou mij? (A 2;3) toen loopt een man. (Mo/A 3.2) where’s fork now me? then walks a man da valt stoel wel om. (A 2;3) waar woon jij? (Mo/A 3.7) there falls chair indeed flat where live you? hier rijtte auto’s? (A 2;4) omdat daar zit mij vrienden. (Mo/A 3.7) here drive cars? because there sits my friends

166 

 The functional stage

zo doet Cynthia. (A 2;4) so does C

toen zit achter hem man. (Mo/A 3.9) then sits behind him man

Utterances as in (9) and (10) are used all of a sudden. They show that the language system has created a functional topic position for adverbials and a lexicalsemantic position for an agent in (9) or a theme in (10). A representation of the grammatical structure underlying the evidence in agentive utterances as in (6), (7) and (9) is given in Figure 5. It shows that the utterance structure at the relevant stage is the result of the interaction between three components of grammatical structure (1) the full projection of F, (2) the full projection of V and (3) a complex predicate (Pred′) serving as the complement of V. F is the head of FP, V is the head of VP and Pred is the head of the lexical entity Pred′. ScP is the position of a scope particle as, for example, niet (not), wel (indeed), ook (also), weer (again) etc. The representation in Figure 5 shows that the lexical projection VP serves as the complement of F, while Pred′ is the complement of V. Furthermore, there is a ‘hold for’-relation between the head-complement structure and its specifier. Figure 5 accounts for the fact that at the relevant stage learner utterances as in (6), (7) and (9) have both a position SpecFP for a topic constituent as part of a functional projection FP and a position SpecVP for the agent as the subject of a lexical projection VP. NPs in topic position receive their semantic interpretation from the projection of V. With causal action verbs, an empty object position as ei in Figure 5 (a) accounts for the semantic interpretation of the topic element as the object of the action. An empty subject position as ei in Figure 5 (b) accounts for the semantic interpretation of the topic element as the agent of the action. Elements in the position of F are used as carriers of finiteness. In Figure 5, it is the modal verb that serves this function. Its position ej as the head of VP is empty. This empty V position accounts for the semantic interpretation of the modal in F as a deontic modal, which means that it is used to express the volition, ability, permission or obligation of an agent to perform an action.⁵¹ Finally, note that in Figure 5 both the functional projection FP and the predication VP occur with head-complement structure, while the complex predicate Pred′ which serves as the complement of V has complement-head structure. It

51 In the target language, it is possible for aw epistemic modal to occur in F (head), while a deontic modal occurs in V (head). Examples are ik moet dat wel kunnen doen (I must that indeed can do), ik kan het moeten doen (I can it must do), dat moet hij niet mogen doen (that must he not may do).

Evidence of the projection of F  

 167

(1) FP SpecFP ↑ topic

←

[ F (head) + complement ] F′ ‖ (2) VP SpecVP ← ↑ subject

a.

b.

c.

topicOBJ-i Modj

subject

broodje bread

(+ ScP) [ V (head) + complement ] V′ ‖ (3) Pred′ | (complement) + Pred ej

ei – Pred[CAUSAL ACTION]

Cynthia wel C indeed

ej

opete ei – up-eat

topicSUBJ-i Modj

ei

ScP

ej

NP – Pred[CAUSAL ACTION]

Cynthia Ci

mag mayj

ei

wel indeed

ej

broodje opete bread up-eat

topicADV daa there

Modj kan can

subject ik I

ScP niet meer not anymore

mag mayj

ScP

ej ej

Pred[AGENTIVE MOTION] lope walk

FP = (functional) projection of F VP = (lexical) projection of V Pred′ = complex predicate F′ = functional head + complement V′ = lexical head + complement Pred′ = complement + Pred ← = ‘hold for’-relation ScP = scope particle Figure 5: Utterance structure with FP and agentive VP

shows that in Dutch, the head-complement structure in syntax is established differently from the head-complement structure in the lexicon. In sum, utterances as in (3) play a key role in the acquisition of the functional projection FP. They occur with an initial, topic position and no structural position for the subject. The creation of a subject position for the agent as in (6), (7) and (9), is due to fact that the learner grammar has established two head positions,

168 

 The functional stage

a functional head position for the projection of F and a lexical head position for the projection of V. With the functional projection FP and the lexical projection VP, utterance structure accommodates a division of labour. The functional projection FP serves to express the pragmatic properties of information structure, the lexical projection VP serves to express the syntactic properties of verb-argument structure.

6.2.2 The functional category F The representation in Figure 5 shows that, due to the development of the agentive predicate into a full projection of V, the grammatical function of the modal predicate (Mod) in second position has changed. Initially, this modal predicate served as the head of a lexical projection. Reanalysis turned it into the head of the functional projection FP. Evidence of this is the fact that the set of elements that used to occur as instantiations of the modal predicate at the lexical stage has changed significantly. As shown in Hoekstra and Jordens (1994), at the functional stage, unanalysed lexical modal expressions have disappeared all of a sudden. That is, due to the acquisition of the functional category F, phrasal expressions such as ulle (want), mag-ikke (may-I) or nee (want-not), kanwel (can-indeed), kanniet (can-not) or handigniet (handy-not), magwel (may-indeed) or magniet (may-not) and moe(t)nie(t) (must-not) or hoefnie (must-not) are no longer present. Evidence of this sudden, qualitative change is found not only in child L1 Dutch but also in adult L2 Dutch. Learners appear to give up ulle (want), mag-ikke (may-I) and nee (want-not) in favour of wil(t) (want) and wil niet (want not). The phrasal expressions kanwel (can-indeed) and magwel (may-indeed) become analysed as kan (…) wel and mag (…) wel, and kanniet (can-not), magniet (may-not), moe(t)nie(t) (must-not) and hoefnie (must-not) become analysed as kan (…) niet, mag (…) niet, moet (…) niet and hoef(t) (…) niet. The examples in (11) and (12) illustrate the distributional differences in use between the unanalysed modal expressions at the lexical stage and their analysed counterparts at the functional stage.

6.2.2.1 Analysis of the modal expressions ‘ulle’, ‘mag-ikke’ and ‘nee’ With the acquisition of F in child L1 Dutch, unanalysed ulle, mag-ikke and nee have disappeared. The examples in (11) show that they have been given up in favour of wil, wilt and wil niet with a typically deictic subject, i.e. the subject pronouns ik, ikke (I) and je (you) and proper nouns such as Poes, Pino, mama and Jaja.

Evidence of the projection of F  

 169

(11) The analysis of ulle, mag-ikke and nee in child L1 Dutch The lexical stage

The functional stage

ulle tore make. (J 1;10) I-want tower make ulle ook zitte, stoel. (J 1;10) I-want also sit-down, chair nee poesje deze hebbe. (J 1;10) I-want-not kitty this get ulle Cynthia bed ligge. (J 1;11) I-want C bed lie ulle evetjes dit doen. (J 1;11) I-want just this do nee afdoen. (J 1;11) I-want-not off-do mag-ikke paartie rije?(A 2;0) may-I horsie ride? papa, nee opete. (A 2;0) daddy, I-want-not up-eat mag-ikke ijssie hebbe? (A 2;1) may-I ice-cream get? mag-ikke Mijnie wakker make? (A2;1) may-I M awake make? mag-ikke ook bank zitte? (A 2;1) may-I also couch sit? nee bad zitte. (A 2;1) I-want-not bath sit-down

ik wilt opslaan. (J 2;0) I want on-hit Poes wil kijke naa boter. (J 2;0) kitty wants look at butter wil niet oppe straat lope, Pino. (J 2;0) wants not on-the street walk, P wil je’s opedoen? (J 2;1) want you-just open-do? die wilt ik hebbe. (J 2;2) that-one want I get mama wilt zo eve kijke. (J 2;2) mommy wants so just look Jaja wilt jou kijke. (A 2;2) J wants you look ikke wil deur dicht make. (A 2;3) I want door closed make ik wilt óok plante water geve. (A 2;3) I want too plants water give Poesje wilt op trap zitte. (A 2:3) kitty wants on stairs sit nou wilt Jaja naarre kippetjes kijke. (A 2;4) now wants J to-the chicken look Poekie wíl nie kome. (A 2;4) kitty wants not come

A survey of the L1 Dutch data shows that, with the sole exception of ik wil melluk pakke (I want milk get. J 1;11), examples with wil, wilt, wil niet do not occur at the lexical stage, i.e. neither in Jasmijn (1;10–1;11) nor in Andrea (2;0–2;1).

6.2.2.2 Analysis of the unanalysed modal expressions At the lexical stage, unanalysed modal expressions such as kanwel (can-indeed) and kanniet (can-not), magwel (may-indeed) and magniet (may-not), moe(t)nie(t) (must-not) and hoefnie (must-not) are commonly used as the lexical head of a VP. They are used to express the willingness, the ability, the permission or the obligation of the agent to carry out a particular action. With the acquisition of the functional category F, these lexical entities are analysed in a targetlike manner,

170 

 The functional stage

i.e. as carriers of the functional properties of finiteness. As a consequence, the elements wel (indeed) and niet (not) are going to function in a targetlike fashion, too, i.e. as particles that are used with scope over the predication. Examples that are illustrative for this process of restructuring are given in (12). (12) Reanalysis of unanalysed modal expressions The lexical stage

The functional stage

child L1 Dutch kanniet pakke zelf. (J 1;11) can-not get myself magwel poesje opete. (J 1;11) may-indeed kitty up-eat kanwel papa zitte. (A 2;1) can-indeed daddy sit-down da kanniet zitte. (A 2;1) there can-not sit-down

da kan ik niet meer lope. (J 2;2) there can I not anymore walk mag ik wel hebbe. (J 2;2) [that] may I indeed get dà mag papa wel doen. (A 2;3) that may daddy indeed do mag ik eve wel deze aankome? (A 2;3) may I just indeed this touch?

adult L2 Dutch die magwel kope. (Os/T 2.5) that may-indeed buy ik magwel alles. (Os/T 2.6) I may-indeed everything [do] kanniet lezen op school. (Mo/A 1.4) can-not read at school zij ook magniet gaan bij ouders. (Mo/A 2.1) she too may-not go with parents

ja, mag ik wel. (Os/T 3.1) yes, [that] may I indeed [do] dan kan ik wel. (Os/T 3.4) then can I indeed [do] hier kan ik niet lezen. (Mo/A 2.8) here can I not read hoeft toch niet? (Mo/A 3.5) [it] must indeed not?

6.2.2.3 Epistemic modals The acquisition of the functional projection FP makes it possible for epistemic modals to serve as carriers of finiteness. The difference between deontic and epistemic modals in adult L1 Dutch is illustrated in (13). (13) Deontic and epistemic modality deontic

epistemic

De monteur kan het voor je maken. the fitter can it for you repair

De baby kan eruit vallen. the baby can it-out fall

Evidence of the projection of F  

Je moet deze brief nog op de post doen. you must this letter still on the mail put Hij mag geen zwaar werk doen. he may no heavy work do

 171

Je moet om 8 uur in de stad zijn. you must at 8 o’clock in the town be Hij mag niet achterblijven. he may not back-stay

Deontic modals serve as the head of the lexical projection VP. They require an agentive predicate as their complement and are used to express the willingness, the ability, the permission or the obligation of the agent to carry out an action. Epistemic modals are the head of the functional projection FP. They may be used with both an agentive and a non-agentive VP as their complement and are used to express that an event or a situation is possible, optional or necessary. With the acquisition of F, modal verbs can be used not only as deontic modals with predicates of (causal) action and agentive motion, but also as epistemic modals with state and change-of-state predicates. Examples are given in (14). (14) Deontic and epistemic modality at the functional stage deontic

epistemic

child L1 Dutch ik mag niet zelf pakke chocola. (J 2;1) I may not self get chocolate ik kanne bij pakke. (J 2;2) I can it-with get da mag Jaja wel mee peje. (A 2;3) there may J indeed with play Jaja moet Ruudje pakke. (A 2;3) J must R get

ik kan niet Herrie vinde. (J 2;0) I can not H find kannie niet staan. (J 2;1) can-he not stand auto mag niet kome. (A 2;2) car may not come zo kan Jaja niks zien. (A 2;3) this-way can J nothing see

adult L2 Dutch dan moet ik huis zoeken. (Os/T 3.6) then must I house look-for dan kan ik niks anders doen. (Os/T 3.6) then can I nothing else do hier kan ik niet lezen, hè? (Mo/A 2.8) here can I not read, ok?

dan moeten jullie wachten. (Os/T 2.7) then must you wait dan ga ik mijn familie wonen. (Os/T 3.6) then go I my family live tot hoelang moet ik hier blijven. (Mo/A 2.9) till how-long must I here stay

172 

 The functional stage

6.2.2.4 The auxiliary verbs ‘heb, heeft’ and ‘ben, is’ With the acquisition of F, the learner grammar has a functional position for the expression of finiteness as it occurs in the target language. This explains why, simultaneously with the use of both deontic and epistemic modals as in (14), learners all of a sudden appear to be able to use the auxiliary verbs heb, heeft (have, has) and ben, is (am, is) in this position, too. In the acquisition process, these auxiliary verbs are the first words without lexical meaning. They provide the clearest evidence that the learner has reached the functional stage of acquisition. In the target language, the auxiliary verbs heb, heeft and ben, is occur in utterances with past participle structures. At the lexical stage however, there is no position for F and, thus, no possibility for these auxiliary verbs to be used. This explains the distributional difference between the use of past participle structures with no auxiliary at the lexical stage and past participle structures with an auxiliary at the functional stage. Evidence from child L1 Dutch is given in (15). (15) Utterances with past participles in child L1 Dutch The lexical stage

The functional stage

bal weg. topt. (J 1;10) ball gone. hidden dit Cynthia maakt. (J 1;10) this C. made poppie haartie wast. (J 1;10) doll hair washed dit Cynthia weest. (J 1;10) this C been poes opgete. (J 1;11) kitty up-eaten ikke ook boot hees. (A 2;0) I too boat been Jaja kamd. (A 2;0) J combed papa potmaakt. (A 2;0) daddy kaput-made mama lekker aapt? (A 2;1) mommy nice slept? jou hege dees. (A 2;1) you got this aap goonmaakt. (A 2;1) monkey clean-made

ikke hè dit pakt. (J 2;1) I have this got heb je visje gehad? (J 2;1) have you fish had? ik heb wonne. (J 2;1) I have won ik heef afspoeld. (J 2;2) I have washed die is altijd opde televisie geweest. (J 2;2) that-one is always on tv been Jaja hemme al goonmaakt. (A 2;2) J has already up-cleaned hemme nogge fippo vonne? (A 2;2) have-we another flippo found? ikke hemme deze tekend. (A 2;3) I have this drawn da ben ikke ook wees. (A 2;3) there am I too been Jaja heef met de haartjes zo doet. (A 2;4) J has with the hairs so done. ik heef óók appel gete. (A 2;4) I have too apple eaten

Evidence of the projection of F  

 173

The examples in (15) by Jasmijn (1;10–1;11) and Andrea (2;0–2;1) are evidence of the lexical stage. At the relevant stage, utterances with a past participle occur without an auxiliary verb. The examples in (15) by Jasmijn (2;0–2;2) and Andrea (2;2–2;4) are evidence of the functional stage. At the relevant stage, utterances are produced with a past participle and an auxiliary verb. The auxiliary verbs heb, heeft and ben, is are the first linguistic elements without lexical meaning. Evidence from adult L2 Dutch is given in (16). (16) Utterances with past participles in adult L2 Dutch The lexical stage

The functional stage

hij zegt okee, jij goed gewerkt. (Os/T 2.1) he says ok, you good worked ik ook getrouwd, tien. (Os/T 2.1) I too married, ten vandaag school ook afgelopen. (Os/T 2.1) today school too ended en dan vier uur geslapen. (Os/T 2.3) and then four hours slept dan helemaal kapot gemaakt. (Os/T 2.5) then totally kaput made politie zien alles wat hij gedaan. (Os/T 2.6) police see everything what he done och man ik vorige week gekocht. (Os/T 2.8) oh man I last week bought hij zegt hij niet veel slaap. (Mo/A 1.3) he says he not much slept ik uh geboren bij de las cotes. (Mo/A 1.6) I eh born at the las cotes dan hij un brossis gedaan polices. (Mo/A 1.9) then he a summon done police ja twee dagen ik niet werk. (Mo/A 2.1)

ik heb gehoord de film. (Os/T 2.8) I have heard the movie eten is afgelopen. (Os/T 2.9) dinner is ended ik heb dokter geweest. (Os/T 3.1) I have doctor been hij heeft rijbewijs haald. (Os/T 3.1) he has drivers-license got wie heeft verzekeringen maken? (Os/T 3.3) who has insurances made? ik heb naam vergeet van de winkel. (Os/T 3.4) I have name forgotten of the shop wij hebben fabriek opgebeld. (Os/T 3.4) we have factory called ik heb beetje geld sparen. (Mo/A 2.2) I have some money saved toen hij heeft een ongeluk gehad. (Mo/A 2.9) then he has an accident had vrijdag nacht ik heb niet geslapen. (Mo/A 3.5) friday night I have not slept ik heb daar geslapen buiten. (Mo/A 3.6)

174 

 The functional stage

yes two days I not worked ik toen die auto gekocht. (Mo/A 2.2) I then that car bought

I have there slept outside ik heb jou vorig keer niet gezien. (Mo/A 3.8) I have you last time not seen

The examples in (16) by Osman/T (2.1–2.8) and Mohammed/A (1.3–2.2) are evidence of the lexical stage. At the relevant stage, utterances with a past participle occur without an auxiliary verb. The examples in (16) by Osman/T (2.8–3.4) and Mohammed/A (2.2–3.8) are evidence of the functional stage. At the relevant stage, utterances are produced with a past participle and an auxiliary verb. As in child L1 Dutch, the auxiliary verbs heb, heeft and ben, is are the first linguistic elements without lexical meaning. The acquisition of the auxiliary verbs heb, heeft and ben, is is a developmental process that takes place rather suddenly. This is shown in Table 1 and Figure 6. The figures in Table 1 show the frequency with which past participle forms occur both in Jasmijn from (1;10) to (2;2) and in Andrea from (2;0) to (2;4). The past participle forms are categorized in terms of the contexts in which they appear: without or with an auxiliary verb, i.e. as Vpp or Aux + Vpp.

Jasmijn (1;10–2;2)

1;10 1;11 2;0 2;1 2;2

Vpp 23 32 21 7 2

Andrea (2;0–2;4) Aux + Vpp – = .00 2 = .06 2 = .09 5 = .42 11 = .85

Total 23 34 23 12 13

2;0 2;1 2;2 2;3 2;4

Vpp 19 13 25 8 1

Aux + Vpp – = .00 1 = .07 12 = .32 24 = .75 17 = .94

Total 19 14 37 32 18

Table 1: The acquisition of the auxilary verbs heb, heeft and ben, is in child L1 Dutch

The data in Table 1 are represented graphically in Figure 6. The representation of the relevant data in Table 1 and in Figure 6 shows that initially both children produce past participles without an auxiliary verb most of the time. Evidence of a sudden increase in the use of past participles with auxiliaries is shown in Jasmijn after (2;0) and in Andrea after (2;3). The auxiliaries heb, heeft and ben, is are functional elements. Hence, they do not occur at the lexical stage. However, when past participle forms are used with the auxiliary heb or heeft, as in heb or heeft gekregen (have, has gotten), or less frequently with ben or is, as in ben or is gevallen (am, is fallen), the language

Evidence of the projection of F  

Jasmijn (1;10–2;2).

 175

Andrea (2;0–2;4) 100

90

90

80

80 70

70 60

60

50

Aux + Vpp

40 30

50

Aux + Vpp

40

20

30 20

10

10 0

0 1;10 1;11 2;0

2;1

2;2

2;0 2;1 2;2 2;3 2;4

Figure 6: The acquisition of the auxilary verbs heb, heeft and ben, is in child L1 Dutch

system has changed.⁵² Utterance structure has developed the functional position F as the head of FP. As noted above, it is the acquisition of the functional projection FP that marks the change from the lexical system to the functional system. Once the structural properties of FP have been established, they hold for every utterance. It explains the abruptness of the process as it appears from the figures in Table 1 and Figure 6 The developmental process in L1 acquisition can also be observed with the adult L2 learners. Here, too, this process takes place rather suddenly. This becomes apparent from the figures in Table 2. These figures show the type/token ratios of the past participles that are used with the auxiliary verb heb, heeft.

52 A past participle can be preceded by a form of the auxiliary verb hebben (have), i.e. heb, heeft (have, has), or zijn (be), i.e. ben, is (am, is). However, due to the fact that the use of ben, is is restricted to a relatively small category of intransitive verbs, it occurs rather infrequently. The available data are in line with what has been found for the use of the auxiliary heb, heeft. That is, at the lexical stage, Jasmijn and Andrea produced utterances in which the auxiliary is is typically absent. Examples are: chicke little valle (chicken little fallen. J 1;10), dit Cynthia weest (this C been. J 1;10), Nog niet? Afberope? (not yet? finished. A 2;0), Jaja óók gijbaan hees (J too slide been. A 2;0). At the functional stage, both Jasmijn and Andrea produced utterances with the auxiliary is, even with the same participles they used at the lexical stage. Examples are: die is altijd opde televisie geweest (that-one is always on tv been. J 2;2), nou’s weer aflope (now is again finished. J 2;2), da issie varre (there is-he fallen. A 2;2), die isse Jaja gete (that is J forgotten. A 2;4), Ruudje is uittefaapt (R is finished-slept. A 2;4), Jaja is zelf naa bove gaan naa papa (J is herself upstairs gone to daddy. A 2;4), isse Barnies affehope, mam? (is B finished, mommy? A 2;4). The same is true for adult L2 learners of Dutch. At the lexical stage, the auxiliary ben is absent as in ik niet trouwd. geen kinderen (I not married. no children, Os/T 2.3). At the functional stage, it occurs as zijn, is in ze zijn vandaag niet gewerkt (they are today not worked. Os/T 3.5), hier vrouw is vald (here woman is falled [= fallen]. Mo/A 3.2).

176 

 The functional stage

Osman/T cycle 1.1–1.5 1.6–2.0 2.1–2.5 2.6–3.0 3.1–3.5

Mohammed/A heb(t)+Vpp – –  2/2  6/7 17/43

heeft+Vpp – – – 6/11 → 2.7–2.8 6/10

cycle 1.6–2.0 2.1–2.5 2.6–3.0 3.1–3.5 3.6–3.9

heb(t)+Vpp – 11/15  7/15 19/48 19/38

heeft+Vpp – – → 2.1–2.2 10/34 11/25 20/40

→ = establishment of F Table 2: The acquisition of the auxiliary verb heb, heeft in adult L2 Dutch

Given that type/token ratios are a measure of productivity, the relevant figures show that in both adult learners the spontaneous acquisition of the auxiliary heb, heeft occurs within a limited period of time, i.e. within weeks rather than months. For Osman/T this is between Cycle 2.7 and 2.8, while for Mohammed/A this is between Cycle 2.1 and 2.2. To conclude, the use of the auxiliaries heb or heeft and ben or is is evidence that the learner grammar has the functional position F. The fact that such a position exists explains why heb, heeft and ben, is are immediately used productively. Of course there is the phenomenon of backsliding. That is, under particular circumstances learners may occasionally fall back onto the grammatical system of an earlier stage of development. However, the point is that a learner grammar with the structural position F does not discriminate between types of lexical verb that the auxiliary verbs are used with. In other words, there is no learner grammar that uses the auxiliary verbs heb, heeft or ben, is with some past participles, while with others it does not. The structural position F, once established, holds across the board. While in the developmental process of the acquisition of F, utterances with Vpp will be given up in favour of Aux + Vpp, utterances with Vinf will be given up in favour Mod + Vinf. The relevant process can be observed as it takes place in Jasmijn (2;0) and Andrea (2;3). At this point in acquisition, both children may vary between using Vpp and Vinf, on the one hand, and Aux + Vpp and Mod + Vinf, on the other. Even with the same lexical verbs, they use both options. Examples of these are given in (17).

Evidence of the projection of F  

 177

(17) Variation in child L1 Dutch: Vpp, Vinf vs. Aux + Vpp, Mod + Vinf Vpp, Vinf

Aux + Vpp , Mod + Vinf

Jasmijn (2;0) rommel maakt. mess made anne boek pakke. other book get opemake, danoontje. open-make, danoontje ik aa Cynthia geve. I to C give Daphnie ook uitkijke. D also out-look

heef Cynthia maakt. has C made ‘k ga boter pakke. I will butter get doe je Pino make? do you P make? mag ik Tompoes geve? may I T give? poes wil kijke na boter. kitty will look to butter

Andrea (2;3) papa ook doen? daddy also do? papa Jaja same passe daan. daddy J together pee done papa ook wanneke. daddy too walk mama eve opzoeke. mommy just for-look papa eve torentje make. daddy just tower make Jaja eve make? ja, maakt. J just make? yes made kijke es, vonne. rietje vonne. look, found. straw found

dà mag papa wel doen. that may daddy indeed do dà heefe Jaja daan. that has J done papa, gaan wanneke. daddy, [we] will walk Mijnie doet barretje zoeke. M does ball look-for Jaja gaat liedje make van jou. J will song make of you Jaja hém al goonmaakt. J has already clean-made ikke heef Bert vonne. I have B found

To summarize, in both child L1 and adult L2 learners of Dutch, the use of auxiliary verbs is evidence of the acquisition of F. The functional category F is established as the result of a reanalysis of modal and aspectual elements that initially occur in lexical head position. As soon as F has been established, auxiliary and modal verbs come to be used with all the lexical verbs that at an earlier stage were used as past participle verbs or as infinitives alone.

178 

 The functional stage

6.2.2.5 The auxiliary verbs ‘doe, doet’ and ‘ga, gaat’ The acquisition of the functional position F can also be observed with the use of phrasal elements such as doetie (does-he) and gaatie (will-he). At the lexical stage, i.e. in Jasmijn (1;10–1;11) and Andrea (2;0–2;1), doetie and gaatie are used as unanalysed elements in lexical head position to express the semantic function of control. At the functional stage, i.e. in Jasmijn (2;0–2;2) and Andrea (2;2–2;4), these elements start to be used as auxiliary verbs serving as carriers of finiteness. This explains why they become analysed. A few examples are given in (18). (18) Utterances with doe(t) and (gaa(t) in child L1 Dutch The lexical stage

The functional stage

doetie huile. (J 1;11) does-he cry doetie omdraaie. (J 1;11) does-he over-turn deetie aflope. (J 1;11) did-he off-walk doetie bijte. (J 1;11) does-he bite doetie alles opete. (J 1;11) does-he everything up-eat doetie nou? opzitte kruk. (J 1;11) does-he now? on-sit-down stool doetie renne. (J 1;11) does-he run doetie kiekeboe. (J 1;11) does-he peekeboo [do] doetie berg op. (J 1;11) does-he mountain on [go]

doe je Pino make? (J 2;0) do you P make? ik doet neusje snuite. (J 2;0) I does nose blow ik gaat Pino make. (J 2;0) I will P make ik ga eve die glije. (J 2;0) I will just that slide poesje doet likke. (J 2;1) kitty does lick ga je ook school toegaan? (J 2;1) will you too school to-go? gaat Cynthia slape? (J 2;1) will C go-sleep? doet mama mij … (J 2;2) does mommy me … [do] nou gaatie weer naar huis toe gaan. (J 2;2) now will-he again to home to go

papa, doen g[l]ijbaan heg zette. (A 2;0) daddy, do-[we] slide away put papa boekie leze. gaan boekie leze.(A2;0) daddy booklet read. will[-we] booklet read zo doek ik jurk ui(t)doen. (A 2;0) this-way do-I I dress off-do gaan autorije? (A 2;1) will-[we] car-drive?

doen ze same hope. (A 2;2) do they together walk doense same zitte. (A 2;2) do-they together sit-down Jaja gaat daar zitte … J will there sit-down … en papa gaat daar zitte. (A 2;2) and daddy will there sit-down

Evidence of the projection of F  

doetie [sl]ape nou? (A 2;1) does-he go-sleep now? gaatie [sl]ape. (A 2;1) will-he go-sleep doetie viesmake. (A 2;1) does-he dirty make doetie hantie geve. (A 2;1) does-he hand give kountje huire. doetie huire. (A 2;1) clown-little cry. does-he cry

 179

Jaja doet kitkat opete. (A 2;3) J does kitkat up-eat Jaja gaat liedje make van jou. (A 2;3) J will song make of you gaan ze almaal ete, zie? (A 2;3) will they all eat, see? ik doe mij fesje aan mij jamaboek doen. (A 2;4) I do my vest on my pyjamas do gaat ikke ook mee naa paardrije? (A 2;4) will I also with to horse-ride

The use of doet and gaat as carriers of finiteness is a temporary phenomenon. This is because, in the system of the target language, the functional properties of these auxiliary verbs are carried by the finite lexical verb. With the use of lexical verbs as carriers of finiteness in F, they will eventually disappear.

6.2.3 Topicalization Further evidence of the acquisition of FP is the fact that simultaneously with the use of the auxiliaries heb, heeft and ben, is, both child and adult learner varieties show the systematic use of topicalization. Evidence of this is given in (19) and (20). (19) Auxiliary verbs and the acquisition of topicalization in child L1 Dutch heef Cynthia maakt. (J 2;0) [this] has C made die heb ik wel geplakt. (J 2;2) that have I indeed glued die heef Cynthia gemaakt. (J 2;2) that has C made heb ik oppegete. (A 2;2) [this] have I up-eaten dà heefe Jaja daan. (A 2;2) that has J done da issie varre. (A 2;2) there is-he fallen da hem papa hijm oppedaand? (A 2;3) there has daddy glue on-done?

180 

 The functional stage

die hem ik van J kege. (A 2;4) that have I from J got die isse Jaja gete. (A 2;4) that is J forgotten (20) Auxiliary verbs and the acquisition of topicalization in adult L2 Dutch Izmir heb ik niet geweest. (Os/T 3.4) Izmir have I not been daar heeft mij zus van vader twee uur rijden … (Os /T 3.4) there has my sister of father two hours drive daarom ik heb uitgemaakt. (Os/T 3.4) therefore I have finished die heb ik allemaal vergeten. (Mo/A 3.2) that have I all forgotten toen heeftie teruggegeven aan. (Mo/A 3.6) then has-he back-given to ben ik daar nooit geweest. (Mo/A 3.7) [there] am I there never been dat heb ik nooit gezegd. (Mo/A 3.8) that have I never said As noted in Section 6.1.2, topicalization at the lexical stage occurs with structures as in (3). (3) Object or adverb in topic position, agent missing⁵³ child L1 Dutch

adult L2 Dutch

nee losmake. (J 1;10) [this] wantAGnot loose-make

dan kanniet lezen, kan begrijpen niet. (Os/T 1.7) then canAGnot read, canAG understand not da kanwel opzitte. (J 2;0) en dan moet teruggeven. (Os/T 2.3) there canAGindeed on-sit and then [this] mustAG back-give deze magniet teke. (A 2;2) die magwel kopen. (Os/T 2.5) this mayAGnot draw that mayAGindeed buy hier moet poesje eve kamme. (A 2;3) charlie zegt: nou magwel weglopen. (Os/T 2.9) here mustAG kitty just comb C says: now mayAGindeed away-run

53 AG indicates that the agent is used implicitly with the modal predicate.

Evidence of the projection of F  

 181

In (3), the topic position is taken by the object or an adverb, while there is no subject position for the agent. The agent is expressed implicitly with the modal head. Topicalization with the auxiliary heb, heeft or ben, is, however, always occurs as in (19) and (20). That is, given the position F as the head of FP, there is a position for the subject as SpecVP of the lexical projection VP. Hence, there are no examples of topicalization with the auxiliary heb, heeft or ben, is and a missing subject as for example in (21), neither in child L1 nor in adult L2 Dutch. (21) Utterances not attested *die heb geplakt. that have glued *dat heb nooit gezegd. that have never said *hier heb opgedaan here have on-put *nou heeft teruggegeven now has back-given This is precisely what could be expected on the basis of the acquisition process as described. At the lexical stage, modal predicates allow the agent to be left unexpressed. This is because at the relevant stage, modal predicates are used to express the willingness, the ability, the possibility or the obligation of an agent to perform an action. Reanalysis affects the linguistic status of these modal predicates. As the functional head of FP, they serve as the prerequisite for the use of the auxiliaries heb, heeft and ben, is as instantiations of F. This reanalysis of the modal predicate as the head of the functional projection FP has become possible due to the development of the agentive complement into the full projection of V. With both FP and VP there are two specifier positions: SpecFP as the topic position and SpecVP as the subject position. Consequently, the acquisition of the auxiliaries heb, heeft and ben, is in F can only occur given that this subject position is structurally available. It explains why utterances as in (21) do not occur. To summarize, the rapid increase in the use of auxiliaries as shown in child L1 Dutch in Table 1 and Figure 6 and in adult L2 learners as shown in Table 2 provides evidence that the learners are in the process of acquiring the functional projection FP of the target language. This process entails the acquisition of two structural positions: A structural position F for auxiliary verbs – and later also for lexical verbs – to express the functional properties of finiteness and a structural position SpecFP to account for the phenomenon of topicalization. Achievement

182 

 The functional stage

of the relevant syntactic features shows that the learners have reached the functional stage.

6.3 The topic position 6.3.1 The affix in F as a topicalization device The developmental process as described in Figure 4 shows that the initial position becomes a position exclusively projected as a topic position as soon as the learner system has created a position for the subject in SpecVP. As a topic position, the initial position is available for any type of constituent. In the default case, however, i.e. in the absence of an element with topic function, word order is determined by the semantic principle of control asymmetry. According to this principle, the initial position is taken by the subject as the element that is highest in control. This explains why “ (…) fronted nonsubjects have the discourse interpretation ‘topic’, while a sentence-initial subject may or may not have such an interpretation” (Anderson 1993: 93). In the target language, there is an alternative means to express that the subject has topic function. It occurs particularly in spoken language varieties. Examples are given in (22). (22) Topicalization of the subject in Dutch Doetie opeens de deur dicht. does-he suddenly the door closed Gaatie vol op zijn plaat. will-he fully on his face Kannie trots op zijn. can-he be proud of Heeftie effe rust. has-he a-moment rest In (22), the affix -ie has subject function. Within the given context, it refers to an entity implicitly serving as the topic. However, the topic position remains empty, because reference to an entity cannot occur twice. Given that the affix -ie serves as the subject, it seems that the type of utterance as in (22) is in fact a topicalization device specifically developed for constituents with the subject function.

The topic position 

 183

In learner languages, topicalization of the subject is very frequent. Examples of child L1 and adult L2 Dutch are given in (23) and (24). The use of the affixes -ie, -’k, -se, -e, -we reflects the spoken variety that learners receive as input. (23) Topicalization of the subject in child L1 Dutch gaatie plantjes geve water, Daphnie. (J 2;0) [NPi] will-hei plants give water, Di doetie haaltje telle. (J 2;2) [NPi] does-hei story tell mag’k niet die hebbe. (J 2;2) [NPi] may-Ii not that get doense same hope. (A 2;2) [NPi] do-theyi together walk magtie papa zitte. (A 2;2) [NPi] may-hei daddy sit moete nieuwe make, vokke. (A 2;2) [NPi] must-wei new-ones make, flakes gaatie draaie. (A 2;3) [NPi] will-hei turn (24) Topicalization of the subject in adult L2 Dutch A: B: A: B: A: B:

Als je me nodig hebt, moet je me maar roepen. If you me need, must you me just call OK. Roep’k straks jou. OK. [NPi] call-Ii later you Maggie weg? may-it away? Maggie weg ja. [NPi] may-iti away yes Het is 85 bij 85. it is 85 to 85 Zullewe zien of klopt dat. [NPi] shall-wei see if is-all-right that

It seems interesting to note that in L2 Dutch there is also a non-targetlike version of utterances with a topicalized subject. An example of this is shown in (25).

184 

 The functional stage

(25) Topicalization of the subject in adult L2 Dutch Spotje, komtie pas hier. (Slawek 14.10.06) spotlighti, comes-hei only here Examples as in (25) are not targetlike because the subject is explicitly referred to both with an NP in topic position (spotje, spotlight) and with the affix -ie in F.

6.3.2 Yes/no- and wh-questions With the acquisition of FP, learner language has both the functional positions of F and SpecFP. SpecFP provides a position for elements to refer to the topic situation, i.e. to the situation that the utterance holds for. The position of SpecFP also plays a role, if the speaker wants to find out whether the utterance can be linked to a particular situational context at all. In such a situation, i.e. a situation referred to with a yes/no-question, the position of SpecFP remains empty. Furthermore, if the speaker wants to know for which entity, location, time or situation a particular predication may hold, SpecFP is the position for question words such as wat (what), waar (where), wanneer (when) and hoe (how). These wh-words indicate the focus of a question. Thus, with the acquisition of SpecFP the language system has the possibility to express not only THAT an assertion applies to a particular situational context, but also WHETHER and if so, HOW the assertion applies to this context. Examples of the first occurrences of yes/no- and wh-questions in child L1 and adult L2 Dutch are given in (26) and (27). (26) Yes/no- and wh-questions in child L1 Dutch zulle lego spele? (J 2;1) shall[-we] lego play? wil’jes opedoen? (J 2;1) want-you open-do? ga je staan? (J 2;1) will you stand-up? heb je visje gehad? (J 2;1) have you fish got? mag ik wel uit bedje klimme? (J 2;2) may I indeed out-off bed climb? mag ik die plakke? (J 2;2) may I that glue?

The topic position 

 185

mag ik afknippe? (J 2;2) may I off-cut? waa ben je nou geweest? (J 2;2) where are you now been? waar heb je chocola gelate? (J 2;2) where have you chocolate left? mag ik ook doen? (A2;2)⁵⁴ may I too do? mag kikker ook mij vasthoue? (A2;2) may frog also me tight-hold? hemme nogge fippo vonne? (A 2;2) have-we any flippo found? heeftie ook een tandje poetse tanne krege? (A 2;2) has-he also a tooth brush teeth got? moetie pakke, deze? (A 2;3) must-he get, this? mag beetje verfe Ernie Bert? (A 2;3) may a-bit paint Ernie Bert? mag Jaja kaas ete? (A 2;3) may J cheese eat? één, tee. zumme terre? (A 2;3) one, two. shallwe draw? heef anne mense ook doend? (A 2;3) have other people also done? isse barnies affehope, mam? (A 2;4) is barnies ended, mommy? ga je nou hare? (A 2;4) [what] will you now get? doet papa voor mij nou doen? (A 2;4) [what] does daddy for me now do? (27) Yes/no- and wh-questions in adult L2 Dutch waar heb jij geweest? (Os/T 3.4) where have you been?

54 At the lexical stage, Andrea uses utterances with mag-ikke as in mag-ikke paartie rije? (may-I horsie ride? A 2;0) relatively frequently. They function as phrasal expressions which give way to the analysed structures at the functional stage.

186 

 The functional stage

nou hoe moet ik doen? (Os/T 3.6) now how should I do? dan wat moet ik doen? (Os/T 3.6) then what should I do? heb jij zaterdag televisie gezien jij? (Mo/A 2.2) have you saturday tv seen? heb jij geen klok gezien? (Mo/A 2.3) have you no clock seen? heb je die auto gezien? (Mo/A 2.5) have you that car seen? wat heb ik gedaan? (Mo/A 3.2) what have I done? wat moet ik doen dan? (Mo/A 3.3) what should I do then? As shown in (26) and (27), it seems not accidental that yes/no- and wh-questions occur in utterances with either an auxiliary or a modal verb. The reason is that the SpecFP position for yes/no- and wh-questions has been created with the projection of F. Wh-question words occur in initial, focus position regardless of their syntactic function. As is the case with constituents in topic position, the semantic function of these wh-question words is determined by the structure of the projection of V. (28) Topicalization and focalization at the functional stage FP | SpecFP F′ | F (head)

topic / focus a.

die thati

VP | SpecVP V′ | V

complement ‖ object – Pred[CAUSAL ACTION]

subject wilt ik wantj I

ej

hebbe. (J 2;2) ei get

The topic position 

b. c. d.

topic / focus

subject

[] ei

heef hasj

Cynthia C

[] ei

ga willj

je you

[]

mag mayj

kikker ook frog also

mag mayj

e.

da there

f.

waa(r) ben where arej

 187

object – Pred[CAUSAL ACTION] ej

maakt. (J 2;0) ei made

ej

nou hare? (A 2;4) now ei get?

ej

mij vasthoue?(A 2;2) me tight-hold?

papa wel daddy indeed ej

zitte van Jaja. (A 2;3) sit-down of J

je you

nou geweest? (J 2;2) been?

ej

As shown in (28a), die (that) in initial, SpecFP position is interpreted as the object, because the object position of VP is empty (ei). This is also the case in (28b) and (28c), in which both the object position ei and the SpecFP position are empty. In (28d), only SpecFP is empty. This indicates that there is no situational context that the assertion may apply to. It turns the utterance into a yes/no-question. In both (28e) and (28f), the SpecFP position is taken by an adverbial. With the deictic pronoun da (there) in (28e) the speaker refers to the contextual situation that the utterance applies to. With the wh-word waar (where) in (28f), the speaker wants to find out which contextual situation it is.

6.3.3 Verb-third in adult L2 The acquisition of the functional category system in Dutch seems similar for children learning Dutch as their mother tongue and adults learning Dutch as their second language. It is a developmental process of stagewise progression. Nevertheless, children and adults behave differently. Normally, all children will learn the adult model of their mother tongue within a relatively short period of time. Adult L2 learners, however, show a lot of variation both with respect to the speed of the acquisition process and the ultimate level they achieve. That is, some adult learners are much faster than others and some adult learners become more like native speakers than others. These differences between child and adult learners are due to the communicative and cognitive function of the target language system. For child L1 learners, the model of the input language is the only option there is. Acquisition of the target language interacts with the child’s social and

188 

 The functional stage

cognitive development. For an adult L2 learner, however, language learning takes place after the system of the mother tongue has been acquired. The target language is just a means to make oneself understood. Therefore, as soon as the adult learner feels comfortable with a specific level of competence, the acquisition process may slow down or even come to a standstill. As argued before, the communicative function of topicalization is the driving force in the acquisition of the functional category system. For the expression of topicalization, children have no other option than the linguistic means provided by the system of the target language. For adult learners, however, given their linguistic and social prerequisites, there is room for some variation that is not targetlike. A typical feature of this kind is the phenomenon referred to as ‘verb-third’. In utterances with verb-third the adverbial occurs before the initial subject as in (29). (29) Verb-third in adult L2 Dutch toen ik kom hier, ik heb een jaar nederlands leren. (Mo/A 2.3) then I come here, I have a year dutch learn toen politie hebben die meisje gevat. (Mo/A 3.9) then police have that girl taken This type of structure can be explained by the fact that a subject in initial position as argued in Section 6.3.1 may not have topic function. As shown in (29), this allows the creation of an XP position with topic function before the subject. Taking the utterance structure in (29) into acount, the L2 learner grammar at the relevant stage can be characterized with the examples in (30) produced by Mohammed/A. The use of the auxiliary heb shows that this L2 learner has reached the functional stage in the acquisition of L2 Dutch. (30) Mohammed/A. Utterance structure at the functional stage. * = not attested (a)

ik heb daar geslapen buiten. (Mo/A 3.6) I have there slept outside ik heb jou vorig keer niet gezien. (Mo/A 3.8) I have you last time not seen

(b)

die heb ik allemaal vergeten. (Mo/A 3.2) there have I everything forgotten die heb ik hier gedaan. (Mo/A 3.5) that have I here done

The topic position 

(c)

wat heb ik gedaan? (Mo/A 3.2) what have I done?

(d)

heb jij geen klok gezien? (Mo/A 2.3) have you no clock seen? heb je die auto gezien? (Mo/A 3.5) have you that car seen?

(e)

vrijdag nacht ik heb niet geslapen. (Mo/A 3.5) friday night I have not slept s’morgens ik heb gesnipperd. (Mo/A 3.5) in-the-morning I have taken-a-day-off

*(f)

ik die heb allemaal vergeten. (not attested) I that have all forgotten

*(g)

ik toen heb jou niet gezien. (not attested) I then have you not seen

(h)

toen hier ik heb beetje geslapen. (Mo/A 3.5) then here I have a-little slept

(i)

toen die ik heb opgedronken alleen. (Mo/A 3.6) then that I have up-drunk alone

 189

The examples in (30) show that at the relevant stage, there is an initial, SpecFP position for a subject as in (30a), for a topicalized object as in (30b) and for a focus element as in (30c). In case of a yes/no-question as in (30d), the position of SpecFP is empty. As pointed out before, due to the fact that a subject in SpecFP may not have topic function, topicalization may also occur in utterances as in (30e). In (30e) there is an XP-topic position before the subject. Utterances as in *(30f) and *(30g) are not attested, however. They are in conflict with the present system, because the constituent with topic function is not in initial position.⁵⁵ The XP-topic position is usually taken by an adverbial as in (30e). This seems to be the result of an interaction between the language system at the functional stage and the relevant input as in (31).

55 Utterances as in (30a) and (30e) also occur in the L2 grammar of Osman/T at the functional stage. Examples are ik heb alleen zes uur daar gewerk (I have only six hours there worked) and misschien ik heb ja ik ook werken (maybe I have yes I too work).

190 

 The functional stage

(31) Subordination with wanneer (when), toen (then) and als (if) Wanneer hij komt, weet ik niet. when he comesSCl, know I notMCl Toen hij sliep, heb ik hem niet gewekt. when he sleptSCl, have I him not woken-upMCl Als hij belt, ben ik niet thuis. If he ringsSCl, am I not homeMCl Examples as in (31) are common in Dutch. They are complex sentences with a subordinate clause before the main clause. Evidence of subordination is the use of the finite verb in final position. However, due to its position, it may look as though the subordinate clause were a main clause with the finite verb in the position of verb-third. Given the frequency with which utterances as in (31) occur in the input, the L2 learner might conclude that in main clauses there are two structural positions before the finite verb, i.e. a topic/focus position for an adverbial and a second position for the subject. As argued before, this accounts for the fact that Mohammed/A produces (30a), (30b), (30c) and (30d) simultaneously with (30e). The same is true for (30h) and (30i). (32)

Mohammed/A. Utterance structure at the functional stage.

(h)

Toen hier ik heb beetje geslapen. (Mo/A 3.5) then here I have little slept

(i)

Toen die ik heb opgedronken alleen. (Mo/A 3.6) then that I have up-drunk alome

These examples show that topicalization is also possible with two adverbials as in (30h) or with an adverbial and an object as in (30i). In sum, utterances with an adverbial before an initial subject are typically produced by second language learners of Dutch. These utterances are not targetlike. They are due to the fact that a subject in initial, SpecFP position does not necessarily function as a topic. Subordinate clauses with wanneer (when), toen (then) and als (if) in initial position seem to be an additional reason for why the finite verb may occur in the position of verb-third in L2 learner production data.

Finiteness 

 191

6.4 Finiteness 6.4.1 Auxiliary verbs 6.4.1.1 Verb placement Evidence provided in developmental studies by Becker on L2 German (2005) and Verhagen on L2 Dutch (2009a) shows the crucial role of the auxiliary verb as a prerequisite for the acquisition of finiteness and placement of the lexical verb in the position of verb-second. In her study on the acquisition of L2 Dutch by L1 speakers of Turkish and Morrocan Arabic, Verhagen (2009a) investigated the placement of the lexical verb with respect to the position of the sentence negator niet. In Dutch, as in German, the lexical verb is placed either before or after niet. Examples with the verb lopen (walk) are given in (33a) and (33b). These examples show that the position of the lexical verb lopen depends on whether there is an auxiliary verb or not. (33a)

Het meisje heeft niet op het strand gelopen. the girl has not on the beach walked

(33b)

Het meisje loopt niet op het strand. the girl walks not on the beach

In (33a) the lexical verb gelopen (walked) occurs after niet due to the fact that the position of the finite verb before niet is taken by the auxiliary verb heeft (has). In (33b) there is no auxiliary verb, hence the lexical verb loopt (walks) occurs in ‘raised’ position, i.e. its position is before niet. With an auxiliary verb as in (33a), the lexical verb after niet is non-finite (gelopen-Pp). In the absence of an auxiliary verb as in (33b), the lexical verb before niet is finite (loopt-3Sg). Verhagen (2009a) carried out a cross-sectional study with two groups of L2 learners of Dutch. That is, L2 learners who did not yet produce the auxiliary heb, heeft and learners who did: the −Aux group and the +Aux group, respectively. Verhagen found that learners of the −Aux group only rarely placed the lexical verb before the negation, whereas learners of the +Aux group did so significantly often. The relevant data are shown in Table 3. Thus, it seems that, as long as L2 learners have not yet acquired a position for the auxiliary verb, the lexical verb is placed after the negation. However, as soon as the position of AUX has been established, learners appear to be able to use the lexical verb before the negation. The acquisition of AUX is, thus, a prerequisite for the placement of the lexical verb before niet.

192 

− Aux + Aux

 The functional stage

L1 Turkish

L1 Moroccan Arabic

 3.5% (6 / 176) 37.0% (71 / 192)

 8.6 % (8 / 93) 78.4 % (138 / 176)

Table 3: The use of the lexical verb before niet in L2 Dutch (Verhagen 2009a: 64, 67)⁵⁶

Verhagen (2009a) also carried out an elicited imitation task. With this task, she investigated the frequency with which L2 learners changed a given utterance with an incorrect, i.e. non-targetlike, niet + V order as in (34a) into a correct V + niet order as in (34b). (34a)

* De jongens en meisjes niet lopen op het strand. the boys and girls not walk on the beach

(34b)

De jongens en meisjes lopen niet op het strand. the boys and girls walk not on the beach

The relevant data are given in Table 4.

− Aux + Aux

L1 Turkish

L1 Moroccan Arabic

  0 % (0 / 54) 7,2 % (5 / 69)

 1,4 % (1 / 70) 13,2 % (12 / 91)

Table 4: The position of the lexical verb. From niet + V to V + niet (Verhagen 2009a: 97,99f.)⁵⁷

As in the experiment on production, learners of the −Aux group appear not to be able to place the lexical verb before niet. Only learners of the +Aux group are able to do so. Thus, again, the acquisition of the auxiliary verb is a prerequisite for the placement of the lexical verb before niet. In sum, with both a production and an elicited imitation task, Verhagen (2009a) has shown that with the acquisition of the category AUX, the learner grammar has established a functional position for the placement of the lexical verb. In the target system, this position is linked to the expression of finiteness and, hence, the functional projection FP.

56 The relevant data are taken from Tables 5,6,7 and 8. 57 The relevant data are taken from Tables 4, 5, 6 and 7.

Finiteness 

 193

6.4.1.2 ‘Light verbs’ A developmental approach to the acquisition of finiteness in L2 acquisition is taken in Parodi (2000). Parodi notes that from early on in L2 acquisition light verbs (modals, auxiliaries and possessive have) occur in raised position, i.e. before the negator and with targetlike morphology. She interpretes this observation as evidence of the early presence of the functional category INFL (or: F). Furthermore, she argues that the presence of INFL and, thus, FP is evidence of either transfer from the L1 or access to innate linguistic knowledge of UG. Despite the early presence of INFL, Parodi also notes that, initially, lexical verbs are not raised to INFL. Finiteness with lexical verbs seems to be the result of a relatively slow process of acquisition. Unlike the conclusion put forward in Parodi (2000), the spontaneous production data from L2 Dutch and German as presented in Chapter 5 and the experimental data from L2 Dutch in Verhagen (2009a) show that, initially, the functional projection FP is absent. FP becomes established only with the acquisition of the auxiliary verb heb, heeft. Thus, as long as this auxiliary verb has not been acquired, raising seems not possible. Given the fact that auxiliary verbs are learned at more advanced stages of acquisition, it seems questionable whether the occurrence of light verbs as observed in Parodi (2000) can be interpreted as evidence of FP. Verhagen (2009) also investigated the use of ‘light verbs’ similar to those in Parodi (2000) in L2 learners of Dutch who did not yet produce the auxiliary heb, heeft. These L2 learners used the copula or a modal predicate as in (35a) and (35b), modals or gaan with an agentive predicate as in (35c), and a particular use of is with an agentive predicate as in (35d). (35a)

Maar dat is nog niet laat. but that is yet not late (Verhagen 2009a: 64)

(35b)

Dat kan niet. that can not (Verhagen 2009a)

(35c)

Groene man niet wil springen. green man not wants jump-Inf Meneer rood ook wil niet springen. mister red also wants not jump-Inf (Verhagen 2009a: 65) Charlie zegt . Charlie says Vrouw zeggen < Niet man, dit is vrouw gaat pakken >. woman say-Inf (Verhagen 2009a)

194 

 The functional stage

(35d)

Charlie Chaplin is steel de brood. Charlie Chaplin is steal-0 the bread De jongens is spelen. the boys is play-Inf (Verhagen 2009a, 53) Hij is moet beneden. he is must down [go] Misschien hier hij is wil slapen. maybe here he is wants go-sleep-Inf (Verhagen 2009a, 62)

Usually (see fn. 83), these ‘light verbs’ occur before niet. According to Parodi, examples as in (35a) to (35d) should provide evidence that INFL is available. However, at the relevant stage, the auxiliary heb, heeft is absent and, as shown in Chapter 5, so is the functional projection FP. As long as FP is absent, learner utterances are produced with the structure of lexical projections only. Therefore, I would argue that the utterances in (35) are produced on the basis of lexical projections as they occur at the lexical stage. At the lexical stage, as shown in Chapter 5, utterance structure can be accounted for in terms of two types of lexical projections: lexical projections with an action predicate as the complement of a modal head, and lexical projections with a state- or change-of-state predicate in the position of the lexical head. Within this account, utterances with a copula as in (35a), are evidence of state predicates. They occur with a theme as the subject. This also holds for the example in (35b). Here the modal predicate kan refers to a possible state: a possibility. Modal predicates as in (35b) are used to express a possibility (kan, can), an option (mag, may) or a necessity (moet, must). As in the target language the pragmatic function of the copula and these modal predicates is to express the modality of an assertion. Similar observations have also been made for L2 German. Becker (2005: 373, 378) compares the use of is in (36a) with the use of the negator nix (not, nothing) in (36b). She argues that “the form is only appears in affirmative utterances in [the] pre-basic variety. Moreover, it occupies the same position as the negator: it is placed between the topic constituent and the focus constituent. This suggests that the learner recognizes the semantic relationship between assertion and negation” (379). (36a)

mein mann is in arbeite. my husband is in work

Finiteness 

(36b)

 195

nee, mein kind nix in schul no, my child nothing in school

As shown in Verhagen (2009a), examples as in (36a) and (36b) can also be found in adult L2 Dutch. They are given in (37a) to (37d). (37a)

De meneer rood niet springen. the mister red not jump-Inf De meneer blauw is springen. the mister blue is jump-Inf (Verhagen 2009a: 61)

(37b)

Die man niet doen, de zij is doen. that man not do-Inf, the she is do-Inf (Verhagen 2009a: 61)

(37c)

De bang niet daar, is daar. the scared (=danger) not there, is there (Verhagen 2009a: 64)

(37d)

Een meisje is niet krijgen een brood. a girl is not get-Inf a bread Die man niet is stolen [=stelen]. that man not is steal-Inf (Verhagen 2009a: 65)

As in L2 German, here too, is is used to mark the default pragmatic function of assertion. Hence, it is used together or in complementary distribution with the negator niet.⁵⁸ Modal predicates as in (35c) and (35d) typically occur in agentive types of utterance. They are used to express the assertion of the willingness (wil, want), the ability (kan, can), the permission (mag, may) or the obligation (moet, must) of an agent to perform an action. This also holds for the use of the predicate gaat (will) as in (35c), which occurs in agentive types of utterance with the lexical-aspectual meaning is van plan (is going to). The use of is in (35d) has been accounted for in the literature with a range of proposals (see Verhagen 2009a: 58ff.). Fact is, that this particular use of is appears in complementary distribu-

58 At the lexical stage, the copula is serves as a state predicate with a theme as its subject. As shown in Chapter 5, this also holds for change-of-state predicates such as valt (falls) or komt (comes). They occur with the same distribution as the state predicate is. However, they belong to a small and infrequent type of predicate. This may explain why, for example in the study by Parodi (2000), these verb forms have not been noticed.

196 

 The functional stage

tion with the use of modal predicates and that in some cases it occurs even in combination with a modal as in is moet (is must) and is wil (is want) in (35d). It is evidence that, as in (35a), is is used to express the default pragmatic function of assertion. This explains why it occurs in opposition to niet in (37a), (37b) and (37c) or together with niet as in (37d). In sum, at the lexical stage, the so-called ‘light verbs’ in Parodi (2000) and Verhagen (2009a) are mainly modal and state predicates used to express the pragmatic function of assertion. As argued in Chapter 5, Section 5.3.1, in utterances with an agent as the subject, they are used to express the assertion of a modality, while in utterances with a theme as the subject, they are used to express the modality of an assertion. In both types of utterance, is is used to express assertion as the default. Given that at the relevant stage, the lexical head is used to carry the pragmatic function of assertion, this explains why in examples as in (38) the modal head may occur with wel (indeed) in opposition to niet (not). See Jordens and Dimroth (2006, 181ff.). (38) Modal head with wel (indeed) vs. niet (not) child L1 Dutch

adult L2 Dutch

kannie bal pakke. (J 1;9) can-not ball get kusse hebbe, magwel. (J 1;10) pillow get, may-indeed poes magniet oppe aanrecht. (J 2;1) kitty may-not on-the sink [climb] kanwel papa zitte. (A 2;1) can-indeed daddy sit-down

sigaretten roken, magniet. (Os/T 1.8) cigarettes smoke, may-not ja magwel hier gaan. (Os/T 1.8) yes may-indeed here go jij magniet hier slapen. (Os/T 3.6) you may-not here go-sleep pilske magwel hè, een twee per dag. (Mo/A 2.8) beer may-indeed [drink] okay, one two per day

It also explains why, as shown in (39), in the absence of a modal or a state predicate, utterances may simply occur with wel in opposition to niet. (39) Empty head with wel (indeed) vs. niet (not) child L1 Dutch

adult L2 Dutch

poppie niet Jaja help. (A 2;0) doll not J help Ruti hel [= wel] bad zitte. (A 2;1) R indeed bath sit-down

ik niet zien. (Mo/A 1.1) I not see maar hij wel broer van hem. (Mo/A 2.3) but he indeed brother of him

Finiteness 

 197

Finally, ‘light verbs’ do not count as evidence for the presence of INFL. This is because, as shown in Verhagen (2009a), at the relevant stage, raising does not occur. In the absence of raising, the morphology of verb forms such as is (is), kan (can), wil (want), moet (must) and gaat (will) provides no evidence of agreement either. Verb forms at the relevant stage are morphologically unanalysed. They are learned as they occur in the input and, therefore, it only seems as if they occur with targetlike morphology. Research taking ‘light verbs’ and non-agentive lexical verbs such as valt (falls) or komt (comes) as evidence of INFL, does so at face value. It illustrates a common error in acquisition research, i.e. the assumption that learner utterances can be analysed in terms of the system of the target language.

6.4.1.3 Morphological properties of finiteness in L2 Dutch In Dutch as in German, verb placement correlates with morphology. The relation between non-finite lexical verb forms in head-final position and finite lexical verb forms in head-initial position is usually referred to in terms of ‘raising’. The position of the lexical verb as the head of FP is considered to be the landing site of an operation that moves the lexical verb from its underlying, non-raised position into its superordinate, i.e. raised position. As a criterion to establish whether or not a verb can be considered as raised, L2 researchers commonly take the positioning of the lexical verb relative to the sentence negator. Verb forms occurring after the negator are assumed to appear in their default, i.e. ‘non-raised’ position. Verb forms occurring before the negator are assumed to appear in their nondefault, i.e. ‘raised’ position. In L2 learners, verb placement often does not correlate with the acquisition of morphology. That is, finite verbs occur in non-raised position (Meisel 1997) and non-finite verbs occur in raised position (Prévost and White 2000). On the assumption that FP is present, finite verb forms in non-raised position are accounted for in terms of the Impaired Representation Hypothesis (IRH) as stated in Beck (1998), Franceschina (2001) and Hawkins and Chan (1997). According to the IRH, the subsystem which determines the application of features of verb morphology seems to be in disorder. On the other hand, non-finite verb forms in raised position are seen as evidence that it is just verb morphology that does not play a role, yet. Absence of tense and agreement morphology in raised position is accounted for in terms of the Missing Surface Inflection Hypothesis (MSIH) as proposed in Prévost and White (2000). The MSIH states that, on the one hand, verb raising shows that FP has been established, while, on the other hand, absence of tense and agreement shows that FP is underspecified with respect to its morphological properties.

198 

 The functional stage

In the experimental data as presented in Verhagen (2009a: 131, 132) both the phenomena of impairment and underspecification were observed in Moroccan and Turkish learners of L2 Dutch. Evidence from a production experiment is given in Tables 5 and 6.

Beginners (N=24) finite V-NEG (raised) 3 NEG-V (non-raised) 2

Intermediates (N=31) non-finite

finite

non-finite

 6 41

48  0

11 23

Table 5: Finite and non-finite verbs in negated utterances for Moroccan beginners and intermediates

Beginners (N=21) finite V-NEG (raised) 1 NEG-V (non-raised) 7

Intermediates (N=25) non-finite

finite

non-finite

 2 98

12  1

17 54

Table 6: Finite and non-finite verbs in negated utterances for Turkish beginners and intermediates

The data show the same patterns for both language groups: “a typical missing surface inflection pattern was found for the more advanced learners, who produced verb-raising relatively often, but not for the beginning learners, who placed finite forms in raised as well as non-raised position” (2009a: 133). Thus, ‘impairment’ particularly occurs with beginners, whereas ‘underspecification’ is characteristic of L2 learners who are more advanced. Further evidence comes from an imitation experiment with a group of beginners as presented in Verhagen (2009a: 138f.). Examples of target and response sentences are given in (40a) and (40b). (40a)

Target:

De mevrouw niet slaapt op de bank. the woman not sleeps-3Sg on the couch

Response:

De mevrouw niet slapen de bank. the woman not sleep-Inf on the couch

Finiteness 

(40b)

Target:

De collega helpen niet met het werk. the colleague help-Inf not with the work

Response:

De collega helpt niet met het werk. the colleague helps-3Sg not with the work

 199

From the changes made in this elicited imitation task Verhagen concludes that “learners […] show a preference for non-finite verbs forms to occur in a nonraised position, but no preference for finite verbs to occur in either raised or nonraised position” (2009a: 139). In other words, the beginning L2 learners did not accept non-finite forms in raised position, but they accepted finite forms in nonraised position. Why should this be the case? Why does ‘impairment’ occur with beginners, whereas ‘underspecification’ is found with learners who are more advanced? In the target language input, lexical verbs may occur both in head-final (Jan gaat een appel eten) and, as the result of raising, in head-initial position (Jan eet een appel). Initially, however, as shown in Chapter 5, the learner grammar does not have the morphological contraints that apply to raising. That is, learners do not yet acknowledge the syntactic relevance of non-finite- vs. finite verb morphology. Therefore, L2 learners may conclude that the same verb forms that are learned in head-initial position of the predicate can also occur in head-final position of the complement and vice versa. This is similar to the view put forward by Meisel, who notes that verb placement in early L2 acquisition is “independent of the acquisition of finiteness” (1997: 255). With the acquisition of the auxiliary heb, heeft, FP projects a position for F. If the position of F is not taken by an auxiliary verb, the lexical verb must serve as a carrier of finiteness in head-initial position. At the beginning of the functional stage, however, learners may use the lexical verb as it occurs in head-final position, i.e. with its non-finite morphology. This explains why the phenomenon of underspecification that the MSIH attempts to deal with is found with learners of Dutch when they are more advanced.

6.4.2 Grammatical aspect At the lexical stage, as noted in Chapter 5, predicates are used in complementary distribution. That is, state and change-of-state predicates occur in initial position as the head of the utterance, while action and result-state predicates occur in final position as the head of the complement. State and change-of-state predicates occur with finite morphology, while action predicates appear with an infini-

200 

 The functional stage

tival and result-state predicates with a past participle form. At the relevant stage, however, morphology is linguistically irrelevant. It is the formal expression of the lexical predicate as it occurs in the input. Now, given that morphology is irrelevant, why is it that lexical predicates are distributed the way they are? As discussed in Chapter 2, verb morphology that is relevant with respect to the expression of aspectual meaning is subject to processes of lexicalization. Thus, at the lexical stage, when morphology does not yet play a role, learners will see aspect as a lexical property and, consequently, they will see the aspectual properties of lexical predicates in the target language as related to variation in placement rather than to variation in form.⁵⁹ State and change-of-state predicates occur in initial position. State predicates refer to the physical or psychological condition of a person or an object. Changeof-state predicates refer to motions and emotions that are not under control. The opposition between state and change-of-state predicates seems to represent the opposition between imperfective and perfective aspect. Examples are is (is) vs. komt (comes); blijft (stays) vs. gaat (will); staat (stands) vs. valt (falls); heeft (has) vs. krijgt (gets); heeft aan (has on) vs. past (fits); zit (sits) vs. schrikt (is frightened); slaapt (sleeps) vs. wordt wakker (wakes up); ligt (lies) vs. glijdt uit (slips); ruikt lekker (smells nice) vs. prikt (pricks); woont (lives) vs. vliegt weg (flies away). Action and result-state predicates occur in final position. Action predicates refer to causal actions and motions that are under control. Result-state predicates refer to the result state of an action or a change-of-state. The opposition between action and result-state predicates seems to represent the opposition between prospective or progressive aspect and perfect aspect. Typical examples are predicates such as maken (make) and doen (do) with particles referring to a result state as in potmake (kaput-make) vs. potmaakt (kaput-made); goonmake (clean-make) vs. goonmaakt (clean-made); opdoen (on-do) vs. opdoend (on-done); opete (upeat) vs. opgete (up-eaten). But also aaie (caress) vs. aaid (caressed); geve (give) vs. geefd (given); neme (take) vs. nome (taken); tannepoetse (teeth-brush) vs. kamd (combed). There is also a lexical opposition between change-of-state and result-state predicates. They represent the opposition between perfective and perfect aspect.

59 In Klein (1994) ‚aspect‘ is defined in terms of the relation between the time for which an assertion is made, i.e. the topic time (TT), and the time of the situation (TSit). Imperfective: “the TT falls entirely within TSit” (108); Perfective: the TT coincides with a TSit that makes “a transition from a state where a certain situation obtains, to one where this is no longer the case” (110); Prospective: the TT is in the pretime of TSit (114). Perfect: the TT is in the posttime of the situation. (109); Progressive: the TT falls entirely within TSit (9).

Finiteness 

 201

Examples are krijgt (gets) vs. krege (got); neemt (takes) vs. nome (taken); valt (falls) vs. valle (fallen) or vald (*falled); ziet (sees) vs. ziend (*seed); vindt (finds) vs. vonne (found).⁶⁰ The lexical opposition between state and action predicates represents the opposition between imperfective aspect and progressive or prospective aspect. Examples are zit (sit) vs. zitte (sit-down); rije (drive) vs. rije (go-drive). The lexical opposition between state and result-state predicates represents the opposition between imperfective and perfect aspect. Examples are heeft (has) vs. had (had); is (is) vs. weest (been); slaapt (sleeps) vs. *slaapt (=geslapen, slept); doet (does) vs. daan (done); kijkt (looks) vs. *kijkt (=gekeken, looked). In sum, state predicates such as staat (stands) or blijft (stays) are used to express imperfective aspect. Change-of-state predicates such as komt (comes) or valt (falls) are used with perfective aspect. Both types of predicate occur in initial position. Furthermore, action predicates such as rijden (drive) as in gaat rijden (will drive) and eten (eat) as in is aan het eten (is eating) are used to express prospective or progressive aspect. They occur in final position. Predicates such as gevallen as in is gevallen (is fallen) or potmaakt as in heeft potmaakt (has kaputmade) are used to express perfect aspect. They occur in final position, too. Given that, at the lexical stage, aspect serves to express a lexical property, learners cannot but conclude that the complementary distribution of lexical predicates, i.e. the relation between the semantics of the predicate and placement in initial or final position, is based on lexical aspect.

aspect

imperfective

perfective

prospective

perfect

progressive

semantics position

‖ state initial

‖ change of state initial

‖ action final

‖ result state final

Table 7: Lexical expression of aspect in early L1 Dutch

The distribution of lexical predicates in terms of placement is summarized in Table 7. It shows that children are sensitive to aspectual distinctions from early on. At the functional stage, this sensitivity enables them to acquire the formal means to express aspectual differences in terms of grammatical aspect.

60 * = incorrect in the target language.

202 

 The functional stage

The linguistic means to express grammatical aspect seem to develop together with the use of the auxiliary verbs heb, heeft or ben, is. At the lexical stage, however, heb, heeft and ben, is are state predicates. At the relevant stage, heb, heeft are used to express a ‘state of possession’, i.e. possession of either a physical object as in dit hebt wij (this have we. J 1;11) or a physical condition as in poppie heef dors (doll has thurst. J 2;0); ben, is are used to express a ‘state of being’ as in dit is dop (this is cap. J 1;11). Examples of the lexical use of heb, heeft and ben, is are given in (41). (41) The lexical use of heb, heeft and ben, is in child L1 Dutch heb, heeft: state of possession

ben, is: state of being

heb jij? (J 1;11) [what] have you? dit hebt wij. (J 1;11) this have we poes heef tanne hier inne mond. (J 2;0) kitty has teeth here in-the mouth poppie heef dors. (J 2;0) doll has thurst Maria heef blauwe oge. (J 2;1) Maria has blue eyes nou heb ik nie meer koud. (J 2;2) now have I no more cold Jaja heef au. (A 2;1) J has ow Jaja heef koud. (A 2;1). J has cold da heef ik au. (A 2;2) there have I ow jeuk hem ik. (A 2;2) itch have I Jaja hem niet plakhanne. (A 2;2) J has not sticky-hands Mijnie heeft rooie neus. (A 2;3) M has red nose Ruudje hemme koute voete. (A 2;3) R has cold feet dors hem ik. (A 2;3) thurst have I

dit is dop. (J 1;11) this is cap dit is voor mama. (J 1;11) this is for mommy poes is onne tafel. (J 2;0) kitty is under-the table die is leuk. (J 2;0) that-one is nice is de plak nou? (J 2;1) [where] is the glue now? da is niet poes. (J 2;1) there is not kitty deke is nou? (A 2;0) blanket is now? hier issie. (A 2;0) here is-it papa is niet beer. (A 2;1) daddy is not bear nee, is niet zandbak in. (A 2;1) no, is not sandbox in Tita is? (A 2;1) [where] T is? issie jarig. (A 2;1) is-he birthday die is van jou, toti. (A 2;2) that-one is of you, pencil waar is v[l]iegtuig nou? (A 2;3) where is airplane now?

Finiteness 

Jaja hem vieze hanne. (A 2;3) J has dirty hands ikke Jaja hem nog mooie nagels. (A 2;3) I J have still nice nails

 203

almaal is troep. (A 2;3) everything is rubbish die is bove. (A 2;3) that-one is upstairs

The use of heb, heeft and ben, is in (41) is as it occurs in the target language. Hence, utterances as in (41) are used both at the lexical and the functional stage. The data in (42), however, typically appear at the functional stage. (42) The lexical-aspectual use of heb, heeft and ben, is in child L1 Dutch heb/heeft: ‘possession’ as the result state of a causal action

ben/is: ‘being’ as the result state of a change of state

heb jij melkje in? (J 1;11) have you milk in? heef nie kleertjes aan. (J 2;0) has not clothes on papa heef niet nog [thee] op. (J 2;2) daddy has not yet tea up Jaja heb mondje vol. (A 2;2) J has mouth full

is slot. (J 1;11) is closed is Cynthia weg? (J 1;11) is C gone die is almaal pot. (J 2;1) that-one is all broken disse pot. (A 2;0) this-one is broken

Jaja heb deze nodig. (A 2;2) J has this needy Mijnie hem niet zonneb(r)il op? (A 2;2) M has not sunglasses on? papa hemme ook b(r)il op. (A 2;2) daddy has also sunglasses on ik HEM niet broek aan. (A 2;3) I have not pants on ikke heef hoedje op. (A 2;3) I have hat on opa’s hemme bril op. (A 2;3) grandfathers have glasses on baby heefe goene uit. (A 2;4) baby has shoes off

is over. (A 2;1) is covered Mijnie is hakker. (A 2;2) M is awake Ernie is niet pot. (A 2;2) E is not broken hek is nog ope. (A 2;3) gate is still open is de deur al dicht. (A 2;3) is the door already closed dat is k[l]aar. (A 2;4) that is done nou issie uit. (A 2;4) now is-he out

As shown in (42), the lexical verb heb, heeft with its meaning ‘state of possession’ is used in utterances as in ikke heef hoedje op (I have hat on. A 2;3). Here, the physical object of possession hoedje (hat) occurs with a particle op (on) that

204 

 The functional stage

indicates that the state that the physical object is in, is the result state of a causal action.⁶¹ Similarly, the lexical verb ben, is with its meaning ‘state of being’ is used in utterances as in die is almaal pot (those is all kaput. J 2;1). Here, the particle (ka)pot indicates that the state that the theme is in, is the result state of a change of state. The particular use of heb, heeft with a result-state predicate indicating ‘possession’ as the result state of a causal action seems to be the link between the lexical use of heb, heeft and its functional use. The same is true for ben, is with a result-state predicate indicating ‘being’ as the result state of a change of state. A result state, as expressed in the utterances in (42), is the state of affairs that occurs at a time span after the situation time of a particular event. The same is true for the auxiliary verbs heb, heeft and ben, is that are used to express the functional property of perfect aspect. Therefore, it makes sense that the lexicalaspectual use of heb, heeft and ben, is with a result state predicate as in (42) will be the link between the lexical use of heb, heeft and ben, is in (41) and the functional use in (43). (43) The functional use of heb, heeft and ben, is in child L1 Dutch result state / perfect aspect ikke hè dit pakt. (J 2;1) I have this got heb je visje gehad? (J 2;1) have you fish had? ik heb wonne. (J 2;1) I have won ik heef afspoeld. (J 2;2) I have rinsed

die is altijd opde televisie geweest. (J 2;2) that-one is always on-the tv been zo, nou’s weer aflope. (J 2;2) so, now-is again done waa ben je nou geweest? (J 2;2) where are you now been?

Jaja hemme al goonmaakt. (A 2;2) J has already clean-made hemme nogge f[l]ippo vonne? (A 2;2) have-we another flippo found? ikke hemme deze tekend. (A 2;3) I have this drawn

da issie varre. (A 2;2) there is-he fallen da ben ikke ook wees. (A 2;3) there am I too been die isse Jaja gete. (A 2;4) that-one is J forgotten

61 Gilles (2003: 178) has similar examples from Jolien: ik heb die uit (I have that-one off. 2;1); ik wil ook een trui aan (I want also a sweater on. 2;2).

Finiteness 

ik heef óók appel gete. (A 2;4) I have too apple eaten Jaja heef met de haartjes zo doet. (A 2;4) J has with the hairs so done.

 205

Ruudje is uittefaapt. (A 2;4) R is finished-slept Jaja is zelf naa bove gaan. (A 2;4) J is self to upstairs gone

The lexical-aspectual category ‘result state’ and the functional category ‘perfect aspect’ are linguistically different. A lexical category is an open class of linguistic elements that have a particular semantic property in common. A functional category is a closed class of linguistic elements that are grammatically defined relative to one another. Therefore, as soon as learners are able to use the auxiliary verbs heb, heeft and ben, is as morphological elements of the tense-aspect system, they show further evidence of the acquisition of the projection of F. At the functional stage, it is the acquisition of the functional position of F that enables learners to express aspectual distinctions with the use of the grammatical category of auxiliary verbs. At the relevant stage, subcategories of auxiliary verbs occur in grammatical opposition. That is, heb, heeft and ben, is are used to express perfect aspect, doe, doet (do, does) are used to express progressive aspect and ga, gaat (go, will) are used to express prospective aspect. The examples in (44) are evidence of the distributional opposition between heb, heeft or ben, is, doe, doet and ga, gaat in Jasmijn (2;0–2;2). Noteworthy is the use of the lexical verb maken (make), which is used in the aspectual contexts of perfect aspect, progressive aspect and prospective aspect. Thus, Jasmijn produces utterances such as die heeft jou maakt (that-one has you made. J 2;1); doe je Pino make? (do you P make. J 2;0); ik gaat Pino make (I will P make. J 2;0). They show her ability to express aspectual distinctions in terms of a closed-class functional category system. That is, a linguistic system in which the grammatical function of a morphological element is defined relative to the grammatical function of the other members of the same category. (44) Grammatical aspect in child L1 Dutch: Jasmijn perfect: heb, heef(t); ben, is heef Cynthia maakt. (J 2;0) has C made heb je visje gehad? (J 2;1) have you fish had? ikke hè dit pakt. (J 2;1) I have this got ik heb wonne. (J 2;1) I have won

206 

 The functional stage

die heeft jou maakt. (J 2;1) that-one has you made ik heef afspoeld. (J 2;2) I have rinsed die is altijd opde televisie geweest. (J 2;2) that-one is always on-the tv been waar ben je nou geweest? (J 2;2) where are you now been? die heb ik wel geplakt. (J 2;2) that have I indeed glued progressive: doe, doet doe je Pino make? (J 2;0) do you P make? ik doet neusje snuite. (J 2;0) I do nose blow poesje doet likke. (J 2;1) kitty does lick doetie haaltje telle. (J 2;2) does-he story tell prospective: ga, gaat ik gaat Pino make. (J 2;0) I will P make ik ga eve die glije. (J 2;0) I will just that slide ga je ook school toegaan? (J 2;1) will you also school to-go? gaat Cynthia slape? (J 2;1) will C go-sleep? The examples in (45) are evidence of the distributional opposition between hemme, heef, ben, is vs. doe, doet vs. ga, gaat in Andrea (2;2–2;4). Noteworthy is the use of the lexical verb eten (eat). Andrea uses it in the aspectual contexts of perfect aspect, progressive aspect and prospective aspect as in ik heef óók appel gete (I have too apple eaten. A 2;4), Jaja doet kitkat opete (J does kitkat up-eat. A 2;3), gaan ze almaal ete, zie? (will they all eat, see. A 2;3).

Finiteness 

 207

(45) Grammatical aspect in child L1 Dutch: Andrea perfect: hemme, heef; ben, is kijk maa da issie varre. (A 2;2) look there has-he fallen Jaja hemme al goonmaakt. (A 2;2) J has already clean-made ikke hemme deze tekend. (A 2;3) I have this drawn da ben ikke ook wees. (A 2;3) there am I also been isse barnies affehope mam? (A 2;4) is barnies finished mommy? ik heef óók appel gete. (A 2;4) I have too apple eaten progressive: doe, doet, (doen) doen ze same hope (hope = lopen). (A 2;2) do they together walk Jaja doet kitkat opete. (A 2;3) J does kitkat up-eat ik doe mij fesje aan mij jamaboek doen. (A 2;4) I do my vest on my pyjamas do prospective: gaat, (gaan) Jaja gaat daar zitte en papa gaat daar zitte. (A 2;2) J will there sit and daddy will there sit gaan ze almaal ete, zie? (A 2;3) will they all eat, see? Jaja gaat liedje make van jou. (A 2;3) J will song make of you gaat ikke ook mee naa paardrije? (A 2;4) will I also with to horse-ride? In sum, the examples in (44) and (45) are evidence of the acquisition of the morphological means to express aspectual distinctions in L1 Dutch. At the relevant stage, heb, heeft, ben, is, vs. doe, doet vs. ga, gaat are used as the auxiliary verbs to express perfect, progressive and prospective aspect. At the lexical stage, these verb forms were used as predicates each with their own lexical meaning. The

208 

 The functional stage

use of these verb forms as functional elements to express aspectual distinctions becomes possible with the acquisition of the functional position F.

6.4.3 Tense At the relevant stage, there is virtually no evidence of tense. There are some past tense forms that are frequently used, i.e. was (was), had (had), deed (did), liep (walked), ging (went), zag (saw), zei (said), maakte (made) etc. Although these verb forms may refer to a time span before the time of utterance, at the relevant stage, tense is part of their lexical meaning. It is not yet a property of the functional category system.

6.4.4 Agreement As noted in 6.1.2, reanalysis of modal predicates as instantiations of the functional category F serves as a prerequisite for learners to acquire the auxiliary verb forms heb, heeft and ben, is. Auxiliary verbs serve as carriers of both semantic and morphological finiteness. Morphological variation of the auxiliary verbs provides information with respect to person and number of the subject. However, in Dutch the inflectional system of auxiliary verbs shows a high degree of syncretism. This explains why, at the functional stage, modal and auxiliary verbs are used with both targetlike and non-targetlike agreement morphology. Evidence of the amount of variation in inflectional morphology is given in (46) and (47). (46) Agreement morphology in Jasmijn (2;0–2;2) a. b. c. d.

wil, *wilt (want-1Sg); wil? (want-2Sg); wil, *wilt (want-3Sg).⁶² heb, *heef (have-1Sg); heb? (have-2Sg); *heef, heeft (has-3Sg); was (was1Sg); ben? (are-2Sg); is (is-3Sg). *doet (do-1Sg); doe? (do-2Sg); doet (does-3Sg). ga, *gaat (go-1Sg); ga? (go-2Sg); gaat (will-3Sg).

(47) Agreement morphology in Andrea (2;2–2;4) a. b.

wil,*wilt (want-1Sg); wil? (want-2Sg); wil,*wilt (want-3Sg). *heef,*hem (have-1Sg); *hem (has-3Sg); ben (am-1Sg), is (is-3Sg).

62 * = incorrect in the target language

Summary 

c. d.

 209

doe (do-1Sg); doet (does-3Sg); doen (do-3Pl). ga, *gaat (go-1Sg); gaat (go-3Sg); gaan (go-3Pl).

The evidence given in (46) and (47) indicates that is unlikely that the morphological properties of agreement should play a role in the acquisition of verb raising. Morphological inflection at the functional stage shows that agreement would be the result rather than the cause of the acquisition of finiteness.

6.5 Summary In child L1 and in adult L2 Dutch, learner varieties develop from a lexical system to a functional system. At the lexical stage, learner grammar is relatively simple. Functional categories are absent. Hence, topicalization, auxiliary verbs, verb raising, agreement, tense, definiteness and anaphoric pronouns do not yet play a role. Utterance structure is determined by lexical projections that interact with principles of information structuring. In lexical structures in which an agent plays a role, the agent occurs in initial position as in Jaja mag dop opdoen (J may cap on-do. A 2;0). In lexical structures in which there is no role for an agent, the theme occurs in initial position as in deze magwel (this may-indeed. J 1;10) or poppie valt hier (doll falls here. A 2;0). Thus, a non-agentive argument may only occur in initial position if the agent is not available. Both types of utterances, referred to as type A (agentive) and type B (non-agentive), are lexical structures that specify a hold-for relation between a predicate and a constituent that may function either as an agent or a theme. The constituent that the predicate holds for is the subject. At the lexical stage, learner grammar has no structural topic position. Hence, topicalization cannot be expressed with the functional means of the target system. However, at the relevant stage, topicalization can be achieved with the structural possibilities that are available. In utterances with an agentive lexical projection, the object may occur in initial, topic position as in, for example, disse hoeniet meeneme (this-one must-not withtake. A 2;1). This type of structure does not provide a position for the agent and, therefore, a non-agentive argument, i.e. the object, may occur in initial, topic position. As a consequence of this, in the absence of a position for the agent, the modal predicate serves to express the function of the agent implicitly. However, absence of a position for the agent is not targetlike. This explains why at the relevant stage, the function of the agent comes to be expressed with an affix (ie) as in doettie omdraaie (does-he aroundturn. J 1;11). Reanalysis of the affix as a pronoun as in mag IK doen (0 may I do. A 2;3) or as an NP as in broodje mag Cynthia wel opeten (roll may C indeed up-eat.

210 

 The functional stage

A 2;4) establishes a subject position for the agent. In order to provide for both a subject and a topic position, the utterance structure of the learner grammar needs two specifier positions: SpecVP for the subject and SpecFP for the topic constituent. SpecVP is projected by V as the lexical head of the predicate-argument structure. SpecFP is projected by F as the functional head of the information structure. At the lexical stage, the pragmatic function of assertion is carried by the lexical predicate. With agentive predicates the pragmatic function of assertion is expressed by a modal predicate such as wil (want), nee (don’t want), kanwel (canindeed), kanniet (can-not), by an aspectual predicate such as doet (does) or gaat (will), or in the default case by is (is) or an empty position. At the functional stage, the acquisition of the functional projection FP allows verbal elements in the position of F to be used as carriers of finiteness, i.e. they are used to express that the predication is true for a particular time and space. Spatio-temporal anchoring is achieved with elements in the position of SpecFP, i.e. the functional topic position. The establishment of a functional position F for the expression of finiteness leads to the reanalysis of unanalysed lexical modal expressions such as unne (want), mag-ikke (may-I), nee (want-not), kanwel (can-indeed). As elements of a functional category, they come to function as auxiliary verbs. With the functional position F as a position for auxiliary verbs, modal verbs can be used both as deontic and as epistemic modal verbs. As deontic modals they express the assertion of the willingness, the ability, the permission or the obligation of an agent to perform an action, as epistemic modals they express the possibility, the optionality or the necessity that an event occurs. Finally, the functional position of F serves as a prerequisite for the acquisition of the auxiliary verb heb, heeft (have, has) and ben, is (am, is) and, eventually, the possibility for the lexical verb to occur in the position of verb-second. To conclude, at the lexical stage, agentive predicates can only occur with an object in initial, topic position if the agent is not expressed. Reanalysis of the lexical modal head as a functional head provides a position for both a constituent with topic function as the specifier of FP and an agent as the specifier of VP. FP serves as the structural prerequisite for the expression of the functional properties of information structure of the target language. Hence, topicalization is the driving force in the development of learner languages from a lexical system to a functional system. That is, from a system in which utterances are the expression of a lexical projection to a system in which utterances are the expression of a structure with both a functional projection FP and a lexical projection VP as the complement of F.

7 Finiteness in language acquisition research 7.1 Finiteness in first language research 7.1.1 L1 Dutch: Gillis (2003) In his case study on child L1 Dutch, Gillis (2003) investigated the early acquisition of verb morphology in Jolien, a girl growing up in a Flemish-Dutch speaking environment. Gillis collected this girl’s spontaneous production data when she was (1;5) to (2;5) years of age. He analysed her data in terms of what he calls ‘mini-paradigms’. Gillis takes these mini-paradigms to be early evidence of morphological variation. With the emergence of mini-paradigms children are claimed to have reached the so-called ‘proto-morphological stage’. At the proto-morphological stage the child has left the initial ‘pre-morphological stage’ when morphological variation does not yet play a role. Gillis’ study focusses on the use of verb morphology with main verbs, auxiliaries and the copula. He shows that mini-paradigms are the outcome of a developmental process that occurs in Jolien after the age of (2;0). The relevant data of the types of verb that Jolien uses between (1;5) and (2;5) are represented in Table 1 (Table 4, in Gillis 2003: 185).

Age

% Main verb

% Auxiliary

1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2 2;3 2;4 2;5

92.86 97.67 94.12 100.00 97.06 93.33 96.77 80.60 80.20 60.00 40.00 39.90 62.77

7.14 2.33 5.88

% Copula

% Auxiliary + Main verb

16.42 9.90 29.57 31.25 36.36 20.44

2.94 4.44 3.23 2.99 5.94 6.09 21.25 13.64 13.14

2.22

3.96 4.35 7.50 10.10 3.65

Table 1: Percentage of utterances with a main verb, auxiliary, copula and a periphrastic verb phrase.

212 

 Finiteness in language acquisition research

In Table 1, Gillis distinguishes between main verbs, auxiliaries as single (finite) verb forms, the copula and periphrastic verb phrases. He uses the term ‘auxiliary’ for both modal and aspectual auxiliary verbs. Periphrastic verbs are verb phrases with a main verb and an auxiliary verb. The figures in Table 1 show that it is relevant to discriminate between auxiliaries as single (finite) verb forms and auxilaries in periphrastic verb phrases. The former category occurs in small numbers from the beginning of the acquisition process and is used throughout the whole period. The latter category seems to become firmly established between (2;1) and (2;4). The difference between the two types of auxiliaries is due to the fact that initially, i.e. from (1;5), the child produces utterances such as kan niet ‘can not’. In fact, it seems “the only auxiliary that shows up until 1;9” (Gillis 2003: 190f.). After (1;10) the child begins to use kan in verb phrases such as kan hard bijten (can hard bite. 2;0), kan het niet vertellen (can it not tell. 2;2), die kan staan (that-one can stand up. 2;4). The relevant observation is evidence of the use of two types of verb, both of which Gillis calls ‘auxiliaries’. However, the single modal verb kan niet is in fact a thematic verb. That is, it is a lexical verb meaning either ‘I am not able to do it’ or ‘it is impossible’. As a lexical verb, it has to be distinguished from modal verbs that are used in periphrastic verb phrases. While the thematic use of modal verbs by Jolien occurs from age (1;5), the periphrastic use seems to become productive between (2;0– 2;4). The periphrastic use of modal verbs is evidence that the language system at the relevant stage has established both lexical and functional categories. It explains why it is that from (2;1) onwards Jolien is able to use the functional auxiliary verb heb, heeft, too. Examples are utterances such as, for example, mama heeft verteld hier (mama has told [a story] here) (Gillis 2003: 191). Also relevant seems Gillis’ observation that “finite verbs only start being acquired and used regularly once the child has detected two verb positions in the sentence” (Gillis 2003: 198f.). The evidence is given in Figure 1 (Figure 2 ‘Verb lemmas and verb types’, in Gillis 2003: 182). The relevant data show that from (2;1) onwards, with the same verb lemma different verb forms (verb types) are used. Hence, verb forms come to be used with morphological variation. Thus, from (2;1) onwards, i.e. with the periphrastic use of auxiliary verbs as shown in Table 1, lexical verbs that where initially learned as infinitives are used in morphologically finite form, too. It indicates that the acquisition of a functional position for the auxiliary verb serves as a prerequisite for the placement of the lexical verb with finite morphology in sentence-final position. Finally, there is another phenomenon that seems to be part of the same developmental process. Evidence is given in Figure 2 (Figure 5, in Gillis 2003: 186). Figure 2 shows the increase in the use of overt subjects between (2;1–2;4). Furthermore, it indicates that “sentences with a subject overwhelmingly prefer a finite

Finiteness in first language research  

 213

Figure 1: Number of new verb lemmas and forms in the childs cumulative vocabulary.

Figure 2: Percentage of sentences with with/without an overt subject and the accompanying finite / non-finite verb form

verb” (Gillis 2003: 185f.). At the same time “the number of sentences with correct agreement is considerable” (Gillis 2003: 186). Hence, Gillis concludes that the use of overt subjects “seems to coincide with the discovery of agreement between subject and verb” (Gillis 2003: 199). In sum, with the observational data on the acquisition of verb morphology in Jolien, Gillis (2003) shows that around (2;0) and (2;1) there are four processes of language development that occur more or less simultaneously. At the relevant stage, developmental progress occurs (1) from a stage at which there is a 1:1-cor-

214 

 Finiteness in language acquisition research

respondence between verb type and verb form to a stage at which “morphological variation increases”, (2) from a stage at which “the child’s verb lexicon consists almost exclusively of lexical verbs (used mainly as infinitives)” to stage at which “the copula zijn ‘be’ and a growing number of (modal) auxiliaries are acquired”, (3) from a stage with no morphological variation to a stage where “marked finite forms (second and third person singulars, as opposed to bare stems) are used with growing frequency” and, finally, (4) from a stage with no overt subjects to a stage with “overt marking of subject-verb agreement” (Gillis 2003: 195). These developmental phenomena are evidence of what Gillis calls “a relatively clear delineation of a premorphological and a protomorphological stage” (Gillis 2003: 171). Gillis’ pre-morphological stage seems identical to what I refer to as the lexical stage. At the relevant stage, morphological variation is absent. Lexical verbs are used as unanalysed elements in either head-initial or head-final position. Gillis’ proto-morphological stage seems identical to what I call the functional stage, i.e. the stage at which functional elements become integrated into the developing system of learner language. With respect to the focus of research in Gillis (2003), there is one observation in his study that is not central to the discussion. It is the fact that at the premorphological stage children “seem to use only one verb type for each lemma: only a finite or a non-finite type appears in the early stages of acquisition”. Furthermore, verb types with finite morphology occur in initial position and are less frequent than non-finite verb types that are always in final position. Hence, the distribution of verb forms is based on both ‘positional preferences’ and ‘positionform correlations’ (Gilles 2003: 196). Gillis notes that this observation has also been made in the earliest studies on verb acquisition in Dutch by De Haan (1987) and Jordens (1990). He refers to the observation in Jordens (1990) that “verbs with finite morphology in first/second sentence position differ semantically from verbs in sentence final position that take non-finite morphology” (Gilles 2003: 198). Furthermore, as in Jordens (1990), Gillis argues that, initially, infinitives are typically eventive verbs, while finite verbs are typically stative verbs. “Eventives are verbs that refer to dynamic changes that occur within a relatively bounded interval, involving one or more actors. Transactive action verbs, which denote exchanges between an agent and a patient (e.g., ‘hit’, ‘kiss’) are the prototypical examples. Statives such as ‘love’, ‘know’ denote relatively stable, unbounded conditions or situations” (Gillis 2003: 199). Gillis’ analysis of his spontaneous production data, thus, confirms the fact that for lexical verbs there is initially a correlation between position, form and distribution. That is, with lexical verbs there is one subset that occurs in initial position and is morphologically finite, whereas there is another subset that occurs in final position and is morphologically non-finite.

Finiteness in first language research  

 215

As I have argued in Chapter 5, at the relevant stage this distribution can be accounted for in terms of a systematic opposition between agentive and non-agentive types of utterance. In agentive types of utterance, action predicates are the complement of a modal or an aspectual head constituent that is used to express control (Ctrl). This head constituent occurs in initial position and, hence, action predicates are placed in final position. In non-agentive types of utterance, stative predicates serve as the head constituent and as such they occur in initial position. Furthermore, the position of the predicate determines the morphology with which it is learned. Thus, the presence or absence of the expression of control explains not only the positional difference between action predicates and statives but also the fact that, morphologically, action predicates are ‘infinite’ and statives are ‘finite’.⁶³ At the functional stage, i.e. with the acquisition of a functional position for verbal elements to express finiteness, both statives and action predicates may occur in initial position. As pointed out in Chapter 6, it is the acquisition of a functional position for verbal elements that explains the acquisition of variation in verb placement and with it the acquisition of verbal morphology. The relevant stage is what Gillis (2003) refers to as the proto-morphological stage, i.e. the stage at which marked finite verbs occur as part of a morphological mini-paradigm.

7.1.2 L1 German: Bittner (2003) In her study on the emergence of verb inflection, Bittner (2003) collected data from two German speaking girls: Anna at (1;8–2;1) and Caroline at (1;6–2;2). As in Gillis (2003), Bittner (2003) distinguishes between a pre-morphological and a proto-morphological stage. Bittner (2003) also observed that at the pre-morphological stage “most verbs are attested in only one morphological form (…)”. More specifically, she found that, “whereas Anna concentrates on -en and -t forms (…), Caroline doesn’t show any preference for other than -en forms” (60). With the emergence of the first form contrasts, the two girls show a further difference. In

63 Given that action verbs are more frequent than state or change-of-state verbs and given that action verbs in the position of the complement are infinite, this explains why “infinitives have a much higher TTR [type-token ratio] than finite verbs in the input” (Gillis 2003: 197). Wijnen, Kempen and Gillis have come to the same conclusion as they state: “Our analyses have yielded moderate correlations (at best) between frequency of verb forms in the input and age of first appearance” (2001: 657). That is, “[t]hey found that for the verbs that occur in the children’s productions as an infinitive and as a finite form, the infinitive was almost always acquired before the finite form although for a considerable number of verbs the infinitive was not the most frequent form in the input” (Gillis 2003: 197).

216 

 Finiteness in language acquisition research

opposition to the most frequently occurring -en form, Anna uses the -t form in 3Sg-contexts, whereas Caroline uses the -0 form mainly in 1Sg-contexts. Bittner explains this difference between Anna and Caroline as due to the ‘more objectoriented’ behaviour of Anna and the ‘more self-oriented’ behaviour of Caroline (63). Interestingly, this seems to be reflected in the use of modal verbs. With Anna “no modal verbs are found in [her] premorphological stage”. Caroline, however, as shown in (1), uses “muß ‘have to’, soll/sollen ‘should’, dürfe ‘may’, kann ‘can’, mag ‘like’ and will/wollt ‘want’, while “will – her most frequent modal form – is recurrent since 1;8” (59). Examples are given in (1). (1) L1 German: Modal verbs in Caroline (1;6–1;11) muss blau nehm(en. (1;6.13) [I] must blue take-Inf arm will. (1;8.24) on-the-arm [I] want will kein, weg. (1;9.3) [I] want not, gone mami, nochmal will. (1;9.6) mommy, one-more-time [I] want will dis / will de. (1;9.20) [I] want this / [I] want this will / will dis, glosser [= große messer]. (1;9.25) [I] want / want this, big-knife will da bleiben. (1;10.16) [I] want there stay-Inf auch (r)ot will. (1;10.23) also red [I] want dürfe essen. (1;10.29) [I] may eat-Inf will. (1;11.2) [I] want soll nicht / mag nicht. (1;11.4) [I] should not / [I] may not gar nix kann nix. (1;11.8) [I] absolutely nothing can nothing soll(e)n der halten brot. (1;11.8) [I] should that-one hold-Inf bread will nich(t). (1;11.14) [I] want not

Finiteness in first language research  

 217

ich will anhüpfen. (1;11.17) I want hop-Inf These utterances are typical for learners at the lexical stage of language learning. At the relevant stage, modal predicates are used as a device to express the willingness, the ability, the permission or the obligation of an agent to perform an action. Since the agent often refers to the speaker, these utterances serve to express variation with respect to the pragmatic function with which they are used. At the relevant stage, this is particularly the case for Caroline who is claimed to be ‘more self-oriented’. At the lexical stage, as shown in Chapter 5, modal predicates occur as the head of an agentive utterance structure with an action predicate as its complement. In such a context the action predicate is used with -en morphology as it occurs in the input. This explains why, initially, Caroline only uses verbal -en forms. Anna, on the other hand, is characterized as ‘more object-oriented’. This may explain why, instead of utterances with modal verbs, she produces utterances with lexical verb forms with both -en and -t morphology. At the proto-morphological stage, i.e. with the emergence of the first form contrasts, it comes as no surprise that in opposition to the most frequently occurring -en form, Anna, as the child who is ‘more object-oriented’, uses the -t form in 3Sg-contexts, whereas Caroline, as the child who is ‘more self-oriented’, uses the -0 form mainly in 1Sgcontexts. The proto-morphological stage is claimed to occur for Anna at (1;11-2;1) and for Caroline at (2;0–2;2). Evidence for this is the use of periphrastic perfect constructions with the auxiliaries hab (have-1Sg), haben (have-1Pl) and ist (is-3Sg) and periphrastic modal constructions with modal auxiliaries as, for example, will (want-1Sg), sollen (must-1Pl), muß (must-1Sg), kann (can-1Sg) (Bittner 2003: 66). Examples are given in (2) and (3). (2) L1-German: Periphrastic constructions in Anna (1;11–2;1) gestern hab ich zu(ge)guckt. yesterday have I watched (wir) lange spielt haben. we long-time played have ich will essen. I want eat malen sollen wir. draw must we

218 

 Finiteness in language acquisition research

(3) L1-German: Periphrastic constructions in Caroline (2;0–2;2) mami (ich) macht hab. mommy, [look] (I) made have mann brunnen fallen ist. man fountain fallen is muß aber uhu anmalen. [I] must just owl draw reintun kann schon. in-put [I] can already The acquisition of auxiliaries in periphrastic verb phrases is evidence of a language system that has two verb positions, i.e. a functional position to express finiteness and a lexical position for the main verb. This accounts for the phenomena as they appear in Anna at (1;11–2;1) and in Caroline at (2;0–2;2). That is, with the use of periphrastic auxiliaries as in (2) and (3) Bittner sees the emergence of mini-paradigms as evidence that “the finite verb becomes regular”(Bittner 2003: 78). She notes: “With Anna, it seems to happen after age 2;0.5; with Caroline after age 2;1a” (Bittner 2003: 71f.). A mini-paradigm of verb inflection is defined as a primitive system of inflectional contrasts recurrent in (functionally) contrasting contexts. Three member contrasts for Anna (1;11–2;1) occur with haben (have), sein (be), wollen (want), machen (make), essen (eat), gehen (go), hingehen (go-to). Examples are hab 1Sg, hast 2Sg, hat 3Sg*1Sg, haben 1Pl*1Sg, haben Inf; ist 3Sg, sind 1.3Pl, war 3Sg.Past; will 1.3Sg, willst 2Sg, wollte 3Sg.Past, wollen 1.3Pl; mach 1Sg, macht 3Sg, machen *1Sg, mach Imp; eß 1Sg, *eß 2Sg, ißt 2Sg, *eßt,*esse 3Sg, essen *3Sg, essen Inf; geht 3Sg, gehen 3Pl, [ge]gangen Pp; hingehst 2Sg, hingeht 3Sg, hingehen *3Sg. See Bittner (2003: 69f.).⁶⁴ Three member contrasts with Caroline (2;0–2;2) occur with haben (have), sein (be), machen (make), singen (sing), halten (hold), anmalen (on-draw), sagen (say), tanzen (dance). Examples are hast 2Sg, haben *1Sg; bin 1Sg, ist 3Sg, sind 3Pl; mach 1Sg*3Sg*1Pl, mache 1Sg, macht 3Sg*1Sg, machen 3Pl*1Sg, (g)emacht Pp; singt 3Sg; singe Imp; halte 1Sg, hält 3Sg, halten *3Sg; anmal 1Sg, anmalen *1Sg, an(g)emalt Pp; sagt 3Sg, sag 2Sg*3Sg*3Pl, sag Imp; tanz 1Sg, tanzt 3Sg, tanzen 3Pl*3Sg. See Bittner (2003: 70f.). Three member contrasts are evidence of the emergence of mini-paradigms of verb inflection, They appear simultaneously with the acquisition of the aux-

64 * indicates that in the relevant context, the inflectional form is not used in a targetlike way.

Finiteness in first language research  

 219

iliary verbs haben (have) and sein (be). Given the process of the acquisition of functional categories as described in Chapter 6, this is precisely what is to be expected. As argued in Chapter 6, the use of the auxiliary verbs heb, heeft (have, has) and ben, is (am, is) in Dutch shows that the language system has created a functional position for the expression of finiteness. It is the creation of this functional position that is a prerequisite for the acquisition of inflectional morphology with lexical verbs. With the acquisition of a position for the finite verb, the language system also provides the possibility for learners to acquire the morphological means of temporal embedding. This accounts for “the emergence of past forms” (Bittner 2003: 78). Examples are given in (4) and (5). (4) L1 German: Past forms in Anna (zu) groß waren. (1;11.20) too big were (wer) wars? (1;11.30) who was-it? da war jemand. (2;0.5) there was someone wollt(e) nicht kommen / klopfte da (imitation). (2;0.29) wanted not come / knocked there der onkel kam. (2;5.18) the uncle came dachte. (2;6.20) thought aber die hexe konnte, hu die nich(t), schlufen. (2;7.5) but the witch could, hu that-one not, slept du solltest doch mit mir puzzle spielen! (2;7.5) you should indeed with me puzzle play! so was hab(e)n, hatt(e)n wir doch schon mal. (2;7.5) that have, had we indeed already before was hatt(e)n die denn? (2;7.5) what had they then? (5) L1 German: Past forms in Caroline war Eis? ausschneiden wollt ich? (2;2.29) was ice? out-cut wanted I? hase wollt guck eh? (2;3.0) hare wanted look eh?

220 

 Finiteness in language acquisition research

oh, ging so lang. (2;3.11) oh, took so long die löwe kam rein, elefant kam nur ein kleiner elefant. (2;3.12) the lion came in, elephant came only a little elephant aehm, warum war dis oben? (2;5.28) eh, why was this up-there? Furthermore, as argued in Chapter 6, the result of the acquisition of a functional projection for the expression of finiteness is that utterance structure has both a topic and a subject position. Hence, simultaneously with the acquisition of finite lexical verbs, learners systematically use both overt subjects and correct agreement. Evidence of the use of overt subjects in Bittner (2003: 58) is given in (6) and (7). Evidence of correct agreement in Bittner (2003: 78) is the emergence of -st forms in 2Sg.pres.ind contexts. Examples are given in (8) and (9). (6) L1 German: Overt subjects in Anna papa auch schlafen (daddy too sleep); schlafen alle (sleep everybody); papa macht (daddy made-Pp); bär esst [= isst (bear eats); hier kommt die mama hier (here comes that mommy here); mama liegt da (mommy lies there). (1;11.20) und der pieken (and that-one prick); wir beide malen (we both draw); Lucas (hat das) rumschmissen (Lucas has it around-thrown); du auch malen (you too draw); ausschlafen die (hamster) (sleep that-one [hamster]); hamster schlafen alle (hamster sleep all); Anna guckt (Anna looks); wo issa [= ist der] stachel … oder der ist auf toilette (where is the prickle or that-one is on the toilet); elefant ist da (elephant is there); da sind die (räder) (there are these [wheels]); ich möchte (I would-like). (1;11.30) (7) L1 German: Overt subjects in Caroline wasser teht [= geht?]. (2;0.7) water goes [?] is(t) da nase. (2;0.10) is there nose der uhu weint (the owl cries); mutter da is(t) (mommy there is); leim aussieht nomma [= nochmal] (glue looks one-more-time). (2;0.26) asa bringen [= hase springen] (hare jump); Ayche oln [= holen] (saft) (Ayche get [juice]); ich anzieh (I get-dressed); ich sollten um ma macht (I should on-madePp). (2;1.3) ich auch hinsetz (I too down-sit). (2;1.7) und ich weil dinken [= trinken] (and I want drink). (2;1.8)

Finiteness in first language research  

 221

die sind wei (?) (those are we); Susanne tann [= kann] nich(t) (Susanne can not). (2;1.9) die is(t) da (that-one is there); ich darf atz [= jetzt] (I may now). (2;1.14) da elefant ist (the elephant is), ne [=nein], darf er nich(t) (no, may he not); (wenn/ weil) ich schenk [= geschenkt] hab (if/because I given have). (2;1.15) (8) L1 German: 2Sg -st forms in Anna siehste [= siehst du]. (see-you). (2;0.5) du isst das? (you eat that?); du musst (you should); (wo) du hingehs(t)? ([where] you to-go?); du hast (you have); spange has(t) (clasp have [you]). (2;0.29) du willst kaffee? (you want coffee?); du möchtest kaffee? (you would-like coffee?); wenn du reingehst will ich reingehen mit (if you in-go want I in-go with); du nimmst den (you take that-one). (2;1.13) du willst mitbauen, wa? (you want with-build?); du kannst nicht raus? (you cannot out?); kaffee hast du aus(g)etrunken (coffee have you all-drunk); du musst auto fahren? (you must car drive?); du fahrst (you drive); du bist ein rowdy (you are a rowdy); du bist nicht müde (you are not tired); du machst das (you make that). (2;1.27) (9) L1 German: 2Sg -st forms in Caroline (das) weißt du mami (that know you mommy). (2;1.2) hast du. (have you). (2;2.12) mami, auch kuchen schenkst gut (mommy, too cakes give good). (2;2.13) auche [= auch das/brauche?] wasser, weißt du (too/need water, know you). (2;2.18) siehst ja wohl, affe sagt ich ich ich (see indeed yes, monkey says I I I). (2;3.0) weißt du angst vor dem? (know you fear for that-one?); ein dach musst du malen (a roof must you draw). (2;3.2) eine turm siehste, eine tolle (a tower see-you, a nice-one). (2;3.9) sagst du (say you). (2;3.11) weißt du mami (know you, mommy). (2;3.22) With the acquisition of a functional topic position, the learner system is equipped to express the function of contextual anchoring. In Bittner (2003: 58), this seems to account for “the emergence or increase of determiners and pronouns”. Evidence of this can be found in (6), (7), (8) and (9), too.

222 

 Finiteness in language acquisition research

7.1.3 Summary Gillis (2003), in his study on child L1 Dutch, and Bittner (2003), in her study on child L1 German, provide additional evidence for the two-stage model of acquisition as presented in Chapter 5 and Chapter 6. From the point of view of the acquisition of inflectional morphology, they discriminate between a pre-morphological and a proto-morphological stage. At the pre-morphological stage, the distribution of verb forms is based on “position-form correlations” (Gillis 2003: 196) and thus “most verbs are attested in only one morphological form” (Bittner 2003: 60). The proto-morphological stage is established with the detection of two verb positions in the sentence (Gillis 2003: 198f.) and the emergence of the first form contrasts (Bittner 2003: 78). As I have argued in Chapter 5, the lexical stage is a stage at which morphology does not play a role. Verb forms are learned with the form in which they occur in the input. At the relevant stage, there are two types of utterance structure. Utterances with action predicates in which the predicate occurs in final position and utterances with state or change-of-state predicates in which the predicate occurs in initial position. Due to the position in which these types of predicate occur in the input, action predicates appear with -en and state or change-of-state predicates with -t or -0 morphology. Given the fact that at the lexical stage the placement of action verbs, on the one hand, and state or change-of-state verbs, on the other, is due to the semantics of the verb, morphology is irrelevant. It explains the observation by Bittner and Gillis that, at the relevant stage, verb forms usually occur in only one morphological form. The acquisition of auxiliaries in periphrastic verb phrases is evidence of a process of language development in which the initial lexical system is given up in favour of a functional system. At the functional stage, the utterance structure has two verb positions, i.e. a functional position to express finiteness and a lexical position for the main verb. This accounts for the phenomena as Bittner observed them to appear at the proto-morphological stage in Anna at (1;11–2;1) and in Caroline at (2;0–2;2). That is, with the use of the auxiliaries haben (have) and sein (be) there is a functional position to express finiteness. In the absence of an auxiliary verb, this functional position may be taken by the lexical verb. This explains why, at the relevant stage, mini-paradigms of verb inflection emerge to express finiteness morphologically. Furthermore, with the acquisition of a position for the finite verb, verb morphology is used as a means to express temporal embedding. Finally, with the projection of a functional category for the expression of finiteness, utterance structure has both a topic and a subject position. The structural subject position accounts for the emergence or increase of overt subjects and, with it, for the acquisition of agreement morphology. Evidence of this is shown

Finiteness in second language research 

 223

by the emergence of -st forms in 2Sg-contexts. The topic position accounts for the fact that, regardless of its syntactic function, a nominal or adverbial phrase may be used with topic function. The establishment of the topic function as a function of contextual anchoring may also explain the fact that, at the relevant stage, determiners and pronouns will be used. As with topicalization, determiners and pronouns are used to express contextual features of information structure.

7.2 Finiteness in second language research 7.2.1 Verb placement and inflectional morphology 7.2.1.1 The contingency between verb placement and morphology Research on second language development has generated a number of studies on the acquisition of the functional category system, particularly in L2 learners of German, Dutch and French. In these studies, there has been an extensive debate on the acquisition of verb placement and morphology.⁶⁵ In German, Dutch and French there is a contingency between verb placement and morphology such that verb forms that are morphologically finite, occur in sentence-initial (verbsecond) position, while verb forms that are morphologically non-finite, occur in sentence-final position. In the relevant studies, variability in verb placement is usually defined in terms of the position of the verb with respect to the position of the sentence negator ‘not’: nicht in German, niet in Dutch, pas in French. As shown in Figure 3, there are two options. verb ‖ sentence-initial (verb-second) V is pre-Neg or: Neg is postverbal

←

Neg nicht niet pas

→

verb ‖ sentence-final V is post-Neg or: Neg is preverbal

Figure 3: Verb placement in German, Dutch and French

65 Becker (2005); Clahsen and Penke (1992); Eubank (1996); Jansen, Lalleman and Muysken (1981); Meisel (1992, 1997); Parodi (1998, 2000); Rule and Marsden (2006); Vainikka and Young-Scholten (1994, 1996); Verhagen (2009a, 2009b); Weissenborn (1990).

224 

 Finiteness in language acquisition research

On the one hand, when the verb form occurs sentence-initially, the position of the verb is pre-Neg or: Neg is postverbal. Examples are Jaja valt niet (J falls not. A 2;0) in child L1 Dutch and die naam hier staat niet (that name here stands not. Mo/A 2.8) in adult L2 Dutch. On the other hand, when the verb form occurs sentencefinally, the position of the verb is post-Neg, or: Neg is preverbal. Examples are niet omgooie (not overthrow. J 1;11) in child L1 Dutch and ik niet lezen op de huis (I not read on [=in] the house. Mo/A 1.4) in adult L2 Dutch. Under the assumption that the final position of the lexical verb is the base or default position, the position of verb-second is usually accounted for in terms of what in generative theory is commonly referred to as ‘movement’ or ‘raising’. In its base position, sentence-finally, the lexical verb is morphologically non-finite: it is either an infinitive or a past participle. In the position of verb-second, verb forms are morphologically finite, not only with lexical verbs that are accounted for in terms of raising, but also with auxiliary verbs. Finite verb forms appear with the morphological properties of agreement and tense. The contingency between verb placement and morphology with respect to the position of the negator is represented in Figure 4. verb ‖ sentence-initial (verb-second) V is pre-Neg or: Neg is postverbal ‖ movement / raising inflection: finite

←

Neg nicht niet pas

→

verb ‖ sentence-final V is post-Neg or: Neg is preverbal ‖ base position default form: non-finite

Figure 4: Verb placement and morphology

The contingency between verb placement (raising) and inflectional morphology (finiteness) plays a central role in studies on the acquisition of the functional properties of finiteness. The availability of this contingency is seen as evidence in favour of the idea that functional categories are a given property of the human language faculty, regardless if it becomes established at the beginning or later in the acquisition process.

7.2.1.2 Methodological questions Empirical studies on the acquisition of verb placement and inflectional morphology are usually based on either spontaneous or elicited production data. However, the evaluation of these data is not always unproblematic.

Finiteness in second language research 

 225

As far as spontaneous production data are concerned, there is the question of productivity. Utterances or phrases are usually described as generated on the basis of a system of internalized grammatical rules and principles. However, utterances or phrases may equally well be produced as unanalysed parts of speech that are acquired on the basis of rote learning. Thus, with respect to verb placement, how can one be sure that the linguistic data accounted for as verb raising are produced by rule? In some studies it is taken to be sufficient, if finite verb forms are used in sentence-initial position, while non-finite verb forms are used in sentence-final position. However, if rote learning is taken into account, the mere fact that for some verb forms morphology seems to correlate with a particular position does not count as an argument. In other studies, the criterion is the amount of overlap, i.e. the degree to which the same type of verb can occur both in sentence-initial and in sentence-final position. However, if overlap is a criterion for acquisition, the question is, how much overlap is needed to decide that the rules of verb placement are used productively indeed? As I argued in Chapter 5 for L1 and L2 Dutch, lexical verbs (predicates) are initially used in complementary distribution. That is, state and change-ofstate predicates occur in initial position, while action verbs occur in final positon. Given this to be the case, this distributional difference serves as counterevidence to the assumption that at the relevant stage verb raising has been established. A similar problem arises with respect to the acquisition of morphology. How can one be sure that morphological rules are used productively? If verb forms in learner language are morphologically identical to those that are used in the target language, does this mean that the underlying production process is the same, too? Different processes may lead to the same result. More specifically, the same form can be produced as rote learned in learner language, while in the target language it is produced by rule. Thus, the morphology of the verbal element itself is never reliable evidence of the acquisition of inflection. As I argued in Chapter 5 for L1 Dutch, initially lexical verb forms seem to be selective with respect to their morphological form. That is, state and change-of-state verbs that occur in initial position are always finite, while action verbs that occur in final position are always non-finite. Hence, it seems that, at the initial state, verb forms are used with rotelearned morphology in a position that is determined by their semantics. Nevertheless, in acquisition research it is common practice that the mere occurrence of types of utterances with a finite verb form in initial position and a non-finite form in final position is interpreted as evidence of a contingency between verb placement (raising) and morphology (finiteness, inflection) and thus as evidence of a language system that is based on a productive set of morpho-syntactic rules.

226 

 Finiteness in language acquisition research

7.2.1.3 Verb placement and inflectional morphology in second language research In studies on L1 acquisition, it is commonly claimed that access to knowledge of the functional category system such as the contingency between verb raising (i.e. verb placement in second position) and inflection (subject-verb agreement) is available at the initial stages of the acquisition process. This contingency has also been claimed to hold for the process of L2 acquisition, either as the result of transfer from L1 or as given by UG. See, for example, Schwartz and Sprouse (1996), Herschensohn (2001), Haznedar and Schwartz (1997), Lardière (1998), Ionin and Wexler (2002), Prévost and White (2000). However, as far as the developmental process of verb placement and the use of inflectional morphology is concerned, adult L2 learners may differ from child L1 learners. In L2 research various proposals have been put forward to account for this.

7.2.2 Restrictions on the contingency between verb placement and inflection 7.2.2.1 Parodi (1998, 2000) Parodi studied the contingency between raising and agreement in the L2 acquisition of German. She found that speakers of Italian and Spanish acquiring L2 German consistently inflected non-thematic verbs and used them with post-verbal negation. As shown in (10a), (10b), (11a) and (11b) non-thematic verbs are auxiliaries, the copula be, modals and possessive have (2000: 365). (10)

(11)

a.

die familie is nich gekomm von Spanie. the family is-3Sg not come-Pp from Spain

b.

wollen nix arbeit. want-3Pl not work-Inf

a.

ich will nich mit sie gewohnt. I want-1Sg not with them live-Pp

b.

aber er hat nich die papier. but he has-3Sg not the papers

According to Parodi these non-thematic verbs “are, if not semantically empty, underspecified in their lexical content” (2000: 361). Hence, they are called ‘light

Finiteness in second language research 

 227

verbs’.⁶⁶ Thematic, i.e. lexical verbs, on the other hand, were usually left uninflected and were used with preverbal negation as in (12) (2000: 374).⁶⁷ (12)

wenn schiedsrichter nich schreiben, versicherung nich bezahl. (Giovanni III) if referee not write-Inf insurance not pay-Inf

This distributional difference at the early stages of second language acquisition is accounted for as the result of some kind of division of labour. According to this view, Parodi argues that „thematic verbs are responsible for lexical information, while nonthematic verbs [i.e. light verbs] are the main carriers of syntactical information“. That is, light verbs „spell out the functional category I[NFL], specifically its AGR component“ (2000: 373).⁶⁸ Language development occurs when thematic verbs are used with both agreement and postverbal negation, too (2000: 365f.).⁶⁹ (13)

a.

schaffst das nich? manage-2Sg that not?

b.

ich versteh Michael wirklich nich. I understand-1Sg Michael really not

To put this into Parodi’s words: “when pre-verbal negation dominates […] targetlike subject-verb agreement is not present” but “negation occurs post-verbally when the verb shows targetlike subject-verb agreement” (2000: 376). In fact, Parodi argues that “[p]ostverbal negation – i.e. the raising of the finite verb to C – seems to follow […] a learning of subject-verb agreement” (2000: 376). According to Parodi, this process of development is evidence that learners have access to the functional category system and its corresponding features and that it is inflection that causes raising. Parodi, however, also notes that the general picture of this language acquisition process is not without some typical exceptions. For example, Giovanni, par-

66 The examples in (10) and (11) are from Clahsen (1988) Stage I. 67 „In my view […] the learners are making a morphosyntactic distinction between finite and nonfinite verbs and between two different sites for verbs in the syntactic structure, as well as distinguishing verb classes, i.e., thematic and nonthematic verbs” (Parodi 2000: 364). 68 In Jordens (1990: 1434), I argued that this ‘division of labour’ is due to a processing problem. The association between morphological agreement and verb raising is too complex for learners at the relevant stage. 69 The examples in (13a) and (13b) are from Clahsen’s (1988) Stage III.

228 

 Finiteness in language acquisition research

ticularly at the initial stage, produced some non-targetlike utterances as in (14a) and (14b) (2000: 375). In these examples the negator is placed before the verb, although the verb has subject-verb agreement. (14)

a.

aber vielleicht nicht will. meine chef. (Giovanni III) but maybe not want-3Sg. my boss

b.

nicht gefällt essen deutsch. (Giovanni II) not like-3Sg food German

Furthermore, Parodi observed that Giovanni also produced examples of postverbal negation without subject agreement.⁷⁰ Although Parodi’s study seems to provide a fair account of the L2 data, one should be aware of the fact that Italian and Spanish on the one hand, and German on the other, have some structural properties in common. This holds particularly for the auxiliary verb as a carrier of the functional properties of finiteness. The simple use of the auxiliaries have and be – as for example in (9a) die familie is nich gekomm von Spanie (the family is-3Sg not come-Pp from Spain) – or elements referred to as ‘light verbs’ is not enough to justify the conclusion that learners have acquired I[NFL], i.e. finiteness as a functional category of their language system (2000: 373). Evidence of the presence of finiteness as a functional category are the syntactic properties of verb-second. However, Parodi’s study provides no evidence for this. Furthermore, rather than serving as ‘carriers of syntactical information’, verb forms such as, on the one hand, modals, i.e. wollen in particular, and, on the other hand, the copula be and possessive have are used to express the opposition between the presence or absence of control on the part of the speaker. The use of negation with these verb forms as, on the one hand, in (10b) wollen nix arbeit (want-3Pl not work-Inf) and in (11a) ich will nich mit sie gewohnt (I want-1Sg not with them live-Pp) and, on the other hand, in (10b) aber er hat nich die papier (but he has-3Sg not the papers) shows why action verbs as the complement of the modal predicate occur with preverbal negation and why state verbs such as the copula be and possessive have occur with postverbal negation. Therefore, I would argue that at the relevant stage, the contingency between ‘light verbs’ and inflection is not to be interpreted as evidence that ‘light verbs’ are the spell-out of the functional category I[NFL]. The proposal outlined in Chapter 5 and 6 has been put forward as an alternative. It distinguishes between a lexical stage and

70 Evidence is given in Table 5 in Parodi (2000: 374).

Finiteness in second language research 

 229

a functional stage of acquisition and is able to account for the observation that the functional projection FP is the result of a process of acquisition which accommodates the expression of finiteness and topicalization as functional means to express properties of information structure.

7.2.3 The dissociation of verb placement and inflection: The Impaired Representation Hypothesis 7.2.3.1 Meisel (1997) Meisel investigated the spontaneous production data of native speakers of Spanish, Italian and Portuguese learning L2 French and L2 German. He comes to the conclusion that in the relevant data there is no evidence of a contingency between verb placement and inflection. That is, placement of finite and non-finite forms is random. On the one hand, he notes that “[a] learner of French or German who becomes aware of the fact that [the negator] is placed postverbally in the target language is able to do so without taking the [+/− finite] distinction into account” (1997: 258). On the other hand, he finds that the negator is also placed preverbally not only with non-finite verb forms, but in a few cases with finite verb forms, too. In his L2 French data, he found, for example, an utterance as in (15) (1997: 243), while L2 German learners produced utterances as, for example, in (16) and (17) (1997: 249, 258). (15)

Si continue le travail dans cette condition, je pas continuerai le travail dans l’usine. ‘If working conditions continue I, I will not continue to work in the factory’ (M2)

(16)

in deutschlande ich nich verstehst. (Zita rec. 3) in Germany I not understand-2Sg

(17)

auch net hat recht vielleicht. (José) also not has-3Sg right perhaps

The examples in (15), (16) and (17) are evidence that the L2 learner language is unlike the language system of L1 learners. As argued in Rule and Marsden, “[n]onfinite forms are underspecified for finiteness in L2 grammars and can be inserted into the node with a [+ finite feature] but the reverse is not possible because finite forms are specified [+ finite] and cannot be inserted into a terminal node specified for the feature [−finite]” (2006, 194). Meisel therefore concludes that the

230 

 Finiteness in language acquisition research

production data from Romance learners of L2 French and L2 German show that “the underlying knowledge in L1 and L2 acquisition is substantially different in nature” (1997: 257).⁷¹ Under the hypothesis that L2 learners have access to the properties of the functional category system, Meisel’s data have been accounted for as due to a malfunction in the syntactic computational device. More specifically, his data have been interpreted as evidence that in L2 learners access to morphology as a feature property of finiteness is impaired. This dissociation between verb placement and inflection is therefore termed the Impaired Representation Hypothesis (IRH)⁷². Although in Meisel’s learner data the acquisition of verb placement and the acquisition of inflection appear to be dissociated, it seems that verb placement as it occurs in the target language is learned as the result of a stagewise developmental process. For example, the data from Giovanni (1997: 253) show that, initially, as in (18) “preverbal placement [of the negator] dominates quite clearly”, while at the same time “[the negator] sometimes follows modals and finite verb forms” (1997: 252). (18)

eh, nich rauche, oder rauch nicht. (Giovanni, rec. 16) uhm, not smoke-Inf or smoke-1Sg not

It is only in his last recordings that Giovanni has achieved an advanced stage of acquisition that shows a correlation between verb placement and inflection. Thus, Meisel notes that at the relevant stage “one could argue that Giovanni is beginning to learn subject-verb agreement; first and second person markings emerge, and the frequency of non-finite verb forms in contexts requiring finite forms decreases” (1997: 252).

71 Meisel’s conclusion from his analysis of the French L2 data is not without controversy, if he argues that finite verbs may occur not only before but also after pas. On the basis of their analysis of the relevant ESF data, Prévost and White (2000) found that finite verbs consistently occur before pas. Herschensohn (2001), analysing data from American students of L2 French, reaches the same conclusion: finite verbs were always placed before pas. 72 This term is used in Beck (1998), Eubank (1993/94; 1996), Franceschina (2001), Hawkins (2001) and Hawkins and Chan (1997).

Finiteness in second language research 

 231

7.2.4 The separation of verb placement and inflection: The Missing Surface Inflection Hypothesis 7.2.4.1 Prévost and White (2000) Research by Prévost and White on the spontaneous production data from two speakers of Moroccan Arabic learning L2 French (Abdelmalik and Zahra) and two speakers of Spanish and Portuguese learning L2 German (Ana and Zita) has shown that these learners have difficulty with the use of inflection. As it is the case in L2 German, finite forms, as in (19), correctly occur in initial position (2000: 122). (19)

er sagt der herr. (Ana, month 4) he say-3Sg the man ich lese wenig. (Ana, month 7.4) I read-1Sg little wir lernen. (Zita, month 6.4) we learn-1Pl hunderte du bekommst. (Zita, month 15) hundred you get-2Sg

Non-finite verbs however, appear, both targetlike in final position as shown in (20) (2000: 116f.) and non-targetlike in initial position as in (21) (2000: 122)⁷³. (20)

du willst nich arbeite hier. (Zita, month 24.4) you want not work-Inf here ich will diese jahr fahr nach Spanien mit mein eltern. (Ana, month 13.5) I want this year drive-Inf to Spain with my parents

(21)

er kaufe ein blume. (Ana, month 4) he buy-Inf a flower Michael spiele hier. (Zita, month 10) Michael play-Inf here die männer arbeite ganze tag. (Zita, month 23.5) the men work-Inf whole day

73 Prevóst and White (2000: 122) interpret these verb forms as 1Sg-forms. However, they suggest that these forms could also be analysed as non-finite forms. Given the learner grammar at the relevant stage, as, for example, in the utterance für lerne deutsch is nicht gut (to learn-Inf German is not good, in: Ana, month 13.5) and du willst nich arbeite hier (you want not work-Inf here, in: Zita, month 24.4), it seems that this is precisely how they are used.

232 

 Finiteness in language acquisition research

Prévost and White claim that the same is true for L2 French. The examples in (22) and (23) (2000: 116f. 121) come from the ESF project “Second language acquisition by adult immigrants”.⁷⁴ (22)

i mange pas. (Zahra, month 26.7) he eat-3Sg not moi j’aime pas de monde. (Abdelmalek, month 25) me I like-1Sg not some people quand moi est pas là. (Zahra, month 39.5) when me is-3Sg not there

(23)

parce que peut pas donner. (Abdelmalek, month 17.7) because can not give-INF mais on peut pas dormir. (Abdelmalek, month 17.7) but one can not sleep-INF

Finite forms as in (22) are used targetlike. They occur consistently before pas. Non-finite forms systematically appear targetlike after the negator as in (23). However, the data in Table 2 (2000: 119) show that there is an overuse of nonfinite verb forms in finite contexts. It is much larger than the overuse of finite verb forms in non-finite contexts.

context

finite

non-finite

overuse

non-finite verb form

finite verb form

L2 French

Abdelmalek Zahra

24.1% 22.9%

5.8% 1.3%

L2 German

Ana Zita

10.4% 16.4%

8.4% 5.8%

Table 2: Overuse of non-finite vs. finite verb forms

These data are reason for Prévost and White to conclude that “non-finite forms occur in both non-finite and finite positions, [while] finite forms are largely restricted to finite (raised) positions” (2000: 119). For a theoretical account of these data Prévost and White propose that “[v] ariability in morphology [should] reflect difficulties in identifying the appropri-

74 Perdue (1993a, b); Guiliano and Véronique (2005).

Finiteness in second language research 

 233

ate morphological realization of functional categories; that is, the problem lies in mapping from abstract features to their surface morphological manifestation” (2000: 108). This means that while inflection is part of the grammatical system at the surface level, some features may be partially specified or unspecified. More specifically it holds that “in adult L2 acquisition non-finite forms are underspecified with respect to finiteness (…) [and] may be inserted into a node involving the feature [+ finite]” (2000: 128). The non-finite forms in finite contexts are thus in fact some kind of “default substitutes for finite forms” (2000: 119). Underspecification of finiteness explains why finite verbs do not occur in the position of nonfinite verbs. “Finite forms (…) are not underspecified, hence cannot be inserted into non-finite positions” (2000: 128). Examples as in (19) and (22) show that “from the very first instances of negative clauses” finite verbs are systematically placed before the negator. According to Prévost and White this “suggests that verb-movement has occurred and, hence, that the features and feature strength involved in raising (and checking) are known to the learners” (2000: 117). In order to refer to their hypothesis that “at the surface morphological level (…) inflection is assumed to be absent, rather than at the abstract featural level”, Prévost and White (2000: 108) use the term Missing Surface Inflection Hypothesis (MSIH).⁷⁵ This hypothesis may also account for similar data in the L2 Dutch from speakers of Moroccan Arabic and Turkish. Examples are given in (24). (24) L2 Dutch. Non-finite forms in initial position. hij pakken een mayonaise. (Mo/A 1.6) he get-Inf a mayonnaise hij wil kopen een renault. (Mo/A 1.2) he wants buy-Inf a renault die man kopen een auto. (Os/T 1.9) that man buy-Inf a car dan hij vallen achter. (Os/T 2.9) then he fall-Inf behind The hypothesis of the underspecification of inflection is part of a general hypothesis according to which it is claimed that L2 learners, as it is assumed for L1 learners, have access to inflection and movement as properties of the functional category system. The only difference is that, in the L2 learner system, inflection initially surfaces only partly. However, the data in Table 2 show that finite forms,

75 Haznedar and Schwartz (1997) use the term Missing Inflection Hypothesis (MSI).

234 

 Finiteness in language acquisition research

although they are largely restricted to finite (raised) positions, may occasionally also occur in non-finite contexts. This seems to indicate that, initially, these verb forms are used as unanalysed elements. It suggests that there is an initial stage at which lexical verbs are distributed regardless of the morphology of the verb. The question, however is, whether this distribution is random or not. As shown in Chapter 5, evidence from L2 Dutch shows that it is not. At the initial stage, verb forms are mostly unanalysed and in complementary distribution, such that state and change-of-state verbs occur as finite verb forms in initial position, while action verbs occur in default form in final position. Finally, the Missing Surface Inflection Hypothesis accounts for the fact that in L2 acquisition verb raising may occur with non-finite verb forms. In other words, in L2 varieties inflection is not a criterion for movement. The question is, if the morphological properties of inflection do not serve as a criterion, what does? As argued in Chapter 6, it is the notion of finiteness referred to as ‘semantic finiteness’ (Lasser 1997), i.e. the notion of finiteness that accounts for the expression of the pragmatic function of assertion.

7.2.5 The acquisition of auxiliaries as a prerequisite for verb movement The data presented in Section 7.2.3 and in Section 7.2.4 seem to suggest that the acquisition of agreement in L2 German, L2 French and L2 Dutch is a two-stage process. At the initial stage, finite and non-finite verb forms may occur both in non-raised and in raised position. This is true even though non-finite verb forms in final position dominate and finite verb forms in final position are rare. It shows that verb morphology is initially unanalysed. At a more advanced stage of acquisition, verb forms with the morphological features of subject-verb agreement occur in second position. This is evidence of the acquisition of a functional position for the expression of finiteness. The study by Verhagen (2009a) on the acquisition of L2 Dutch provides additional experimental evidence of this two-stage process of acquisition. In her study, Verhagen tested L1 speakers of Turkish and Moroccan on their ability to produce finite and non-finite verb forms in raised and non-raised contexts. Both groups of learners were divided into a group of beginning and a group of intermediate L2 learners. The acquisition of auxiliary verbs was used as a criterion. L2 learners who did not use the auxiliary verb heb, heeft (have, has) with a past-participle phrase were categorized as beginners, L2 learners who did use this auxiliary verb were categorized as intermediates. The results of the production experiment are given in Tables 3 and 4 (2009a: 132).

Finiteness in second language research 

Beginners (n=24)

 235

Intermediates (n=31)

Finite

Non-finite

Finite

Non-Finite

V-Neg Neg-V

3 2

 6 41

48  0

11 23

Total

5

47

48

34

Table 3: Moroccan learners of Dutch. Finite and non-finite verbs in negated utterances (3Sgcontexts, lexical verbs only)

Beginners (n=24)

Intermediates (n=31)

Finite

Non-finite

Finite

Non-Finite

V-NEG NEG-V

1 7

  2  98

12  1

17 54

Total

8

100

13

71

Table 4: Turkish learners of Dutch. Finite and non-finite verbs in negated utterances (3Sg- contexts, lexical verbs only)

The data in Tables 3 and 4 show that “[b]eginning Moroccan and Turkish learners of Dutch who had not acquired auxiliary verbs or consistent agreement marking overwhelmingly produced non-finite verbs in non-raised [Neg-V] position” (2009a: 142), i.e. for Moroccan learners in 41 out of 47 instances and for Turkish learners in 98 out of 100 instances. Finite forms were used much less frequently. As in Meisel’s (1997) study, they occurred in raised [V-Neg] as well as non-raised [Neg-V] position. This absence of a “preference for finite verbs in raised [V-Neg] position (…) suggests that finite verbs are not ‘truly finite’ at the earliest stages of acquisition” (2009a: 142). Observations such as these are evidence for the absence of a functional category system at the initial stage. The examples in (25) produced by Turkish learners of Dutch may illustrate the point in case (Verhagen 2009a: 127f.). (25)

en meneer rood en de niet hoort. and mister red and the not hears-3Sg maar ook niet hoor / niet hoort. but also not hear-Inf / not hears-3Sg zij niet pakt. she not takes-3Sg.

236 

 Finiteness in language acquisition research

de man niet pakt. the man not takes-3Sg. rode man niet loopt. red man not walks-3Sg niet komt. not comes-3Sg daarom niet komt in de doek. therefore not comes-3Sg in the cloth niet valt. not falls-3Sg It does not seem to be a coincidence that ‘finite’ verb forms such as hoort (hear3Sg), komt (come-3Sg) and valt (fall-3Sg) are either state or change-of-state verbs. As I argued in Chapter 5, initially, that is, when the functional category system is not yet available, the learner system is a lexical system. At the lexical stage, state and change-of-state verbs occur in initial head position. In the input model, verb forms in this position are used with finite morphology. L2 learners use these forms as unanalysed elements, not only in initial but also in final position. This seems to be the case as long as the language system has no position for the expression of the functional properties of finiteness. The same conclusion appears from the early L2 learner data in Meisel (1997). The fact that finite morphology is limited to state and change-of-state verbs is evidence against Eubank (1993/94), who argues that functional categories are initially present but locally impaired. With respect to her data in Tables 3 and 4, Verhagen, furthermore, notes that “[m]ore advanced Moroccan and Turkish learners of Dutch who could use auxiliary verbs and produced agreement marking in a more consistent manner showed a different pattern: clear contingency patterns were found” (2009a: 142f). That is, these learners “produced verb-raising [V-Neg] relatively often” (2009a: 133), both with finite and non-finite verbs, and, “unlike the beginners, they did not place finite verbs in non-raised position” (2009a: 143).⁷⁶ As argued before, it shows that the language system has a position for the expression of the functional properties of finiteness, while the morphological properties of finiteness are in the process of acquisition.

76 As noted before, this phenomenon of the use of non-finite verbs in a raised position has been accounted for with the Missing Surface Inflection Hypothesis (Haznedar and Schwartz 1997; Prévost and White 2000).

Finiteness in second language research 

 237

7.2.6 The stagewise acquisition of the functional category system 7.2.6.1 Rule and Marsden (2006) In their study on the acquisition of functional categories in L2 French by speakers of English in an instructional context, Rule and Marsden argue for a stagewise developmental process, too. They discriminate between an early stage with no properties of a functional category system and a later stage at which the functional category system of the L2 seems to be established. Examples of the early stage are given in (26). (26) Early stage. No functional category system. il ne pas jouer au basketball. he not play-Inf basketball elle ne pas fumer. she not smoke-Inf elle ne pas aller au piscine. she not go-Inf to the swimming pool il ne pas prendre le bus. he not take-Inf the bus At the relevant stage, the non-finite verb typically occurs in its position after the negator pas. Athough “[t]here was a very low production of pas V[+fin] (…,) [t]his suggested that the placement of finite verbs is not random” (2006: 212). Examples of the advanced stage are given in (27). (27) Advanced stage. The establishment of the functional category system il ne joue pas. he play-3Sg not elle ne mange pas. she eat-3Sg not elle ne travaille pas. she work-3Sg not elle ne fait pas. she do-3Sg not il ne regarde pas. he watch-3Sg not When utterances as in (27) appear, there is a “clear drop” in the use of utterances as in (26). This process takes place “fairly rapidly”, while “the number of ‘non-

238 

 Finiteness in language acquisition research

chunk’ ne V[+fin] pas combinations, i.e. those that could certainly be considered as productive use, increased over the years” (2006: 202, 204, 211). At the same time, learners produced “only a very few verbs” with a non-finite form as in il ne manger pas (2006: 206). According to Rule and Marsden this non-finite form functions as, what they refer to as, a ‘default’ finite form. They argue that these ‘default’ finite forms are due to “processing pressures”, “particularly amongst learners with the most exposure to instruction” (2006: 213, 209). Evidence of the finite syntactic behaviour of these default finite forms is the fact that “they always occurred with subject clitic pronouns”. Further evidence is the observation that simultaneously with the use of these default forms, learners “have acquired the fully specified finite forms of the same verb types” (2006: 213, 207). With respect to the developmental process over time, Rule and Marsden note that “individual learners do not alternate between pas followed by a non-finite verb [as in (26)] and a finite verb followed by pas [as in (27)]” (2006: 203). That is, “learners who produced utterances like *il ne pas manger ‘he not eat’ did not also produce il ne mange pas or *il ne manger pas for ‘he doesn’t eat’ “ (2006: 211). These observations are evidence of “the existence of separate stages in grammatical development: an early stage without functional categories, in which, if verbs are present, negation is preverbal and a later stage with functional categories, in which negation is postverbal” (2006: 203). Rule and Marsden, furthermore, note that “once the functional categories are projected […] the learners evidence knowledge of finiteness as indicated by their correct placement of verbs in relation to the negative particle pas. The evidence suggests that there is no syntactic deficit in their grammars and this knowledge of finiteness appears to be irrespective of the surface morphology of the verb”. The implication of this finding is not only that it is “further support to the MSIH” (2006: 214), it also shows that the syntax of verb placement and inflectional morphology are acquired as separate systems of linguistic knowledge. It provides empirical evidence for the conclusion that finiteness is a linguistic notion that functions independently of the morphological properties of the verb. An important observation in this context applies to the learners “who consistently demonstrated correct negative placement”. Learners who achieved this advanced stage of acquisition, not only “used correct agreement in a range of contexts”, they also used “finite lexical verb inversion in questions and generally used tense correctly”. It leads Rule and Marsden to conclude that “the functional category I/T is fully specified when it emerges” (2006: 212). In sum, the data as they are analysed in Rule and Marsden show that the acquisition of the contingency between verb placement and inflection, as is demonstrated with correct negative placement and agreement, is a two-stage process. Furthermore, they show that simultaneously with the acquisition of this contin-

Finiteness in second language research 

 239

gency, learners are also able to use other properties of the functional category system such as inversion of the finite lexical verb and tense. However, it remains unclear, why it is that finiteness is acquired as a function of verb placement and why the relevant contingencies appear. As argued in Chapter 6, it is finiteness and topicalization as the functional properties of information structure that appear to be the driving force. Finiteness and topicalization set the stage for the morphosyntactic features of word order variation and verbal inflection.

7.2.6.2 Vainikka and Young-Scholten (1996, 2011): The Organic Grammar approach The Organic Grammar approach put forward by Vainikka and Young-Scholten (2011) holds that both children and uninstructed second language learners begin with what is called the ‘core’ of the utterance, i.e. the projection of a bare VP structure. The theory of Organic Grammar, furthermore, entails the piecemeal development of syntactic structure: “the gradual appearance of parts of syntactic structure (i.e. the specific functional projections) that gives rise to stages of acquisition” (2011: 9f.). The structure of a VP projection is either head-initial or head-final. Vainikka and Young-Scholten note that at the initial state of L2 acquisition headedness of the VP projection is transferred from the L1. Thus, at the initial state, the VP projections of Turkish and Korean learners of L2 German are head-final as in (28), while the VP projections of Italian and Spanish learners of German are head-initial as in (29). (28) L2 German. Transfer of the L1 head-final VP ja alles hier kaufen. (Memduh / L1 Turkish) yes everything here buy-Inf eine katze fisch alle essen. (Changsu / L1 Korean) a cat fish all eat-Inf (29) L2 German. Transfer of the L1 head-initial VP in fabriken eh nis soviel spreche deutsch. (Salvatore / L1 Italian) in factory uh not so much speak-Inf German ich wohnen (en) la grenza hier. (Jose / L1 Spanish) I live-Inf [in the border Sp.] here A head-initial VP as in (29) is not targetlike. Under the influence of L2 input, it will be given up in favour of a head-final VP projection as in (30) (2011: 128ff., 135ff.).

240 

 Finiteness in language acquisition research

(30) L2 German. Head-final VP in L2 learners with a head-initial VP in L1 vielleicht schule essen. (Salvatore / L1 Italian) maybe school eat-Inf ich immer nur eine Tag in de woche gucken. (Jose / L1 Spanish) I always only one day in the week look-Inf Vainikka and Young-Scholten found that, both in child L1 and in adult L2 acquisition, functional projections developed gradually and were triggered by positive evidence, i.e. by overt syntactic distribution and overt morphology. Vainikka and Young-Scholten note that “the productive usage of a functional element, particularly one lacking in semantic content such as the auxiliary, indicates the acquisition of the syntactic position in which this element occurs” (1996: 168). Furthermore, they claim that once a syntactic position for the auxiliary has been established, learners are in the position to acquire both verb raising and agreement as in (31) (2011: 185f., 200). (31) L2 German. Verb raising and agreement ich habe auf italienisch gesagt. (Bruno / L1 Italian) I have-1Sg in Italian said jetzt brau wohnungsamt fragen. (Sevinc / L1 Turkish) now need-0 housing authority ask gut mach ich ihm eine cappuccino mit alles voll. (Jose / L1 Spanish) good make-1Sg I him a cappuccino with everything The acquisition model of Vainikka and Young-Scholten is similar to the model of language development as proposed for the Dutch learner data given in Chapter 6. Both are based on the assumption that functional projections are initially absent. However, absence of functional projections does not necessarily mean that learners are only able to produce the type of non-finite VP structures as in (28), (29) and (30). „In the data there are some instances of early IP-level elements such as modals, auxiliaries, forms of sein ‘to be’, and raised verbs with (the third person singular) -t are also attested” (2011: 87). Examples of the kind of data that Vainikka and Young-Scholten refer to are given in (32). They are produced by Changsu (L1 Korean) at the VP stage and qualified as “precursors of the next stage” (1994: 284).

Finiteness in second language research 

 241

(32) Head-initial structures in early L2 German ich möchte nicht gut da. (Changsu / L1 Korean) I want-1Sg not good there schnee ist da. (Changsu / L1 Korean) snow is-3Sg there ich bin Frankfurt zug fahren. (Changsu / L1 Korean) I am-1Sg Frankfurt train go die frau alleine trinkt saft. (Changsu / L1 Korean) the women alone drinks-3Sg juice Furthermore, Vainikka and Young-Scholten (2011: 87) refer to examples from the literature on the L1 acquisition of German as in (33). (33) Head-initial structures in early L1 German. weint die katze. (Hannah 2;0–2;4) cry-3Sg the cat will lala habe. (Simone 1;8–2;0) want-1Sg/3Sg dummy have With respect to the data in (32) and (33), Vainikka and Young-Scholten argue that “[i]t should be noted that the spontaneous oral production data on which the L1 German analyses are based were not specifically collected to study the initial syntactic stage, and thus involve data from a randomly selected (at least with respect to this stage) point onwards” (2011, 88). Furthermore, they refer to Tracy (2002), who “believes that the potential IP-related constructions in the earliest data such as the FP are [in fact] ‘V2 mimicry’, i.e. they should be treated as unanalyzed, memorized chunks” (2011: 88). With Vainikka and Young-Scholten, I agree that the utterances in (32) and (33) may not serve as evidence of the acquisition of a functional projection. On the other hand, however, I would claim that this type of data is not as marginal as suggested. As argued in Chapter 5, these utterances are examples of a VP structure which is head-initial. Moreover, I have argued in Chapter 5 that utterances with a non-finite VP are in fact head-initial, too. This is because in utterances with a non-finite VP, the VP is projected as the complement of a head-initial V. Although the position of V may remain empty, it structurally serves the expression of control.

242 

 Finiteness in language acquisition research

7.2.6.3 Becker (2005) Becker studied the acquisition of negation and finiteness in three Italian learners of L2 German. The spontaneous production data she used were gathered longitudinally within the framework of the ESF project “Second language acquisition by adult immigrants”⁷⁷. Becker’s study focusses on the elements that are used to express the functional properties of finiteness. These elements are placed in the position that precedes the negator. According to Becker, the linguistic means to express finiteness are acquired in three stages. The initial stage is characterized by the fact that finiteness as a functional property is not overtly marked. Early examples of utterances by Angelina (An) that are used to express a negation are given in (34) (Becker 2005: 281). (34)

nix gut. (An 1.1) no good nee, nee, nix andere kind. (An 1.1) no other child nee, mein kind nix in schul. (An 1.5) no, my child not in school

Becker notes that Angelina also uses the invariant form is, for example in (35), as a means to express affirmation (2005: 285). (35)

mein mann is in arbeite. (An 1.1) my husband is in work mein mann is *geloso*. (An 1.1) my husband is jealous

The fact that is occurs in the same position as the negator is evidence that the learner uses this invariant form is in opposition to the negator to explicitely express the default function of an assertion as opposed to its functioning as a negation⁷⁸. Exactly the same phenomenon has been observed in L2 Dutch. The examples in (36) are given in Verhagen (2009a: 60f.).

77 Perdue (1993a, b). 78 As in Becker (2005: 285), I use the term ‘assertion’ here as opposed to ‘negation’.

Finiteness in second language research 

 243

(36) L2 Dutch. The opposition between is and niet as the expression of an assertion a.

de meneer rood niet springen / de meneer blauw is springen. the mister red not jump-Inf / the mister blue is jump-Inf

b.

die man niet doen, de zij is doen. that man not do-Inf, the she is do-Inf

c.

hij blauwe is bed slapen / hij groene 0 bed slapen / hij rode 0 bed slapen. he blue is bed sleep-Inf / he green 0 bed sleep / he red 0 bed sleep

The example in (36a) comes from a Moroccan learner describing two adjacent scenes in a story-telling task; (36b) is produced by a Turkish learner describing a film scene; (36c) is taken from a Turkish learner in a story-telling task, too. Both (36a) and (36b) show that, as in L2 German, is and niet are used in complementary distribution; (36c) is evidence that, as an alternative to is, the position of is may remain empty. These examples show that, at the initial stage, the function of an utterance as an assertion is expressed lexically: is as the expression of the positive default function of an assertion and niet as the expression of its functioning as a negation. Hence, is is in fact the first verb form used to express finiteness. As Becker phrases it: “The form is (…) is finite insofar as it is understood to be the carrier of AST [assertion]” (2005: 288). As the expression of an assertion, early is is used as a lexical means to express the hold-for relation between the predicate and the subject. In other words, it is a lexical element that indicates that the predicate ‘is true for’ the subject. This explains why it is used not only like a copula of the target language as in (37a), but also in non-targetlike utterances with a non-finite verb as in (37b, c, d) (Verhagen 2009a: 53, 61, 64).⁷⁹ (37) L2 Dutch. The use of is to express an assertion a.

de bang niet daar, is daar. the danger not there, is there

b.

de meneer blauw is springen. the mister blue is jump-Inf

c.

Charlie Chaplin is steel de brood. Charlie Chaplin is steal-0 the bread

79 The fact that is in utterances as in (37) is a lexical element, explains the observation that “is was frequently produced by learners, who do not yet use hebben” (Verhagen 2009b: 220).

244 

 Finiteness in language acquisition research

d.

de jongens is spelen. the boys is play-Inf

The use of is not as a copula but as a lexical means to express an assertion both in (37a) and in (37b, c, d) shows that a distinction between a category ‘adverb’ for the element daar (there) and a category ‘non-finite verb’ for elements such as springen (jump), steel de brood (steal the bread) and spelen (play) has no function in the learner grammar at the relevant stage. In fact, at this lexical stage, it is only the functioning of these elements as predicates that is relevant.⁸⁰ Becker notes that developmental progress occurs when finiteness comes to be expressed morpho-syntactically. The relevant process takes place when, as an alternative to utterances as in (38), learners use aspectual and modal auxiliaries as in (39) (2005: 287, 293f.).⁸¹ (38)

mein vater nicht schlafen. (An 3.1) my father not sleep ich nicht sprechen deutsch gut. (An 3.1) I not speak German well

(39)

in deutschland vielleicht ich habe nicht gesehe. (Mo 2.1) in Germany maybe I have not seen er hat nicht die zug gesehen. (Ti 2.3) he has not the train seen für moment du kannst nicht die ferien haben. (Mo 3.5) for moment you can-2Sg not the holidays get die papa sage: du muss nich sehn. (Ti 2.3) the dad say: you must-1/3Sg not look

With the acquisition of a position for the auxiliary as in (39), this position becomes available for verbal elements to express the semantic function of finiteness. Furthermore, as shown in (39), the negator occurs after the auxiliary. This is because finiteness has scope over the negation of the assertion. This structural ordering of

80 The interpretation of the use of is as the expression of perfective or durative aspect in Starren (2001), is determined by the perspective of the fully fledged system of the target language. It has no bearing on the system of the learner language at the relevant stage. The suggestion that is is “a structural device that helps (…) learners acquiring finite syntax” (cited in Verhagen 2009b: 220) has no bearing on the learner system either. If it were to create a position for the finite verb, why is it that learners do not also use heb or heeft? 81 Mo = Marcello, Ti = Tino.

Finiteness in second language research 

 245

the expression of finiteness and negation seems a general feature of learner languages. As Bernini has pointed out, it also occurs in Swedish learners of Italian, in “utterances with non-target post-copular/post-auxiliary negation” (2003: 179) as, for example, in un problema de cambiatore e non buono (‘a problem of changer is not good’) and ma eeh ho no fatto (‘but I-have not done’) (2003: 178). He argues that the deviation from the target structure cannot be accounted for with transfer, because in the L2 Italian of these Swedish learners an utterance such as jag läser inte (I don’t read) is never produced as *leggo non. The acquisition of the functional category of the auxiliary verb is a prerequisite for a prolonged stage of acquisition in which lexical verbs come to be used in the position of the auxiliary verb. A few examples are given in (40). (40)

ich sage nicht deine name. (Ti 2.7) I tell not your name sie geben nicht die geld wie sie mussen geben. (Ti 3.6) they give not the money as they must give ich mache nicht auf. (Ti 3.5) I make not on er arbeit nicht gut. (Ti 2.6) he work not well

Lexical verbs in the position of the auxiliary serve as carriers of the functional property of finiteness. They may occur in this position, i.e. before the negator, only after the acquisition of the auxiliary verb. Verhagen (2009a) has made exactly the same observation in her investigation of the acquisition of L2 Dutch by speakers of Moroccan Arabic and Turkish. Before the acquisition of the auxiliary heb, heeft (have-1Sg, has-3g) both groups of learners rarely produced the lexical verb before the negation. The relevant production data as given in Verhagen (2009a: 95, Table 2 and 3) are summarized in Table 5.

Moroccan

V-Neg

Turkish

−Aux (n=24)

+Aux (n=31)

−Aux (n=21)

+Aux (n=25)

9%

78%

3%

37 %

Table 5: L2 Dutch. Placement of the lexical verb (V) before the negator (Neg)

The data in Table 5 show that in learner grammars that have a position for the auxiliary verb (+Aux), lexical verbs may occur before the negator. The difference between speakers of Moroccan Arabic (78%) and Turkish (37%) is claimed

246 

 Finiteness in language acquisition research

to be due to the position of the lexical verb in the L1. In Moroccan Arabic the lexical verb occurs in sentence-initial position. For learners who have acquired the auxiliary, the position of the auxiliary sentence-initially is used as a position for the lexical verb as it is in the L1. L1 influence from Turkish may explain why, in Turkish learners of L2 Dutch, the lexical verb is used less frequently in initial position, even though with the auxiliary a position for it is available. At the relevant stage, correct agreement marking is also a matter of time. Becker shows that learners may use the same verb with both correct and incorrect agreement as in (41) (2005: 299). (41)

sie sehe eine bäckerei. (Ti 2.7) she see-Inf a bakery eine alte dame seht das. (Ti 2.7) an old lady see-3Sg that

It seems noteworthy that the examples in (41) provide evidence of the same phenomenon that has been accounted for by the Missing Surface Inflection Hypothesis (Prévost and White 2000). In sum, for the linguistic marking of finiteness, Becker (2005) discriminates three stages of acquisition. Initially, there is no morphological marking of finiteness: is (is) and niet (not) are used as lexical means to express the pragmatic function of an assertion. Then, auxiliary and modal verbs are used as carriers of finiteness. Finally, lexical verbs are used for the expression of finiteness, too. In addition to her findings on the acquisition of finiteness, Becker comes up with an observation that seems quite different in nature. She points out that the use of lexical verbs in the position before the negator co-occurs with “another major syntactic development (…), namely the emergence of topicalization” (2005: 299). Examples are given in (42) (2005: 300). (42)

weil die deutsch zeitung ich kann nich noch lesen. (Ti 3.5) because the German newspaper I can not yet read die amerikan ich versteh nich …. (Ti 3.5) the american I understand not aber die zweite seite kenns du nicht. (Ti 3.6) but the second side know you not

This observation by Becker is evidence for the conclusion in Chapter 6 that, at the final stage as the result of the acquisition of FP, both the verb-second position for the expression of finiteness and the initial position for the expression of topicalization become established more or less simultaneously.

Finiteness in second language research 

 247

It should be noted that in Becker (2005), focus is on the acquisition of the morpho-syntactic expression of finiteness, i.e. on the formal expression of the pragmatic function of assertion. Evidence of the formal means to express this functional property is the use of aspectual and modal auxiliary verbs. However, fact is that modal verb forms are already used at the initial stage, i.e. before the acquisition of auxiliaries. The same is true for finite verb forms that are used as unanalysed elements and are not ‘truly finite’. Both these modal and finite verb forms are used with postverbal negation. Although these verb forms, as in ich weiss nicht (I don’t know), ich versteh nich (I don’t understand), gefällt mir nicht (don’t like it), das stimmt nicht (this is wrong), das geht nicht (it doesn’t work) etc. (2005: 298) are indeed not ‘truly finite’, they do not necessarily qualify as part of a formulaic expression. Rather, at the relevant stage, as shown in Chapter 5, they belong to the category of state and change-of-state verbs. In head-initial position, these types of verb systematically occur with postverbal negation. What Verhagen calls ‘light verbs’ in Turkish and Moroccan learners of Dutch are also examples of unanalysed finite verb forms. At the initial stage, when auxiliary verbs are not yet part of the learner grammar, these ‘light verbs’, i.e. the copula, modals, the use of gaan meaning ‘going to’ and a particular use of is, “predominantly preceded negation” (2009a: 65).⁸² Examples are given in (43) (2009a: 64f.). (43)

maar dat is nog niet laat. but that be-3Sg yet not late meneer rood ook wil niet springen. mister red also want-3Sg not jump-Inf een meisje is niet krijgen een brood. a girl be-3Sg not get-Inf a bread

These so-called ‘light verbs’ are either modal verbs used to express the willingness, the ability, the permission or the obligation of the agent to carry out a particular action or, as in the case of utterances with is, they are used to express by default the assertion of a hold-for relation between the predicate and the subject. They serve as lexical predicates in head-initial position and, as is the case in the target-language, they precede negation.⁸³

82 See the data in Tables 5 and 6 vs. Tables 7 and 8 in Verhagen (2009a: 64, 67). 83 At the lexical stage, as shown in (34) vs. (35) and in (36), there is an opposition between the assertion markers is meaning ‘is true’ and the negator nix or niet meaning ‘is not true’. Furthermore, instead of targetlike is niet learners may also produce niet is as in die man niet is stelen (that man not steal-Inf). As assertion markers, is and nix or niet are used with scope

248 

 Finiteness in language acquisition research

Finally, Becker provides no explanation for why auxiliaries are acquired all of a sudden, why “the morpho-syntactic marking of finiteness is developed in two steps” (2005: 292) and why topicalization and finiteness are acquired simultaneously. These questions have been accounted for in Chapter 6. Given that the auxiliary is used to express the pragmatic function of finiteness, the position of the auxiliary is a prerequisite for the placement of the lexical verb. Furthermore, given that both the verb-second and the topic position are part of the functional projection FP, they are acquired simultaneously. It is the pragmatic function of this projection to express aspects of information structure that are linked to the functioning of an utterance as an assertion. Both the position of verb-second and the topic position provide the structural requisites with which this pragmatic function is expressed. Hence, they are projected simultaneously. In other words, the expression of the pragmatic function of an assertion explains the contingency between the acquisition of the finite verb in head-initial position and a specifier position for elements with topic function.

7.2.7 Summary Language acquisition is a developmental process. That means that some properties of the system of the target language are learned earlier than others. The development of the learner system proceeds stagewise. That is, a simple learner system at the initial stage of language acquisition will be given up in favour of a more complex system at an advanced stage. Studies of both child L1 and adult L2 learners of Dutch have shown that, at the initial stage, the linguistic system of the learner language is purely lexical. That is, particular linguistic categories such as the auxiliary verb, determiners and question words are lacking. Also lacking are structural properties of the target language such as movement (or raising) and inflection (or agreement). The question is, why is this the case? What does it tell us about language acquisition? For example, why is it that initially auxiliary verbs do not occur? Is this because they are not noticeable for semantic and/or phonological reasons, or is it because the learner grammar at the relevant stage has no structural position available yet?

over the lexical predicate. Modal elements at the relevant stage are lexical elements. They may, therefore, be treated as part of the predicate. With the assertion markers is and niet, this explains the incidental use of non-targetlike utterances such as groene man niet wil springen (green man not wants jump-Inf) and hij is moet beneden (he is must down) and misschien hier hij is wil slapen (maybe here he is want sleep-Inf) (Verhagen 2009a: 62, 65).

Finiteness in second language research 

 249

Another question concerns the absence of verb movement (or raising) and with it the absence of inflection (or agreement). In L1 acquisition, verb movement and inflection are linked. In L2 acquisition, however, this contingency does not exist. Verb movement in L2 learners may occur without the acquisition of inflection. In a series of studies on the acquisition of finiteness, two types of proposals have been put forward to account for this.

7.2.7.1 The contingency between verb placement and morphology In their study on the acquisition of functional categories in early L2 French, Rule and Marsden (2006) have shown that the distribution of the verb forms as they occur with learners at the initial stage differs from the distribution of verb forms as they occur with learners who are more advanced. They argue that, at the initial stage, the morpho-syntactic use of verb forms can be accounted for by the Impaired Representation Hypothesis (IRH). With this IRH, it is suggested that, initially, L2 learners may not have access to morphology as a feature property of inflection, while they do have access to a position that makes raising possible. This explains why finite and non-finite verb forms may occur both in final and in initial position. Furthermore, they claim that, at a more advanced stage, the morpho-syntactic use of verb forms can be accounted for by the Missing Surface Inflection Hypothesis (MSIH). With this MSIH, it is claimed that L2 learners have access to morphology as a feature property of inflection and that there is a position available for raising to take place. However, in L2 learners, morphology of the raised verb form may be underspecified, i.e. it may not necessarily be realized at the surface-structure level. This explains why, at the relevant stage, non-finite verb forms occur in final position, while both finite and non-finite verb forms may occur in initial position. Raising as it is used in IRH and MSIH is a metaphorical term for the relation between the two positions of an utterance in which a verb form may appear: a default position and a secondary position. This secondary position is a position which can only be taken under specific conditions. Verb forms occurring in this position are characterized as ‘raised’. According to the UG approach to language acquisition, the condition for verb raising is the inflection of the verb. That means: if there is no inflection on the verb, the verb occurs in its default (base) position. If, however, there is inflection on the verb, the verb appears in its secondary position. Thus, morphology is a prerequisite for raising, or, put differently, there is a contingency between verb placement and inflection. In UG based approaches, it is this relation of contingency which is the driving force in the acquisition of raising. It entails the notion of underspecification which serves as a theoretical means to account for the fact that morphology may play a role only at a more abstract level of representation.

250 

 Finiteness in language acquisition research

7.2.7.2 The acquisition of the auxiliary as a carrier of finiteness The alternative hypothesis within the present functional approach aims to deal with the question how learners are able to arrive at their knowledge of a language system solely on the basis of an analysis of the distributional properties of the input. As pointed out before, the term ‘raising’ in Dutch and German is used to account for the particular relation between the two positions of an utterance in which a verb form may appear. Learners have to find out that these positions are related in terms of a primary or default position and a secondary or raised position. Furthermore, they have to find out under what conditions verb forms are placed in either default or raised position. The idea is that it is the acquisition of the auxiliaries heb, heeft (have, has) and ben, is (am, is) that establishes the relation between both positions. Hence, as long as there is no auxiliary, raising is not possible and, thus, there is neither a raised nor a default position. Experimental L2 data as presented in Verhagen (2009a, b) show that the absence or presence of the functional category of the auxiliary verb distinguishes between learner systems at two stages of language development. At the initial, lexical stage, when auxiliary verbs are absent, neither raising nor inflectional morphology plays a role. At the relevant stage, as shown in Chapter 5, unanalysed lexical verb forms (predicates) are used in complementary distribution. In adult L2 learners, this complementary distribution correlates with the semantics of the head-complement structure essentially the same way as in child L1 learners, i.e. action verbs occur in the position of the complement, while state and changeof-state verbs occur as the lexical head. At the initial stage of L1 acquisition, morphology is irrelevant and, hence, not productive. Morphological correctness mirrors the form in which the lexical verb appears in the input (see Clahsen 1986, Jordens 1990, Ingram and Thompson 1996). In adult learners, morphology is irrelevant, too. However, here verb forms in a particular position do not necessarily mirror the form in which they occur in the input. This explains the observation of ‘random’ placement. It is this observation that the Impaired Representation Hypothesis (IRH) is based on. At the functional stage, with the establishment of a position for the auxiliaries heb, heeft (have, has) and ben, is (am, is), learner grammar has two verb positions. Placement depends on the presence or absence of the auxiliary. The position of the auxiliary is used to express the functional properties of finiteness. With an auxiliary verb, the lexical verb occurs in default position. If the auxiliary is absent, its position is taken by the lexical verb. The lexical verb is placed in this secondary or raised position for functional reasons, i.e. for the expression of finiteness as a functional category that serves the pragmatic function of assertion. If the auxiliary is present, the lexical verb in default position has its default, i.e. non-finite, morphology. If the auxiliary is absent, the lexical verb in raised

Finiteness in second language research 

 251

position will come to be used with the inflectional morphology of the target language. Inflectional morphology, however, does not serve a semantic function. In L2 learners, this explains the observation of inflection as an optional feature. It is this observation that the Missing Surface Inflection Hypothesis (MSIH) is based on. It is the assumption of both the IRH and the MSIH that inflectional morphology plays a crucial role in the acquisition of verb movement. However, I would argue it is not. With the acquisition of the auxiliary the learner grammar has a functional position for the expression of finiteness. Having established the relation between a functional and a lexical position of the verb, this relation is a prerequisite for the acquisition of the syntactic relation referred to as verb raising. With the use of lexical verbs as carriers of finiteness, learners are in the position to acquire the morphological properties of verbal inflection. Hence, it is the acquisition of finiteness as a category of information structure, i.e. finiteness as category that is used to express the pragmatic function of assertion, that leads to the acquisition of verb raising and, as a consequence, to the acquisition of inflectional morphology with the lexical verb. One question, however, remains. Why is it that L1 and L2 learners behave differently with respect to the acquisition of the inflectional morphology of the verb? More specifically, why is it that only L2 learners produce non-finite verb forms in finite position and to a lesser extent finite verbs in non-finite position, too. To me it seems that this is due to the way in which L1 and L2 learners analyse the target language input. For L1 learners the relevant category of analysis is the predicate, whereas for L2 learners it is the verb. A verb form that is learned as part of an unanalysed predicate will occur in a syntactic position in which it appears as formally correct. A verb form that is learned as a member of a verbal category however, may be used in another verbal position in which it may appear as formally incorrect. This explains why only L2 learners produce utterances with a non-finite verb in finite position as in dan hij vallen achter (then he fall-Inf behind) and ik maken de fiets (I make-Inf the bicycle) and to a lesser extent also finite verbs in non-finite position, as in ik wil altijd zit met Nabil (I want always sit-1Sg with N).

8 Conclusion 8.1 Basic languages Basic languages or basic varieties are learner languages at an intermediate state of language acquisition. They are symbolic systems, i.e. “self-standing, independent systems showing their own regularities (…) on the structural, semantic and informational level” (Perdue 2006: 859). In the present study, the spontaneous production data from two children learning L1 Dutch and two adults learning L2 Dutch have shown that learner languages are the outcome of processes of creative construction. The child data in this study were collected by the present author. They are diary data from two children: Jasmijn at (1;8–2;2) and Andrea at (1;8–2;4). The second language data come from the European Science Foundation (ESF) project “Second Language Acquisition by Adult Immigrants” (Perdue 1993a, b). They were collected in three cycles of 10 months each.⁸⁴ The analysis of the longitudinal data from both these child L1 and adult L2 learners has revealed the principles of language acquisition that determine the development of a simple, basic learner system into the more complex system of the target language. As described in Chapter 5 and Chapter 6, this developmental process can be accounted for in terms of a stage model that distinguishes between an initial, lexical stage and an advanced, functional stage.

8.2 The lexical stage At the lexical stage, the structure of the learner utterances in both child L1 and adult L2 Dutch is determined by the lexical category of the predicate. At the relevant stage, there are two types of utterance depending on whether the predicate is agentive or non-agentive. Agentive utterances occur with action predicates referring to a causal action such as maken (make), pakken (get) and kopen (buy) or to an agentive motion such as lopen (walk), gaan (go) and gaan zitten (sit-down). Non-agentive utterances occur with state predicates such as blijf[t] (stays), lust (likes), pas(t) (fits) and kan (can) or change-of-state predicates such as komt (comes), gaat (goes) and valt (falls). Morphologically, causal action and agentive

84 In the examples given, a reference such as, for example, ‘J 1;9’ means that the utterance occurred when Jasmijn was 1 year and 9 months of age. ‘Mo/A 3.9’ means that this utterance by Mohammed with L1 Moroccan Arabic occurred at month 9 of cycle 3. ‘Os/T 1.3’ means that this utterance by Osman with L1 Turkish occurred at month 3 of cycle 1.

254 

 Conclusion

motion predicates appear as non-finite. They occur with an agent as the external argument and are placed in final position. State and change-of-state predicates appear as morphologically finite. They occur with a theme as the external argument and are placed in initial position. In agentive utterances, the predicate may occur as an infinitive. These utterances are the so-called ‘root infinitives’. The similarity between child L1 and adult L2 learners becomes evident with the examples in (1). (1) Type A1. Infinite verb form (root infinitive) child L1 Dutch

adult L2 Dutch

Mijnie zelf doen. (J 1;11) M self do ik chocomel hebbe. (J 2;10) I chocolate-milk get jíj gijbaan make. (A 2;1) you slide make

ik alles doen. (Mo/A 2.2) I everything do veel geld hebben. (Os/T 1.7) much money get hier koekfabriek pakket maken. (Os/T 1.3) here cookie-factory package make

At the relevant stage, agentive predicates may also occur as infinitives preceded by a modal or an aspectual element or by the default verb form doet. As shown in (2), the utterances in child L1 and adult L2 learners are similar. (2) Type A2. Modal / Aspectual element + infinitive child L1 Dutch

adult L2 Dutch

ulle ijsje hebbe. (J 1;8) want ice cream get kannie[t] bal pakke. (J 1;9) can-not ball get Jaja mag dop opdoen. (A 2;0) J may cap on-do Peter moet zitte. (J 1;11) P must sit-down ik ga eve die glije. (J 2;0) I go just that slide-Inf gaatie ape. (A 2;1) goes-he sleep-Inf doetie viesmake. (A 2;1) does-he dirty-make-Inf

hij wil kopen een renault. (Mo/A 1.2) he wants buy a renault kan met hem slaap. (Mo/A 1.5) can with him go-sleep mag ook hier werke? (Os/T 2.3) may also here work? en dan wij moet zo doen. (Os/T 2.2) and then we must so do hij gaat snel lopen. (Mo/A 1.6) he goes fast run ikke komt hier werken. (Os/T 1.3) I comes here work-Inf ik lezen doen. (Os 2.7) I read-Inf do-Inf

The lexical stage  

 255

In utterances as in (2) the modal and aspectual elements are used to express the willingness (ulle, wil, want), the ability (kan, can), the permission (mag, may) or the obligation (moet, must) of the agent to perform an action, gaat (goes) and komt (comes) refer to the intention of the agent, and it is the function of the default doet (does) only to express that an agent is involved. It should be noted that at the lexical stage, these elements are predicates, not auxiliaries. They serve the expression of ‘control’. At the relevant stage, the predicate is not necessarily a verb. Examples of a non-verbal predicate preceded by a modal element are given in (3). (3) Type A3. Modal + non-verbal predicate child L1 Dutch

adult L2 Dutch

magniet oppe dak. (J 1;11) may-not on-the roof kannie zellef. (J 1;11) can-not self

ik magwel alles. (Os/T 2.6) I may-indeed everything moet zelf hè. (Os/T 3.2) must self

At the lexical stage, particles are also used as predicates. Examples of particles as agentive predicates are given in (4). (4) Type A4. Particle child L1 Dutch Particle

variant: Particle + infinitive

variant: Modal + particle

tittat om. (J 1;7) watch on [do] Mijnie in, tiktak. (J 1;8) M in [do], watch dop op. (A 1;8) cap on [do] goene aan. (A 1;11) shoes on [do] zak uit. (A 1;11) bag out [take]

[tiktak] omdoen. (J 1;11) watch on-do da [melk] ingenke. (J 1;10) there milk in-pour dop opdoen. (A 1;11) cap on-do goene aandoen. (A 2;0) shoes on-do dit uithale. (J 1;11) this out-get

minne tittat om. (J 1;7) want watch on

hoeniet plak op. (A 1;11) must-not glue on

minne hoene uit. (J 1;17) want shoes out

The examples in (4) also show that, at the relevant stage, agentive particles may occur as predicates followed by an infinitive with doen (do), schenken (pour),

256 

 Conclusion

halen (get), or preceded by a modal element such as minne (want) and hoeniet (must-not). Non-agentive utterances occur with a finite verb form. They are used to express a state or a change of state. The similarity between child L1 and adult L2 learners is evident in (5). (5) Type B1. Finite verb form child L1 Dutch

adult L2 Dutch

da poes blijf hier. (J 1;11) there kitty stays here uil zo komt. (J 1;10) owl immediately comes Mijnie valt om. (J 1;11) M falls over deze kanniet ope. (A 2;1) this can-not open gaatie af. (A 2;1) goes-it off hij doe(t) niet. (J 1;11) he does (= works) not

hij blijft met mij. (Mo/A 1.5) he stays with me ik komt hier. (Os/T 1.6) I comes here ik woont in casablanca. (Mo/A 1.4) I lives in casablanca die kan ook. (Os/T 2.6) that is-possible too die man gaat naar ander stad. (Os/T 1.6) that man goes to other town hij doet niet. (Mo/A 1.9) it does (= works) not

Examples of particles as predicates of non-agentive utterances are given in (6). (6) Type B4. Particle child L1 Dutch Particle

variant: Finite form + prepositional phrase

fiets in. (J 1;8) bicycle [goes] in [boekje] op [schoot]. (A 1;8) book [is] on lap die deruit. (A 1;9) that [comes] there-out Mijnie buggy in. (J 1;8) M buggy in

kom es inne huisje. (J 2;0) come in-the house [paard] zittie oppe dak. (J 1;11) horse sits-he on-the roof kommie da-uit. (A 2;0) comes-he there out gaattie inne garage. (J 2;2) goes-it into garage

The lexical stage  

A: agentive

B: non-agentive

Ctrlʺ agent Ctrlʹ (carries out)

 257

Predʺ theme Predʹ (occurs / undergoes)

Ctrl complement | action (causation, motion): e.g. dop opdoen (cap on-do), tiktak om (watch on-do), bad zitte (bath sit-down)

modal: e.g. wil (want), kan (can), mag (may), moet (must) | aspectual: e.g. gaat (goes), komt (comes) | default: e.g. doet (does), 0 = to be inferred | assertion: e.g. is (is), 0 (indeed)

Pred ( complement ) | state: e.g. zit (sits), lust (likes), past (fits), kan (is possible), doet (works) | change of state: e.g. valt (falls), komt (comes), gaat (goes) | assertion85 : is (is), 0 (indeed)

Figure 1: The syntactic structure of agentive (A) and non-agentive (B) types of utterance

The examples in (6) also show that at the lexical stage non-agentive particles may occur as the head of a prepositional phrase in utterances with finite verb forms such as ga(at), kom(t) and zit.

85 In early L2 learner varieties, is is used to express the hold-for relation between the predicate and the subject not only in non-agentive utterances as in de bang is daar (the danger is there), but also in agentive utterances as in de jongens is spelen (the boys is play-Inf). At the relevant stage, is is the lexical means to express the pragmatic function of assertion.

258 

 Conclusion

In sum, at the initial stage, the basic language system of child L1 and adult L2 learners of Dutch is a lexical system. Utterances are the expression of two types of predication that are used in complementary distribution. As represented in Figure 1, predications are either agentive or non-agentive. In agentive utterances the predicate refers to an action carried out by an agent. In non-agentive utterances the predicate refers state or a change of state that applies to a theme. This lexical learner system is the result of a creative process of language learning in which learners at the initial stages of the acquisition process use their ability to analyse the target language input in terms of a structural opposition between two types of lexical categories. This analysis seems to be based on the relevance of a cognitive distinction between situations that are under control and situations that are not. The representation in Figure 1 provides of an account of all the data, i.e. of all the utterances with a predicate, not just of utterances with verbs. The verb is a category of the target language. It becomes relevant only when learners reach the functional stage of language development. As shown in Figure 1A, the agentive predicate projects a position for an agent serving as the participant that carries out a causal action or an agentive motion. The presence of a modal or aspectual element or an element with an agentive default function indicates that there is an agent serving as the controller of an action (e.g. dop opdoen, cap on-do; tiktak om, watch on-do; bad zitte, bath sitdown). The modal elements wil (want), kan (can), mag (may), moet (must) are used to express the semantics of control, i.e. the willingness, the ability, the permission or the obligation of the agent to perform an action. The negative modal equivalents wilniet (want-not), kanniet (can-not), magniet (may-not) and hoefniet (must-not) are, of course, used to express the opposite. In Figure 1B, the nonagentive predicate projects a position for a theme that refers to the element that either occurs in a state or undergoes a change of state. In agentive utterances as in type A, modal elements are lexicalizations of variation in control. Aspectual elements such as gaat or komt are also lexicalizations of control. They are used to refer to the intention of the agent to perform an action. If doet is used, it expresses the default meaning of control. Modal and aspectual elements with the meaning of ‘control’ do not occur with state or change-of-state predicates. Thus, utterances such as deze kan vallen (this-one can fall), hij mag komen (he may come), deze gaat vallen (this-one goes fall), hij komt zitten (he comes sit), hij doet blijven (he does stay) do not occur. At the functional stage however, as in the target language, the language system is such that this type of utterance is commonly used.

The lexical stage  

 259

Modal elements, komt, gaat and doet are also used in non-agentive utterances. However, in this type of utterance as in (5) they are used to refer to a state (modal, doet) or a change of state (komt, gaat). At the lexical stage, as represented in Figure 1, is is used as the lexical head of both agentive and non-agentive types of utterance. In agentive utterances, is is used to express the hold-for relation between the predicate and the agent. Similarly, it is used in non-agentive utterances with respect to the relation between the predicate and the theme. It serves as a means to express the default function of assertion. At the lexical stage, utterance structure is determined by the semantics of the predicate. Hence, there is no head movement (raising). Verb morphology is determined by the form in which the verb occurs in the input. Thus, although there seems to be a contingency between verb placement and morphology, there is no role for morphological variation of the lexical verb. Verb morphology is unanalysed. It is the position that determines the morphology of the verb and not vice versa. Typical for adult L2 Dutch is the fact that the contingency between verb placement and morphology is less pronounced. Thus, L2 learners may produce both (7a) and (7b).⁸⁶ (7) a.

b.

hij pak trui. (Mo/A 1.3) he take[s] sweater ik komt hier. (Os/T 1.6) I comes here

hij pakken een mayonaise. (Mo/A 1.6) he get-Inf a mayonnaise dan hij vallen achter. (Os/T 2.9) then he fall-Inf behind

Furthermore, there is some variation in adult L2 learners that child L1 learners do not show. For example, Moroccan learners of Dutch use more verb forms, both finite and non-finite, in initial position than Turkish learners. Turkish learners, on the other hand, use more verb forms, both finite and non-finite, in final position

86 Klein and Perdue note that initially „ (…) lexical items typically occur in one invariant form. It corresponds to the stem, the infinitive or the nominative in the target language: but it can also be a form, which would be an inflected form in the target language. Occasionally, a word shows up in more than one form, but this rare variation does not seem to have any functional value“ (1997: 113). For a similar view, see, among others, Ingram and Thompson (1996) and Bittner (2003).

260 

 Conclusion

than Moroccan learners.⁸⁷ Thus, although morphology seems irrelevant, placement in initial or final position is not random. It shows the linguistic influence of the L1.⁸⁸ To conclude, at the initial stage of child L1 and adult L2 Dutch, utterance structure is the expression of the projection of a lexical category. Functional categories are not yet established. At the relevant stage, absence of the auxiliary verb heb, heeft (have, has) is evidence that there is no functional position for the expression of (semantic) finiteness. As summarized in Table 1, absence of the establishment of the functional projection of finiteness means (1) that functional categories such as auxiliaries are absent, (2) that head movement (verb-second) does not occur, (3) that topicalization and focalization do not play a role, (4) that neither the determiner system (definite vs. indefinite) nor the inflectional system (agreement, tense) has been established and that wh-words and anaphoric pronouns do not occur.⁸⁹ The data as presented in (1) to (7) show that this is exactly what can be observed.

finiteness (auxiliary) head movement (verb-second) topicalization / focalization determiners (definiteness) wh-words inflection (agreement, tense) anaphoric pronouns

Lexical stage

Functional stage

Basic Variety

Target system

− − − − − − −

+ + + + + + +

Table 1: Functional categories in the acquisition of child L1 and adult L2 Dutch

87 According to Verhagen (p.c.), Moroccans typically produce complex, head-initial verb forms as in die hond hij wil eten de taart (that dog he wants eat the cake); hij gaat pakken de taart op tafel (he goes get the cake on the table); zij is pakt de taart in de oven (she is got the cake in the oven). However, with the acquisition of the auxiliary heb, heeft this becomes: de hond wil taart eten (the dog wants cake eat). 88 See Verhagen (2009a: 133f.) Tables 6 and 7. This confirms the Alternation Hypothesis as put forward in Jansen et al. (1981: 315). It states that “when the target language offers an alternation between two patterns (e.g., verb final and verb second, as in Dutch), a second language learner will tend to overgeneralize the pattern existing in his or her first language (e.g., verb final in Turkish, verb second in Moroccan Arabic).” 89 Each of these functional categories consists of a closed-class set of elements. The meaning of a closed-class element depends on its status as member of a system as a whole.

The lexical stage  

 261

VP subject

V′

Spec

V | Headlex

complement | ComplVP

Type A:

subjectagent poes (kitty

PredCtrl wil wants

Type B:

subjecttheme

Predstate/ change of state (Compl)

poes (kitty

vindt finds

Complaction brokjes eten cat-food eat)

deze lekker this tasty)

Figure 2: Utterance structure at the lexical stage: the full projection of V

The representation in A and B of Figure 1 indicates the semantic difference between the two types of utterance in the learner languages at the lexical stage. Structurally, however, both these representations are identical. For language learners it might therefore be not too difficult to discover that both are variants of one abstract underlying structure as in Figure 2.⁹⁰ This underlying structure is in fact the full projection of V. In studies on L1 acquisition, it is commonly assumed that, initially, utterance structure is head-final. The reason for this is the frequency of learner utterances as in (1). However, as shown in Figure 2, at the lexical stage utterance structure is head-initial. The head-final structure in (1) is in fact the structure of the complement. In Figure 2, this holds for both brokjes eten (cat-food eat) in type A and deze lekker (this tasty) in type B. At the lexical stage, utterances are predications, i.e. expressions of a hold-for relation between a predicate and a subject (external argument). The selection of a constituent as the subject is based on the degree of activity. Thus, if there is an agent as in type A, it becomes the subject. A non-agent as in type B may become the subject only if there is no agent.

90 Both in A and B the complement is a complex lexical entity. Lexical entities are head-final. The position of V however is to the left of this complement and, hence, the syntactic structure of the utterance is head-initial.

262 

 Conclusion

At the lexical stage, the subject simultaneously serves the function of the topic. That is, the subject establishes the relation with the situation that the utterance applies to. This explains why the subject always occurs in initial position. It also explains why at the relevant stage topicalization of, for example, the object or an adverb as it occurs in the target language, does not exist.

8.3 The functional stage A major process of restructuring occurs in agentive utterances as represented in type A of Figure 1. In the target language, these utterances frequently occur with a topicalized object or a topic adverbial. At the lexical stage, however, topicalization as it occurs in the target language is not an option. This is because the initial position is a subject position. Nevertheless, given the constraints of the learner language at the lexical stage, some kind of topicalization is possible in utterances as in (8). (8) Agentive utterances with an object or an adverbial in topic position⁹¹ child L1 Dutch

adult L2 Dutch

dit nee afdoen. (J 1;10) this noAG off do da kanwel opzitte. (J 2;0) there canAGindeed on-sit disse hoeniet meeneme. (A 2;1) this mustAGnot with-take

hier schoenen uitdoen moet. (Os/T 1.8) here shoes off-do mustAG die wijn magwel drinken. (Os/T 2.3) that wine mayAGindeed drink dan moet daar helemaal opruimen. (Os 2.5) then mustAG there all up-clean

Utterances as in (8) are not targetlike. However, they may serve the function of topicalization due to the fact that the relevant structure does not provide a position for a subject. It is a creative solution that is possible within the constraints given by the learner system at the lexical stage. This solution, however, does not hold vis-à-vis the target language input. Hence, restructuring occurs as represented in Figure 3. As shown in Figure 3, reanalysis of the VP-complement in terms of a full projection of V establishes a subject position for the agent. This allows the modal element to become reanalysed as the functional head F. The establishment of this

91 In the following examples, AG indicates that the agent is implicitly used with the modal predicate.

The functional stage 

 263

functional head position serves the expression of finiteness, i.e. the expression of the pragmatic function of assertion. Furthermore, the full projection of F creates a specifier position for the expression of the topic function to establish the relation between the utterance and the situation that the utterance applies to. Thus, at the functional stage, learner grammar proceeds towards a complex, fully-fledged system which accounts for the embedding of the predication within a discoursesituational context. FP Spec

F′ F (head)

complement ‖ VP Spec

V′ complement

topic brokjes poes

poes

Aux ↑ mag wil ↑ Fin ↑ eet

Headlex

subjectagent object

Predaction

poes

niet eten eten

brokjes

brokjes

Figure 3: Utterance structure at the functional stage

Reanalysis of the lexical head as a functional head becomes evident in utterances with a past participle. At the functional stage, these utterances come to be used with the auxiliary heb, heef (have, has) and ben, is (am, is). This acquisition process occurs rather abruptly. It is due to the fact that, given a position for the auxiliary verb, this position is structurally available. Examples of this developmental process are shown in (9).

264 

 Conclusion

(9) Utterances with past participles at the lexical and the functional stage child L1 Dutch The lexical stage

The functional stage

poppie haartie wast. (J 1;10) doll hair washed papa potmaakt. (A 2;0) daddy kaput-made

ikke hè dit pakt. (J 2;1) I have this got Jaja hemme al goonmaakt. (A 2;2) J has already up-cleaned

adult L2 Dutch The lexical stage

The functional stage

vandaag school ook afgelopen (Os/T 2.1) today school too ended hij zegt hij niet veel slaap. (Mo/A 1.3) he says he not much slept

eten is afgelopen. (Os/T 2.9) dinner is ended ik heb daar geslapen buiten. (Mo/A 3.6) I have there slept outside

Reanalysis of the lexical head as a functional head explains the contingency between the use of auxiliaries (aspectual and modal) and the variation in word order (inversion) as a function of topicalization or focalization. It accounts for the examples as given in (10) and serves as a prerequisite for the acquisition of the determiner system, wh-words and anaphoric pronouns. (10) The establishment of topic and focus child L1 Dutch

adult L2 Dutch

object topic heef Cynthia maakt. (J 2;0) [that] has C made die heb ik wel geplakt. (J 2;2) that have I indeed glued die mag jij zo opdrinke. (J 2;2) that may you later up-drink die heef Cynthia gemaakt. (J 2;2) that has C made dà heefe Jaja daan [=gedaan]. (A 2;2) that has J done da mag papa wel doen. (A 2;3) that may daddy indeed do

moet jij zelf betalen. (Os/T 3.2) [that] must you yourself pay kan hij helemaal vergeten. (Os/T 3.2) [that] can he all forget kan ik proberen. (Os/T 3.6) [that] can I give-a-try dat wil ik niet [doen]. (Mo/A 2.6) that want I not [do] die heb ik allemaal vergeten. (Mo/A 2.9) that have I all forgotten heeft ie gekocht. (Mo/A 3.1) [that] has he bought

The functional stage 

die isse Jaja gete [=vergeten]. (A 2;4) that is J forgotten broodje mag Cynthia opete. (A 2;4) bun may C indeed up-eat

 265

die heb ik hier gedaan. (Mo/A 3.5) that have I here done dat heb ik nooit gezegd. (Mo/A 3.8) that have I never said

adverb topic hier mag je neus snuite. (J 2;4) here may you nose blow da issie varre [=gevallen]. (A 2;2) there is-he fallen da kanne kindere inzitte. (A 2;4) there can children in-sit

dan moet ik huis zoeken. (Os/T 3.3) then must I house look-for hier kan ik niet lezen, hè? (Mo/A 2.8) here can I not read, right? toen heeftie teruggegeven aan. (Mo/A 3.2) the has-he back-given to

adverb / object focus waar ben je nou geweest? (J 2;2) where are you now been? waar heb je chocola gelate? (J 2;2) where have you chocolate left

waar heb jij geweest? (Os/T 3.4) where have you been wat moet ik doen dan? (Os/T 3.3) what should I do then?

0-focus heb je visje gehad? (J 2;2) have you fish had? mag ik eve wel deze aankome? (A 2;3) may I just indeed this touch?

heb je die auto gezien? (Mo/A 2.5) have you car seen? kan ik niks anders doen? (Os/T 3.3) can I nothing else do?

Verbal elements in the position of the auxiliary verb are used to express the pragmatic function of assertion. In the absence of an auxiliary verb, it is the lexical verb that serves this function. Hence, the lexical verb may occur in either of two positions: as the lexical head in its sentence-final position or as carrier of the pragmatic function of assertion in the functional position of verb-second. Placement of the lexical verb in the position of verb-second is referred to as ‘head movement’. With head movement the lexical verb is used to carry out two grammatical functions simultaneously. That is, the semantic function of the head of a lexical category and the pragmatic function of finiteness. Given the linguistic complexity of this double function, (‘dummy’) auxiliaries such as gaat (goes), komt (comes), doet (does) and is (is) may serve as a means to avoid head movement in examples as in (11).

266 

 Conclusion

(11) ‘Dummy’ auxiliaries: gaat (goes), komt (comes), doet (does) and is (is) in child L1 and adult L2 Dutch⁹² child L1 Dutch

adult L2 Dutch

ik ga eve die glije. (J 2;0) I go just that slide-Inf gaatie ape. (A 2;1) goes-he sleep-Inf

hij gaat hier werken. (Mo/A 3.9) he goes here work-Inf ikke komt hier werken. (Os/T 1.3) I comes here work-Inf

ik doete opzitte. (J 1;11) I does-it on-sit-Inf doetie viesmake. (A 2;1) does-he dirty-make-Inf ⁹³

doe dan die trekken. (Os/T 2.9) do then that-one draw-Inf hij doet krant (…) opendoen. (Os/T 3.3) he does newspaper opendo-Inf

toen is Hans eve Erre ophaje. (A 2;3) then is H just E up-pick-Inf toen is e vogel da vliegen.⁹⁴ then is a bird there fly-Inf

die is nou werken. (Os/T 1.3) that-one is now work-Inf … hij is crossfiets spelen. (Mo/A 1.3) … he is crossbike play-Inf

Finally, the contingency in the input between the position of the verb and its morphology makes it possible for the learner to discover the regularities of the variation in verbal morphology and thus to acquire the morphological properties of finiteness. Hence, it is the acquisition of the auxiliary as the head of a functional projection that serves as a prerequisite for – not the cause or the trigger of – the acquisition of inflection (agreement and tense). The acquisition of inflection itself is the result of a process of morphological analysis that follows the acquisition of variation in verb placement (movement). The morphological analysis of the lexical verb forms is a gradual process that is irrelevant with respect to the expression of the functional properties of finiteness. Crucial for the expression of finiteness as a functional category is the position of the lexical verb, not its morphology. It explains why, at the relevant stage of L2 acquisition, non-finite verb forms may occur in second (or third) position as, for example, in ik maken die fiets

92 Some variation occurs in examples as in: ik was poepie doen (I was pooh do-Inf. J 2;0); kwam alle familie in auto zitte (came all family in car sit-Inf. Mo/A 3.9); ikke staan hard werken (I stand hard work-Inf. Mo/A 2.4) 93 See also Van Kampen (1997): ik doe afwassen (I do dishwash-Inf. Sarah 2;5); ik doe ook praten (I do also talk-Inf. Sarah 2;5). 94 In Jolink (2009: 253)

Driving forces  

 267

alles (I make-Inf that bicycle everything. Mo/A 1.3) and dan hij vallen achter (then he fall-Inf behind. Os/T 2.9). In sum, utterances such as disse hoeniet meeneme (this mustAGnot with-take. A 2;1) typically occur at the lexical stage. They are not targetlike. They serve the function of a provisionary kind of topicalization. Substitution of the action predicate (meenemen, with-take) with a full blown lexical projection (ik deze meeneme, I this-one with-take) allows the creation of two functional positions: a position for the expression of finiteness and a position for a constituent with topic or focus function. These functional positions are a prerequisite for the acquisition of function words: auxiliary verbs, determiners, wh-words and anaphoric pronouns; the acquisition of the morphological properties of inflection: tense and agreement; and the acquisition of word order variation: topicalization and focalization and, hence, inversion. The developmental process as visualized in Figure 3 accounts for the observations summarized in Table 1. The functional position of finiteness (F) is a prerequisite for the acquisition of auxiliary verbs, head movement (verb-second) and the inflectional systems of agreement and tense. The functional specifier position (SpecFP) is a prerequisite for the acquisition of topicalization and focalization and both the determiner system and the use of wh-words.

8.4 Driving forces Lexical categories are relevant with respect to the content of an utterance. They are used to refer to actions, states and changes-of-state. Functional categories are relevant with respect the properties of information structure. They are the means with which an utterance is embedded into its discourse-situational context. At the initial stage, the learner grammar is shaped to express the hold-for relation between a predicate and the subject (the external argument). This type of relation, referred to as a predication, is expressed in terms of a lexical projection in which the head-complement relation serves to express the predicate (VP) and the specifier (SpecVP) serves to express the subject. Furthermore, if the subject is the agent of a modal predicate, reference to the speaker expresses the willingness, the ability, the permission or the obligation of the speaker to carry out a particular action. The subject is the theme if it refers to the entity that either occurs in a state or undergoes a change of state. At the functional stage, finiteness plays a central role as far as its informational function is concerned. Finiteness is used to express that an utterance is true, i.e. that it holds for a particular situation. At the initial stage, the learner system lacks the functional means for the expression of the pragmatic function of finiteness. At the relevant stage, this particular func-

268 

 Conclusion

tion of an utterance and its contextual embedding is expressed either lexically or it is left to be inferred. In the language acquisition process, the linguistic means for the expression of the function of finiteness becomes available as the result of a process of restructuring. It is the pragmatic function of finiteness and topicalization to refer to the situation that the utterance applies to. For the acquisition of these functions, the position of the lexical head has to be reanalysed, such that it serves as the functional category F. F is the head of the functional projection FP. FP provides a position F for the expression of finiteness and a position SpecFP for elements with either topic or focus function. In sum, it is the pragmatic function of contextual embedding that serves as the driving force in the acquisition of the functional projection of finiteness. In the system of the target language, the functional role of finiteness is exerted with a verbal element in second position. This verbal element is either an auxiliary or in absence of an auxiliary, it is the lexical verb. For the expression of the functional properties of finiteness, its morphological properties are irrelevant. This explains why learners, L2 learners in particular, may not observe the contingency between placement and morphology as native speakers of the target language system have learned to do. Finiteness and topicalization are the functional means with which an utterance is embedded into a discourse-situational context. They are the linguistic means that make it possible for other people to know what the utterance is about and what the intention is with which it is used. A prerequisite for this on the cognitive side is the capacity of a human being to place himself into the position of another human being. As long as learners have available the linguistic means of the basic language system, they are not or only to a limited extent equipped with the ability to express this capacity. For second language learners, this is due to the fact that they simply do not yet have the appropriate linguistic means. For children, it seems plausible to assume that the relation between cognitive development and language development goes hand in hand. Hence, it is claimed that children develop from an egocentric phase with a lexical system that serves the expression of a predication to a context-sensitive phase with a functional system that is tuned to the relevant situational context (see also Platzack 1996).⁹⁵ For

95 With respect to the frequent occurrence of non-finite utterances in early stages of language acquisition, Platzack argues: „Note that I consider finiteness to express the anchoring of the clause in time and space (…), hence lacking finiteness, the nonfinite sentences are not supposed to be as closely bound to the speech act as the finite ones. Speculating about the deeper reasons for the abundant use of nonfinite sentences in early child language, and

Driving forces  

 269

children with a specific language impairment (SLI) it has been noted that functional aspects of language are most difficult to acquire (Johnston 1997; Clahsen and Hansen 1997; Reilly et al. 2004). In this context, it is noteworthy that Johnston points out that “since our work in the 1970s (…) a number of investigators have confirmed that SLI children learning English do not use inflectional morphemes, auxiliaries, articles and other functors as well as would be expected from MLU [mean length of utterance]” (Johnston 1997: 172). It seems most likely that it is the limited capacity of SLI children to place themselves into the position of another person that explains why it is that precisely these functional aspects of language are most difficult to acquire.

noticing that the option seems to fade away around the age of three, it is tempting to relate it to the egocentric standard observed by Piaget, which is replaced by a contextsensitive standard at age four (…) (1996: 388)“. The conclusion from this should be that the use of finite forms at the initial stages of language acquisition does not serve to express the communicative function of finiteness. However, Platzack does not seem to be willing to draw this conclusion. On the contrary, he sees the use of finite and nonfinite forms as evidence that „children know the syntactic specificities of their first language at an early age“ (1996: 405). Whatever position is taken, it seems that one cannot have both at the same time.

References Anderson, Stephen R. 1993. Wackernagel’s revenge: Clitics, morphology, and the syntax of second position. Language 69: 68–98. Baker, Carl Lee. 1979. Syntactic theory and the projection problem. Linguistic Inquiry 10: 533–581. Bannard Colin, Elena Lieven and Michael Tomasello. 2009. Modeling children’s early grammar knowledge. Proceedings of the National Academy of Sciences, Vol. 106: 17284–17289. Barbiers, Sjef. 2002. Modality and polarity. In Sjef Barbiers, Frits Beukema and Wim van der Wurff (eds.), Modality and its interaction with the verbal system, 51–73. Amsterdam: Benjamins. Beck, Marie-Luise. 1998. L2 acquisition and obligatory head movement: English-speaking learners of German and the local impairment hypothesis. Studies in Second Language Acquisition 20: 311–348. Becker, Angelika. 2005. The semantic knowledge base for the acquisition of negation and the acquisition of finiteness. In Henriëtte Hendriks (ed.), The structure of learner varieties, 263–314. Berlin/New York: Mouton de Gruyter. Behrens, Heike. 1993. Temporal reference in German child language: Form and function of early verb use. Ph.D. dissertation. Amsterdam: University of Amsterdam. Bernini, Giuliano. 2003. The copula in learner Italian: Finiteness and verbal inflection. In Christine Dimroth and Marianne Starren (eds.), Information structure and the dynamics of language acquisition, 159–185. Amsterdam: John Benjamins. Bittner, Dagmar. 2003. The emergence of verb inflection in two German-speaking children. In Dagmar Bittner, Wolfgang U. Dressler and Marianne Kilani-Schoch (eds.), Development of verb inflection in first language acquisition, 53–88. Berlin: Mouton de Gruyter. Bley-Vroman, Robert. 1990. The logical problem of foreign language learning. Linguistic Analysis 20: 3–49. Booij, Geert E. 1996. Inherent versus contextual inflection and the split morphology hypothesis. In Geert E. Booij and Jaap van Marle (eds.), Yearbook of morphology 1995, 1–16. Dordrecht: Kluwer. Booij, Geert E. 2000. Inflection and derivation. In Geert E. Booij, Christian Lehmann and Joachim Mugdan (eds.), Morphology. An international handbook on inflection and word-formation. Vol. 1, 360–369. Berlin: Walter de Gruyter. Bowerman, Melissa. 1974. Learning the structure of causative verbs: A study in the relationship of cognitive, semantic and syntactic development. Papers and reports on child language development, 142–178. Stanford University Committee on Linguistics. Bowerman, Melissa. 1990. Mapping thematic roles onto syntactic functions: Are children helped by innate linking rules? Linguistics 28: 1253–1289. Clahsen, Harald. 1986. Verb inflections in German child language: Acquisition of agreement markings and the functions they encode. Linguistics 24: 79–121. Clahsen, Harald. 1988. Kritische Phasen der Grammatikentwicklung. Eine Untersuchung zum Negationserwerb bei Kindern und Erwachsenen. Zeitschrift für Sprachwissenschaft 7: 3–31. Clahsen, Harald. 1989. The grammatical characterization of developmental dysphasia. Linguistics 27: 897–920.

272 

 References

Clahsen, Harald and Detlef Hansen. 1997. The grammatical agreement deficit in specific language impairment: Evidence from therapy experiments. In Myrna Gopnik (ed.), The inheritance and innateness of grammars, 141–160. New York: Oxford University Press. Clahsen, Harald, Jürgen M. Meisel and Manfred Pienemann. 1983. Deutsch als Zweitsprache: Der Spracherwerb ausländischer Arbeiter. Tübingen: Gunter Narr. Clahsen, Harald and Pieter Muysken. 1986. The availability of Universal Grammar to adult and child learners. A study of the acquisition of German word order. Second Language Research 2: 93–119. Clahsen, Harald and Pieter Muysken. 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 Jürgen M. Meisel (ed.), The acquisition of verb placement. Functional categories and V2 phenomena in language acquisition, 181–223. Dordrecht: Kluwer. Clahsen Harald and Monika Rothweiler. 1992. Inflectional rules in children’s grammars: Evidence from German participles. In Geert E. Booij and Jaap van Marle (eds.), Yearbook of morphology 1992, 1–34. Dordrecht: Kluwer. Cook, Vivian J. and Mark Newson. 1996. Chomsky’s universal grammar. An introduction. Oxford: Basil Blackwell. Corder, S. Pit. 1978. ‘Simple codes’ and the source of the second language learner’s initial heuristic hypothesis. Studies in Second Language Acquisition 1: 1–10. de Bleser, Ria and Josef Bayer. 1988. On the role of inflectional morphology in agrammatism. In Michael Hammond and Michael Noonan (eds.), Theoretical morphology, 45–70. San Diego, CA: Academic Press. de Haan, Ger. 1987. A theory-bound approach to the acquisition of verb placement in Dutch. In Ger de Haan and Wim Zonneveld (eds.), Formal parameters of generative grammar. Yearbook 1987, 15–30. Dordrecht: ICG. de Villiers, Jill G. and Peter A. de Villiers. 1986. The acquisition of English. In Dan I. Slobin (ed.), The cross-linguistic study of language acquisition. Vol. 1, 27–139. Hillsdale N.J.: Lawrence Erlbaum. Dik, Simon C. 1978. Functional Grammar. Amsterdam: North Holland. 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 Christine Dimroth and Marianne Starren (eds.), Information structure and the dynamics of language acquisition, 65–93. Amsterdam: John Benjamins. Ertel, Suitbert. 1977. Where do the subjects of sentences come from? In Sheldon Rosenberg (ed.), Sentence production. Developments in research and theory, 141–167. Hillsdale, N.J.: Lawrence Erlbaum. Eubank, Lynn. 1993/94. On the transfer of parametric values in L2 development. Language Acquisition 3: 184–208. Eubank, Lynn. 1996. Negation in early German-English interlanguage: More valueless features in the L2 initial stage. Second Language Research 12: 73–106. Franceschina, Florencia. 2001. Morphological or syntactic deficits in near-native speakers? An assessment of some current proposals. Second Language Research 17: 213–247. Gillis, Steven. 2003. A case study of the early acquisition of verbs in Dutch. In Dagmar Bittner, Wolfgang U. Dressler and Marianne Kilani-Schoch (eds.), Development of verb inflection in first language acquisition, 171–203. Berlin: Mouton de Gruyter.

References 

 273

Guiliano, Patrizia and Daniel Véronique. 2005. The acquisition of negation in French L2. An analysis of Moroccan Arabic and Spanish “learner varieties”. In Henriëtte Hendriks (ed.), The structure of learner varieties, 355–404. Berlin/New York: Mouton de Gruyter. Hatch, Evelyn M. 1983. Psycholinguistics. A second language perspective. Rowley, Mass.: Newbury House. Hawkins, Roger. 2001. Second language syntax. A generative introduction. Blackwell. Hawkins, Roger and Cecilia Yuet-hung Chan. 1997. The partial availability of universal grammar in second language acquisition: The ‘failed functional features hypothesis.’ Second Language Research 13: 187–226. Haznedar Belma and Bonnie D. Schwartz. 1997. Are there optional infinitives in child L2 acquisition? In Elizabeth Hughes, Mary Hughes and Annabel Greenhill (eds.), Proceedings of the 21st annual Boston University conference on language development, 257–268. Somerville, MA: Cascadilla Press. Herschensohn, Julia. 2001. Missing inflection in second language French: Accidental infinitives and other verbal deficits. Second Language Research 17: 273–305. Hoekstra, Teun and Peter Jordens. 1994. From adjunct to head. In Teun Hoekstra and Bonnie D. Schwartz (eds.), Language acquisition studies in generative grammar, 119–149. Amsterdam: John Benjamins. Hopper, Paul and Sandra Thompson. 1980. Transitivity in grammar and discourse. Language 56: 251–299. Hutz, Matthias. 2004. Is there a natural process of decay? A longitudinal study of language attrition. In Monika S. Schmid, Barbara Köpke, Merel Keijzer and Lina Weilemar (eds.), First language attrition. Interdisciplinary perspectives on methodological issues, 189–206. Amsterdam: John Benjamins. Ingram, David and William Thompson. 1996. Early syntactic acquisition in German: Evidence for the Modal Hypothesis. Language 72: 97–120. Ionin, Tania and Kenneth Wexler. 2002. Why is ‘is’ easier than ‘-s’: Acquisition of tense/ agreement morphology by child second language learners of English. Second Language Research 18: 95–136. Jansen, Bert, Josien Lalleman and Pieter Muysken. 1981. The Alternation Hypothesis: Acquisition of Dutch word order by Turkish and Moroccan foreign workers. Language Learning 31: 315–336. Johnston, Judith R. 1997. Specific language impairment, cognition and the biological basis of language. In Myrna Gopnik (ed.), The inheritance and innateness of grammars, 161–180. New York: Oxford University Press. Jolink, Anke. 2009. Finiteness in children with SLI. A functional approach. In Christine Dimroth and Peter Jordens (eds.), Functional categories in learner language, 235–259. Berlin: Mouton de Gruyter. Jolink, Anke. 2010. The occurrence of the dummy verb zijn in L1 Dutch: Longitudinal data from normally developing children and children with SLI. Workshop on dummy auxiliaries in (a) typical first and second language acquisition. July 1–2. Nijmegen: Radboud University. Jordens, Peter. 1990. The acquisition of verb placement in Dutch and German. Linguistics 28: 1407–1448. Jordens, Peter. 1991. Linguistic knowledge in second language acquisition. In Lynn Eubank (ed.), Point counterpoint. Universal Grammar in the second language, 199–218. Amsterdam: John Benjamins. Jordens, Peter. 2002. Finiteness in early child Dutch. Linguistics 40: 687–765.

274 

 References

Jordens, Peter and Christine Dimroth. 2006. Finiteness in children and adults learning Dutch. In Natalia Gagarina and Insa Gülzow, The acquisition of verbs and their grammar: The effect of particular languages, 173–198. Dordrecht: Springer. Klein, Wolfgang. 1994. Time in language. London: Routledge. Klein, Wolfgang. 1997. Learner varieties are the normal case. The Clarion. Magazine of the European Second Language Association 3: 4–6. Klein, Wolfgang. 1998. Assertion and finiteness. In Norbert Dittmar and Zvi Penner (eds.), Issues in the theory of language acquisition, 225–245. Bern: Peter Lang. Klein, Wolfgang. 2008. The topic situation. In Bernt Ahrenholz, Ursula Bredel, Wolfgang Klein, Martina Rost-Roth and Romuald Skiba (eds.), Empirische Forschung und Theoriebildung. Beiträge aus Soziolinguistik, Gesprochene Sprache- und Zweitspracherwerbsforschung. Festschrift für Norbert Dittmar zum 65. Geburtstag, 287–305. Klein,Wolfgang and Norbert Dittmar. 1979. Developing grammars. Heidelberg: Springer Verlag. Klein, Wolfgang and Clive Perdue. 1997. The Basic Variety (or: Couldn’t natural languages be much simpler?). Second Language Research 13: 301–347. Klein, Wolfgang and Clive Perdue. 1992. (eds.) Utterance structure. Developing grammars again. Amsterdam: John Benjamins. Lardière, Donna. 1998. Dissociating syntax from morphology in a divergent L2 end-state grammar. Second Language Research 14: 359–375. Lasser, Ingeborg. 1997. Finiteness in adult and child German. MPI Series in Psycholinguistics. Nijmegen: Max Planck Institute. Lenneberg, Eric. 1967. Biological foundations of language. New York: John Wiley. Meisel, Jürgen M. 1997. The acquisition of the syntax of negation in French and German: Contrasting first and second language development. Second Language Research 13: 227–263. Meisel, Jürgen M. (ed.). 1992. The acquisition of verb placement. Functional categories and V2 phenomena in language acquisition. Dordrecht: Kluwer. Müller, Natascha and Zvi Penner. 1996. Early subordination: The acquisition of free morphology in French, German, and Swiss German. Linguistics 34: 133–165. Parodi, Teresa. 1998. Der Erwerb funktionaler Kategorien im Deutschen. Tübingen: Gunter Narr. Parodi, Teresa. 2000. Finiteness and verb placement in second language acquisition. Second Language Research 16: 355–381. Perdue, Clive (ed.). 1993a. Adult language acquisition: Cross-linguistic perspectives. Vol. I. Field methods. Cambridge: Cambridge University Press. Perdue, Clive (ed.). 1993b. Adult language acquisition: Cross-linguistic perspectives. Vol. II. The results. Cambridge: Cambridge University Press. Perdue, Clive. 2006. “Creating language anew”: Some remarks on an idea of Bernard Comrie’s. Linguistics 44: 853–871. Perdue, Clive, Sandra Benazzo and Patrizia Giuliano. 2002. When finiteness gets marked: The relation between morphosyntactic development and use of scopal items in adult language acquisition. Linguistics 40: 849–890. Piaget, Jean. 1983. Piaget’s Theory. In Paul H. Mussen (ed.), Handbook of child psychology. Vol. 1. History, theory, and methods, 103–128. New York: Wiley. Pienemann Manfred. 1986. Psychological constraints on the teachability of languages. In Carol Pfaff (ed.), First and second language acquisition processes, 143–168. Cambridge, Mass.: Newbury House. Also 1988 in: Studies in Second Language Acquisition 6: 186–214.

References 

 275

Platzack, Christer. 1996. The Initial Hypothesis of syntax. A minimalist perspective on language acquisition and attrition. In Harald Clahsen (ed.), Generative perspectives on language acquisition. Empirical findings, theoretical considerations and crosslinguistic comparisons, 368–414. Amsterdam: John Benjamins. Poeppel, David and Kenneth Wexler. 1993. The Full Competence Hypothesis of clause structure in early German. Language 69: 1–33. Prévost, Philippe and Lydia White. 1999. Accounting for morphological variation in second language acquisition: Truncation or missing surface inflection? In Marc-Ariel Friedemann and Luigi Rizzi (eds.), The acquisition of syntax, 202–235. London: Longman. 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. Reilly, Judy, Jill Weckerly and Beverly Wulfeck. 2004. Neuroplasticity and development: The acquisition of morphosyntax in children with early focal lesions and children with specific language impairment. In Ludo Verhoeven and Hans van Balkom (eds.), Classification of developmental language disorders. Theoretical issues and clinical implications, 39–59. New Jersey: Erlbaum. Rizzi, Luigi. 1993. Some notes on linguistic theory and language development: The case of root infinitives. Language Acquisition 3: 371–393 Rule, Sarah and Emma Marsden. 2006. The acquisition of functional categories in early French second language grammars: The use of finite and non-finite verbs in negative contexts Second Language Research 22: 188–218. Schaerlaekens, Annemarie and Steven Gillis. 1987. De taalverwerving van het kind. Groningen: Wolters-Noordhoff. Schwartz, Bonnie D. 1996. Parameters in non-native language acquisition. In Peter Jordens and Josine Lalleman (eds.), Investigating second language acquisition, 211–235. Berlin, New York: Mouton de Gruyter. Schwartz, Bonnie D. and Rex A. Sprouse. 1996. L2 cognitive states and the Full Transfer / Full Access model. Second Language Research 12: 40–72. Slobin, Dan I. 1996. From “thought and language” to “thinking for speaking”. In John J. Gumperz and Stephen C. Levenson (eds.), Rethinking linguistic relativity, 70–96. Cambridge: CUP. Slobin, Dan I. (ed.). 1986. The cross-linguistic study of language acquisition. Vol. 1. Hillsdale N.J.: Lawrence Erlbaum. Snow, Catherine. 1976. Semantic primacy in first and second language acquisition. Interlanguage Studies Bulletin 1: 137–165. Starren, Marianne. 2001. The second time. The acquisition of temporality in Dutch and French as a second language. Utrecht: LOT. Tracy, Rosemarie. 2002. Growing (clausal) roots. All children start out (and may remain) multilingual. Linguistics 40: 653–686. Vainikka, Anne and Martha Young-Scholten. 1994. Direct access to X’-theory: Evidence from Korean and Turkish adults learning German. In Teun Hoekstra and Bonnie Schwartz (eds.), Language acquisition studies in generative grammar, 265–316. Amsterdam: John Benjamins. Vainikka, Anne and Martha Young-Scholten. 1996. The early stages in adult L2 syntax: Additional evidence from Romance speakers. Second Language Research 12: 140–176.

276 

 References

Vainikka, Anne and Martha Young-Scholten. 2011. The acquisition of German. Introducing Organic Grammar. Berlin: Mouton de Gruyter. van Kampen, Jacqueline and Frank Wijnen. 2000. Hoofdstuk 6. Grammaticale ontwikkeling. In Steven Gillis and Annemarie Schaerlaekens (eds.), Kindertaalverwerving. Een handboek voor het Nederlands, 225–285. Groningen: Martinus Nijhoff. Verhagen, Josje. 2009a. Finiteness in Dutch as a second language. Ph.D. dissertation: Amsterdam, VU University / Nijmegen, Max Planck Institute for Psycholinguistics. Verhagen, Josje. 2009b. Light verbs and the acquisition of finiteness and negation in Dutch as a second language. In Christine Dimroth and Peter Jordens (eds.), Functional categories in learner language, 203–234. Berlin: Mouton de Gruyter. Verrips, Maaike. 1996. Potatoes must peel. The acquisition of the Dutch passive. Ph.D. dissertation: University of Amsterdam. Weissenborn, Jürgen. 1990. Functional categories and verb movement: The acquisition of German syntax reconsidered. In Monika Rothweiler (ed.), Spracherwerb und Grammatik. Linguistische Untersuchungen zum Erwerb von Syntax und Morphologie (Linguistische Berichte Sonderheft 3), 190–224. Opladen: Westdeutscher Verlag. Weissenborn, Jürgen. 1992. Null subjects in early grammars: Implications for parameter-setting theories. In Jürgen Weissenborn, Helen Goodluck and Thomas Roeper (eds.), Theoretical issues in language acquisition: Continuity and change in development, 269–299. Hillsdale, NJ: Lawrence Erlbaum. Wexler, Kenneth. 1994. Optional infinitives, head movement and the economy of derivations. In David Lightfoot and Norbert Hornstein (eds.), Verb movement, 305–350. Cambridge: Cambridge University Press. Wexler, Kenneth. 1998. Very early parameter setting and the Unique Checking Constraint: A new explanation for the Optional Infinitive stage. Lingua 106: 23–79. Wijnen, Frank. 1997. Temporal reference and eventivity in root infinitivals. In Jeannette C. Schaeffer (ed.), The interpretation of root infinitives and bare nouns in child language. MIT Occasional Papers in Linguistics 12, 1–25. Wijnen, Frank, Masja Kempen and Steven Gillis. 2001. Root infinitives in Dutch early child language: An effect of input? Journal of Child Language 28: 629–660.

Index Agent first 161 agreement 9, 82–84, 197, 208, 209, 213, 214, 220, 226–228, 238, 240, 246–249 see also inflection anchoring – spatial 4, 72–74, 78 – temporal 4, 51, 70, 71, 74, 78, 210 aspect – grammatical 199–202, 205, 207 – lexical 195, 201, 203, 205 assertion 4, 42–45, 52–55, 70, 74, 194–196, 242–244 – affirmative 127–129, 194, 242 – negative 73, 127–129, 194, 242–245 auxiliary 53, 172–181, 191, 192, 202–207, 211, 212, 217–219, 234–236, 244–248, 250, 260, 263–269 Baker’s paradox 148 Basic Variety 1, 30, 41, 111, 253 see also learner language Becker, Angelika 242–248 Bittner, Dagmar 215–221 Booij, Geert E. 24–29 change-of-state verb 15, 60, 61, 89, 90, 110, 222, 225, 234, 236 see also predicate Clahsen, Harald 4, 34, 82, 87–90, 95–97, 99, 107 coherence, phrasal 15 complementary distribution 82, 85, 96, 103, 118, 131, 195, 201, 225, 243 complement-head 109, 166 complementizer 86, 87 compounding 11, 12, 24, 27 conceptualization 15, 57, 58, 61, 63, 65, 77 contextual embedding 4, 16, 19–21, 34, 66, 145, 146, 153, 155, 268 control 5, 36, 124–129, 133, 139, 150–152, 160, 182, 255, 258 Corder, S. Pit 111 definiteness 2, 3, 19, 21–23, 38, 72, 73, 209, 260

derivation 11, 12, 22–29, 33 determiner 1, 32, 113, 144, 221, 223, 248, 260, 264, 267 driving force 2, 3, 36, 78, 150, 158, 188, 210, 239, 249, 267, 268 ergatives 129–131, 148 European Science Foundation (ESF) 30, 38, 116, 253 finiteness 3, 4, 7, 19, 21–23, 27, 96, 99, 110, 155, 191–193, 218–220, 222, 224, 228–230, 238, 239, 248–251, 260, 266–269 – morphological 43, 44, 82, 84, 91, 98, 103, 114, 142, 154, 199, 248 – semantic 43, 44, 66, 70, 73, 114, 143, 153, 234, 242–247, 260 focalization 154, 186, 260, 267 focus 5, 36, 51, 136–138, 154, 184–187, 264–268 see also position Full Competence Hypothesis 4, 81–84, 99, 108 function words 111–113, 267 functional category 21–23, 26–29, 112, 113, 143 functional category system 3–5, 84, 87, 110, 113, 145, 187, 188, 205, 226, 227, 235–237 Gillis, Steven 6, 211–215, 222 Hatch, Evelyn M. 36 head 3–7, 10–13, 15, 16, 50–56 – functional (F) 7, 167, 168, 181, 210, 262–264 – lexical (V) 10, 11, 15, 16, 104, 128, 133, 152, 153, 167–169, 178, 263–265, 268 head movement 49, 55, 82–86, 95–97, 99, 103, 107, 259, 260, 265 see also verb raising, verb movement head-complement 52, 109, 133, 139, 140, 142, 166, 167 head-final 11, 13, 14, 98–102, 107, 108, 140–142, 239, 240

278 

 Index

head-initial 13, 14, 81, 98–102, 107, 108, 110, 138–142, 239, 240 Heidelberg Project 30 identifiability 19–22, 72, 73 Impaired Representation Hypothesis (IRH) 6, 197, 230, 249, 250 imperative 42, 43 inflection 6, 7, 11, 12, 22–29, 82–84, 90, 98, 112, 142, 144, 198, 215–234, 248–251, 260, 266–269 see also agreement Ingram, David 4, 82, 90–92, 95–97, 99, 107, 122, 143, 150, 152 initial state 4, 79, 84, 95, 225, 239 Klein, Wolfgang 1, 20, 30, 31, 34, 36, 52, 66, 70–79, 111, 114, 136, 200, 259 language acquisition – first 1, 30–38 – second 1, 30–38, 41 Lasser, Ingeborg 43, 44, 66, 234 learner language 1, 2, 4, 79, 107, 110, 129, 143, 144, 210, 225, 245, 248, 261, 262 see also Basic Variety learner system 3, 6, 30, 32, 36, 79–81, 110, 127, 153, 157, 158, 221, 248, 250, 262, 267 learner variety 79, 114, 144, 150, 152 lexical category 27, 36, 127, 205, 253, 260, 265 lexical projection – agentive 125, 134, 141, 151 – non-agentive 125, 134, 141, 151 lexical selection 3, 15 lexicalization 25, 26, 57–59, 61–63, 65, 77, 114, 125, 200, 258 Lexical-Semantic Hypothesis 90–92, 94 light verbs 193–197 Marsden, Emma 229, 237, 238, 249 Meisel, Jürgen M. 30, 81, 146, 197, 199, 229, 230 mini-paradigms 6, 211, 218, 222 Missing Surface Inflection Hypothesis (MSIH) 6, 197, 231–234, 238, 249, 251 modal expressions 168–170, 210 see also predicate

Modal Hypothesis 4, 82, 91, 92, 122 modality 54, 96, 97, 107 – deontic 67, 69, 170, 171 – epistemic 68, 69, 170, 171 non-subjects in first position 82, 84, 86 Optional Infinitive Hypothesis 85, 91, 97 Organic Grammar 239 Parodi, Teresa 193–196, 226–228 particle verb 44, 62, 85, 95, 109, 129, 131–133 see also predicate Perdue, Clive 1, 30, 31, 34, 36, 38, 41, 111, 144, 145, 149, 161, 253, 259 perspective taking 3, 16–18 phrase structure 14, 15, 51 Piaget, Jean 37, 110, 269 Poeppel, David 4, 82–86, 88, 90–92, 94–97, 99, 107–109, 118, 129, 152 position – focus 3, 137, 186, 190 – subject 18, 48, 49, 152, 162, 163, 166, 167, 181, 210, 262 – topic 145–147, 157, 160, 164–166, 180–182, 221, 223, 248 predicate – action 104, 133, 148, 152, 199–201, 215, 217, 222, 253, 267 – agentive 60, 105, 107, 151, 152, 155, 158, 160, 168, 171, 210, 254, 255, 258 – aspectual 119, 120, 254, 258 – change-of-state 59–61, 104–107, 110, 128 – modal 119–121, 125, 254, 255, 258, 259 – non-agentive 67, 130, 133, 141, 144, 151 – particle 122–124, 255, 256 – state 104–107, 110, 128 predication 9, 10, 52, 55, 57, 62–66, 73, 77, 127, 128, 258, 261, 267, 268 pre-morphological stage 211, 214, 215, 222 productivity 62, 96, 97, 114, 144, 148, 158, 176, 225, 240 projection – functional 3, 6, 50, 51, 66, 96, 110, 142–144, 150, 166–168, 175, 229, 240, 248, 260, 266, 268

Index 

– lexical 51, 54, 55, 124 see also lexical projection agentive and non-agentive pronominalization 19, 21–23, 38 pronoun – anaphoric 20, 113, 209, 260, 264, 267 – deictic 20, 23, 111, 113, 137, 144, 168, 187 proto-morphological stage 211, 214–217, 222 question 3–6, 42–45, 50, 86, 113, 142, 144 – wh-question 184–186 – yes/no-question 184–187 right branching 46 root infinitives 83, 85, 92, 94, 96–98, 105–109, 115, 120, 254 see also verb form infinite Rule, Sarah 229, 237, 238, 249 scrambling 74, 113, 127, 143, 144, 153 state verb 60, 89, 90, 101, 110, 222, 225, 228, 234, 236, 247, 250, 256–258 see also predicate structure – information 5, 7, 16, 20, 36, 50, 78, 114, 137, 138, 143–145, 210, 229, 239, 248, 251, 267 – predicate-argument 2, 4, 5, 21, 51, 56, 62, 104, 110, 114, 151, 152, 210 – utterance 3, 16, 31, 150, 155, 156, 167 subject 5, 9, 16–19, 48, 49, 63–66, 77, 96, 130–138, 142–145, 150, 155, 156, 162–164, 166–168, 181–184, 186–190, 212–214, 220, 222, 261–263, 267 see also position, agreement Subject first 135, 149 subject selection 17–19, 63, 65, 77 tense 1, 3, 6, 19–23, 25–29, 38, 72, 78, 113, 114, 144, 260, 266, 267 Thompson, William 4, 82, 90–92, 95–97, 99, 107, 122, 143, 150, 152 topic 3–6, 136–138, 143–145, 155, 159, 161–167, 182, 186–190, 210, 262–268 see also position, topic situation Topic first 136, 137, 149, 161 topic situation 20, 22, 27, 52, 74–78, 136, 137, 142, 184

 279

topic time (TT) 20, 22, 71, 72, 200 topicality 1 topicalization 2–6, 18–23, 108, 113, 143, 144, 148, 154, 158–160, 179–190, 209, 210, 239, 246, 248, 260, 262, 267, 268 transitivity 4, 82, 87–90, 96, 97, 107 underspecification 148, 197, 198, 229, 233, 249 utterance – agentive 253–257 see also predicate – non-agentive 65, 128, 135, 153, 253, 256–259 see also predicate Vainikka, Anne 239–241 verb form – finite 80, 81, 95, 100, 103, 116, 118, 232, 235, 256 – infinite 80, 81, 95, 100, 103, 115, 254 see also root infinitives verb morphology 80–83, 90, 91, 94, 95, 197, 199, 200, 211–213, 222, 259 verb movement 224, 233, 234, 248–251 see also head movement, verb raising verb placement 81, 82, 95, 197, 199, 215, 223–226, 229–231, 238, 239, 249, 259 verb raising 224–227, 240, 248–250, 259 see also head movement, verb movement verb-second 4, 23, 41–45, 48, 113, 153, 154, 210, 223, 224, 228, 248, 260, 265 verb-third 187–190 Wexler, Kenneth 4, 82–99, 107–109, 118, 129, 152, 226 wh-words 260, 267 word formation 22, 26, 28, 33 word order 2, 3, 13, 14, 32, 47, 110–112, 134–136, 138–142, 152, 159, 182 – basic 41–45, 50, 81, 97, 98, 103, 107, 108 – variation 46, 49, 113, 239, 264, 267 Young-Scholten, Martha 239–241 ZISA study 30